The Extension of Biology

Through Culture
Edited by Andrew Whiten, Francisco J. Ayala, Marcus W. Feldman,
and Kevin N. Laland

Arnold and Mabel

Beckman Center

Irvine, CA

November 16–17, 2016

This work is reprinted from the Proceedings of the National Academy of
Sciences of the United States of America, vol. 114, no. 30, pp. 7734–7737 and
pp. 7775–7922, July 25, 2017, and includes articles from the Arthur M.
Sackler Colloquium The Extension of Biology Through Culture, held
November 16–17, 2016, at the Arnold and Mabel Beckman Center in
Irvine, CA. The articles appearing in these pages were contributed by
speakers and attendees at the colloquium and were anonymously peer
reviewed. Any opinions, findings, conclusions, or recommendations
expressed in this work are those of the authors and have not been
endorsed by the National Academy of Sciences.

The National Academy of Sciences is a private, nonprofit, self-perpetuating

society of distinguished scholars engaged in scientific and engineering
research, dedicated to the furtherance of science and technology and to
their use for the general welfare. Upon the authority of the charter granted
to it by the US Congress in 1863, the Academy has a mandate that requires
it to advise the federal government on scientific and technical matters.

ISBN 10: 0-309-46316-5

ISBN 13: 978-0-309-46316-4

© Copyright by the National Academy of Sciences, USA

All rights reserved. Published 2017
Printed in the United States of America
The Extension of Biology
Through Culture

National Academy of Sciences

Washington, DC
Arthur M. Sackler, M.D.
1913–1987

B orn in Brooklyn, New York, Arthur M. Sackler was edu-

cated in the arts, sciences, and humanities at New York
University. These interests remained the focus of his life, as he
became widely known as a scientist, art collector, and philan-
thropist, endowing institutions of learning and culture through-
out the world.
He felt that his fundamental role was as a doctor, a vocation
he decided upon at the age of four. After completing his
internship and service as house physician at Lincoln Hospital in
New York City, he became a resident in psychiatry at Creed-
moor State Hospital. There, in the 1940s, he started research
that resulted in more than 150 papers in neuroendocrinology,
psychiatry, and experimental medicine. He considered his
scientific research in the metabolic basis of schizophrenia his
most significant contribution to science and served as editor of
the Journal of Clinical and Experimental Psychobiology from 1950 to 1962. In 1960 he started
publication of Medical Tribune, a weekly medical newspaper that reached over one million
readers in 20 countries. He established the Laboratories for Therapeutic Research in 1938, a
facility in New York for basic research that he directed until 1983.
As a generous benefactor to the causes of medicine and basic science, Arthur Sackler built
and contributed to a wide range of scientific institutions: the Sackler School of Medicine
established in 1972 at Tel Aviv University, Tel Aviv, Israel; the Sackler Institute of Graduate
Biomedical Science at New York University, founded in 1980; the Arthur M. Sackler Science
Center dedicated in 1985 at Clark University, Worcester, Massachusetts; and the Sackler School
of Graduate Biomedical Sciences, established in 1980, and the Arthur M. Sackler Center for
Health Communications, established in 1986, both at Tufts University, Boston, Massachusetts.
His pre-eminence in the art world is already legendary. According to his wife Jillian, one of
his favorite relaxations was to visit museums and art galleries and pick out great pieces others
had overlooked. His interest in art is reflected in his philanthropy; he endowed galleries at the
Metropolitan Museum of Art and Princeton University, a museum at Harvard University, and
the Arthur M. Sackler Gallery of Asian Art in Washington, DC. True to his oft-stated
determination to create bridges between peoples, he offered to build a teaching museum in
China, which Jillian made possible after his death, and in 1993 opened the Arthur M. Sackler
Museum of Art and Archaeology at Peking University in Beijing.
In a world that often sees science and art as two separate cultures, Arthur Sackler saw them
as inextricably related. In a speech given at the State University of New York at Stony Brook,
Some reflections on the arts, sciences and humanities, a year before his death, he observed:
‘‘Communication is, for me, the primum movens of all culture. In the arts. . . I find the emotional
component most moving. In science, it is the intellectual content. Both are deeply interlinked
in the humanities.’’ The Arthur M. Sackler Colloquia at the National Academy of Sciences pay
tribute to this faith in communication as the prime mover of knowledge and culture.
 July 25, 2017 | vol. 114 | no. 30
Proceedings of the National Academy of Sciences of the United States of America
Cover image: A juvenile capuchin
monkey observes a skilled adult male
eating a nut it has just broken using
a hammerstone. Articles in the Sackler
Colloquium on the Extension of Biology
Through Culture explore social learning
and cultural transmission in humans and
nonhuman animals as well as the
interplay between cultural and genetic
evolution. See the Introduction to the
Sackler Colloquium by Andrew Whiten
et al. on pages 7775–7781. Image
courtesy of Luca Antonio Marino
(EthoCebus Project, Brazil).
PNAS
The Extension of Biology Through Culture
www.pnas.org
The papers collected here result from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, The Extension of Biology Through Culture. Complete information about this colloquium
and video recordings of most presentations are available on the NAS website at http://www.
nasonline.org/Extension_of_Biology_Through_Culture.
Contents
INTRODUCTION
7775 The extension of biology through culture
Andrew Whiten, Francisco J. Ayala, Marcus W. Feldman,
and Kevin N. Laland
COLLOQUIUM PAPERS
7782 Cultural evolutionary theory: How culture evolves and why it matters
Nicole Creanza, Oren Kolodny, and Marcus W. Feldman
7790 Culture extends the scope of evolutionary biology in the great apes
Andrew Whiten
7798 Synchronized practice helps bearded capuchin monkeys learn to extend
attention while learning a tradition
Dorothy M. Fragaszy, Yonat Eshchar, Elisabetta Visalberghi, Briseida Resende, Kellie Laity,
and Patrícia Izar
7806 Older, sociable capuchins (Cebus capucinus) invent more social behaviors,
but younger monkeys innovate more in other contexts
Susan E. Perry, Brendan J. Barrett, and Irene Godoy
7814 Gene–culture coevolution in whales and dolphins
Hal Whitehead
7822 Song hybridization events during revolutionary song change provide
insights into cultural transmission in humpback whales
Ellen C. Garland, Luke Rendell, Luca Lamoni, M. Michael Poole, and Michael J. Noad
7830 Conformity does not perpetuate suboptimal traditions in a wild
population of songbirds
Lucy M. Aplin, Ben C. Sheldon, and Richard McElreath
7838 A social insect perspective on the evolution of social learning mechanisms
Ellouise Leadbeater and Erika H. Dawson
7846 Cultural macroevolution matters
Russell D. Gray and Joseph Watts
7853 Pursuing Darwin’s curious parallel: Prospects for a science of cultural evolution
Alex Mesoudi
7861 Evolutionary neuroscience of cumulative culture
Dietrich Stout and Erin E. Hecht
7869 Identifying early modern human ecological niche expansions and associated
cultural dynamics in the South African Middle Stone Age
Francesco d’Errico, William E. Banks, Dan L. Warren, Giovanni Sgubin, Karen van Niekerk,
Christopher Henshilwood, Anne-Laure Daniau, and María Fernanda Sánchez Goñi
7877 Cumulative cultural learning: Development
and diversity
Cristine H. Legare
7884 Young children communicate their ignorance
and ask questions
Paul L. Harris, Deborah T. Bartz, and Meredith L. Rowe
7892 Changes in cognitive flexibility and hypothesis
search across human life history from childhood
to adolescence to adulthood
Alison Gopnik, Shaun O’Grady, Christopher G. Lucas,
Thomas L. Griffiths, Adrienne Wente, Sophie Bridgers,
Rosie Aboody, Hoki Fung, and Ronald E. Dahl
7900 How language shapes the cultural inheritance
of categories
Susan A. Gelman and Steven O. Roberts
7908 Coevolution of cultural intelligence, extended life
history, sociality, and brain size in primates
Sally E. Street, Ana F. Navarrete, Simon M. Reader,
and Kevin N. Laland
7915 The evolution of cognitive mechanisms in response
to cultural innovations
Arnon Lotem, Joseph Y. Halpern, Shimon Edelman,
and Oren Kolodny
NEWS FEATURE
7734 Can animal culture drive evolution?
Carolyn Beans
 COLLOQUIUM INTRODUCTION

INTRODUCTION
COLLOQUIUM
The extension of biology through culture
Andrew Whitena,1, Francisco J. Ayalab, Marcus W. Feldmanc, and Kevin N. Lalandd

Biology is the study of life. How our understanding of the
nature and evolution of living systems is being enriched
and extended through new discoveries about social
learning and culture in human and nonhuman animals is
the subject of the collection of articles we introduce here.
Recent decades have revealed that social learning
and the transmission of cultural traditions are much
more widespread in the animal kingdom than earlier
suspected, affecting numerous forms of functional
behavior and creating a secondary form of evolution,
built onto the better-known primary, genetically based
form. New scientific approaches to the study of human
cultural evolution have also emerged and become
productive. However, these developments in the study
of cultural phenomena in both human and nonhu-
man animals have yet to be seriously integrated into
mainstream evolutionary biology. Here we offer an
introductory overview of the background and scope of
a collection of articles that report recent progress in
these fields, and outline their proposed significance
for biology at large.
The theoretical backbone of the life sciences, its
central organizing principle, is of course evolution, by
now rich in both theory and empirical support (1–3).
The great synthesis of Darwin’s and Wallace’s evolu-
tionary insights and early 20th century understanding
of genetics that became known as the “Modern Syn-
thesis” was achieved by a brilliant set of biologists
mainly in the period 1938–1946 (4), and its principles
have provided the core of evolutionary theory since
that time (5). Thus, contemporary texts on “evolu-
tion” focus on such topics as mutation, genetically
based inheritance, population genetics, genomics,
and the natural and sexual selection pressures that
shape gene frequencies, genotypes, and pheno-
types (1, 2, 6, 7). Genes and their role in inheritance
have come to be celebrated as the pivotal elements
in evolution (8).
However, a second form of evolution was also rec-
ognized long ago, in the ways that cultural phenomena
have changed in the course of human history, through
a different form of inheritance: that in which people learn
from others (social learning), including from previous
generations. Darwin himself recognized the parallels
between the evolution of culturally inherited languages
and organic evolution (9, 10); indeed, evolutionary fam-
ily trees of languages proposed by philologists long
predated the Origin of Species, although they were
further spurred by its publication (11–13).
During the 1970s and 1980s, first by Cavalli-Sforza
and Feldman (14–16) and then Boyd and Richerson
(17), the implications of the existence of the two forms
of evolution, organic and cultural, was at last explored
systematically and formally, through conceptual and
mathematical modeling that formed a foundation for
later empirical investigations. The present collection
of papers opens with a contribution by Creanza et al.
(18) that offers an overview of both the foundational
studies in (human) cultural evolution and major devel-
opments in the period since. The early body of 20th
century work laid out some of the ways in which cul-
tural evolution: (i) echoes many core principles of or-
ganic evolution, yet (ii) also differs from it in dramatic
ways that change evolutionary dynamics, and (iii) in-
teracts with the genetically based phenomena to
create new complexities (“gene–culture coevolu-
tion”). We return to discuss these further, below.
From a somewhat different perspective Maynard-
Smith and Szathmary (19) distinguished a series of
major transitions in the nature of evolution, such as
the emergence of multicellularity and of sex, the
most recent major transition being the emergence
of (human) culture; and Dawkins (20) gave a name
to cultural elements suggested to be the analogs
of genetic replicators—“memes”—which has been
assimilated into popular culture. Other authors sug-
gested “semes” (21), echoing semiotics, the study
of signs and symbols.
We shall discuss such developments and subse-
quent related scientific progress further below, but for

a
Centre for Social Learning and Cognitive Evolution, School of Psychology and Neuroscience, University of St. Andrews, St. Andrews KY16 9JP,
United Kingdom; bDepartment of Ecology and Evolutionary Biology, University of California, Irvine, CA 92697; cDepartment of Biology,
Stanford University, Stanford, CA 94305; and dCentre for Social Learning and Cognitive Evolution, School of Biology, University of St. Andrews,
St. Andrews KY16 9JP, United Kingdom
This paper results from the Arthur M. Sackler Colloquium of the National Academy of Sciences, “The Extension of Biology Through Culture,” held
November 16–17, 2016, at the Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering in Irvine, CA. The complete
program and video recordings of most presentations are available on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: A.W., F.J.A., M.W.F., and K.N.L. wrote the paper.
The authors declare no conflict of interest.
1
To whom correspondence should be addressed. Email: a.whiten@st-andrews.ac.uk.

www.pnas.org/cgi/doi/10.1073/pnas.1707630114 PNAS | July 25, 2017 | vol. 114 | no. 30 | 7775–7781

the moment we make one observation: all of these writings on
culture were focused on a single species, our own. That other
species might exhibit the core properties of cultural transmission,
and that this was worthy of scientific investigation, began to be
recognized in a number of different lines of animal behavior
research only around the middle of the last century. Moreover,
evidence for animal social learning and tradition started to
become substantial only in the most recent decades, and thus
was not only unremarked in the Modern Synthesis but gained only
minimal mention in the foundational works of cultural evolution
(14, 17). The earliest reports for nonhuman animals (henceforth
“animals”) were of novel traditions arising and spreading, nota-
bly bottle-top opening to drink milk by titmice (22) and washing
food items in the sea by Japanese monkeys, referred to conser-
vatively at the time as “pre-cultures” (23). An additional revela-
tion was the discovery of bird song learning and the existence of
local song dialects (24). Since these foundational studies, there
has been a proliferation of studies documenting social learning
and traditions in animals (25, 26), often referred to as “animal
culture” (27).
The Discovery of Widespread Animal Culture
Research over the last half-century has led to the revelation that
learning from others (social learning) is widespread in the animal
kingdom and spans a great range of important functional con-
texts, including diet, feeding techniques, travel route selection,
predator avoidance, vocal communication, migration, and mate
and breeding site choices (26, 28). Hundreds of laboratory ex-
perimental studies have demonstrated social learning and trans-
mission in a wide variety of animals. Social learning is now extensively
documented in mammals (29), with a particular intensity of re-
search studies in primates (30–33) and cetaceans (34–37), in birds
(38–41), in fish (42), and in insects (43, 44). The fact that social
learning has been shown to play important roles spanning mul-
tiple functional contexts (25–28) suggests that many animals are
not simply acquiring one or a few behavioral patterns socially,
but rather that social learning is central to their acquisition of
adaptive behavior.
Social learning may lead to the spread of a behavior to other
individuals, which is what defines cultural transmission, and the
establishment of traditions that come to characterize whole
groups, subgroups, or populations. However, social learning may
also be transient, and lead to no such substantial population-level
effects; for example, a monkey may learn from others that a par-
ticular tree is in fruit or that a snake is in a particular bush, which
shape its behavior for only a short period thereafter. Evidence for
social learning, so widespread in animals, is thus not sufficient to
demonstrate the larger phenomena of traditions and culture.
Nevertheless, cultural diffusion has been further documented not
only in all of the vertebrate research reviews listed above, but also
in insects, at least in the laboratory. For example, Alem et al. (45)
trained demonstrator bumble bees to pull a small piece of string
to access an artificial flower providing food, and experiments then
showed that many other bees that observed them acquired the
novel technique, with this learning not shown by control bees that
had no model from which to learn. Moreover, other bees learned
from the earliest learners, and transmission across several such
“cultural generations” was demonstrated. The extent to which this
kind of transmission occurs in the wild is now a question that
demands focused study.
In vertebrates, plentiful evidence exists that such cultural
transmission occurs in nature, with substantial and indeed striking
7776 | www.pnas.org/cgi/doi/10.1073/pnas.1707630114
impact. For example, archaeological excavations have shown re-
mains of nut-cracking materials dated to 4,300 y below the sur-
face of the Taï Forest, where modern chimpanzees continue this
practice, absent in most parts of Africa (31, 46); 27 y of observa-
tions have documented the spread from a few to over 600
humpback whales of a new form of hunting technique, lob-tail
fishing (47); and new humpback songs have been found to
emerge, spread quickly to whole populations, and pass in waves
across the Pacific in consecutive years (36, 37).
That it has taken so long to begin to discover the broad scope
of animal culture has several possible explanations. One impor-
tant source of the progress made has been the achievement of
long-term field studies. For example, just a half-century ago re-
searchers knew next to nothing about the behavior of our closest
primate relatives in the wild, but in recent times it has been pos-
sible to assemble data on each of several species from multiple,
decades-long field studies to discover putative cultural differ-
ences amenable to more focused study, as now achieved for
different genera of the great apes (31). The example of the long-
term observations that identified the spread of the lob-tail hunting
technique in humpback whales has already been mentioned (47).
That study also illustrates the application of increasingly sophis-
ticated statistical analyses to identify social learning through the
way in which a novel behavior spreads along the lines of social
networks [network-based diffusion analysis (26)], an approach now
applied in other contexts, such as tool use in chimpanzees (48).
This is one among a growing armory of such statistical approaches
bearing fruit in the discipline (26).
The most compelling evidence of social learning comes from
experiments, the most basic (but powerful) forms of which involve
comparing an experimental condition in which animals have wit-
nessed a trained model complete some novel task, with a control
condition lacking a model, or comparing two experimental con-
ditions in which models display different task solutions. Dyadic
studies of this kind have a century-long history, but in recent de-
cades modifications have been made to allow testing for the
more culturally relevant phenomenon of the diffusion of behav-
iors across multiple individuals or even between groups. Several
different experimental designs have since identified diffusion
through social learning, both in laboratory and field conditions,
and in an accelerating number and diversity of mammalian, avian,
piscine, and insect species (49, 50).
However, this growing array of widespread cultural phenom-
ena in animals appears to have gone largely unrecognized in
mainstream texts on evolution (e.g., refs. 1, 6, 7, 19) or receives
only minimal mention (2). One major goal of the present PNAS
issue is to illustrate the scope of the findings on animal culture that
now merit integration into evolutionary biology at large.
Building on a recent book-length review making the case that
culture pervades numerous aspects of the life of whales and
dolphins (34), two papers in the present collection describe recent
progress in studies of these cetaceans. Garland et al. (37) focus on
the transmission of complex song in humpback whales, present-
ing a painstaking and highly illuminating analysis of the process of
song hybridization during the remarkable periods of “cultural
revolution” found in this species. Behavioral hybridization has
often been highlighted as a phenomenon differentiating cultural
from genetically based evolution, although hybridization is far
from unknown at the species level and recent research in-
creasingly suggests that it has been much more common than
previously suspected at the level of genetic transfer (51). The
new humpback data reveal two different, specific ways in which
Whiten et al.
hybridization occurs, involving the application of systematic
structural rules that the authors propose are similar to—and pro-
vide additional insights into—those identified in both birdsong
and human language. Whitehead (35) goes on to review the evi-
dence for gene–culture coevolution in both whales and dolphins,
appraising the evidence that matrilineal cultural inheritance has
been particularly influential in creating ecologically specialized
communities in species, such as killer whales, in turn explaining
low diversity in matrilineal mitochondrial DNA and regional vari-
ation in mitochondrial DNA haplotype distribution. We return to
this study in addressing the issue of gene–culture coevolution
further below.
Cultural transmission has long been recognized in the sphere
of birdsong (24, 38) but birds have tended to be seen as “one-trick
(cultural) ponies” in this respect. However, recent studies have
identified cultural transmission across a much broader span of
behavior (39, 40, 52, 53). A striking case is the high-fidelity spread
of alternative foraging behaviors experimentally seeded in sub-
stantial communities of great tits (40). Here, Aplin et al. (41) show
that these behaviors will evolve adaptively as payoffs change,
and the authors present evidence that this occurs through an in-
triguing combination of conformist social learning and payoff-
sensitive individual learning.
Recent evidence that insects also show not only social learn-
ing, but a capacity for cultural transmission spreading across
communities, is reviewed here by Leadbeater and Dawson (44),
leading them to appraise the potential consequences for the
evolution of learning processes and the brain. The authors con-
clude that “Social insects are distant relatives of vertebrate social
learners, but the research we describe highlights routes by which
natural selection could coopt similar cognitive raw material within
the animal kingdom.”
Primates have long been at the forefront in research on animal
culture (27). In the present collection Whiten (31) reviews the di-
versity of complementary observational and experimental evi-
dence for social learning and multiple-tradition cultures in the
great apes. Although ape (and other nonhuman) culture does not
encompass the elaborate levels of cultural evolution evident in
humans, Whiten concludes that the accumulated evidence now
exists for the principal implications of culture for evolutionary bi-
ology alluded to earlier: cultural evolution displays a number of
properties evident in genetic evolution but through the different
means of social learning. The interactions of genetic and cultural
transmission are evolutionarily consequential.
Cultural evolution rests upon the complementary processes of
innovation and selective transmission, but concentration of re-
search to date on testing primates’ capacity for transmission has
arguably resulted in neglect of the innovation (“mutation”) ele-
ment, which is less susceptible to experimental manipulation in
the laboratory, and challenging to record in the wild. Here Perry
et al. (33) report on 10 y of observations of over 200 capuchin
monkeys (Cebus capucinus) in 10 groups, with research explicitly
focused on innovations as well as their transmission. Distinguish-
ing four main functional categories of behavior, these researchers
report 17 innovations in foraging and drinking, 9 in hygienic and
other self-directed behaviors, 53 classed as investigative, and
49 as social behaviors. Just 21% of these large totals were picked
up by others, indicating marked selectivity, which these authors
dissect further. This research begins to identify the Darwinian
processes of variation and selective transmission underlying cul-
tural transmission in nonhuman species. As noted above, these
processes can continue throughout life, contrasting with the
Whiten et al.
genetic package transmitted at conception; this is the focus of a
third primate study, on a different genus of capuchin monkey
(Sapajus libidinosus) renowned for their tool-assisted nut-cracking
behavior. Fragaszy et al. (32) trace the acquisition of this skill
through the course of development over the several years needed
to acquire competence, revealing the complex cycles of practice
and attention to experts’ nut-cracking, with adults’ behavior
helping to focus attention on rare elements of the skill that are
critical to success.
Human Culture Is Special
The present issue includes several papers on a single species: our
own. This focus clearly does not correspond with any proper
proportional representation among animal species; the explana-
tion is simply that the scope and penetration of culture is excep-
tional in humans (54–57), and because of this, human culture
extends biology in many additional and extraordinary ways. In-
deed, some of the consequences of human culture, like the de-
struction of other species’ habitats, climate change, and pollution,
are already having (and in many cases have already had) major
effects on the evolution, distribution, and extinction of major
segments of the world’s biota (58, 59).
Studies of human cultures have been pursued by an even
greater diversity of approaches than those sketched for animal
culture above. There is of course a whole discipline of social and
cultural anthropology for which, as the name implies, the target of
study is culture, and this has often striven for forms of participant
observation extending to self-immersion in different cultures, an
approach actively avoided by most students of primate behavior,
and often not even thinkable for more distantly related species. In
contrast to common approaches in cultural anthropology, those
working within the field of cultural evolution have created an array
of different approaches and methods that are often more con-
ventionally scientific. These include a greater emphasis on such
elements as formal and mathematical modeling, quantification
and statistical analysis of numerical data, hypothesis testing, and
systematic experimentation (18, 25, 56, 60–63). There is not the
space here to offer anything like a comprehensive review of the
resulting discoveries, but we can outline something of their range,
with selected illustrations.
Those studies that examine the earliest evidence for human
culture have shown a continuing trend for markers of change to be
found at ever earlier dates. For example, the earliest evidence of
stone tool use has recently been pushed back from 2.6 to
3.4 million y ago (64), roughly the half-way point since our shared
ancestry with Pan. Here, Stout and Hecht (65) develop models of
early lithic culture that integrate its distinctive human elements
and primate foundations, both behavioral and neural. Similarly,
the beginnings of what has been labeled “symbolic culture,”
indexed by such features as decorative items, like beads, has
been pushed back from the previous “cave art” dates of around
30 ka to closer to 100 ka, or in the case of some elements like
ochre, to even earlier dates, through a diversity of striking ar-
chaeological finds (66). Here, d’Errico et al. (67) provide a rich and
detailed account of the ways in which cultural repertoires and
their associated ecological niches differentiated and evolved in
these periods.
Within historical times, the records of human culture have
become amenable to some of the systematic and quantitative
methods developed within evolutionary biology to reconstruct
phylogenetic relationships at scales ranging from macroevolution
to finer-grained speciation patterns (68). Such approaches have
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7777
been particularly powerfully applied to the differentiation and
evolution of language groups (68, 69), but also to such diverse
topics as the evolution of socio-political organization (70) and folk
tales (71). Here, Gray and Watts (60) apply this approach to the
evolution of religion, using this example to explore the analysis of
cultural macroevolution.
As in the capuchin study of Fragaszy et al. (32), the psycho-
logical and social processes that allow human culture to be so
distinctive need to be examined as individuals’ life histories un-
fold, and the affordances of the culture in which they develop are
selectively assimilated and further modified. On the one hand,
these processes are part of our species’ biology, their properties
shaped during the millennia of evolutionary time over which our
ancestors became increasingly and intensively dependent on
cumulative cultural inheritance (54, 56, 57). In turn, these unique
cultural processes operating in humans generate forms of life not
hitherto witnessed in the natural world. To highlight and dissect
some of these special cultural phenomena, we include in this issue
four contributions that share a focus on ontogenetic development.
Legare (72) provides an overview of core features of human
development that facilitate the adaptive transmission and re-
finement of culture, including the concept of “natural pedagogy,”
whereby adults provide active support to cultural assimilation and
children are predisposed to recognize and respond to this in
particular ways, such as selective and discriminating copying (for
example with respect to alternative cultural models), conformity,
including the recognition of norms, and innovative flexibility.
Other, complementary contributions in the present collection
focus on more specific topics, including the active role that chil-
dren come to play in recognizing their ignorance, as well as their
knowledge, and systematically seek information to remedy this
(73); the ways in which related hypothesis testing changes through
the long period of human development in relation to the stage of
cognitive development and socio-ecological context (74); and the
significance of language as both a product and medium of culture,
illustrated by the linguistic labels and generics that provide spe-
cial forms of both the transmission fidelity and affordance for in-
novation that permit cumulative culture (75).
How Culture Extends Biology
How the existence of culture extends our understanding of the
scope and nature of living systems and their evolution was initially
analyzed in three major respects (14, 15, 17). First, cultural phe-
nomena provide a second inheritance system (76) built on the
foundations of the primary, genetically based system, and this can
generate a second form of evolution in the sphere of culturally
transmitted behaviors and artifacts. Second, because cultural
transmission is mechanically different from genetic transmission in
particular ways, such as horizontal diffusion among nonrelatives, it
can have new and drastically different evolutionary consequences
(62). Third, the two systems may interact in complex ways, the
phenomenon of gene–culture coevolution (16, 56, 77, 78). To
these three we can now add two other important dimensions. One
is that the accumulating evidence that social learning and cultural
transmission are much more widespread and consequential
across the animal kingdom than earlier suspected, extends much
more broadly the implications of the three effects outlined above,
which were originally conceived with a focus on human culture. A
second is that studies increasingly dissect and delineate the
richness of the consequences of cultural evolution, and resulting
diversification of life forms. Examples of recent such discoveries
range from the elucidation of functional forms of teaching in
7778 | www.pnas.org/cgi/doi/10.1073/pnas.1707630114
nonhuman animals (79) to “rational imitation” (80) and “over-
imitation” (81) in children. The contributions to this issue address
all of these prospects conceptually and empirically in diverse and
important ways. Here, we offer a brief introductory overview of
the background foundations to these new explorations and
updated reviews.
Cultural Phenomena Create a New Form of Evolution. The core
of adaptive evolution through natural selection involves the triad
of variation in characters, competitive selection of the best
adapted to current circumstances among them (“survival of the
fittest,” although relative reproductive success is what ultimately
counts), and inheritance of those selected characters by descen-
dants (82). Interwoven in cycles of these processes are three ad-
ditional principles, notably the refinement of adaptations suited to
the properties of ecological niches, the accumulation of com-
plexes of these, and differentiation of descendant populations
where they are sufficiently separated, for example by geography,
ultimately leading to speciation. The latter three effects are
manifested in the picture of organic evolution with which we are
familiar, involving a broadly progressive complexification in life—
from early bacteria to the sophisticated animals of today—and a
vast diversity of living species, all displaying a remarkable fit to the
ecological niche they so successfully inhabit. Current thinking in
cultural evolution suggests that all these principles are active also
in human cultural evolution (14, 16, 56, 61). Social learning and
transmission provide the inheritance element and human in-
vention the variants, the most successful of which are transmitted
to future generations, generating cultural adaptations to envi-
ronments around the world; and progressive, cumulative cultures
show immense regional differentiation. Empirical evidence in
support of these contentions has accumulated over recent de-
cades, reviewed for example in refs. 61, 62, 78, 83, and 84, and is
pursued further in the present collection (18, 63).
Such questions about cultural evolution have remained little
studied in the animal culture literature, which has instead been
focused on the more fundamental matter of establishing what
cultural phenomena exist in a diversity of species, and what
transmission processes underpin these (25, 27, 34). Initial explo-
rations of Darwinian dynamics in the case of animal culture (53)
have taken the list of eight key properties extracted from the
Origin of Species (9) for testing with human data [the six listed
above, plus changes of function and convergent evolution (83)]
and through examining studies of animal culture, concluded there
is evidence for all of them (although minimal and slow-developing
compared with the most recent, cumulative cultures of humans).
However, there is evidence that animal traditions with suboptimal
payoffs are sometimes, although seemingly not always (85), vul-
nerable to decay (41), implying the working of the core Darwinian
triad, and it seems likely that those animals for which there is now
evidence of multiple-tradition cultures are the descendants of
lines of ancestors among whom these traditions were pro-
gressively added, surviving through their success, as in the case of
over 4,300-y-old nut-cracking in chimpanzees, mentioned earlier
(46). Nonetheless, experimental studies, for example, of mate-
choice copying, show that animal social transmission can be
evolutionarily consequential, even if short lived (86).
In any case, it is becoming apparent that cultural phenomena
play an important role in shaping many species’ adjustment to and
exploitation of their environments, with likely significant evolu-
tionary consequences that are the focus of current research.
Whiten et al.
Cultural Evolution Includes Characteristics Absent in Geneti-
cally Based Evolution. Cultural evolution may display analogs of
organic evolution outlined above, but it is also different in many
fundamental respects, further extending the scope of the evolu-
tionary processes that shape biological systems (14, 17). Notably,
transmission is not only vertical, as in genetic inheritance from
parent to offspring, but can be horizontal, between unrelated
peers, or oblique, from unrelated individuals in the parental
generation (14); moreover, because this involves neurally based
learning rather than genetic change, such transmission can be
quite rapid, as well illustrated by the case of “revolutions” in the
songs of humpback whales, which may change annually yet
quickly come to be sung by whole populations (37), and in a va-
riety of human and animal cases further explored in this issue.
Furthermore, unlike the genetically packaged adaptive in-
formation inherited at conception, social information may be
gathered throughout ontogeny and indeed across the lifespan,
and in interaction with individual learning and practice, it can thus
permit iterative and flexible forms of adaptation as circumstances
change. Here, this is illustrated in analyses of extended ontogeny
of difficult skills, like nut-cracking in primates (32, 48). Moreover,
even innovations, the analog of mutations that become subject to
selective adoption and further transmission to others, may be far
from random; instead, they are often immediately functional, most
clearly in the example of intelligent, goal-oriented human inven-
tions, but also in the closest animal counterparts.
The social transmission process may itself be adaptively sha-
ped by different biases in what is selectively assimilated, variously
referred to as transmission biases (17) or social learning strategies
(87). Examples include biases to copy behavioral routines, where
there is evidence they are successful, conformist copying of the
majority (exploiting “the wisdom of the crowd”), and indirect
biases, such as copying individuals on the basis of their reputation
or group identity. Evidence for an array of such biases has accu-
mulated in studies of both human and animal cultural transmission
(88) and are further addressed in this collection (48, 72, 73).
Gene–Culture Coevolution. Empirical evidence for cultural
practices creating selection pressures that feed back to affect
biological evolution have been known for some time in the human
case (17). Such ideas reach back further to the Baldwin effect,
which proposed that a measure of plasticity in animals’ adjustment
to their worlds during their lifetimes, including by learning, could
create selection pressures for corresponding organic change (89),
as well as to later notions of “behavioral drive” and “cultural
drive” (90, 91) and niche construction (92). However, these ideas
are becoming much refined and supported by extensive data in
the age of genomics (93, 94).
Using a diversity of evidence from archaeology to neurosci-
entific investigations, Stout and Hecht (65) analyze how the
cultures of the Stone Age shaped brain size and structure to create
the cognitive and manual skills required for the evolution of
greater sophistication in tool making. This kind of feedback may
have been in existence for a very long time. Following a com-
parative phylogenetic analysis of primate brain and behavior data,
Street et al. (95) conclude that cultural processes may have gen-
erated such selective feedback (see also ref. 31). Their analyses
suggest that both brain expansion and high reliance on culturally
transmitted behavior coevolved with sociality and extended life-
span in primates. This coevolution is consistent with the hypoth-
esis that the evolution of large brains, sociality, and long lifespans
has promoted reliance on culture, with reliance on culture in turn
driving additional increases in brain volume, cognitive abilities,
and lifespans in some primate lineages. Lotem et al. (96) further
describe an explicit model that accommodates the shaping of
cognition by culture, from basic building blocks of learning and
data acquisition to phenomena such as language and tool use.
The authors illustrate how learning and cognition will evolve in
response to human cultural activities.
It has been proposed that cultural differentiation between
groups can have knock-on effects on genetic differences, in the
case of birdsong leading to segregated communities between
which courtship and mating break down, ultimately leading to
speciation (97). The most comprehensive analyses of such effects
in cultural evolution among animals have been in whales and
dolphins (34, 35). In some species, different migratory routes
appear to be culturally transmitted from mothers to calves, thence
reflected in diverging genetic make-ups. Most striking is the case
of the differentiation of killer whale ecotypes characterized by
alternative hunting strategies (targeted at seals vs. salmon and
other fish, for example), song types, and residence patterns, that
are proposed to be responsible for anatomical changes, such as
different jaw types suited to alternative prey (35, 98).
The Present Issue: The Extension of Biology
Through Culture
The papers that follow in this collection address the multiple
topics alluded to in the introduction above. Papers in the earlier
parts of the collection have a predominant focus on studies of
animals, and the remainder a focus on the human case. These are
sandwiched between an opening paper coauthored by one of the
founders of the subject of cultural evolution, that offers an over-
view of core cultural evolution theory and empirical findings
across human demography, population dynamics, and ecology
(18), and a final complementary overview appraising progress
made and prospects for the future of these endeavors (63).
Acknowledgments
The authors acknowledge the supplementary financial support to the Sackler
Colloquium of November 2016, provided by the John Templeton Foundation.

1 Barton N, et al. (2007) Evolution (Cold Spring Harbor Lab Press, Cold Spring Harbor, NY).
2 Futuyma DJ (2013) Evolution (Sinauer Associates, Sunderland, MA), 3rd Ed.
3 Losos JB (2013) Princeton Guide to Evolution (Princeton Univ Press, Princeton, NJ).
4 Huxley JS (1942) Evolution, the Modern Synthesis (Allen & Unwin, London).
5 Mayr E (1982) The Growth of Biological Thought: Diversity, Evolution and Iheritance (Harvard Univ Press, Cambridge, MA).
6 Mayr E (2002) What Evolution Is (Weidenfield and Nicholson, London).
7 Ridley M (2004) Evolution (Blackwell, Cambridge, MA), 3rd Ed.
8 Laland K, et al. (2014) Does evolutionary theory need a rethink? Nature 514:161–164.
9 Darwin C (1859) On the Origin of Species by Natural Selection (Murray, London).
10 Darwin C (1871) The Descent of Man and Selection in Relation to Sex (Murray, London).
11 Jones W (1798) The third anniversary discourse, delivered 2nd February 1786: On the Hindus. Asiatick Researches 1:415–431.

Whiten et al. PNAS | July 25, 2017 | vol. 114 | no. 30 | 7779
12 Schleicher A (1850) Linguistische Untersuchungen. 2. Teil: Die Sprachen Europas in systematischer Übersicht (HB König, Bonn).
13 Schleicher A (1869) Darwin Tested by the Science of Language (JC Hoten, London).
14 Cavalli-Sforza LL, Feldman MW (1973) Cultural versus biological inheritance: Phenotypic transmission from parents to children. (A theory of the effect of parental
phenotypes on children’s phenotypes.). Am J Hum Genet 25:618–637.
15 Cavalli-Sforza LL, Feldman MW (1981) Cultural Transmission and Evolution: A Quantitative Approach (Princeton Univ Press, Princeton, NJ).
16 Feldman MW, Cavalli-Sforza LL (1976) Cultural and biological evolutionary processes, selection for a trait under complex transmission. Theor Popul Biol
9:238–259.
17 Boyd R, Richerson P (1985) Culture and the Evolutionary Process (Univ of Chicago Press, Chicago).
18 Creanza N, Kolodny O, Feldman MW (2017) Cultural evolutionary theory: How culture evolves and why it matters. Proc Natl Acad Sci USA 114:7782–7789.
19 Maynard-Smith J, Szathmary E (1995) The Major Transitions in Evolution (Freeman, Oxford).
20 Dawkins R (1976) The Selfish Gene (Oxford Univ Press, Oxford).
21 Hewlett BS, de Silvestri A, Guglielmino CR (2002) Genes and semes in Africa. Curr Anthropol 43:313–321.
22 Fisher J, Hinde RA (1949) The opening of milk bottles by birds. Br Birds 42:347–357.
23 Kawai M (1965) Newly acquired pre-cultural behaviour of the natural troop of Japanese monkeys on Koshima Islet. Primates 2:1–30.
24 Marler P, Tamura M (1964) Culturally transmitted patterns of vocal behavior in sparrows. Science 146:1483–1486.
25 Whiten A, Hinde RA, Stringer CB, Laland KN, eds (2012) Culture Evolves (Oxford Univ Press, Oxford).
26 Hoppitt W, Laland KN (2013) Social Learning: An Introduction to Mechanisms, Methods and Models (Princeton Univ Press, Princeton, NJ).
27 Laland KN, Galef BG, eds (2009) The Question of Animal Culture (Harvard Univ Press, Cambridge, MA).
28 Galef BG, Whiten A (2017) The comparative psychology of social learning. APA Handbook of Comparative Psychology, eds Call J, Burghardt G, Pepperberg I,
Snowdon C, Zentall T (American Psychological Association, Washington, DC), pp 411–440.
29 Thornton A, Clutton-Brock T (2011) Social learning and the development of individual and group behaviour in mammal societies. Philos Trans R Soc Lond B Biol Sci
366:978–987.
30 Whiten A (2012) Social learning, traditions and culture. The Evolution of Primate Societies, eds Mitani J, Call J, Kappeler PM, Palombit RA, Silk JB (Chicago Univ
Press, Chicago), pp 682–700.
31 Whiten A (2017) Culture extends the scope of evolutionary biology in the great apes. Proc Natl Acad Sci USA 114:7790–7797.
32 Fragaszy DM, et al. (2017) Synchronized practice helps bearded capuchin monkeys learn to extend attention while learning a tradition. Proc Natl Acad Sci USA
114:7798–7805.
33 Perry SE, Barrett BJ, Godoy I (2017) Older, sociable capuchins (Cebus capucinus) invent more social behaviors, but younger monkeys innovate more in other
contexts. Proc Natl Acad Sci USA 114:7806–7813.
34 Whitehead H, Rendell L (2015) The Cultural Lives of Whales and Dolphins (Chicago Univ Press, Chicago).
35 Whitehead H (2017) Gene–culture coevolution in whales and dolphins. Proc Natl Acad Sci USA 114:7814–7821.
36 Garland EC, et al. (2011) Dynamic horizontal cultural transmission of humpback whale song at the ocean basin scale. Curr Biol 21:687–691.
37 Garland EC, Rendell L, Lamoni L, Poole MM, Noad MJ (2017) Song hybridization events during revolutionary song change provide insights into cultural
transmission in humpback whales. Proc Natl Acad Sci USA 114:7822–7829.
38 Catchpole CK, Slater PJB (2008) Bird Song: Biological Themes and Variations (Cambridge Univ Press, Cambridge, UK), 2nd Ed.
39 Slagsvold T, Wiebe KL (2011) Social learning in birds and its role in shaping a foraging niche. Philos Trans R Soc Lond B Biol Sci 366:969–977.
40 Aplin LM, et al. (2015) Experimentally induced innovations lead to persistent culture via conformity in wild birds. Nature 518:538–541.
41 Aplin LM, Sheldon BC, McElreath R (2017) Conformity does not perpetuate suboptimal traditions in a wild population of songbirds. Proc Natl Acad Sci USA
114:7830–7837.
42 Laland KN, Atton N, Webster MM (2011) From fish to fashion: Experimental and theoretical insights into the evolution of culture. Philos Trans R Soc Lond B Biol Sci
366:958–968.
43 Grüter C, Leadbeater E (2014) Insights from insects about adaptive social information use. Trends Ecol Evol 29:177–184.
44 Leadbeater E, Dawson EH (2017) A social insect perspective on the evolution of social learning mechanisms. Proc Natl Acad Sci USA 114:7838–7845.
45 Alem S, et al. (2016) Associative mechanisms allow for social learning and cultural transmission of string pulling in an insect. PLoS Biol 14:e1002564.
46 Mercader J, et al. (2007) 4,300-year-old chimpanzee sites and the origins of percussive stone technology. Proc Natl Acad Sci USA 104:3043–3048.
47 Allen J, Weinrich M, Hoppitt W, Rendell L (2013) Network-based diffusion analysis reveals cultural transmission of lobtail feeding in humpback whales. Science
340:485–488.
48 Hobaiter C, Poisot T, Zuberbühler K, Hoppitt W, Gruber T (2014) Social network analysis shows direct evidence for social transmission of tool use in wild
chimpanzees. PLoS Biol 12:e1001960.
49 Whiten A, Mesoudi A (2008) Review. Establishing an experimental science of culture: Animal social diffusion experiments. Philos Trans R Soc Lond B Biol Sci
363:3477–3488.
50 Whiten A, Caldwell CA, Mesoudi A (2016) Cultural diffusion in humans and other animals. Curr Op Psychol 8:15–21.
51 Shapiro JA (2017) Biological action in read-write genome evolution. Interface Focus, in press.
52 Mueller T, O’Hara RB, Converse SJ, Urbanek RP, Fagan WF (2013) Social learning of migratory performance. Science 341:999–1002.
53 Whiten A (2017) A second inheritance system: The extension of biology through culture. Interface Focus, in press.
54 Tomasello M (1999) The Cultural Origins of Human Cognition (Harvard Univ Press, Cambridge, MA).
55 Pagel M (2012) Wired For Culture: The Natural History of Human Communication (Allen Lang, London).
56 Henrich J (2015) The Secret of Our Success: How Culture is Driving Human Evolution, Domesticating Our Species, and Making Us Smarter (Princeton Univ Press,
Princeton, NJ).
57 Laland KN (2017) Darwin’s Unfinished Symphony: How Culture Made the Human Mind (Princeton Univ Press, Princeton, NJ).
58 Boivin NL, et al. (2016) Ecological consequences of human niche construction: Examining long-term anthropogenic shaping of global species distributions. Proc
Natl Acad Sci USA 113:6388–6396.
59 Alberti M, et al. (2017) Global urban signatures of phenotypic change in animal and plant populations. Proc Natl Acad Sci USA, 10.1073/pnas.1606034114.
60 Gray RD, Watts J (2017) Cultural macroevolution matters. Proc Natl Acad Sci USA 114:7846–7852.
61 Mesoudi A, Whiten A, Laland KN (2006) Towards a unified science of cultural evolution. Behav Brain Sci 29:329–347, discussion 347–383.
62 Mesoudi A (2011) Cultural Evolution: How Darwinian Theory Can Explain Culture and Sythesize the Social Sciences. (Univ of Chicago Press, Chicago).
63 Mesoudi A (2017) Pursuing Darwin’s curious parallel: Prospects for a science of cultural evolution. Proc Natl Acad Sci USA 114:7853–7860.
64 Harmand S, et al. (2015) 3.3-million-year-old stone tools from Lomekwi 3, West Turkana, Kenya. Nature 521:310–315.
65 Stout D, Hecht EE (2017) Evolutionary neuroscience of cumulative culture. Proc Natl Acad Sci USA 114:7861–7868.
66 d’Errico F, Stringer CB (2011) Evolution, revolution or saltation scenario for the emergence of modern cultures? Philos Trans R Soc Lond B Biol Sci 366:1060–1069.
67 d’Errico F, et al. (2017) Identifying early modern human ecological niche expansions and associated cultural dynamics in the South African Middle Stone Age. Proc
Natl Acad Sci USA 114:7869–7876.
68 Gray RD, Atkinson QD, Greenhill SJ (2011) Language evolution and human history: What a difference a date makes. Philos Trans R Soc Lond B Biol Sci
366:1090–1100.
69 Bouckaert R, et al. (2012) Mapping the origins and expansion of the Indo-European language family. Science 337:957–960.

7780 | www.pnas.org/cgi/doi/10.1073/pnas.1707630114 Whiten et al.

70 Currie TE, Mace R (2011) Mode and tempo in the evolution of socio-political organization: Reconciling ‘Darwinian’ and ‘Spencerian’ evolutionary approaches in
anthropology. Philos Trans R Soc Lond B Biol Sci 366:1108–1117.
71 Ross RM, Atkinson QD (2016) Folktale transmission in the Arctic provides evidence for high bandwith social learning among hunter-gatherer groups. Evol Hum
Behav 37:47–53.
72 Legare CH (2017) Cumulative cultural learning: Development and diversity. Proc Natl Acad Sci USA 114:7877–7883.
73 Harris PL, Bartz DT, Rowe ML (2017) Young children communicate their ignorance and ask questions. Proc Natl Acad Sci USA 114:7884–7891.
74 Gopnik A, et al. (2017) Changes in cognitive flexibility and hypothesis search across human life history from childhood to adolescence to adulthood. Proc Natl
Acad Sci USA 114:7892–7899.
75 Gelman SA, Roberts SO (2017) How language shapes the cultural inheritance of categories. Proc Natl Acad Sci USA 114:7900–7907.
76 Whiten A (2005) The second inheritance system of chimpanzees and humans. Nature 437:52–55.
77 Feldman MW, Laland KN (1996) Gene-culture coevolutionary theory. Trends Ecol Evol 11:453–457.
78 Richerson PJ, Boyd R (2005) Not by Genes Alone: How Culture Transformed Human Evolution (Univ of Chicago Press, Chicago).
79 Hoppitt WJE, et al. (2008) Lessons from animal teaching. Trends Ecol Evol 23:486–493.
80 Gergely G, Bekkering H, Király I (2002) Rational imitation in preverbal infants. Nature 415:755.
81 Lyons DE, Damrosch DH, Lin JK, Macris DM, Keil FC (2011) The scope and limits of overimitation in the transmission of artefact culture. Philos Trans R Soc Lond B
Biol Sci 366:1158–1167.
82 Lewontin RC (1970) The units of selection. Annu Rev Ecol Syst 1:1–18.
83 Mesoudi A, Whiten A, Laland KN (2004) Perspective: Is human cultural evolution Darwinian? Evidence reviewed from the perspective of the Origin of Species.
Evolution 58:1–11.
84 Richerson PJ, Christiansen MH (2013) Cultural Evolution: Society, Technology, Language and Religion (MIT Press, Cambridge, MA).
85 Warner RR (1988) Traditionality of mating-site preferences in a coral fish. Nature 335:719–721.
86 Gibson RM, Bradbury JW, Vehrencamp SL (1991) Mate choice in lekking sage grouse revisited: The roles of vocal display, female site fidelity, and copying. Behav
Ecol 2:165–180.
87 Laland KN (2004) Social learning strategies. Learn Behav 32:4–14.
88 Price EE, Wood LA, Whiten A (2016) Adaptive cultural transmission biases in children and nonhuman primates. Infant Behav Dev 48:45–53.
89 Baldwin JM (1902) Development and Evolution (Macmillan, New York).
90 Wyles JS, Kunkel JG, Wilson AC (1983) Birds, behavior, and anatomical evolution. Proc Natl Acad Sci USA 80:4394–4397.
91 Wilson AC (1985) The molecular basis of evolution. Sci Am 253:164–173.
92 Odling-Smee FJ, Laland KN, Feldman MW (2003) Niche Construction: The Neglected Process in Evolution (Princeton Univ Press, Princeton, NJ).
93 Richerson PJ, Boyd R, Henrich J (2010) Colloquium paper: Gene-culture coevolution in the age of genomics. Proc Natl Acad Sci USA 107:8985–8992.
94 Laland KN, Odling-Smee J, Myles S (2010) How culture shaped the human genome: Bringing genetics and the human sciences together. Nat Rev Genet
11:137–148.
95 Street SE, Navarrete AF, Reader SM, Laland KN (2017) Coevolution of cultural intelligence, extended life history, sociality, and brain size in primates. Proc Natl
Acad Sci USA 114:7908–7914.
96 Lotem A, Halpern JY, Edelman S, Kolodny O (2017) The evolution of cognitive mechanisms in response to cultural innovations. Proc Natl Acad Sci USA
114:7915–7922.
97 Grant BR, Grant PR (2002) Simulating secondary contact in allopatric speciation: An empirical test of premating isolation. Biol J Linn Soc Lond 76:545–556.
98 Foote AD, et al. (2016) Genome-culture coevolution promotes rapid divergence of killer whale ecotypes. Nat Commun 7:11693.

Whiten et al. PNAS | July 25, 2017 | vol. 114 | no. 30 | 7781
Cultural evolutionary theory: How culture evolves and
why it matters
Nicole Creanzaa,1, Oren Kolodnyb,1,2, and Marcus W. Feldmanb
a
Department of Biological Sciences, Vanderbilt University, Nashville, TN 37235; and bDepartment of Biology, Stanford University, Stanford, CA 94305
Edited by Kevin N. Laland, University of St. Andrews, St. Andrews, United Kingdom, and accepted by Editorial Board Member Andrew G. Clark April 29, 2017
(received for review January 16, 2017)
Human cultural traits—behaviors, ideas, and technologies that can
be learned from other individuals—can exhibit complex patterns
of transmission and evolution, and researchers have developed
theoretical models, both verbal and mathematical, to facilitate
our understanding of these patterns. Many of the first quantita-
tive models of cultural evolution were modified from existing con-
cepts in theoretical population genetics because cultural evolution
has many parallels with, as well as clear differences from, genetic
evolution. Furthermore, cultural and genetic evolution can interact
with one another and influence both transmission and selection.
This interaction requires theoretical treatments of gene–culture
coevolution and dual inheritance, in addition to purely cultural
evolution. In addition, cultural evolutionary theory is a natural
component of studies in demography, human ecology, and many
other disciplines. Here, we review the core concepts in cultural
evolutionary theory as they pertain to the extension of biology
through culture, focusing on cultural evolutionary applications in
population genetics, ecology, and demography. For each of these
disciplines, we review the theoretical literature and highlight rel-
evant empirical studies. We also discuss the societal implications of
the study of cultural evolution and of the interactions of humans
with one another and with their environment.
|
cultural evolution mathematical models | gene–culture coevolution |
|
niche construction demography
H uman culture encompasses ideas, behaviors, and artifacts
that can be learned and transmitted between individuals and
can change over time (1). This process of transmission and
change is reminiscent of Darwin’s principle of descent with
modification through natural selection, and Darwin himself drew
this explicit link in the case of languages: “The formation of
different languages and of distinct species, and the proofs that
both have been developed through a gradual process, are curiously
parallel” (2, 3). Theory underpins most scientific endeavors, and, in
the 1970s, researchers began to lay the groundwork for cultural
evolutionary theory, building on the neo-Darwinian synthesis of
genetics and evolution by using verbal, diagrammatic, and mathe-
matical models (4–8). These models are, by necessity, approxima-
tions of reality (9), but because they require researchers to specify
their assumptions and extract the most important features from
complex processes, they have proven exceedingly useful in ad-
vancing the study of cultural evolution (10). Here, we review the
field of cultural evolutionary theory as it pertains to the extension of
biology through culture. We focus on human culture because the
bulk of cultural evolutionary models are human-centric and certain
processes such as cumulative culture seem to be unique to humans.
However, numerous nonhuman species also exhibit cultural trans-
mission, and we consider the areas of overlap between models of
human and animal culture in Discussion.
The study of cultural evolution is important beyond its aca-
demic value. Cultural evolution is a fundamentally interdisci-
plinary field, bridging gaps between academic disciplines and
facilitating connections between disparate approaches. For ex-
ample, the advent of technologies for revealing genomic varia-
tion has led to a plethora of studies that measure association
7782–7789 | PNAS | July 25, 2017 | vol. 114 | no. 30
between DNA variants and traits that have major cultural
components, such as years of schooling, marriage choices, IQ test
results, and poverty. Perhaps because of the perceived greater
precision of the genomic data, these culturally transmitted com-
ponents have been relegated to the deep background, creating a
misleading public portrayal of the traits as being predetermined by
genetics (see, e.g., ref. 11). Models of the dynamics of interaction
among culture, demography, and genetics, which uncover the
complexities in the determination of these behaviors and traits, are
crucial to remedy this potentially dangerous misinterpretation.
Here, we explore the ways in which cultural evolutionary
theory and its applications enhance our understanding of human
history and human biology, focusing on the links between cul-
tural evolutionary theory and population genetics, human be-
havioral ecology, and demography. Throughout, we give examples
of efforts to apply theory to data, linking models of cultural evo-
lution to empirical studies of genetics, language, archaeology, and
anthropology. For example, studies of cultural factors, including
language and customs, help biologists interpret patterns of genetic
evolution that might be misinterpreted if the cultural context were
not taken into account. Finally, we outline several societal impli-
cations of cultural evolutionary theory.
Population Genetics and Cultural Evolution
Many of the first models of cultural evolution drew explicit
parallels between culture and genes by modifying concepts from
theoretical population genetics and applying them to culture.
Cultural patterns of transmission, innovation, random fluctua-
tions, and selection are conceptually analogous to genetic pro-
cesses of transmission, mutation, drift, and selection, and many
of the mathematical techniques used to study genetics can be
useful in the study of culture (1, 12). However, these mathe-
matical approaches had to be modified to account for the dif-
ferences between genetic and cultural transmission. For example,
we do not expect cultural transmission to follow the rules of genetic
transmission strictly. Indeed, cultural traits are likely to deviate from
all three laws of Mendelian inheritance: segregation, independent
assortment, and dominance (13).
The simple observation that cultural traits need not conform
to Mendelian inheritance is sufficient to produce complex evo-
lutionary dynamics: If children are likely to reject a cultural trait
that both of their parents possess, the frequency of that trait in
the population may oscillate between generations (4). In addi-
tion, if two biological parents have different forms of a cultural
trait, their child is not necessarily equally likely to acquire the
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: N.C., O.K., and M.W.F. designed research, performed research,
analyzed data, and wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. K.N.L. is a guest editor invited by the Editorial
Board.
1
N.C. and O.K. contributed equally to this work.
2
To whom correspondence should be addressed. Email: okolodny@stanford.edu.
www.pnas.org/cgi/doi/10.1073/pnas.1620732114

Human genes Human culture
Parents
Parents
Kin Societal
norms
Non-kin Peers
Offspring
Offspring
Fig. 1. Cultural transmission is more complex than genetic transmission and
may occur on short timescales, even within a single generation.
mother’s or father’s form of that trait (14). Further, a child can
acquire cultural traits not only from its parents (vertical trans-
mission) but also from nonparental adults (oblique) and peers
(horizontal) (1, 12); thus, the frequency of a cultural trait in the
population is relevant beyond just the probability that an individ-
ual’s parents had that trait (Fig. 1). In most cases, the more
common a cultural trait is in the population, the more likely it is for
an individual to have the opportunity to acquire it through social
learning (15). However, the size of the population may also in-
fluence the continuing transmission, and thus survival, of a cultural
trait (16). The relative importance of a population’s size, and its
environmental context, for the retention and perhaps expansion of
the cultural repertoire constitutes an ongoing debate (16–20).
The Roles of Transmission and Innovation in Cultural Evolution. Thus
far, we have made the analogy between alleles of a gene and
forms of a cultural trait, implying that the cultural trait in
question can be represented in a binary or discrete manner.
Although this approximation is appropriate for some culturally
transmitted traits, such as knowing or not knowing how to use a
certain tool, or smoking or not smoking, some cultural traits are
more naturally regarded as continuous or quantitative traits. For
example, cultural norms and preferences, such as degree of risk
tolerance, have been modeled as continuous traits (e.g., ref. 21), and
knowledge of a tool or technique has usefully been represented in
terms of a quantitative “skill level” (e.g., refs. 16, 22, and 23).
Like genes, cultural traits can be more or less adaptive
depending on the environment and spread accordingly. An in-
teresting question is the following: If a certain behavior may be
either innate (i.e., genetically determined) or culturally acquired
(and thus potentially responsive to the environment), which
environmental patterns would favor the genetic transmission?
Models predict that spatially varying environments will favor
cultural transmission, whereas only highly stable environments
would favor the genetic determination of the behavior (24–26).
Cavalli-Sforza and Feldman note an important reason that
genes, cultural traits, and environments should all be considered
together: “Given the existence of individual plasticity in response
to the environment, correlations between biological relatives are
expected even if there is no genetic variation whatsoever” (14).
Unlike in genetics, where mutations are the source of new
traits, cultural innovations can occur via multiple processes and
at multiple scales (1, 27–29). Most of the models described above
include the cultural transmission of existing traits without pro-
viding a mechanism for novel traits to be introduced to the
population. In many models of social learning, new information
enters a population via trial-and-error learning or individual in-
teractions with the environment, and this information can then
be culturally transmitted (30, 31). New cultural traits can also
originate when existing traits are combined in novel ways, which
can lead to exponential rates of cultural accumulation (32).
Recent models represent innovation as the result of multiple
interacting processes (27–29), and cultural traits can accumulate
in punctuated bursts when these processes of innovation are
interdependent: A truly groundbreaking innovation can pave the
Creanza et al.
COLLOQUIUM
PAPER
way for many related innovations and novel combinations (28).
Such dynamics may explain some of the punctuated bursts that
are observed in the archaeological record of stone tools, like
the dramatic increases in complexity near the transitions from
the Middle to the Upper Paleolithic and from the Paleolithic
to the Neolithic (33–35), and may provide an account of the
dynamics of technological development in historical times (36–
39). In many models of cultural evolution, the frequency of one
or more cultural traits is tracked over time, and the equilibrium
properties are sought. However, recent research highlights the
dynamics of cultural accumulation that occur in the transient
phase before the system approaches an equilibrium (28). For
example, if innovation processes are interdependent, as de-
scribed above, the cultural repertoire can fluctuate dramatically
before approaching an equilibrium because the loss or gain of a
groundbreaking innovation can lead to the loss or gain of its related
innovations as well (28). In addition, these models demonstrate how
innovation processes can change the parameters, and therefore the
dynamics, of cultural evolution, possibly altering the cultural equi-
librium, if there is one (29). For example, a game-changing in-
novation, such as the transition from foraging to agriculture, could
allow a population to feed many more people; thus, a cultural in-
novation can alter the size of the population, which is generally set
as a fixed parameter in cultural evolutionary models (29). Such
nonequilibrium dynamics arise, for example, in a recent com-
EVOLUTION
parison between modeling predictions and the archaeological
record that showed that the frequencies of Neolithic pottery
features over time are not consistent with a cultural system at
equilibrium (40).
Linking Genetic and Cultural Evolution. As mentioned above, the-
oretical treatments of cultural transmission and evolution can
usefully draw on concepts from theoretical population genetics,
ANTHROPOLOGY
extending them to accommodate cultural processes. However,
cultural and genetic evolutionary processes can also interact with
one another and with the environment (Fig. 2), and elucidating
the relative contributions of genes, culture, and environment to a
phenotype can be very difficult (41). Extensive theoretical work
has been devoted to characterizing these interactions, termed
gene–culture coevolution (1, 42), culture–gene coevolution (43),
dual inheritance theory (12, 44), or cultural niche construction
(45, 46). When cultural and genetic evolution interact, the dy-
namics of both genetic and cultural traits are likely to be very
different from those characteristic of only one mode of trans-
mission (47, 48). Further, cultural traits can alter the selection
pressures on genetic traits and vice versa: For example, genetic
traits that are adaptive in one cultural background might not be
adaptive in another (49, 50). The classic example of these in-
teractions between cultural and genetic evolution is lactase
persistence in adulthood: For much of human history, there was
little reason to digest milk after weaning, and adults did not typically
produce the enzyme that digests lactose. However, with the cultural
Climate Niche construction
Environmental change Selection pressures
Ultraviolet exposure Pathogens
Altitude Demography Microbiome
Transmission dynamics
Population size
Innovations
Flora and fauna
Food sources Migration
Subsistence strategy Genetic admixture
Available resources Cultural connectivity
Fig. 2. Cultural, genetic, and environmental factors influencing evolution.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7783
practice of cattle domestication and dairying, a genetic mutation
that enabled the production of the lactase enzyme in adulthood was
strongly favored by selection (51, 52).
Theoretical analyses show that gene–culture coevolution can
be dynamically complex and surprisingly unpredictable. For ex-
ample, a well-known finding in population genetics is that a fit-
ness advantage to heterozygote genotypes maintains genetic
variation in a population. However, it is not sufficient to main-
tain genetic variation for heterozygote offspring to be superior to
homozygotes in their ability to acquire an advantageous cultural
trait that is transmitted culturally by a parent (12). In fact, the
fitness advantage to the culturally transmitted trait has to be
sufficiently large that it overcomes imperfection in vertical cul-
tural transmission. In a similar vein, Aoki et al. (26, 53) modeled the
evolution of a genetic trait that increased the efficiency of teaching,
defined as vertical transmission of a cultural trait. Genetic variation
at this teaching locus could not be maintained with asexual haploid
genetics and uniparental cultural transmission, but sexual haploid
genetics and biparental cultural transmission could preserve both
genetic polymorphism of the teaching locus and polymorphism
of the cultural trait. These examples illustrate the theoretical
complexity that emerges when standard population genetic the-
ory is extended to include the interactions between genetic and
cultural traits; the result is a highly nonlinear theory with com-
plications not seen in purely biological theory.
The theoretical literature on gene–culture interactions has
become increasingly relevant in the genomic era. Genome-wide
association studies (GWAS) have shown many genomic associ-
ations with a wide array of complex phenotypes and have allowed
detection of signals of genetic adaptation (54). However, GWA
studies of behavioral phenotypes such as IQ, educational at-
tainment, and life history should be interpreted with care (55–
58). As the authors of one such study state: “Studies of genetic
analyses of behavioural phenotypes have been prone to mis-
interpretation, such as characterizing identified associated vari-
ants as ‘genes for education.’ Such characterization is not correct
for many reasons: Educational attainment is primarily deter-
mined by environmental factors” (55). Statistical relationships
between genetic variants and behaviors need not be causal be-
cause assortative mating, spatial autocorrelation, and a shared
environment can influence such relationships (55, 59–61). Twin
studies of tobacco smoking point to interacting roles of genetics,
environment, and assortative mating in the initiation and con-
tinuance of smoking (62). In large-scale studies of human health,
environmental and cultural factors should also be considered
because these could conflate the effects of genetics and ancestry
with those of poverty, stress, racism, or socioeconomic status
(63–65). For example, data from the large-scale Health and
Retirement Study showed an association between African an-
cestry and hypertension: The prevalence of hypertension was
eight percentage points higher in respondents with the highest
quartile of African ancestry compared with those with the lowest
quartile (63). However, controlling for a subset of factors related
to socioeconomic status (childhood disadvantage, education,
income, and wealth) explained ∼38% of this disparity, reducing
it to a five-percentage-point difference (63).
Nonrandom Assortment and Biased Transmission
Many theoretical population genetic studies make the assump-
tion that mating is random within a population. However, in real
human populations, this assumption is often violated, as indi-
viduals tend to prefer mates with similar phenotypes, such as eye
color (66), height, IQ (67), education level (61), and smoking
status (68). Cultural evolutionary theory has led to significant
advances in our understanding of the effects of nonrandom mat-
ing, revealing that the transmission and dynamics of cultural traits
can be sensitive to both phenotypic and environmental assorting
(41). Assortative mating, leading to an increased correlation be-
tween mates for genetic or cultural traits, can increase both ge-
notypic and phenotypic variance in a population (69, 70). In
addition, assortative mating (and other forms of homophily) acting
7784 | www.pnas.org/cgi/doi/10.1073/pnas.1620732114
on one cultural trait can influence the evolutionary dynamics of
other cultural traits, facilitating the spread of rare cultural or ge-
netic variants (71, 72). More generally, assorting can affect not just
mate choice but many types of cultural interactions, termed
“assortative meeting” (73). Empirical work supports this theoret-
ical finding; for example, beneficial health behaviors spread more
readily through a social network when individuals’ social contacts
were more similar to themselves (74, 75). Culturally mediated
assortment can also lead to biological differences: Partners that
are more similar tend to have more offspring (76), thus increasing
fitness, and assortative mating within highly homophilic groups
affects the average length of homozygous DNA segments (59, 77),
leading to the appearance of higher levels of inbreeding than
might actually exist. Humans can also assort by language; however,
studies of the interactions between language and genetic pop-
ulation structure show that the resulting dynamics can differ by
population. For example, in some geographic regions, language
boundaries do not seem to act as barriers to gene flow (78–80)
whereas, in other places, assorting with respect to language seems
to have had a large effect, and genetic similarity is more closely
associated with language than with geographic distance (80–83).
Assortative mating has had a measurable effect on human geno-
mic architecture, and genetic and phenotypic correlations between
partners are substantial (84).
In addition to choosing their mates nonrandomly, individuals
can also choose their cultural role models; these cultural trans-
mission biases affect the relationship between a trait’s frequency
in the population and its likelihood of transmission (Fig. 3). For
example, conformity bias is an exaggerated preference for the
cultural variant practiced by the majority of the population,
which can lead to an increasingly large majority over time (85,
86). Alternatively, individuals might preferentially seek out novel
cultural traits, termed rarity bias or novelty bias (30). These
frequency-dependent biases can lead to patterns of cultural dif-
fusion in which the prevalence of a cultural trait can change
dramatically over short timescales, producing logistic growth
(“S-shaped” curves) of trait frequency over time (87, 88). Ex-
amples of cultural traits that are likely to exhibit frequency-
dependent transmission are fashion trends (89), career choices
(12), and baby names (90). Conformist transmission is likely to
dominate when the environment is relatively stable and common
cultural traits are well adapted to that environment (86, 91).
Other types of transmission biases reflect not how common a
trait is in a population, but the characteristics of the people who
have the trait. In the case of prestige bias, individuals attempt to
acquire cultural traits that are perceived to be high quality by
selectively learning from those individuals with high social rank
(92). For example, in an experimental test, children were much
more likely to choose an adult cultural role model if they had
observed bystanders attending to the potential model rather
than ignoring him or her (93); thus, even at a very early age,
humans can assess such characteristics as prestige or social
standing. Individuals can also use observations of success asso-
ciated with a cultural trait, such as a fruitful hunt with a certain
tool, to develop a preference for cultural role models that are
Population Population
Learner Learner Learner Learner
Conformist Novelty Prestige Success
bias bias bias bias
Fig. 3. Biased cultural transmission mechanisms, where orange and purple
represent two forms of an arbitrary cultural trait. Conformity bias predicts
that learners will copy the most common trait, and novelty bias predicts they
will copy the most rare. Prestige bias predicts learners will copy an individual
of high social status (indicated by a crown) whereas success bias predicts they
will copy a successful individual (indicated by a gold medal).
Creanza et al.

demonstrably successful (30). This bias has been demonstrated
experimentally (94, 95); for example, when individuals partici-
pated in simulated hunting with virtual arrowheads and then
modified their arrowheads either by trial and error or imitation,
copying successful individuals gave significantly better results
than trial and error (94).
Models of Culture and Human Ecology
For thousands of generations humans have been carving their
existence in the world with cultural tools that have become in-
tegral to their livelihoods, thereby shaping their environment at
all scales, both intentionally and unintentionally. Attempting to
answer the question of what are the extensions of human biology
through culture leads to a striking conclusion: There are few
aspects of human biology that have not been shaped by our
culture. Human culture has also affected the biology, even the
survival, of nonhuman species (96). In this section, we review a
number of cases for which incorporating culture into models of
ecoevolutionary dynamics has proven valuable for the interpre-
tation, prediction, and, in some cases, direction of human ecol-
ogy and of human impact on the ecosystem.
Human Niche Construction. Niche construction is a process in
which organisms modify their environment in a way that alters
the selective pressures that these organisms experience, thus
affecting evolution (97). A special case of niche construction is
cultural niche construction: the alteration of the environment
through cultural practices, which may themselves evolve. Cul-
tural niche construction involves complex dynamics in which
selective pressures act on the culture itself, interacting with ge-
netic evolution and the environment to influence the spread of
both genetic and cultural traits (71). Because cultural change has
the potential to occur faster than genetic adaptation, dynamics of
niche construction that are driven by cultural traits play a
prominent role in human evolution; yet, only in recent decades
has cultural evolution begun to be explicitly incorporated into
human evolutionary ecology (98). Studies that pioneered this
approach showed how it can provide insight into the dynamics of
the demographic transition in postindustrialized societies (e.g.,
refs. 1 and 99). For example, the reduction in birth rate during
the demographic transition is often characterized as a paradox
because, from a Darwinian fitness perspective, individuals should
prefer to have more offspring, not fewer (100). However, if a
cultural norm favoring small family size spreads, the fertility rate
can drop as well, resulting in a culturally induced demographic
transition (99, 101), which is a case where natural selection and
cultural transmission seem to be in opposition.
The niche-construction approach has been productive in many
other studies, such as those that describe culturally driven change
at the ecosystem level: for example, the extinction of megafauna
after the arrival of humans (102), the change of broad-scale
landscapes as a result of cultivation in early and recent times
(103–105), and the traditional use of fire as a means to manip-
ulate the environmental dynamics in a way beneficial for humans
(106, 107). Niche construction is also important in understanding
the evolutionary dynamics driven by changes in the immediate
environment that humans experience, such as via construction of
shelters and production of clothing that enabled the expansion of
humans into otherwise uninhabitable regions (108), and the use
of fire for food handling, which allowed dramatic changes in
subsistence and may even have led to significant change to the
anatomy of the human jaw (109).
Major Cultural Shifts. A key aspect of human evolution is the change
over time in human subsistence strategies. Several models con-
sider the interaction of hunter-gatherers with the populations of
organisms that they consume and how these interact over time.
They propose that predation pressure can decrease a prey species’
population and exert selective pressures in favor of early re-
production at a smaller body size, potentially leaving a tell-
tale pattern in the archaeological record. The result may be the
Creanza et al.
COLLOQUIUM
PAPER
prey species’ extinction, which forces humans to shift their diet in
response. Such models as the Diet Breadth Model, the Broad
Spectrum Revolution, and Nutritional Ecology (110–113) capture
some of these processes, and, although they differ in many im-
portant dimensions, such as in the role they assign to plants in the
diet, they share the realization that cultural dynamics, genetic evo-
lution, and ecological processes must be considered jointly to un-
derstand human evolution. Studies in this tradition have also
proposed how gradually changing cultural practices may have cre-
ated the conditions that culminated in the Neolithic revolution, with
the domestication of multiple plant and animal species and the
subsequent changes in almost every aspect of human existence
(114–116). An interesting niche-construction perspective of these
topics is proposed by Smith and Zeder (117).
Models in Human Behavioral Ecology. Human behavioral ecology
applies approaches that were developed with a focus on non-
human species to the interpretation of human behavior (118).
One of these approaches is based on optimality in behavior, and
studies frequently devise models that capture human behavioral
constraints and alternatives, as well as their associated payoffs,
which are then considered jointly in predicting behavior or
explaining the evolutionary underpinnings of observed behaviors,
often under the assumption that humans behave in a way that
maximizes their fitness. A broad range of empirical and theo-
EVOLUTION
retical studies of culturally determined behaviors bear directly on
human fitness, past and present. Human ecological traits, such as
life history profiles, subsistence strategies, mating preferences,
economic decision making, and social structures (119–122), have
been analyzed to predict individual behavior and to support
potential intervention that might alter human behaviors at the
societal level.
Interestingly, few studies in human ecology consider the dy-
ANTHROPOLOGY
namics of cultural evolution on which the studied behaviors
depend; thus, for example, it is frequently assumed that alter-
native possible behaviors are available to the human group of
interest when they might not be, such as different subsistence
strategies. Similarly, with some notable exceptions (e.g., refs.
123–128), human behavioral ecology models often do not con-
sider ecological and evolutionary dynamics that may depend on
the studied behavior and that play out on intermediate and long
timescales: For example, how would prey populations evolve
over long periods of time in response to a certain human for-
aging strategy, and how would that feed back onto human
strategy choice? We suggest that these aspects are promising
avenues for further exploration.
Interspecies and Intergroup Dynamics. One of the hotly debated
topics in human prehistory is the replacement of Neanderthals
by modern humans ∼40,000 y ago. A recent study (129) proposed
an ecocultural model that incorporated cultural differences be-
tween two competing species into Lotka–Volterra competition
dynamics and showed that a difference in culture between moderns
and Neanderthals could have driven the latter’s extinction. This
model explicitly includes cultural evolutionary dynamics and
shows that a difference in population sizes between moderns in
Africa and Neanderthals in Eurasia could have led to a differ-
ence in the cultural complexity between the two populations,
allowing the small groups of moderns that migrated out of Africa
to gradually outcompete the larger population of Neanderthals
that they encountered.
This pattern—with one group replacing another as a result of
a culturally derived advantage—is likely to have taken place
repeatedly throughout human history. Thus, for example, genetic
evidence largely supports a scenario in which the Neolithic rev-
olution spread throughout the world not by diffusion of farming
practices among groups but by replacement of hunter-gatherer
groups by farmers (130) (see also refs. 34, 131, and 132). A
second revolution occurred 6,000 to 4,000 y ago, when the early
Neolithic farmers were overwhelmed by Yamnaya invaders from
the Russian Steppe, who had the cultural advantage of
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7785
transportation by horses (133, 134). Such dynamics, in which
cultural adaptation to temporally variable conditions may play an
important role, are also pervasive more recently: For example,
competition between pastoralists and agriculturalists and re-
placement of one by the other are documented from biblical
times to the present (135, 136).
Culture and Microbes. Models are also important in analyzing
humans’ cultural and genetic coevolution with pathogens, the
realm in which many of our species’ harshest evolutionary
challenges have occurred. Some of the clearest signals of natural
selection in the human genome are found near genes that are
directly related to coping with diseases such as malaria (137,
138), Kuru (139), and others (140–142), and the understanding
of their evolutionary dynamics is greatly enhanced when we are
able to couple such genetic evidence with cultural dynamics that
influenced them. Durham (143), for example, argues that yam
farming practices in West Africa significantly increased standing
water, thus increasing breeding sites for malaria-carrying mos-
quitoes, which led to high exposure to malaria and exerted se-
lective pressure in favor of genetic variants that increase
resistance to malaria. In the New Guinea highlands, cannibalism
practices that were widespread until the 1940s drove the Kuru
epidemic among the people of this region (144). A model of
culture–pathogen interactions demonstrated that different be-
havioral regimes could shape dynamics of pathogenic bacteria,
leading to nonintuitive outcomes (145). For example, antibiotic-
resistant strains will spread throughout the population in the
presence of ubiquitous antibiotic use whereas the WT bacteria
have a fitness advantage if antibiotics are not used; however, if
people modify their behavior by decreasing use of antibiotics
when they become less effective, both WT and resistant patho-
gens can coexist (145).
A fast-growing body of research focuses on the host-associated
microbiome: the communities of organisms, mostly bacteria, that
live in and on eukaryotes. The dynamics of the microbiome can
interact with those of its host, including genetic variation, cul-
tural practices, and environmental context, further complicating
the study of evolutionary processes. Thus, for example, the in-
teraction between dairy farming and selection on the lactase
persistence gene has become the poster child of gene–culture
coevolution; however, lactose-using bacteria in humans’ digestive
tracts are very likely to have played a prominent role in the
emergence of dairy farming (146). Moreover, these bacteria
continue to affect individuals who do not carry a genetic muta-
tion that allows them to efficiently digest dairy in adulthood.
Understanding how cultural practices influence human–microbe
interactions may provide us not only with insight into the Neo-
lithic farming revolution or early cattle domestication and re-
lated human evolution since then, but also with the necessary
tools to make informed nutritional choices, such as those related
to dairy utilization in our present lives. Thus, worldwide dietary
recommendations stand to benefit significantly from an im-
proved understanding of microbe–human interactions (147).
Demography and Cultural Evolution
The growth and age structure of human populations are both
affected by norms and beliefs of their members. A predominantly
agricultural lifestyle produced higher population growth than the
hunting-gathering lifestyle it replaced (148, 149). This increased
growth was most likely due to the spread of a complex of cultural
traits (150) whose adoption may have created conditions that
favored the accumulation of subsequent culturally transmitted
behaviors (151, 152). Beginning in the late 19th century, parts of
Europe, Asia, the United States, Australia, and New Zealand
began to undergo a second demographic transition, which in-
volved a change from a high birth rate, high mortality regime to a
lower birth rate, low mortality regime. These changes were due
to the spread of fertility-reducing and survival-increasing be-
haviors that became part of the developed countries’ cultures.
7786 | www.pnas.org/cgi/doi/10.1073/pnas.1620732114
Standard quantitative models of demographic change do not in-
clude within-population variation in behaviors that affect fecundity
or mortality. Projections usually use fixed values for birth and death
rates; however, religious preferences, marriage customs, dietary
choices, population subdivision, and mortality profiles may af-
fect fecundity but are usually not part of demographic models.
Further, aspects of cultural transmission, such as prestige bias
and the choice of nonparental cultural role models, can facili-
tate the spread of fertility-reducing behaviors (12, 153). Thus,
cultural evolutionary approaches should be integrated into
demography, especially the processes that have led to fertility
decline (154).
Many models for life history analysis of humans divide the
lifespan into an ordered series of age classes. These models first
define the fertility rates of each age class and the survival rates
from one age class to the next. Then, they iterate the number in
each age class produced by these parameters to determine the
dynamics of the population, including whether the number in
each age class approaches a stable equilibrium, termed the sta-
tionary age distribution, or whether the population will grow or
go extinct and at what rate (155).
Carotenuto et al. proposed a demo-cultural framework for
such an age-structured population, in which each individual
carried one variant of a dichotomous trait, say H or h, where H
represents the presence of a socially learned behavior (for example,
fertility control) and h is its absence (156). An individual of type H
might also be more likely to survive into the next age class. This
integration of demography and culture yields complex dynamics; for
example, the trait H can persist in the populations even if it lowers
fertility, as long as the cultural transmission of H is reliable enough,
or if H also sufficiently increases the chance of survival. Addi-
tional learning steps can also be added to age-structured models,
such that vertical and horizontal transmission can occur at dif-
ferent rates for different age classes (101). In this case, hori-
zontal learning accelerated the trait’s spread and led to faster
population growth than vertical transmission alone.
An important outgrowth of demo-cultural modeling has been
its application to the sex-ratio problem. In many places, the sex
ratio at birth is strongly biased in favor of males and, in China
and parts of India, has resulted in up to 120 male births for every
100 female births (157). This cultural preference for sons can be
manifested in sex-selective abortion or withholding of resources
from daughters. This bias has both economic and socio-cultural
antecedent, as well as important ethical and demographic
consequences (158).
Data on cultural transmission of son preference can be in-
corporated into formal demographic analysis (159), linking these
data to real-world policy applications (160). Theoretical models
can also aid in predicting the effects of policies: For example,
one such model tracked the cultural transmission of the perceived
present value of a son relative to a daughter, the sex ratio at birth,
and their effects on demographic change (161). The results of this
model suggest that interventions focused on peer-to-peer cultural
transmission of a perceived higher value of daughters might
complement existing economic incentives to support and educate
daughters, with the goal of mitigating the effects of son preference.
The literature on the interaction between cultural transmission
and formal demography is quite sparse. Given the large variety
of customs that relate to birth and death rates in different hu-
man societies, population projections for the future needs of di-
verse populations should incorporate more cultural dynamics than
is currently standard practice.
Discussion
With the extensive body of theoretical and empirical literature
on cultural evolution, researchers in this field are now combining
information from multiple disciplines and integrating disparate
approaches. Part of this new frontier involves more fully bridging
the divide between theory and data, as well as developing
mathematical models than can aid in the interpretation of an-
thropological and archaeological information. In addition to
Creanza et al.

aiding our understanding of human history, the study of cultural
transmission and evolution is extremely relevant in the modern
era. Insights from cultural evolution and the diffusion of inno-
vations have been coopted in advertising and social media to
quantify the viral spread of information (e.g., ref. 162). How can
these cultural evolutionary insights be better used for positive
action and public health? In addition, how can we better use
knowledge about cultural evolution to more fully understand
patterns of human genetic variation and population structure?
As we continue to understand more about the human genome, it
becomes increasingly important to consider environmental and
cultural contexts as well as genetic variation; however, in the
study of gene-culture interactions, faulty logic or racial biases
about “causes” of human differences may be used and must be
cautiously guarded against (reviewed in ref. 163).
In this paper, we have reviewed aspects of human cultural
evolutionary theory, focusing on those that are most closely
linked to the extension of biology through culture. With this
focus, we could not do adequate justice to many important do-
mains of cultural evolutionary theory. In brief, many models of
cultural evolution focus primarily on the transmission of cultural
traits and not on their interactions with genes or fitness. These
models include, but are not limited to, models of social learning
(e.g., refs. 164–167), models of language evolution (e.g., refs.
168–171), empirically driven verbal models of human evolution
based on patterns in material culture (e.g., refs. 172–174), and
models of cultural dynamics within and between groups (e.g.,
refs. 86 and 175–178)). In addition, we focused on human studies,
although cultural processes are present in many other species. For
example, social learning has been extensively studied in non-
human animals, in which behavioral strategies, such as producer
and scrounger, and cultural trajectories can be more clearly
defined than in humans (166, 179). Cultural transmission also
has large-scale evolutionary implications for some nonhuman
animals: For example, theoretical studies suggest that nonrandom
1. Cavalli-Sforza LL, Feldman MW (1981) Cultural Transmission and Evolution: A
Quantitative Approach (Princeton Univ Press, Princeton).
2. Darwin C (1859) On the Origin of Species by Means of Natural Selection (Murray,
London).
3. Darwin C (1888) The Descent of Man, and Selection in Relation to Sex (Murray,
London).
4. Feldman MW, Cavalli-Sforza LL (1976) Cultural and biological evolutionary processes,
selection for a trait under complex transmission. Theor Popul Biol 9:238–259.
5. Feldman MW, Cavalli-Sforza LL (1975) Models for cultural inheritance: A general
linear model. Ann Hum Biol 2:215–226.
6. Blum HF (1978) Uncertainty in interplay of biological and cultural evolution: Man’s
view of himself. Q Rev Biol 53:29–40.
7. Cavalli-Sforza L, Feldman MW (1973) Models for cultural inheritance. I. Group mean
and within group variation. Theor Popul Biol 4:42–55.
8. Alland A, Jr (1972) Cultural evolution: The Darwinian model. Soc Biol 19:227–239.
9. Burnham KP, Anderson DR (1998) Model Selection and Inference: A Practical
Information-Theoretic Approach (Springer, New York).
10. Haldane JBS (1964) A defense of beanbag genetics. Perspect Biol Med 7:343–359.
11. Guedes Jd, et al. (2013) Is poverty in our genes? Curr Anthropol 54:71–79.
12. Boyd R, Richerson PJ (1985) Culture and the Evolutionary Process (Chicago Univ Press,
Chicago).
13. Mesoudi A (2017) Pursuing Darwin’s curious parallel: Prospects for a science of cul-
tural evolution. Proc Natl Acad Sci USA 114:7853–7860.
14. Cavalli-Sforza LL, Feldman MW (1973) Cultural versus biological inheritance: Phe-
notypic transmission from parents to children. (A theory of the effect of parental
phenotypes on children’s phenotypes). Am J Hum Genet 25:618–637.
15. Giraldeau L-A (1994) Social foraging: Individual learning and cultural transmission of
innovations. Behav Ecol 5:35–43.
16. Henrich J (2004) Demography and cultural evolution: How adaptive cultural pro-
cesses can produce maladaptive losses—The Tasmanian case. Am Antiq 69:197–214.
17. Henrich J, et al. (2016) Understanding cumulative cultural evolution. Proc Natl Acad
Sci USA 113:E6724–E6725.
18. Vaesen K, Collard M, Cosgrove R, Roebroeks W (2016) Population size does not explain
past changes in cultural complexity. Proc Natl Acad Sci USA 113:E2241–E2247.
19. Collard M, Ruttle A, Buchanan B, O’Brien MJ (2013) Population size and cultural
evolution in nonindustrial food-producing societies. PLoS One 8:e72628.
20. Collard M, Buchanan B, Morin J, Costopoulos A (2011) What drives the evolution of
hunter-gatherer subsistence technology? A reanalysis of the risk hypothesis with
data from the Pacific Northwest. Philos Trans R Soc Lond B Biol Sci 366:1129–1138.
21. Bisin A, Verdier T (2010) The economics of cultural transmission and the dynamics of
preferences. Handb Soc Econ 319:339–416.
Creanza et al.
COLLOQUIUM
PAPER
mating in birds based on culturally transmitted songs could ac-
celerate speciation (180, 181) and that sexual selection on
learned songs could influence evolution of the neural under-
pinnings of learning (182). Recently, studies in a range of animal
species have shown that cultural practices can emerge, spread,
and change over time, potentially influencing individuals’ fitness
(183–187). Tool use among chimpanzees and capuchins (188–
190) is one such example, which also provides insight regarding
the possible origins of the early phases of our own species’ ad-
aptation to the “cultural niche” (191, 192).
In recent years, models that are used for decision making in
various fields, such as economics and public health, have begun to
take cultural evolution into account, and a growing number also
incorporate the modeling—verbal or mathematical—of the hu-
man ecosystem’s expected coevolution with the spread of cultural
practices. These models play a prominent role in planned strate-
gies related to climate change and reduction of carbon emissions
(193), in predicting global food shortages and requirements (194),
and in assessing the distribution of new practices and technologies
in agriculture (195, 196). In addition, models in epidemiology
have begun to integrate cultural transmission of health practices
and pathogen ecological dynamics with regard to drug distribution
and combating epidemics (e.g., ref. 197).
Deeper analysis of how human culture, human ecology, and
EVOLUTION
the human environment coevolve is necessary for understanding
historical and present dynamics, and for predicting future trends.
These analyses will provide much-needed tools for the planning
and direction of such dynamics. Humans’ worldwide well-being
and that of the ecosystem we live in depend on our ability to
make such predictions and act accordingly.
ACKNOWLEDGMENTS. We thank the John Templeton Foundation and
ANTHROPOLOGY
Stanford Center for Computational, Evolutionary, and Human Genomics
for funding.
22. Kobayashi Y, Aoki K (2012) Innovativeness, population size and cumulative cultural
evolution. Theor Popul Biol 82:38–47.
23. Baldini R (2015) Revisiting the effect of population size on cumulative cultural
evolution. J Cogn Cult 15:320–336.
24. Boyd R, Richerson PJ (1983) The cultural transmission of acquired variation: Effects
on genetic fitness. J Theor Biol 100:567–596.
25. Aoki K, Feldman MW (2014) Evolution of learning strategies in temporally and
spatially variable environments: A review of theory. Theor Popul Biol 91:3–19.
26. Aoki K, Wakano J, Feldman M (2005) The emergence of social learning in a tem-
porally changing environment: A theoretical model. Curr Anthropol 46:334–340.
27. Fogarty L, Creanza N, Feldman MW (2015) Cultural evolutionary perspectives on
creativity and human innovation. Trends Ecol Evol 30:736–754.
28. Kolodny O, Creanza N, Feldman MW (2015) Evolution in leaps: The punctuated accu-
mulation and loss of cultural innovations. Proc Natl Acad Sci USA 112:E6762–E6769.
29. Kolodny O, Creanza N, Feldman MW (2016) Game-changing innovations: How cul-
ture can change the parameters of its own evolution and induce abrupt cultural
shifts. PLOS Comput Biol 12:e1005302.
30. Henrich J, McElreath R (2003) The evolution of cultural evolution. Evol Anthropol 12:
123–135.
31. Rendell L, et al. (2010) Why copy others? Insights from the social learning strategies
tournament. Science 328:208–213.
32. Enquist M, Ghirlanda S, Jarrick A, Wachtmeister C-A (2008) Why does human culture
increase exponentially? Theor Popul Biol 74:46–55.
33. Klein RG, Edgar B (2002) The Dawn of Human Culture (Wiley, New York).
34. Bar-Yosef O (1998) On the nature of transitions: The Middle to Upper Palaeolithic
and the Neolithic revolution. Camb Archaeol J 8:141–163.
35. Roebroeks W (2008) Time for the Middle to Upper Paleolithic transition in Europe.
J Hum Evol 55:918–926.
36. Darmstaedter L, Du Bois-Reymond R (1904) 4000 Jahre Pionier-Arbeit in den Exakten
Wissenschaften (JA Stargardt, Berlin).
37. Aiyar S, Dalgaard C-J, Moav O (2008) Technological progress and regress in pre-
industrial times. J Econ Growth 13:125–144.
38. Kuhn SL (2012) Emergent Patterns of Creativity and Innovation in Early Technologies:
Origins of Human Innovation and Creativity (Elsevier, Oxford), pp 69–88.
39. Lehman HC (1947) The exponential increase of man’s cultural output. Soc Forces 25:
281–290.
40. Crema ER, Kandler A, Shennan S (2016) Revealing patterns of cultural transmission
from frequency data: Equilibrium and non-equilibrium assumptions. Sci Rep 6:39122.
41. Feldman MW, Cavalli-Sforza LL (1979) Aspects of variance and covariance analysis
with cultural inheritance. Theor Popul Biol 15:276–307.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7787
42. Feldman MW, Laland KN (1996) Gene-culture coevolutionary theory. Trends Ecol
Evol 11:453–457.
43. Chudek M, Henrich J (2011) Culture-gene coevolution, norm-psychology and the
emergence of human prosociality. Trends Cogn Sci 15:218–226.
44. Richerson PJ, Boyd R (1978) A dual inheritance model of the human evolutionary
process. I. Basic postulates and a simple model. J Soc Biol Struct 1:127–154.
45. Laland KN, Odling-Smee J, Feldman MW (2000) Niche construction, biological evo-
lution, and cultural change. Behav Brain Sci 23:131–146. discussion 146–175.
46. Odling-Smee J, Laland KN, Feldman MW (2003) Niche Construction: The Neglected
Process in Evolution (Princeton Univ Press, Princeton).
47. Laland KN, Kumm J, Van Horn JD, Feldman MW (1995) A gene-culture model of
human handedness. Behav Genet 25:433–445.
48. Mesoudi A, Whiten A, Laland KN (2006) Towards a unified science of cultural evo-
lution. Behav Brain Sci 29:329–347, discussion 347–383.
49. Rendell L, Fogarty L, Laland KN (2011) Runaway cultural niche construction. Philos
Trans R Soc Lond B Biol Sci 366:823–835.
50. Laland KN, O’Brien MJ (2012) Cultural niche construction: An introduction. Biol
Theory 6:191–202.
51. Feldman MW, Cavalli-Sforza LL (1989) On the theory of evolution under genetic and
cultural transmission with application to the lactose absorption. Mathematical
Evolutionary Theory, ed Feldman MW (Princeton Univ Press, Princeton), pp 145–173.
52. Ingram CJ, Liebert A, Swallow DM (2012) Population genetics of lactase persistence
and lactose intolerance. eLS, 10.1002/9780470015902.a0020855.pub2.
53. Aoki K, Wakano J, Feldman M (2016) Gene-culture models for the evolution of al-
truistic teaching. On Human Nature: Biology, Psychology, Ethics, Policy, and Religion,
eds Tibayrenc M, Ayala F (Academic, Amsterdam), pp 279–296.
54. Berg JJ, Coop G (2014) A population genetic signal of polygenic adaptation. PLoS
Genet 10:e1004412.
55. Okbay A, et al.; LifeLines Cohort Study (2016) Genome-wide association study
identifies 74 loci associated with educational attainment. Nature 533:539–542.
56. Benyamin B, et al.; Wellcome Trust Case Control Consortium 2 (WTCCC2) (2014)
Childhood intelligence is heritable, highly polygenic and associated with FNBP1L.
Mol Psychiatry 19:253–258.
57. Davies G, et al. (2011) Genome-wide association studies establish that human in-
telligence is highly heritable and polygenic. Mol Psychiatry 16:996–1005.
58. Minkov M, Bond MH (2015) Genetic polymorphisms predict national differences in
life history strategy and time orientation. Pers Individ Dif 76:204–215.
59. Abdellaoui A, et al. (2015) Educational attainment influences levels of homozygosity
through migration and assortative mating. PLoS One 10:e0118935.
60. Piffer D (2015) A review of intelligence GWAS hits: Their relationship to country IQ
and the issue of spatial autocorrelation. Intelligence 53:43–50.
61. Domingue BW, Fletcher J, Conley D, Boardman JD (2014) Genetic and educational
assortative mating among US adults. Proc Natl Acad Sci USA 111:7996–8000.
62. Maes HH, et al. (2006) Genetic and cultural transmission of smoking initiation: An
extended twin kinship model. Behav Genet 36:795–808.
63. Marden JR, Walter S, Kaufman JS, Glymour MM (2016) African ancestry, social fac-
tors, and hypertension among non-Hispanic Blacks in the Health and Retirement
Study. Biodemogr Soc Biol 62:19–35.
64. Paradies Y, et al. (2015) Racism as a determinant of health: A systematic review and
meta-analysis. PLoS One 10:e0138511.
65. Nugent NR, Tyrka AR, Carpenter LL, Price LH (2011) Gene-environment interactions:
Early life stress and risk for depressive and anxiety disorders. Psychopharmacology
(Berl) 214:175–196.
66. Laeng B, Mathisen R, Johnsen JA (2007) Why do blue-eyed men prefer women with
the same eye color? Behav Ecol Sociobiol 61:371–384.
67. Keller MC, et al. (2013) The genetic correlation between height and IQ: Shared genes
or assortative mating? PLoS Genet 9:e1003451.
68. Treur JL, Vink JM, Boomsma DI, Middeldorp CM (2015) Spousal resemblance for
smoking: Underlying mechanisms and effects of cohort and age. Drug Alcohol
Depend 153:221–228.
69. Feldman MW, Cavalli-Sforza LL (1977) The evolution of continuous variation. II.
Complex transmission and assortative mating. Theor Popul Biol 11:161–181.
70. Rice J, Cloninger CR, Reich T (1978) Multifactorial inheritance with cultural trans-
mission and assortative mating. I. Description and basic properties of the unitary
models. Am J Hum Genet 30:618–643.
71. Creanza N, Fogarty L, Feldman MW (2012) Models of cultural niche construction with
selection and assortative mating. PLoS One 7:e42744.
72. Creanza N, Feldman MW (2014) Complexity in models of cultural niche construction
with selection and homophily. Proc Natl Acad Sci USA 111(Suppl 3):10830–10837.
73. Eshel I, Cavalli-Sforza LL (1982) Assortment of encounters and evolution of co-
operativeness. Proc Natl Acad Sci USA 79:1331–1335.
74. Centola D (2010) The spread of behavior in an online social network experiment.
Science 329:1194–1197.
75. Centola D (2011) An experimental study of homophily in the adoption of health
behavior. Science 334:1269–1272.
76. Thiessen D, Gregg B (1980) Human assortative mating and genetic equilibrium: An
evolutionary perspective. Ethol Sociobiol 1:111–140.
77. Abdellaoui A, et al. (2013) Association between autozygosity and major depression:
Stratification due to religious assortment. Behav Genet 43:455–467.
78. Hunley K, et al. (2008) Genetic and linguistic coevolution in Northern Island Mela-
nesia. PLoS Genet 4:e1000239.
79. Hunley K, Long JC (2005) Gene flow across linguistic boundaries in Native North
American populations. Proc Natl Acad Sci USA 102:1312–1317.
80. Srithawong S, et al. (2015) Genetic and linguistic correlation of the Kra–Dai-speaking
groups in Thailand. J Hum Genet 60:1–10.
7788 | www.pnas.org/cgi/doi/10.1073/pnas.1620732114
81. Barbujani G, Sokal RR (1990) Zones of sharp genetic change in Europe are also lin-
guistic boundaries. Proc Natl Acad Sci USA 87:1816–1819.
82. Karafet TM, et al. (2016) Coevolution of genes and languages and high levels of
population structure among the highland populations of Daghestan. J Hum Genet
61:181–191.
83. de Filippo C, et al. (2011) Y-chromosomal variation in sub-Saharan Africa: Insights
into the history of Niger-Congo groups. Mol Biol Evol 28:1255–1269.
84. Robinson MR, et al. (2017) Genetic evidence of assortative mating in humans. Nat
Hum Behav 1:0016.
85. Efferson C, Lalive R, Richerson PJ, McElreath R, Lubell M (2008) Conformists and
mavericks: The empirics of frequency-dependent cultural transmission. Evol Hum
Behav 29:56–64.
86. Henrich J, Boyd R (1998) The evolution of conformist transmission and the emer-
gence of between-group differences. Evol Hum Behav 19:215–241.
87. Rogers EM (2010) Diffusion of Innovations (Simon and Schuster, New York).
88. Henrich J (2001) Cultural transmission and the diffusion of innovations: Adoption
dynamics indicate that biased cultural transmission is the predominate force in be-
havioural change. Am Anthropol 103:992–1013.
89. Acerbi A, Ghirlanda S, Enquist M (2012) The logic of fashion cycles. PLoS One 7:
e32541.
90. Acerbi A, Alexander Bentley R (2014) Biases in cultural transmission shape the
turnover of popular traits. Evol Hum Behav 35:228–236.
91. Kendal J, Giraldeau LA, Laland K (2009) The evolution of social learning rules:
Payoff-biased and frequency-dependent biased transmission. J Theor Biol 260:
210–219.
92. Henrich J, Gil-White FJ (2001) The evolution of prestige: Freely conferred deference
as a mechanism for enhancing the benefits of cultural transmission. Evol Hum Behav
22:165–196.
93. Chudek M, Heller S, Birch S, Henrich J (2012) Prestige-biased cultural learning: By-
stander’s differential attention to potential models influences children’s learning.
Evol Hum Behav 33:46–56.
94. Mesoudi A, O’Brien MJ (2008) The cultural transmission of Great Basin projectile-
point technology II: An agent-based computer simulation. Am Antiq 73:627–644.
95. Mesoudi A (2011) An experimental comparison of human social learning strategies:
Payoff-biased social learning is adaptive but underused. Evol Hum Behav 32:
334–342.
96. Alberti M, et al. (2017) Global urban signatures of phenotypic change in animal and
plant populations. Proc Natl Acad Sci USA, 10.1073/pnas.1606034114.
97. Laland KN, Brown GR (2006) Niche construction, human behavior, and the adaptive‐
lag hypothesis. Evol Anthropol 15:95–104.
98. Laland KN, Odling‐Smee J, Feldman MW (2001) Cultural niche construction and
human evolution. J Evol Biol 14:22–33.
99. Ihara Y, Feldman MW (2004) Cultural niche construction and the evolution of small
family size. Theor Popul Biol 65:105–111.
100. Borgerhoff Mulder M (1998) The demographic transition: Are we any closer to an
evolutionary explanation? Trends Ecol Evol 13:266–270.
101. Fogarty L, Creanza N, Feldman MW (2013) The role of cultural transmission in hu-
man demographic change: An age-structured model. Theor Popul Biol 88:68–77.
102. Barnosky AD, Koch PL, Feranec RS, Wing SL, Shabel AB (2004) Assessing the causes of
late Pleistocene extinctions on the continents. Science 306:70–75.
103. Lansing JS, Cox MP, Downey SS, Janssen MA, Schoenfelder JW (2009) A robust
budding model of Balinese water temple networks. World Archaeol 41:112–133.
104. Erickson CL (1992) Prehistoric landscape management in the Andean highlands:
Raised field agriculture and its environmental impact. Popul Environ 13:285–300.
105. Delcourt PA, Delcourt HR (2004) Prehistoric Native Americans and Ecological
Change: Human Ecosystems in Eastern North America Since the Pleistocene (Cam-
bridge Univ Press, Cambridge, UK).
106. Boyd R (1999) Indians, Fire, and the Land in the Pacific Northwest (Oregon State Univ
Press, Corvallis, OR).
107. Bliege Bird R, Bird DW, Codding BF, Parker CH, Jones JH (2008) The “fire stick
farming” hypothesis: Australian Aboriginal foraging strategies, biodiversity, and
anthropogenic fire mosaics. Proc Natl Acad Sci USA 105:14796–14801.
108. Roebroeks W, et al. (1992) Dense forests, cold steppes, and the palaeolithic settle-
ment of Northern Europe. Curr Anthropol 33:551–586.
109. Wrangham RW (2009) Catching Fire: How Cooking Made Us Human (Basic Books,
New York).
110. Stiner MC (2001) Thirty years on the “broad spectrum revolution” and paleolithic
demography. Proc Natl Acad Sci USA 98:6993–6996.
111. Davis S, Rabinovich R, Goren-Inbar N (1988) Quaternary extinctions and population
increase in western Asia: The animal remains from Biq’at Quneitra. Paéorient 14:
95–105.
112. Hockett B, Haws JA (2005) Nutritional ecology and the human demography of Ne-
andertal extinction. Quat Int 137:21–34.
113. Hardy BL (2010) Climatic variability and plant food distribution in Pleistocene
Europe: Implications for Neanderthal diet and subsistence. Quat Sci Rev 29:662–679.
114. Flannery KV (1969) Origins and ecological effects of early domestication in Iran and
the Near East. The Domestication and Exploitation of Plants and Animals, eds
Ucko PJ, Dimbleby GW (Gerald Duckworth, London), pp 73–100.
115. Valla FR, Bar-Yosef O, eds (1991) The Natufian Culture in the Levant (International
Monographs in Prehistory, Ann Arbor, MI).
116. Rowley-Conwy P, Layton R (2011) Foraging and farming as niche construction: Stable
and unstable adaptations. Philos Trans R Soc Lond B Biol Sci 366:849–862.
117. Smith BD, Zeder MA (2013) The onset of the Anthropocene. Anthropocene 4:8–13.
118. Winterhalder B, Smith EA (2000) Analyzing adaptive strategies: Human behavioral
ecology at twenty-five. Evol Anthropol Issues News Rev 9:51–72.
Creanza et al.

119. Henrich J, et al. (2001) In search of homo economicus: Behavioral experiments in
15 small-scale societies. Am Econ Rev 91:73–78.
120. Kaplan H, Hill K, Lancaster J, Hurtado AM (2000) A theory of human life history
evolution: Diet, intelligence, and longevity. Evol Anthropol Issues News Rev 9:
156–185.
121. Winterhalder B, Lu F, Tucker B (1999) Risk-senstive adaptive tactics: Models and
evidence from subsistence studies in biology and anthropology. J Archaeol Res 7:
301–348.
122. Voland E (1998) Evolutionary ecology of human reproduction. Annu Rev Anthropol
27:347–374.
123. Belovsky GE (1988) An optimal foraging-based model of hunter-gatherer population
dynamics. J Anthropol Archaeol 7:329–372.
124. Winterhalder B, Baillargeon W, Cappelletto F, Daniel IR, Prescott C (1988) The
population ecology of hunter-gatherers and their prey. J Anthropol Archaeol 7:
289–328.
125. Broughton JM (1997) Widening diet breadth, declining foraging efficiency, and
prehistoric harvest pressure: Ichthyofaunal evidence from the Emeryville Shell-
mound, California. Antiquity 71:845–862.
126. Low BS, Heinen JT (1993) Population, resources, and environment: Implications of
human behavioral ecology for conservation. Popul Environ 15:7–41.
127. Stiner MC, Munro ND, Surovell TA, Tchernov E, Bar-Yosef O (1999) Paleolithic pop-
ulation growth pulses evidenced by small animal exploitation. Science 283:190–194.
128. Stiner MC, Munro ND, Surovell TA (2000) The tortoise and the hare. Curr Anthropol
41:39–79.
129. Gilpin W, Feldman MW, Aoki K (2016) An ecocultural model predicts Neanderthal
extinction through competition with modern humans. Proc Natl Acad Sci USA 113:
2134–2139.
130. Skoglund P, et al. (2012) Origins and genetic legacy of Neolithic farmers and hunter-
gatherers in Europe. Science 336:466–469.
131. Aoki K, Shida M, Shigesada N (1996) Travelling wave solutions for the spread of
farmers into a region occupied by hunter–gatherers. Theor Popul Biol 50:1–17.
132. Patterson MA, Sarson GR, Sarson HC, Shukurov A (2010) Modelling the Neolithic
transition in a heterogeneous environment. J Archaeol Sci 37:2929–2937.
133. Allentoft ME, et al. (2015) Population genomics of bronze age Eurasia. Nature 522:
167–172.
134. Goldberg A, Günther T, Rosenberg NA, Jakobsson M (2017) Ancient X chromosomes
reveal contrasting sex bias in Neolithic and Bronze Age Eurasian migrations. Proc
Natl Acad Sci USA 114:2657–2662.
135. Wossink A (2009) Challenging Climate Change: Competition and Cooperation
Among Pastoralists and Agriculturalists in Northern Mesopotamia (c. 3000-1600
BC) (Sidestone, Leiden, The Netherlands).
136. Spielmann KA, Eder JF (1994) Hunters and farmers: Then and now. Annu Rev
Anthropol 23:303–323.
137. Kwiatkowski DP (2005) How malaria has affected the human genome and what
human genetics can teach us about malaria. Am J Hum Genet 77:171–192.
138. Tishkoff SA, et al. (2001) Haplotype diversity and linkage disequilibrium at human
G6PD: Recent origin of alleles that confer malarial resistance. Science 293:455–462.
139. Mead S, et al. (2003) Balancing selection at the prion protein gene consistent with
prehistoric kurulike epidemics. Science 300:640–643.
140. Bustamante CD, et al. (2005) Natural selection on protein-coding genes in the hu-
man genome. Nature 437:1153–1157.
141. Sabeti PC, et al.; International HapMap Consortium (2007) Genome-wide detection
and characterization of positive selection in human populations. Nature 449:
913–918.
142. Enard D, Cai L, Gwennap C, Petrov DA (2016) Viruses are a dominant driver of
protein adaptation in mammals. eLife 5:e12469.
143. Durham WH (1991) Coevolution: Genes, Culture, and Human Diversity (Stanford
Univ Press, Stanford, CA).
144. Lindenbaum S (2015) Kuru Sorcery: Disease and Danger in the New Guinea
Highlands (Routledge, Abingdon, UK).
145. Boni MF, Feldman MW (2005) Evolution of antibiotic resistance by human and
bacterial niche construction. Evolution 59:477–491.
146. Walter J, Ley R (2011) The human gut microbiome: Ecology and recent evolutionary
changes. Annu Rev Microbiol 65:411–429.
147. Szilagyi A, Galiatsatos P, Xue X (2016) Systematic review and meta-analysis of lactose
digestion, its impact on intolerance and nutritional effects of dairy food restriction
in inflammatory bowel diseases. Nutr J 15:67.
148. Bocquet‐Appel J (2002) Paleoanthropological traces of a Neolithic demographic
transition. Curr Anthropol 43:637–650.
149. Gage TBB, DeWitte S (2009) What do we know about the agricultural demographic
transition? Curr Anthropol 50:649–655.
150. Ammerman AJ, Cavalli-Sforza LL (1984) The Neolithic Transition and the Genetics of
Populations in Europe (Princeton Univ Press, Princeton).
151. Henn BM, Cavalli-Sforza LL, Feldman MW (2012) The great human expansion. Proc
Natl Acad Sci USA 109:17758–17764.
152. Powell A, Shennan S, Thomas MG (2009) Late Pleistocene demography and the
appearance of modern human behavior. Science 324:1298–1301.
153. Richerson PJ, Boyd R (1984) Natural selection and culture. Bioscience 34:430–434.
154. Colleran H (2016) The cultural evolution of fertility decline. Philos Trans R Soc Lond B
Biol Sci 371:20150152.
155. Leslie PH (1948) Some further notes on the use of matrices in population mathe-
matics. Biometrika 35:213–245.
156. Carotenuto L, Feldman MW, Cavalli-Sforza L (1989) Age structure in models of cul-
tural transmission. Working paper (Morrison Institute for Population and Resource
Studies, Stanford, CA), No 16.
Creanza et al.
COLLOQUIUM
PAPER
157. Banister J (2004) Shortage of girls in China today. J Popul Res 21:19–45.
158. Tuljapurkar S, Li N, Feldman MW (1995) High sex ratios in China’s future. Science
267:874–876.
159. Li N, Feldman MW, Li S (2000) Cultural transmission in a demographic study of sex
ratio at birth in China’s future. Theor Popul Biol 58:161–172.
160. Bhattacharjya D, Sudarshan A, Tuljapurkar S, Shachter R, Feldman M (2008) How can
economic schemes curtail the increasing sex ratio at birth in China? Demogr Res 19:
1831–1850.
161. Fogarty L, Feldman MW (2011) The cultural and demographic evolution of son
preference and marriage type in contemporary China. Biol Theory 6:272–282.
162. Barkow JH, O’Gorman R, Rendell L (2012) Are the new mass media subverting cul-
tural transmission? Rev Gen Psychol 16:121–133.
163. Creanza N, Feldman MW (2016) Worldwide genetic and cultural change in human
evolution. Curr Opin Genet Dev 41:85–92.
164. Enquist M, Ghirlanda S (2007) Evolution of social learning does not explain the or-
igin of human cumulative culture. J Theor Biol 246:129–135.
165. Rendell L, Fogarty L, Laland KN (2010) Rogers’ paradox recast and resolved: Pop-
ulation structure and the evolution of social learning strategies. Evolution 64:
534–548.
166. Arbilly M, Weissman DB, Feldman MW, Grodzinski U (2014) An arms race between
producers and scroungers can drive the evolution of social cognition. Behav Ecol 25:
487–495.
167. Rendell L, et al. (2011) Cognitive culture: Theoretical and empirical insights into
social learning strategies. Trends Cogn Sci 15:68–76.
168. Hruschka DJ, et al. (2009) Building social cognitive models of language change.
Trends Cogn Sci 13:464–469.
169. Nowak MA, Krakauer DC (1999) The evolution of language. Proc Natl Acad Sci USA
96:8028–8033.
170. Gray RD, Greenhill SJ, Ross R (2007) The pleasures and perils of Darwinizing culture
(with phylogenies). Biol Theory 2:1–26.
EVOLUTION
171. Kolodny O, Lotem A, Edelman S (2015) Learning a generative probabilistic grammar
of experience: A process-level model of language acquisition. Cogn Sci 39:227–267.
172. Hovers E (2012) Invention, reinvention and innovation: The makings of Oldowan
lithic technology. Origins of Human Innovation and Creativity, ed Elias S (Elsevier,
Oxford).
173. Bar-Yosef O (2002) The Upper Paleolithic revolution. Annu Rev Anthropol 31:363–393.
174. Klein RG (2008) Out of Africa and the evolution of human behavior. Evol Anthropol
17:267–281.
175. Boyd R, Richerson PJ (2009) Voting with your feet: Payoff biased migration and the
ANTHROPOLOGY
evolution of group beneficial behavior. J Theor Biol 257:331–339.
176. Wiens JJ, Hollingsworth BD (2000) War of the Iguanas: Conflicting molecular and
morphological phylogenies and long-branch attraction in iguanid lizards. Syst Biol
49:143–159.
177. Borgerhoff Mulder M, et al. (2009) Intergenerational wealth transmission and the
dynamics of inequality in small-scale societies. Science 326:682–688.
178. Fogarty L, Strimling P, Laland KN (2011) The evolution of teaching. Evolution 65:
2760–2770.
179. Fehér O, Wang H, Saar S, Mitra PP, Tchernichovski O (2009) De novo establishment
of wild-type song culture in the zebra finch. Nature 459:564–568.
180. Verzijden MN, et al. (2012) The impact of learning on sexual selection and specia-
tion. Trends Ecol Evol 27:511–519.
181. Lachlan RF, Servedio MR (2004) Song learning accelerates allopatric speciation.
Evolution 58:2049–2063.
182. Creanza N, Fogarty L, Feldman MW (2016) Cultural niche construction of repertoire
size and learning strategies in songbirds. Evol Ecol 30:285–305.
183. Aplin LM, et al. (2015) Experimentally induced innovations lead to persistent culture
via conformity in wild birds. Nature 518:538–541.
184. Rendell L, Whitehead H (2001) Culture in whales and dolphins. Behav Brain Sci 24:
309–324, discussion 324–382.
185. Whiten A, et al. (1999) Cultures in chimpanzees. Nature 399:682–685.
186. Whitehead H (2017) Gene–culture coevolution in whales and dolphins. Proc Natl
Acad Sci USA 114:7814–7821.
187. Perry SE, Barrett BJ, Godoy I (2017) Older, sociable capuchins (Cebus capucinus) in-
vent more social behaviors, but younger monkeys innovate more in other contexts.
Proc Natl Acad Sci USA 114:7806–7813.
188. Fragaszy D, Izar P, Visalberghi E, Ottoni EB, de Oliveira MG (2004) Wild capuchin monkeys
(Cebus libidinosus) use anvils and stone pounding tools. Am J Primatol 64:359–366.
189. Whiten A, Horner V, de Waal FBM (2005) Conformity to cultural norms of tool use in
chimpanzees. Nature 437:737–740.
190. Ottoni EB, Izar P (2008) Capuchin monkey tool use: Overview and implications. Evol
Anthropol. 17:171–178.
191. Whiten A (2011) The scope of culture in chimpanzees, humans and ancestral apes.
Philos Trans R Soc Lond B Biol Sci 366:997–1007.
192. Whiten A (2017) Culture extends the scope of evolutionary biology in the great apes.
Proc Natl Acad Sci USA 114:7790–7797.
193. Seneviratne SI, Donat MG, Pitman AJ, Knutti R, Wilby RL (2016) Allowable CO2
emissions based on regional and impact-related climate targets. Nature 529:477–483.
194. Fischer RA, Byerlee D, Edmeades G (2014) Crop Yields and Global Food Security
(ACIAR, Canberra, Australia).
195. Garibaldi LA, et al. (2017) Farming approaches for greater biodiversity, livelihoods,
and food security. Trends Ecol Evol 32:68–80.
196. Kassam A, Friedrich T, Shaxson F, Pretty J (2009) The spread of conservation agri-
culture: Justification, sustainability and uptake. Int J Agric Sustain 7:292–320.
197. Rhines AS (2013) The role of sex differences in the prevalence and transmission of
tuberculosis. Tuberculosis (Edinb) 93:104–107.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7789
Culture extends the scope of evolutionary biology in
the great apes
Andrew Whitena,b,1
a
Centre for Social Learning and Cognitive Evolution, University of St. Andrews, St. Andrews, KY16 9JP, United Kingdom; and bScottish Primate Research
Group, School of Psychology and Neuroscience, University of St. Andrews, St. Andrews, KY16 9JP, United Kingdom
Edited by Kevin N. Laland, University of St. Andrews, St. Andrews, United Kingdom, and accepted by Editorial Board Member Andrew G. Clark April 29, 2017
(received for review January 14, 2017)
Discoveries about the cultures and cultural capacities of the great
apes have played a leading role in the recognition emerging in
recent decades that cultural inheritance can be a significant factor
in the lives not only of humans but also of nonhuman animals. This
prominence derives in part from these primates being those with
whom we share the most recent common ancestry, thus offering
clues to the origins of our own thoroughgoing reliance on cumulative
cultural achievements. In addition, the intense research focus on these
species has spawned an unprecedented diversity of complementary
methodological approaches, the results of which suggest that cultural
phenomena pervade the lives of these apes, with potentially
major implications for their broader evolutionary biology. Here I
review what this extremely broad array of observational and
experimental methodologies has taught us about the cultural lives
of chimpanzees, gorillas, and orangutans and consider the ways in
which this knowledge extends our wider understanding of primate
biology and the processes of adaptation and evolution that shape it.
I address these issues first by evaluating the extent to which the
results of cultural inheritance echo a suite of core principles that
underlie organic Darwinian evolution but also extend them in new
ways and then by assessing the principal causal interactions
between the primary, genetically based organic processes of
evolution and the secondary system of cultural inheritance that
is based on social learning from others.
social learning | culture | evolutionary biology | chimpanzee | orangutan
R ecent decades have revealed social learning (learning from
others) to be pervasive across the animal kingdom, with
important implications for evolutionary biology at large (1) and
the subject of the Sackler Colloquium published here (2). This
article focuses on great apes: chimpanzee (Pan troglodytes), go-
rilla (Gorilla gorilla), and orangutan (Pongo pygmaeus). Other
primates are dealt with elsewhere in the issue (3, 4). Despite an
early report (5), we still know little about cultural phenomena in
chimpanzees’ rarer sister species, the bonobo (Pan paniscus) so
bonobos are omitted here. I also make only limited reference to
human culture, although we are technically also great apes. Human
culture is extensively treated in other papers in this issue.
I first survey the nature and scope of social learning and as-
sociated aspects of cultural transmission in great apes, concluding
that the depth and diversity of observational and experimental
evidence for cultural phenomena are unparalleled among non-
human species. The evidence thus accumulated suggests that
culture permeates the lives of the great apes in the breadth of
behavioral repertoires affected and also in their time-depth. These
properties may be evolutionarily significant. The authors of a
comprehensive recent review of cetacean culture concluded that
“Culture . . . is a major part of what the whales are” (ref. 6, p. 7;
and see ref. 7). Such a statement is obviously true for our own
species (8–11); here I examine the justifications for thinking the
phrase also has validity for great apes.
Following a sister review ranging much more widely across
both vertebrates and invertebrates (1), I take eight core princi-
ples of evolution illuminated by Darwin (12) and assess the ex-
tent to which they apply to cultural phenomena in the great apes
7790–7797 | PNAS | July 25, 2017 | vol. 114 | no. 30
(henceforth simply “apes”), as they do in humans (13). I then
explore ways in which cultural inheritance goes yet further be-
yond these principles, creating new evolutionary phenomena.
Finally I address interactions between the primary manifesta-
tions of organic evolution based on genetic inheritance and the
“second inheritance system” (14) based on social learning. In a
now long-standing body of literature for humans, this inter-
action has been called “gene–culture coevolution” (15); the
logic of such coevolution (10, 16) may apply to other cultural
animals (1, 7).
Diverse and Convergent Evidence for the Scope of Great
Ape Culture
Geographic Variation in Traditions in the Wild. In 1986 Goodall
began to chart differences in behavior patterns among chim-
panzee study sites across Africa (17), proposing these differences
as cultural variants when no genetic or environmental explana-
tion was apparent (later called the “method of exclusion”). The
approach became more comprehensive with time (18, 19),
eventually benefitting from a systematic collaboration between
multiple long-term research groups (20, 21). Similar collaborative
analyses were soon achieved by orangutan field researchers (22)
and more recently by a gorilla consortium (23). These analyses
converged in reporting multiple cultural variants in all three gen-
era: 39 in Pan; 24 in Pongo, and 23 in Gorilla. The variants spanned
apes’ behavioral repertoires, including a great variety of tool use,
food processing, and social behavior, as discussed further below.
Further variants have continued to be reported intermittently for
Pan (24) and Pongo, in the latter case leading to a revised tally of
26–35 variants, depending upon the criteria applied (25).
These surveys are vulnerable to false positives (it can be dif-
ficult to be sure that all alternatives to social learning have been
excluded) and also to false negatives (cultural adaptations to
local environmental properties may be inappropriately excluded)
(26). However, these pioneering efforts provided essential plat-
forms for more refined approaches, some incorporating both
genetic and environmental variables into analyses (27). Other
advances yielded confirmatory evidence for culture through
(i) more focused microhabitat analyses for specific behaviors such
as ant-dipping (28, 29); (ii) comparisons between neighboring
communities sharing genes and habitat properties (30); and
(iii) social learning experiments, as for nut-cracking (31, 32).
The broad geographic surveys thus provide an initially im-
perfect but progressively refined overall picture of ape cultural
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: A.W. wrote the paper.
The author declares no conflict of interest.
This article is a PNAS Direct Submission. K.N.L. is a guest editor invited by the Editorial
Board.
1
Email: a.whiten@st-andrews.ac.uk.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1620733114/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1620733114

repertoires. The approach has been systematically applied to
spider monkeys (Ateles, reporting 23 cultural variants) (33) but not
yet, to my knowledge, to other animals. Evidence exists for mul-
tiple cultural variants in other species such as killer whales, which
display very different hunting repertoires (e.g., for fish versus
seals), song repertoires, and migratory patterns (6), but systematic
tabulations have yet to facilitate direct cross-species comparisons.
Intergroup Variation in Traditions in Captive Communities. A parallel
approach has compared neighboring communities in captive
contexts, with the advantage that genetic and environmental
explanations for group differences can be dismissed more
cleanly. For example in the Chimfunshi chimpanzee sanctuary in
Zambia, a bizarre habit of inserting a blade of grass into one ear
and leaving it there spread in one group but not in others (34).
Moreover a distinctive “hand-clasp” form of grooming was ab-
sent in this and one other group but was customary in others, in
which it additionally took different forms (35). Similar group
contrasts were found in the means by which hard-shelled Strychnos
fruits were opened (36). At the Yerkes Center in the United
States a further contrast in hand-clasp grooming emerged and
spread in one group over several years but remained absent in
another group (37). These results reinforce those derived from the
studies in the wild outlined above.
Quantitative Evidence for Vertical Mother-to-Offspring Transmission.
A study of the ontogeny of using stem-tools for termite-fishing
found that juvenile female chimpanzees spent significantly more
time attending to their mother’s fishing than did their male peers
(38). Consistent with the skills being learned by observation, the
young females tended to master the technique a whole year ahead
of the males, with a significant tendency to match even the length
of probe their mother typically inserted into the mound (38).
Researchers studying orangutans have called the focused vi-
sual attention of juveniles “peering” (Fig.1) (39). Building on
studies documenting correlations between maternal and juvenile
foraging profiles (40, 41), a suite of predictions were confirmed
that were consistent with peering functioning to facilitate learning
key survival skills (39). In foraging and nest-building contexts, in
which peering is most frequent, it was found that (i) the frequency
of peering in foraging contexts was predicted by the quantified
Fig. 1. Peering (39) by a juvenile orangutan as her mother extracts termites
from dead wood. Image courtesy of Christiaan Conradie and Caroline
Schuppli.
Whiten
COLLOQUIUM
PAPER
complexity of processing operations and also by the skill’s rarity;
(ii) peering was followed by a higher rate of exploration of the
item concerned, as was also confirmed specifically for the use of
sticks in foraging (seen only at one of the two sites studied);
(iii) peering rose along with the learning of new skills and di-
minished as competence was achieved; (iv) peering at nest-
building was followed by a rise in nest-building over the next
hour; (iv) developmentally, peering tracked the peak time of
learning to make nests; and (v) by about age 5 y, peering directed
at a juvenile’s mother tipped below 50% and was directed more
toward others, from whom there was still something to learn (39).
Such observations offer a compelling case that juvenile apes’ close
peering facilitates the learning of major life skills.
Quantitative Evidence for Horizontal Transmission. There is both
intracommunity and intercommunity evidence for horizontal
transmission in the wild. An example of the former was tracked
by network-based diffusion analysis, confirming that a novel
chimpanzee behavior, using moss as a water-sponge, spread from
the alpha male along lines of social affiliation, providing quan-
titative circumstantial evidence for transmission (42). Examples
of intercommunity transmission include (i) a significant accel-
eration in habituation to human observers in a chimpanzee
community being newly habituated after two females immigrated
from a well-habituated community (43); and (ii) the spread of
EVOLUTION
ant-fishing to a new community after the immigration of a pro-
ficient individual from a neighboring community where fishing
was habitual (44).
Quantitative and Qualitative Evidence for Investment in Transmission.
Videos of termite-fishing have documented skilled chimpanzee
mothers donating tools to less competent juveniles (Fig. S1), thus
suffering a diminished duration and rate of termite-fishing while
ANTHROPOLOGY
the recipient enjoyed improved fishing (45). The authors pro-
pose these observations meet commonly accepted criteria for a
functional (as opposed to intentional) concept of “teaching.”
They also document mothers orally splitting their tool lengthwise
neatly to make two functional tools or bringing multiple tools
and suggest these behaviors partially buffer mothers from the
costs of youngsters’ demands. Alternatively it might be argued
that these actions are essentially unnecessary and thus represent
the more compelling evidence that the behavior has costs and
therefore counts as teaching, even if the teaching is not as active
as teaching by scorpion provision by meerkats (46) or beaching
to catch seals by killer whales (6). However, the pattern of costs
and benefits suggests that this support has positive fitness ben-
efits for the young, and parallel reports concerning the use of
tools for nut-cracking have also been described (47).
An earlier report described more active involvement in curb-
ing youngsters’ exploration of potentially dangerous food-types.
Haraiwa-Hasegawa reported that when an infant, PN, reached to
touch some fig leaves, “her mother, FT, took PN’s hand and
moved it away from the leaves. As PN continued . . . FT took the
leaves from PN’s hand, plucked all the leaves within her arm’s
reach and dropped them to the ground” (ref. 48, p. 280). At least
one other mother behaved similarly and “prohibited . . . infants
only from feeding on the individual trees that they themselves
never fed on.”
Dyadic Experimental Studies of Social Learning. Experimental re-
ports of social learning by naive individuals from proficient
models multiplied for more than a century in all great ape genera
and have been tabulated and enumerated in successive reviews
(49: n = 19 studies; 50: n = 33 further studies; 51: n = 25 studies
comparing two or more ape species). The more recent experi-
ments often adopt a highly informative two-action approach in
which participants see one of two models, each trained to tackle
a problem such as opening a foraging box in a different way.
Ideally a third, no-model control group is included. This method
has demonstrated social learning in captive chimpanzees (52),
gorillas (53), and orangutans (54) that implies the copying of the
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7791
action (imitation) or movements of the manipulanda (emulation)
that were seen, rather than the simpler process of mere enhance-
ment of the manipulanda (55). These approaches have further
dissected the particular social learning processes at work, a topic
beyond the scope of this review (see refs. 56–59 for in-depth
treatment). More relevant to the present discussion and re-
viewed further below are extensions of these approaches to track
the successive cultural transmissions necessary to sustain traditions.
Nevertheless these dyadic experiments can importantly com-
plement results from the field. For example, nut-cracking with
natural hammer materials, found naturally only in West Africa
(Fig. 2), is shown not to be explicable as an instinct in the West
that is simply absent in the East, because East African chim-
panzees exposed to proficient models became proficient, whereas
controls in a no-model control condition did not, showing that the
behavior is (socially) learned (Fig. 2) (31, 32).
Cultural Diffusion Experiments. Experiments focused on the broader
phenomenon of cultural diffusion typically begin with models
displaying different solutions to a task and then track the potential
spread of the solutions in others who witness them. Alternative
variants provide important complementary information (63, 64).
For example, the “transmission chain” design pairs a first model
(A) with a naive individual (B); then when B achieves some
competence criterion, B becomes a model for a further indi-
vidual (B for C, C for D, and so on). Achieving such configurations
peacefully with apes requires sensitive experimental maneuvering,
but transmission has been demonstrated along chains of up to five
participants in chimpanzees (65) and orangutans (66), as well as
in children (65). These experiments provide important models
of transmission across cultural “generations,” implying a po-
tential for cultural transmission across what would naturally be
decades of ape life.
By contrast, “open diffusion” designs mimic transmission in
the wild in which whole groups are exposed to alternative models;
watching and copying is “open” to any individual in the group.
This design has been applied in chimpanzees and children in
several experiments (67–69), including two experiments in which
tool-use behavioral variants were transmitted across three groups
with significant fidelity (69). These results are important for inter-
preting putative cultures in the wild, such as the nut-cracking dis-
tributed across several hundred kilometers of West Africa, which
would have required repeated intercommunity transmission.
How Pervasive Is the Role of Cultural Inheritance in the Lives
of Great Apes?
Social Learning Shapes a Broad Repertoire of Cultural Variants. We
can first examine how pervasive the role of cultural inheritance is
in the lives of great apes by surveying the range of behaviors
described in the cross-site comparisons summarized above (20–
23, 27).
The chimpanzee lists of 1999–2001 include 30 different kinds
of tool use ascribed to culture, all suggesting functional and
adaptive payoffs beneficial to the performer’s biological fitness
(14, 21, 24). Others have been reported since then, including
sticks bitten and thus sometimes made sharp and used to stab or
evict bush baby prey at Fongoli in Senegal (70, 71); a kit of stout
tools used to make tunnels; and fine stems used to fish down
these tunnels and extract termites from nests deep underground
(72). A majority of such tools are used in food extraction, but
others are used in hygienic actions such as wiping blood or semen
off fur, in protective comfort roles such as creating leaf-cushions
on wet ground, and in local courtship gambits such as bending
small shrubs on the ground (20, 21). Other diverse items include
forms of food processing without tools, ways of dispatching ecto-
parasites located during grooming, and grooming customs such
as the hand-clasp that shows variant forms even in neighboring
communities (73).
The orangutan list of 2003 (22) also includes a dozen different
forms of tool use, several used for food extraction, such as holding a
small stick in the mouth to extract seeds from Neesia fruits or using
“leaf-gloves” to handle spiky fruit. Hygiene/comfort examples
include using a leaf napkin to wipe off sticky latex. The list as a
whole is diverse, incorporating forms of arboreal locomotion,

No nut-cracking, but presence of all necessary raw

materials confirmed
Boesch et al. [61] McGrew et al. [62]

Nut-cracking
recorded at eight
sites in West
Africa across
~500 Km [60] In an experiment in an Island
sanctuary, East African
chimpanzees who do not nut-
crack in the wild learned nut-
cracking through observation
[31,32]

Fig. 2. Convergent evidence for a culture of nut-cracking in chimpanzees. Evidence for nut-cracking is seen at multiple sites in West Africa (20, 21, 60) (white stars)
but is absent at others (black stars). The gray star indicates an early report in Cameroon, which was not subsequently confirmed. Independent studies confirmed
availability of raw materials at two such sites (61, 62). Experiments showed East African chimpanzees (two-letter ID codes) did not initially nut-crack (Phase 1), but
when half of the population was exposed to a proficient model, they began to do so (Phase 2), and all did so once exposed (Phase 3) (31, 32).

7792 | www.pnas.org/cgi/doi/10.1073/pnas.1620733114 Whiten


vocal sounds (some modified using leaves), variant nest con-
structions such as adding sun covers, and whether slow lorises are
eaten (irrespective of lorises’ availability).
The recent gorilla list of putative cultural variants (23) also
displays diversity that includes making bridges across water,
rubbing fruit to clean it or remove spines, incorporating tree-
slapping into displays, using teeth as a “fifth limb” in climbing,
forms of bodily contact while traveling together, and forms of
social play.
The Extent of Vertical Intergenerational Transmission. A detailed
study of the foraging behavior of young wild orangutans before
and after weaning concluded that their “diets were essentially
identical to their mothers’ even though not all mothers had the
same diet” . . . “immatures selectively observed their mothers
during extractive foraging, which increased goal-directed prac-
tice but not general manipulation of similar objects, suggesting
observational forms of learning of complex skills” (ref. 40, p. 62).
This conclusion was reinforced by a later study focused on
peering (39), referred to earlier. At 2–4 y of age, infants foraged
with their mother more than 90% of the time; 94% of their
feeding time occurred when the mother was also feeding; and
96% of their feeding was on the same items as the mother’s (39).
The extent of cofeeding is clearly massive in preweaning years
and is likely to engender the vertical social transmission of
dietary profiles.
These years of mother–offspring association and cofeeding are
typical of all the great apes and appear to lay down dietary
preferences that change relatively little after weaning. Although
the social learning implied may be as simple as enhancement of a
food type by the mother’s feeding on it, such effects are likely to
be profoundly important, because large diet-sets need to be
mastered and selected from the even more vast options a tropical
forest offers. This mastery includes avoiding the numerous plant
parts that are toxic, selecting relatively nutritious options, and
avoiding relatively poor ones. Chimpanzees may eat more than
300 different food types (species × parts) in a year (74); in the
Lopé Park of Gabon, for example, fruit alone is taken from
114 different plants (75) selected from among many hundreds of
potential food types available. The diet of gorillas may be similarly
diverse; gorillas in the Alfi Mountains of Cameroon eat more than
200 different food types, including fruits, seeds, leaves, stems, pith,
flowers, bark, roots, and invertebrates (75). For the orangutans of
Tanjung Putting in Borneo, the figure is again more than 300 dif-
ferent food types (76). However, the dietary profiles of different
populations may vary greatly, as suggested by earlier chimpanzee
studies (77) and confirmed more recently in neighboring orang-
utan populations separated by a large river, which displayed 60%
difference in diet, contrasting with intrapopulation homogeneity
(78). Years of close apprenticeship to a mother who daily displays
her knowledge of such a large but selective diet-set likely provide
an important means of achieving an adaptive response to this
challenging complexity.
Time Depth of Cultural Transmission. Long-term field sites have
shown that techniques such as termite-fishing have continued
across several generations during the half-century of research
now achieved. However, this time-frame pales in comparison
with the discoveries of real archaeological excavation, which in
the Tai Forest of Ivory Coast reached a depth corresponding to
4,300 y, where remains of nut-cracking were identified (illus-
trated in figure 1 of ref. 1) beneath those currently generated on
the surface by chimpanzees (79). Of course, this behavior may be
very much older. Once such a beneficial technology becomes
customary, it may continue in perpetuity, pending major ecological
perturbation. This example suggests that ape cultural inheritance
spans not only the breadth of behavioral repertoires outlined
earlier but also a potentially significant time depth comparable to
that familiar in organic evolution via genetic inheritance.
Whiten
COLLOQUIUM
PAPER
Does Ape Culture Instantiate a New Form of Evolution?
Ape culture may have pervasive effects in shaping the behavioral
repertoires of successive generations in the ways reviewed above,
but do these effects imply an extension of biology in the sense of
instantiating a new form of evolution based not on genetic but on
social inheritance? This development is what Dawkins proposed
in his concept of culturally replicated “memes” as analogies of
genes, creating a new form of evolution in the case of human
culture (80). The idea of aspects of culture such as language
evolving through variation, (cultural) inheritance, and selection
goes back to Darwin’s own writings (81) and was highlighted as the
tenth and latest of the major evolutionary transitions proposed by
Maynard-Smith and Szathmary (82). Mesoudi et al. (13) tackled
the issue in finding abundant evidence for counterparts in human
culture of eight major principles Darwin set out in the Origin
(12): variation, selection, inheritance, adaptation, accumulation of
modifications, geographic variation, convergence, and changes of
function. How does ape culture compare?
In addressing this question we must be clear about the phe-
nomena for which we are querying a potential “evolution.” If a
chimpanzee invents a better hammer for nut-cracking (perhaps
using a stone rather than wood), this innovation may enhance
that individual’s biological fitness, so that its genes are better rep-
resented in future generations, i.e., natural selection shapes bi-
ological evolution. However, if others copy use of the new tool, the
EVOLUTION
fitness (reproductive success) of that cultural entity—stone-tool
use—will be enhanced through its spread, and to that extent
there is cultural evolution of this behavior. It is this second
phenomenon that we are addressing here. Effects on an indi-
vidual culture-bearer’s biological, inclusive fitness are a different
matter to which we return in a later section. We can now con-
sider the eight evolutionary principles noted above.
ANTHROPOLOGY
Variation, Selection, and Inheritance. The three principles of vari-
ation, selection, and inheritance can together be regarded as the
core trinity of Darwinian evolution. Their joint working is an
evolutionary algorithm that has been suggested to have the
power to explain a multitude of phenomena beyond the living
systems to which Darwin applied it (83, 84).
As we have seen above, there is plentiful evidence in the great
apes for the feature of inheritance through social learning that
provides sufficient fidelity to sustain traditions. There is also
cultural variation, in part because, compared with gene replica-
tion, social learning is prone to imperfect copying. In the arrays
of cultural variants among great apes discussed earlier in this
article, there are plenty of behaviors that are displayed by many
but not all individuals in a community (these behaviors are classed
as “habitual” rather than “customary”).
By contrast, as yet there seems to be little direct recording of
cultural evolutionary change through competition and selection
within this variation. This absence of evidence of evolutionary
change is perhaps unsurprising. During the human Stone Age,
even when sophisticated, bilaterally symmetric Acheulian blades
showed an advance over earlier crude Oldowan tools, they
changed relatively little over a million years (85). If, as is plau-
sible, such stability characterizes chimpanzee nut-cracking and
other cultural variants of apes, then we will see little evidence
of cultural selection in human lifetimes. Of course, organic
evolutionary change is itself often slow compared with scientific
lifetimes, although instructive exceptions have often followed
human-caused environmental perturbations that create new se-
lection pressures. The classic example is selection favoring dark
morphs of peppered moths, better camouflaged against the sooty
surfaces of the industrial revolution, and then flipping to favor
light-colored morphs as the world became cleaner again.
Accordingly I have suggested that similar contexts of anthro-
pogenic change may be fruitful for investigating cultural evolu-
tion in animals (1). Scientific experiments may offer a convenient
instance. For example, in a pioneering cultural diffusion study,
three juvenile chimpanzees were confronted with and avoided
two novel objects (86). One youngster was then replaced with a
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7793
naive one, and this replacement was repeated, so after every third
such cycle the triplet contained different individuals. Never-
theless, approaches to the objects increased steadily and in later
generations of triplets became customary (Fig. S2). Accord-
ingly, here there was variation in boldness inheritance insofar
as naive youngsters learned from bolder ones that the objects
could be safely approached and explored, and competitive se-
lection favored small, progressive steps in boldness. Playing
with the objects thus could evolve as the norm in the later gener-
ations composed of different youngsters than the original shy ones.
In a counterpart from the wild, two individuals from a human-
habituated community of wild chimpanzees immigrated into a
neighboring community that scientists were beginning to habituate,
and at that point habituation accelerated significantly (43).
In these examples an initially common variant (caution) was
replaced competitively by another (boldness) which was adap-
tively superior (fear was unnecessary in these contexts). Like-
wise, in all the cultural diffusion experiments with apes cited
earlier, improved foraging techniques spread across test groups
to replace the less competitive behavioral state characterized by
their absence. Here, again, there was variation (although in this
case the variation was engineered by the experimenter), in-
heritance via social learning, and competitive selection favoring
the cultural spread of the new foraging technique.
I suggest experiments such as the one reported in ref. 86 may
allow us to explore the capacity of apes to exemplify, even if in a
limited way, the operation of the Darwinian-trinity algorithm in a
cultural context. Presumably, at some previous time, all the cases
of clearly beneficial cultural variants in the wild, such as nut-
cracking and other tool use, did not exist; so, where they are
customary, their common use is likely to have arisen through the
operation of this algorithm.
Adaptation. The growth of boldness in the studies discussed
above (43, 86) indicates culturally evolved instances of adapta-
tion, although the adaptive payoffs were likely only mild. In the
wild there is evidence that a more crucial level of adaptiveness
has been delivered. Chimpanzees in Bossou, Guinea, were shown
to be reliant on two forms of technology, in particular nut-
cracking and pestle-pounding (a means of extracting nutritious
pulp from the apex of palm trees) during the dry season when
fruit became scarce (87); these cultural variants allow these apes
to inhabit otherwise inadequate habitats. How often culturally
inherited technology is this critical remains difficult to judge at
present, but many forms of tool use allow chimpanzees and
orangutans to gain otherwise unavailable foodstuffs.
Such adaptations concern the local physical environment.
Other adaptations may be societal. In a community of chimpan-
zees that customarily practices hand-clasp grooming, it may be
adaptive to learn this behavior from those already using it; and
where a particular courtship gambit such as leaf-clipping has be-
come common, it will likely be beneficial to adopt this behavior as
an action already recognized by one’s potential mating partner.
Accumulation of Modifications. Human cultural modifications ac-
cumulate in an elaborate fashion that has no match in other
animals and display the most striking analogies with the richness
of the evolved forms of the living world (9, 10, 13, 16, 83, 84).
Many authors assert that we are the only species to exhibit cu-
mulative culture (8, 9, 49, 56), but I suggest this view may be
premature. For example, chimpanzees in Goualougo use a stout
stick to make a deep tunnel to reach subterranean termite nests
and then use long stems to fish down the tunnels, first creating a
distinct brush tip effective for fishing by stripping the stem ends
through their teeth (72). They do so in a context in which the
effective procedure is highly opaque, so it is difficult to see how
the technique could have developed other than by a series of
cumulative steps beginning with the more transparent context of
fishing near the surface. Boesch (47) describes several other
candidates for cumulative cultural evolution in chimpanzees.
Direct evidence for the origins of such routines is lost in the past,
7794 | www.pnas.org/cgi/doi/10.1073/pnas.1620733114
but their complexity suggests elementary forms of cumulative
culture, comparable perhaps to the achingly slow forms that
characterized the early hominin Stone Age.
Geographic Variation. Because the Darwinian algorithm operates
in different regions, organic characteristics differentiate, and
speciation may occur. Parallel effects occur in human cultural
evolution (13, 88). As we have seen, great apes show evidence of
different traditions at geographically separated locations, and
there is evidence from all of the great ape genera that differences
in putative cultural profiles are correlated with the geographic
separation of communities (23, 24). As humans or other apes
disperse over greater distances, one would expect both genetic
and cultural similarities to diminish, and indeed Langergraber
et al. (89) showed that cultural variation in chimpanzees is also
correlated with genetic variation (but this correlation does not
mean genes explain the behavioral differences; see the supple-
mentary information in ref. 79 for further discussion of this study).
Kamilar and Atkinson (90) demonstrated a nested structure in
four samples of human cultural repertoires in North America and
New Guinea, which would occur if, as people disperse, traits are
sequentially added in or lost. Consistent with earlier cladistic
analyses of the branching pattern of chimpanzee profiles (91),
chimpanzees were found to display this pattern of nestedness
across African sites. However, orangutans did not, in agreement
with an earlier detailed orangutan study (27) and possibly reflecting
a greater preponderance of vertical, mother-to-offspring trans-
mission than horizontal transmission between communities.
Convergent Evolution. The Darwinian algorithm delivers some
similar organic evolutionary outcomes in different places, despite
different foundations. Cultural convergences of this kind appear
to occur at different layers of relatedness among apes. An ex-
ample within the same species is hand-clasp grooming, which has
emerged and spread in some chimpanzee communities in the
wild, but not in others (20, 21, 73), as well as in an African
sanctuary (35) and in groups in the United States (37). This
hand-clasp grooming is not simply an individual invention be-
cause within-group spread of the behavior has been documented,
indicating social transmission. Other convergences span different
ape genera, such as use of fly swats and leaf napkins by both
chimpanzees and orangutans; again, these behaviors are neither
species instincts nor individually learned, because they are ha-
bitual at some locations but are absent in the same species at
other locations. Finally there are convergences between apes and
other primates, e.g., the use of stones as hammers to break open
hard-cased food by distantly related long-tailed macaques (92)
and capuchins (3, 93).
Change of Function. Change of function is perhaps the category
for which we are most limited by lack of historical records. In
humans, historical records suggest that, just as morphology can
evolve to serve a new function (arms becoming wings, for ex-
ample), cultural elements may evolve new functions different
from their original one (10). A candidate in great apes is that in
chimpanzees leaf-clipping (noisily shredding leaves with the
teeth) is reported as a courtship bid in some communities but is
used for other functions, such as play, in others (47); such vari-
ations suggest that some of these alternatives may have evolved
from each other or from a common ancestral function. However,
it is possible that the lack of evidence in this category is explained
by the lack of evidence concerning limited cultural cumulation,
noted above.
Culture Extends Biology into New Realms of Evolution
The above discussion focuses on how cultural evolution may
match the template for genetically based Darwinian evolution,
but cultural transmission by social learning also extends the
scope of biological systems by incorporating additional dimen-
sions of inheritance and evolution. Some of these dimensions
have long been recognized in the literature concerning human
Whiten

cultural evolution, including the fact that, in addition to in-
tergenerational transmission shared with genetic inheritance,
cultural transmission can be horizontal (within or between groups,
and extending to nonkin) or oblique (with learning from nonrelatives
in the prior generation) (15). Above I have reviewed some of the
evidence for learning from parents, typically the mother, in apes
(38–41, 45). Horizontal and oblique transmissions are commonly
demonstrated in diffusion experiments (63–69) as well as in ob-
servational studies in the wild (42–44) and in sanctuaries (34).
Such transmission makes cultural learning a powerful adaptive
process, as does the fact that, because it hinges on neural rather
than genetic changes, it can act very much faster; some impor-
tant things can be learned observationally in a matter of minutes
(94), although the acquisition of complex skills may require a
more extended observational apprenticeship (32, 39).
Additionally, although adaptive information is inherited ge-
netically in a package at conception (even if activation is later
adaptively contingent on environmental inputs), the adaptive
information that feeds into social learning can be temporally
distributed in at least two major ways. First, cultural transmission
can be Lamarckian-like, with the adaptive features acquired
through one individual’s lifetime passed on to those who learn
from them. Second, a learner can build up complex skills, such as
some forms of ape tool use, progressively by repeated cycling
through a process of observe, practice, observe again, and practice
again. This pattern can be thought of as a spiral or helical process
of learning in which cycles of observation and practice allow the
learner to assimilate more in later observations than was possible
in the earlier, more naive stages (Fig. 3) (32, 39).
Social learning also may be selective in the assimilation of
information, variously referred to as “directed social learning”
(95), “biased transmission” (15), or “social-learning strategies”
(96), which can in principle shape adaptation and consequent
evolutionary change, with no clear counterparts in the gene-
based processes.
Evidence in great apes has been adduced for a number of the
potential learning rules these analyses highlight (97). Evidence
for a copy-the-majority rule, suggested by the apparent confor-
mity of chimpanzees in diffusion experiments (68), came in
further experiments showing that both children and chimpanzees
would copy the choices of three other conspecifics rather than
those of a single individual repeating the same act three times
Fig. 3. Helical curriculum model of skill development (after ref. 32). Over
repeated cycles of observation-of-expert and practice, the social learner is
able to assimilate more information from the expert and gradually improve
his/her skill level. See the text for more explanation.
Whiten
COLLOQUIUM
PAPER
(98). Orangutans did not show this selectivity (98), possibly
reflecting their less community-based social life. Evidence for
recognizing and copying more successful or productive options
came from further experiments with chimpanzees (99). It has
been suggested that preferentially copying individuals of high
rank could serve this function also, and two studies have shown
chimpanzees preferring to copy a high-ranked rather than a
lower-ranked individual (100, 101). Finally, a tendency to learn
from kin is shown by the studies of peering reviewed earlier,
which showed extensive learning from the mother during apes’
extended preweaning period (39, 41). After weaning, this be-
havior widened to include peering at the activities of others, a
plausibly adaptive shift from initially learning basic information
from parents and then later targeting others to learn more spe-
cific skills, a trend identified in both observational (102) and
experimental (103, 104) studies of human children. However, we
still have only limited understanding of when and why an ape
opts to learn socially (or does not): For example, what deter-
mines when immigrants will either conform to local norms (30)
or instead transmit their habitual skills to others (44)?
Interactions Between Genetic and Cultural Modes of
Inheritance and Evolution
At the broadest level, culture extends biology insofar as some
EVOLUTION
culturally transmitted behaviors are evolutionarily consequential,
i.e., they have implications for practitioners’ survival, reproduc-
tion, and ultimate inclusive fitness (as opposed to the repro-
ductive success of the cultural items themselves, discussed
earlier). Some cultural variants that appear relatively frivolous,
such as staring at one’s reflection in water in gorillas (23) or
applying an autoerotic tool in orangutans (22), may have less
ANTHROPOLOGY
evolutionary significance, but varied forms of tool use by orang-
utans and chimpanzees appear to be highly functional in gaining
access to rich resources such as insect prey, nut kernels, and
honey. Indeed, some of these behaviors appear to be vital for
chimpanzees to exploit niches that would otherwise exclude
them (87). Other culturally transmitted behaviors play functional
roles in grooming, social interactions, and sexual courtship.
Another sense in which culturally transmitted behaviors may
have been evolutionary important concerns their effects on or-
ganic evolution. Cetacean researchers have proposed that cul-
tural differentiation among whales has led to genetic differences
(7, 105). For example, killer whales display eco-types that spe-
cialize in hunting alternative prey such as seals or fish using very
different techniques, and different clans exhibit other behavioral
differences in their songs and migratory/resident patterns, de-
spite often being sympatric (6, 7, 105, 106). Such effects are
suggested to have driven other morphological and genetic dif-
ferentiation, ultimately leading to incipient speciation, because it
becomes difficult for a member of one culture to enter another
and successfully manage the different foraging and courtship
requirements of that culture. This causal pathway would be an
instance of “behavioral drive” (107–109), in which plasticity in
behavior allows a species to exploit or create a new niche, in this
case a culturally dependent one (e.g., hunting fish versus seal, a
cultural drive). This niche in turn may create selection pressures
acting on organic evolution, with effects such as the evolution of
more robust jaws in the seal-hunters (6). Parallel hypotheses
have been developed in the case of birdsong dialects driving
speciation (110, 111).
Dramatically different specialisms such as seen in killer whales
are not apparent among great apes, although the extent to which
similar processes are at work, e.g., in contrasts between nut-
cracking communities of chimpanzees and the nearest neighbors
that do not crack, would repay attention. However, one principal
effect of complex culture on organic evolution in apes has been
proposed concerning encephalization and the cognitive sophis-
tication it can provide: the cultural intelligence hypothesis.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7795
The Cultural Intelligence Hypothesis
In accord with a previously advanced social (or Machiavellian)
intelligence hypothesis, relative brain size in different primate
species was found to be predicted by the typical size of their
social group and the concomitant demands on social cognition
(59, 112). Great apes do not fit this pattern, showing high relative
and absolute brain sizes although gorillas and orangutans do not
live in large communities. However, because all appear to display
relatively complex cultures, the cultural intelligence hypothesis
suggests that this complexity has selected for encephalization,
either in a culture-first or an entwined culture–gene–brain co-
evolution scenario (112–115). One side of this proposition may
be glossed as “culture makes you smart,” as is self-evident in the
human case (9), insofar as present-day humans are smarter than
those a century earlier by virtue of the cumulative cultural
achievements from which they benefit. On a more modest scale,
the same is proposed for the cultural endowment of great apes.
The converse side of the proposal is that there is selection on the
socio-cognitive capacities necessary to assimilate and store all
the potential cultural repertoire available. In turn, it has been
suggested that there will be correlated selection on technical and
general intelligence, so as to benefit from the cultural input, as in
intelligent tool-use, for example (114). One recently offered test
of such ideas showed that when tested on the level playing field
of zoo contexts, Sumatran orangutans scored higher on general
intelligence than their Bornean cousins, as predicted by the
more elaborate cultural repertoires of the Sumatran pop-
ulations in the wild; moreover Sumatrans have brains that are
2–10% larger (116).
1. Whiten A (2017) A second inheritance system: The extension of biology through
culture. Interface Focus, in press.
2. Whiten A, Ayala FJ, Feldman MW, Laland KN (2017) The extension of biology
through culture. Proc Natl Acad Sci USA 114:7775–7781.
3. Fragaszy DM, et al. (2017) Synchronized practice helps bearded capuchin monkeys
learn to extend attention while learning a tradition. Proc Natl Acad Sci USA 114:
7798–7805.
4. Perry SE, Barrett BJ, Godoy I (2017) Older, sociable capuchins (Cebus capucinus) invent
more social behaviors, but younger monkeys innovate more in other contexts. Proc Natl
Acad Sci USA1147806–7813.
5. Hohmann G, Fruth B (2003) Culture in Bonobos? Between species and within species
variation in behavior. Curr Anthropol 44:563–571.
6. Whitehead H, Rendell L (2015) The Cultural Lives of Whales and Dolphins (Univ of
Chicago Press, Chicago).
7. Whitehead H (2017) Gene–culture coevolution in whales and dolphins. Proc Natl
Acad Sci USA 114:7814–7821.
8. Pagel M (2012) Wired For Culture: The Natural History of Human Communication
(Allen Lang, London).
9. Henrich J (2015) The Secret of Our Success: How Culture Is Driving Human Evolution,
Domesticating Our Species, and Making Us Smarter (Princeton Univ Press, Princeton, NJ).
10. Creanza N, Kolodny O, Feldman MW (2017) Cultural evolutionary theory: How cul-
ture evolves and why it matters. Proc Natl Acad Sci USA 114:7782–7789.
11. Legare CH (2017) Cumulative cultural learning: Development and diversity. Proc Natl
Acad Sci USA 114:7877–7883.
12. Darwin CD (1859) On the Origin in Species by Natural Selection (Murray, London).
13. Mesoudi A, Whiten A, Laland KN (2004) Perspective: Is human cultural evolution
Darwinian? Evidence reviewed from the perspective of the Origin of Species.
Evolution 58:1–11.
14. Whiten A (2005) The second inheritance system of chimpanzees and humans. Nature
437:52–55.
15. Boyd R, Richerson P (1985) Culture and the Evolutionary Process (Univ of Chicago
Press, Chicago).
16. Mesoudi A (2017) Pursuing Darwin’s curious parallel: Prospects for science of cultural
evolution. Proc Natl Acad Sci USA 114:7853–7860.
17. Goodall J (1986) The Chimpanzees of Gombe: Patterns of Behavior (Harvard Univ
Press, Boston).
18. McGrew WC (1992) Chimpanzee Material Culture: Implications for Human Evolution
(Cambridge Univ Press, Cambridge, UK).
19. Boesch C, Tomasello M (1998) Chimpanzee and human cultures. Curr Anthropol 39:
591–614.
20. Whiten A, et al. (1999) Cultures in chimpanzees. Nature 399:682–685.
21. Whiten A, et al. (2001) Charting cultural variation in chimpanzees. Behaviour 138:
1489–1525.
22. van Schaik CP, et al. (2003) Orangutan cultures and the evolution of material culture.
Science 299:102–105.
23. Robbins MM, et al. (2016) Behavioural variation in gorillas: Evidence of potential
cultural traits. PLoS One 11:e0160483.
7796 | www.pnas.org/cgi/doi/10.1073/pnas.1620733114
Summary and Conclusions
Research, particularly in the last two decades or so, has shown
that a second inheritance system of social learning is widespread
among animals, extending to all main classes of vertebrate and
also to insects (1, 2). Apes merit a special focus, insofar as they
have been subjected to an unmatched diversity and volume of
observational and experimental studies by multiple research
teams, whose work has revealed what appear to be the richest
nonhuman cultural repertoires identified to date (although some
cetaceans, e.g., killer whales, may show greater cultural differ-
entiation). This article has attempted to indicate the scope of ape
culture research and the key points of its discoveries, particularly
with respect to the theme of the present issue: how these cultural
phenomena may extend biology and its core evolutionary theory
in particular. I have argued that the evidence supports the con-
clusion that the nature of social learning and its consequences in
cultural transmission create new forms of evolution. These new
forms echo well the established core principles of organic evo-
lution but also go beyond them in a number of fundamental
ways, such as horizontal transmission and inheritance of acquired
characteristics, thereby extending the scope of evolutionary
processes we must now entertain. Moreover the primary genet-
ically based forms of evolution shaped and are also shaped by the
consequences of this second inheritance system in complex ways
we are only now starting to uncover.
ACKNOWLEDGMENTS. I thank Sarah Davis, Bill McGrew, Stephanie Musgrave,
Crickette Sanz, and Stuart Watson for comments on early versions of this paper.
24. Whiten A (2012) Social learning, traditions and culture. The Evolution of Primate
Societies, eds Mitani J, Call J, Kappeler P, Palombit R, Silk J (Univ of Chicago Press,
Chicago), pp 681–699.
25. van Schaik CP, et al. (2009) Orangutan cultures re-visited. Orangutans: Geographic
Variation in Behavioral Ecology and Conservation, eds Wich SA, Atmoko SSU,
Setia TM, van Schaik CP (Oxford Univ Press, Oxford, UK), pp 299–309.
26. Laland KN, Janik VM (2006) The animal cultures debate. Trends Ecol Evol 21:542–547.
27. Krützen M, Willems EP, van Schaik CP (2011) Culture and geographic variation in
orangutan behavior. Curr Biol 21:1808–1812.
28. Schöning C, Humle T, Möbius Y, McGrew WC (2008) The nature of culture: Tech-
nological variation in chimpanzee predation on army ants revisited. J Hum Evol 55:
48–59.
29. Möbius Y, Boesch C, Koops K, Matsuzawa T, Humle T (2008) Cultural differences in
army ant predation by West African chimpanzees? A comparative study of micro-
ecological variables. Anim Behav 76:37–45.
30. Luncz LV, Boesch C (2014) Tradition over trend: Neighboring chimpanzee commu-
nities maintain differences in cultural behavior despite frequent immigration of
adult females. Am J Primatol 76:649–657.
31. Marshall-Pescini S, Whiten A (2008) Social learning of nut-cracking behavior in East
African sanctuary-living chimpanzees (Pan troglodytes schweinfurthii). J Comp Psychol
122:186–194.
32. Whiten A (2015) Experimental studies illuminate the cultural transmission of per-
cussive technologies in Homo and Pan. Philos Trans R Soc Lond B Biol Sci 370:
20140359.
33. Santorelli CJ, et al. (2011) Traditions in spider monkeys are biased towards the social
domain. PLoS One 6:e16863.
34. van Leeuwen EJC, Cronin KA, Haun DB (2014) A group-specific arbitrary tradition in
chimpanzees (Pan troglodytes). Anim Cogn 17:1421–1425.
35. van Leeuwen EJC, Cronin KA, Haun DB, Mundry R, Bodamer MD (2012) Neigh-
bouring chimpanzee communities show different preferences in social grooming
behaviour. Proc Biol Sci 279:4362–4367.
36. Rawlings B, Davila-Ross M, Boysen ST (2014) Semi-wild chimpanzees open hard-
shelled fruits differently across communities. Anim Cogn 17:891–899.
37. Bonnie KE, de Waal FBM (2006) Affiliation promotes the transmission of a social
custom: Handclasp grooming among captive chimpanzees. Primates 47:27–34.
38. Lonsdorf EV, Eberly LE, Pusey AE (2004) Sex differences in learning in chimpanzees.
Nature 428:715–716.
39. Schuppli C, et al. (2016) Observational learning and socially induced practice of
routine skills in immature orangutans. Anim Behav 119:87–98.
40. Jaeggi AV, et al. (2010) Social learning of diet and foraging skills by wild immature
Bornean orangutans: Implications for culture. Am J Primatol 72:62–71.
41. Jaeggi AV, van Noordwijk MA, van Schaik CP (2008) Begging for information:
Mother-offspring food sharing among wild Bornean orangutans. Am J Primatol 70:
533–541.
42. Hobaiter C, Poisot T, Zuberbühler K, Hoppitt W, Gruber T (2014) Social network
analysis shows direct evidence for social transmission of tool use in wild chimpan-
zees. PLoS Biol 12:e1001960.
Whiten

43. Samuni L, Mundry R, Terkel J, Zuberbühler K, Hobaiter C (2014) Socially learned
habituation to human observers in wild chimpanzees. Anim Cogn 17:997–1005.
44. O’Malley RC, Wallauer W, Murray CM, Goodall J (2012) The appearance and spread
of ant fishing in the Kasekela chimpanzees of Gombe: A possible case of in-
tercommunity cultural transmission. Curr Anthropol 53:650–663.
45. Musgrave S, Morgan D, Lonsdorf E, Mundry R, Sanz C (2016) Tool transfers are a
form of teaching among chimpanzees. Sci Rep 6:34783.
46. Thornton A, McAuliffe K (2006) Teaching in wild meerkats. Science 313:227–229.
47. Boesch C (2012) Wild Cultures: A Comparison Between Chimpanzee and Human
Cultures (Cambridge Univ Press, Cambridge, UK).
48. Haraiwa-Hasegawa M (1990) A note on the ontogeny of feeding. The Chimpanzees
of the Mahale Mountains: Sexual and Life History Strategies, ed Nishida T (Univ of
Tokyo Press, Tokyo), pp 277–283.
49. Tomasello M, Call J, eds (1997) Primate Cognition (Oxford Univ Press, Oxford, UK).
50. Whiten A, Horner V, Litchfield CA, Marshall-Pescini S (2004) How do apes ape? Learn
Behav 32:36–52.
51. Galef BG, Whiten A (2017) The comparative psychology of social learning. APA
Handbook of Comparative Psychology, eds Call J, Burghardt G, Pepperberg I,
Snowdon C, Zentall T (APA, Washington).
52. Whiten A, Custance DM, Gomez J-C, Teixidor P, Bard KA (1996) Imitative learning of
artificial fruit processing in children (Homo sapiens) and chimpanzees (Pan troglo-
dytes). J Comp Psychol 110:3–14.
53. Stoinski TS, Wrate JL, Ure N, Whiten A (2001) Imitative learning by captive western
lowland gorillas (Gorilla gorilla gorilla) in a simulated food-processing task. J Comp
Psychol 115:272–281.
54. Stoinski TS, Whiten A (2003) Social learning by orangutans (Pongo abelii and Pongo
pygmaeus) in a simulated food-processing task. J Comp Psychol 117:272–282.
55. Whiten A, McGuigan N, Marshall-Pescini S, Hopper LM (2009) Emulation, imitation,
over-imitation and the scope of culture for child and chimpanzee. Philos Trans R Soc
Lond B Biol Sci 364:2417–2428.
56. Tennie C, Call J, Tomasello M (2009) Ratchetting up the ratchet: On the evolution of
cumulative culture. Phil Tran R Soc B 36:2405-2415.
57. Subiaul F (2016) What’s special about human imitation? A comparison with encul-
turated apes. Behav Sci (Basel) 6:13.
58. Whiten A (2017) Social learning and culture in child and chimpanzee. Annu Rev
Psychol 68:129–154.
59. Whiten A, van de Waal E (2016) Social learning, culture and the ‘socio-cultural brain’ of
human and non-human primates. Neurosci Biobehav Rev, 10.1016/j.neubiorev.2016.12.018.
60. Carvalho S, McGrew W (2010) The origins of the Oldowan: Why chimpanzees are still
good models for technological evolution in Africa. Stone Tools and Fossil Bones, ed
Domínguez-Rodrigo M (Cambridge Univ Press, Cambridge, UK), pp 201–221.
61. Boesch C, Marchesi P, Marchesi N, Fruth B, Joulian F (1994) Is nutcracking in wild
chimpanzees a cultural behaviour? J Hum Evol 26:325–338.
62. McGrew WC, Ham RM, White LJT, Tutin CEG, Fernandez M (1997) Why don’t
chimpanzees in Gabon crack nuts? Int J Primatol 18:353–374.
63. Whiten A, Mesoudi A (2008) Review. Establishing an experimental science of culture:
Animal social diffusion experiments. Philos Trans R Soc Lond B Biol Sci 363:3477–3488.
64. Whiten A, Caldwell CA, Mesoudi A (2016) Cultural diffusion in humans and other
animals. Curr Op Psychol 8:15–21.
65. Horner V, Whiten A, Flynn E, de Waal FBM (2006) Faithful replication of foraging
techniques along cultural transmission chains by chimpanzees and children. Proc
Natl Acad Sci USA 103:13878–13883.
66. Dindo M, Stoinski T, Whiten A (2011) Observational learning in orangutan cultural
transmission chains. Biol Lett 7:181–183.
67. Whiten A, Horner V, de Waal FBM (2005) Conformity to cultural norms of tool use in
chimpanzees. Nature 437:737–740.
68. Bonnie KE, Horner V, Whiten A, de Waal FBM (2007) Spread of arbitrary conventions
among chimpanzees: A controlled experiment. Proc Biol Sci 274:367–372.
69. Whiten A, et al. (2007) Transmission of multiple traditions within and between
chimpanzee groups. Curr Biol 17:1038–1043.
70. Pruetz JD, Bertolani P (2007) Savanna chimpanzees, Pan troglodytes verus, hunt with
tools. Curr Biol 17:412–417.
71. Pruetz JD, et al. (2015) New evidence on the tool-assisted hunting exhibited by
chimpanzees (Pan troglodytes verus) in a savannah habitat at Fongoli, Sénégal. R Soc
Open Sci 2:140507.
72. Sanz C, Call J, Morgan D (2009) Design complexity in termite-fishing tools of chim-
panzees (Pan troglodytes). Biol Lett 5:293–296.
73. Nakamura M, Uehara S (2004) Proximate factors of different types of grooming
hand-clasp in Mahale chimpanzees: Implications for chimpanzee social customs. Curr
Anthropol 45:108–114.
74. Inskipp T (2005) Chimpanzee (Pan troglodytes). World Atlas of Great Apes and Their
Conservation, eds Caldecott J, Miles L (Univ of California Press, Berkeley, CA), pp
53–81.
75. Ferriss S (2005) Western gorilla (Gorilla gorilla). World Atlas of Great Apes and Their
Conservation, eds Caldecott J, Miles L (Univ of California Press, Berkeley, CA), pp
105–127.
76. McConkey K (2005) Bornean orangutan (Pongo pygmaeus). World Atlas of Great
Apes and Their Conservation, eds Caldecott J, Miles L (Univ pof California Press,
Berkeley, CA), pp 161–183.
77. Nishida T, Wrangham RW, Goodall J, Uehara S (1983) Local differences in plant
feeding habits of chimpanzees between the Mahale Mountains and Gombe Na-
tional Park, Tanzania. J Hum Evol 12:467–480.
Whiten
COLLOQUIUM
PAPER
78. Bastian ML, Zweifel N, Vogel ER, Wich SA, van Schaik CP (2010) Diet traditions in wild
orangutans. Am J Phys Anthropol 143:175–187.
79. Mercader J, et al. (2007) 4,300-year-old chimpanzee sites and the origins of per-
cussive stone technology. Proc Natl Acad Sci USA 104:3043–3048.
80. Dawkins R (1976) The Selfish Gene (Oxford Univ Press, Oxford. UK).
81. Darwin C (1871) The Descent of Man and Selection in Relation to Sex (Murray,
London).
82. Maynard-Smith J, Szathmary E (1995) The Major Transitions in Evolution (Freeman,
Oxford, UK).
83. Dennett D (1995) Darwin’s Dangerous Idea (Penguin, London).
84. Ridley M (2015) The Evolution of Everything (Fourth Estate, London).
85. Stout D (2011) Stone toolmaking and the evolution of human culture and cognition.
Philos Trans R Soc Lond B Biol Sci 366:1050–1059.
86. Menzel EW, Jr, Davenport RK, Rogers CM (1972) Protocultural aspects of chimpan-
zees’ responsiveness to novel objects. Folia Primatol (Basel) 17:161–170.
87. Yamakoshi G (1998) Dietary responses to fruit scarcity of wild chimpanzees at Bos-
sou, Guinea: Possible implications for ecological importance of tool use. Am J Phys
Anthropol 106:283–295.
88. Gray RD, Watts J (2017) Cultural macroevolution matters. Proc Natl Acad Sci USA
114:7846–7852.
89. Langergraber KE, et al. (2011) Genetic and ‘cultural’ similarity in wild chimpanzees.
Proc Biol Sci 278:408–416.
90. Kamilar JM, Atkinson QD (2014) Cultural assemblages show nested structure in
humans and chimpanzees but not orangutans. Proc Natl Acad Sci USA 111:111–115.
91. Lycett SJ, Collard M, McGrew WC (2009) Cladistic analyses of behavioural variation in
wild Pan troglodytes: Exploring the chimpanzee culture hypothesis. J Hum Evol 57:
337–349.
92. Gumert MD, Malaivijitnond S (2012) Marine prey processed with stone tools by
Burmese long-tailed macaques (Macaca fascicularis aurea) in intertidal habitats. Am
EVOLUTION
J Phys Anthropol 149:447–457.
93. Fragaszy D, Izar P, Visalberghi E, Ottoni EB, de Oliveira MG (2004) Wild capuchin
monkeys (Cebus libidinosus) use anvils and stone pounding tools. Am J Primatol 64:
359–366.
94. Whiten A (1998) Imitation of the sequential structure of actions by chimpanzees
(Pan troglodytes). J Comp Psychol 112:270–281.
95. Coussi-Korbel S, Fragaszy DM (1995) On the relation between social dynamics and
social learning. Anim Behav 50:1441–1450.
96. Laland KN (2004) Social learning strategies. Learn Behav 32:4–14.
ANTHROPOLOGY
97. Price EE, Wood LA, Whiten A (2016) Adaptive cultural transmission biases in children
and nonhuman primates. Infant Behav Dev, 10.1016/j.infbeh.2016.11.003.
98. Haun DB, Rekers Y, Tomasello M (2012) Majority-biased transmission in chimpanzees
and human children, but not orangutans. Curr Biol 22:727–731.
99. Vale GL, Flynn EG, Lambeth SP, Schapiro SJ, Kendal RL (2014) Public information use
in chimpanzees (Pan troglodytes) and children (Homo sapiens). J Comp Psychol 128:
215–223.
100. Horner V, Proctor D, Bonnie KE, Whiten A, de Waal FBM (2010) Prestige affects
cultural learning in chimpanzees. PLoS One 5:e10625.
101. Kendal R, et al. (2015) Chimpanzees copy dominant and knowledgeable individuals:
Implications for cultural diversity. Evol Hum Behav 36:65–72.
102. Henrich J, Broesch J (2011) On the nature of cultural transmission networks: Evi-
dence from Fijian villages for adaptive learning biases. Philos Trans R Soc Lond B Biol
Sci 366:1139–1148.
103. Harris PL, Corriveau KH (2011) Young children’s selective trust in informants. Philos
Trans R Soc Lond B Biol Sci 366:1179–1187.
104. Lucas AJ, et al. (2016) The development of selective copying: Children’s learning
from an expert versus their mother. Child Dev, 10.1111/cdev.12711.
105. Carroll EL, et al. (2015) Cultural traditions across a migratory network shape the
genetic structure of southern right whales around Australia and New Zealand. Sci
Rep 5:16182.
106. Foote AD, et al. (2016) Genome-culture coevolution promotes rapid divergence of
killer whale ecotypes. Nat Commun 7:11693.
107. Wyles JS, Kunkel JG, Wilson AC (1983) Birds, behavior, and anatomical evolution.
Proc Natl Acad Sci USA 80:4394–4397.
108. Wilson AC (1985) The molecular basis of evolution. Sci Am 253:164–173.
109. Bateson PPG (2004) The active role of behavior in evolution. Biol Philos 19:283–298.
110. Grant BR, Grant PR (2002) Simulating secondary contact in allopatric speciation: An
empirical test of premating isolation. Biol J Linn Soc Lond 76:545–556.
111. Grant PR, Grant BR (2002) Adaptive radiation of Darwin’s finches. Am Sci 90:130–139.
112. Whiten A, van Schaik CP (2007) The evolution of animal ‘cultures’ and social in-
telligence. Philos Trans R Soc Lond B Biol Sci 362:603–620.
113. van Schaik CP, Burkart JM (2011) Social learning and evolution: The cultural in-
telligence hypothesis. Philos Trans R Soc Lond B Biol Sci 366:1008–1016.
114. Burkart JM, Schubiger MN, van Schaik CP (2016) The evolution of general in-
telligence. Behav Brain Sci1–65.
115. Street SE, Navarette AF, Reader SM, Land KN (2017) Coevolution of cultural in-
telligence, extended life history, sociality, and brain size in primates. Proc Natl Acad
Sci USA 114:7908–7914.
116. Forss SIF, Willems E, Call J, van Schaik CP (2016) Cognitive differences between
orang-utan species: A test of the cultural intelligence hypothesis. Sci Rep 6:30516.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7797
Synchronized practice helps bearded capuchin
monkeys learn to extend attention while
learning a tradition
Dorothy M. Fragaszya,1, Yonat Eshchara,b, Elisabetta Visalberghic, Briseida Resended, Kellie Laitya, and Patrícia Izard
a
Psychology Department, University of Georgia, Athens, GA 30602; bDavidson Institute of Science Education, Weizmann Institute of Science, Rehovot, Israel
7610001; cInstitute of Cognitive Sciences and Technologies, National Research Council, Italy, Rome, Italy 00197; and dDepartment of Experimental
Psychology, University of São Paulo, Butanta, 05508030, SP, Brazil
Edited by Andrew Whiten, University of St. Andrews, St. Andrews, United Kingdom, and accepted by Editorial Board Member Andrew G. Clark May 29, 2017
(received for review January 23, 2017)
Culture extends biology in that the setting of development shapes
the traditions that individuals learn, and over time, traditions
evolve as occasional variations are learned by others. In humans,
interactions with others impact the development of cognitive
processes, such as sustained attention, that shape how individuals
learn as well as what they learn. Thus, learning itself is impacted
by culture. Here, we explore how social partners might shape the
development of psychological processes impacting learning a
tradition. We studied bearded capuchin monkeys learning a
traditional tool-using skill, cracking nuts using stone hammers.
Young monkeys practice components of cracking nuts with stones
for years before achieving proficiency. We examined the time
course of young monkeys’ activity with nuts before, during, and
following others’ cracking nuts. Results demonstrate that the on-
set of others’ cracking nuts immediately prompts young monkeys
to start handling and percussing nuts, and they continue these
activities while others are cracking. When others stop cracking
nuts, young monkeys sustain the uncommon actions of percussing
and striking nuts for shorter periods than the more common ac-
tions of handling nuts. We conclude that nut-cracking by adults
can promote the development of sustained attention for the crit-
ical but less common actions that young monkeys must practice to
learn this traditional skill. This work suggests that in nonhuman
species, as in humans, socially specified settings of development
impact learning processes as well as learning outcomes. Nonhumans,
like humans, may be culturally variable learners.
primates | attention | development | learning | tool use
T raditions are behaviors shared among members of a group
that are reliably learned by new individuals, in part, with
social support (1). The concept of tradition highlights behavioral
consistency over time and across individuals. Here, we empha-
size the origin of traditions in learning. Learned behavior is one
manifestation of developmental plasticity, and generation of
novel behavior is another. Novel behavioral variants may be
learned by others, thus modifying existing traditions or leading to
new ones (2). In this way, developmental plasticity enables her-
itable modification of behavior through learning, with the result
that culture may extend biology (3, 4). Traditions appear to be
widespread in the animal kingdom (5–7), leading theorists to
propose that cultural evolution, rather than being restricted to
recent human history, has general evolutionary significance (8,
9). Apes (6, 10) and monkeys provide several examples of tra-
ditions. Among monkeys, for example, vervet monkeys learn
their mothers’ food-processing techniques (11) and white-faced
capuchin monkeys learn odd social games, such as poking a
finger in another monkey’s eye, that become established in-
termittently in groups (2, 12).
Given the essential role of learning in traditions, cognitive
processes associated with learning that show developmental plas-
ticity themselves, should be incorporated into our understanding
7798–7805 | PNAS | July 25, 2017 | vol. 114 | no. 30
of how culture can extend biology (13–15). However, cognitive
processes associated with learning are not yet well integrated into
theories of cultural evolution and niche construction (16). Here,
we illustrate how growing up in a group with a prevailing tradition
of using tools in foraging could affect cognitive development in
young monkeys in ways that support their learning this traditional
skill. This work opens a bridge between the learning sciences and
the field of cultural evolution.
Social Experience Influences the Development of Attention
In humans, social experiences are implicated in the development
of attention, memory, and individual learning styles (16–18).
Cultural influences on long-term memory development include,
for example, the development of particular ways of chunking and
rehearsing information to be remembered, such as the con-
struction of “memory palaces” used by the ancient Greeks and
the cultures that succeeded them (19) and the use of written lists
and notes in the present day. Culture also impacts the develop-
ment of working memory, which incorporates structures and
processes used for the temporary storage and organization of
information about events recently heard or seen or about activ-
ities recently performed (20, 21). Working memory is dependent
on sustained attention, and therefore sensitive to attentional
disruption. It is intimately related to motor processes, and lim-
ited in span to perhaps three to four “chunks” of information at
one time. In humans, working memory typically lasts from a few
seconds to more than 1 min depending on emotional salience;
whether the information to be remembered is about events, ac-
tions, or space; and other factors (20).
It is thought that humans share with other primates general at-
tentional and working memory capacity, although humans develop
skills, many of which are cultural, to control attention, and thus to
extend working memory (20, 22). Consider the deep attention
humans develop through meditative practice and in the course of
reading and writing, all of which are explicitly cultural skills. Cul-
tural variation in learned attention and skilled actions in humans
suggests we should seek similar variations in nonhuman primates
and other taxa that, like humans, live in socially constructed
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: D.M.F., Y.E., E.V., B.R., and P.I. designed research; D.M.F., Y.E., and K.L.
performed research; Y.E., and K.L. analyzed data; and D.M.F., E.V., and P.I. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. A.W. is a guest editor invited by the Editorial
Board.
1
To whom correspondence should be addressed. Email: doree@uga.edu.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1621071114/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1621071114

environments. To the extent that the development of sustained
attention and working memory is biased by the actions of social
partners, the process and outcome of social learning are likely to
vary across groups and across species. Other features of socially
constructed niches, such as the creation of artifacts, can also pro-
mote sustained attention by providing opportunities and reminders
to practice particular actions, and thus add another source of
cultural or taxonomic variation in socially biased learning (23).
Small shifts in attentional processes and perceptual biases can shift
the trajectory of learning with far-reaching consequences, as has
been proposed for humans learning language (24).
Developing Sustained Attention for Manual Actions with
Objects
Sustained attention and working memory develop throughout
childhood in humans, in concert with neural development and
social experience (25, 26). Young children extend the duration of
sustained attention to an object while an older person shows
attention to that object, suggesting how social interactions can
support the development of sustained attention in the particular
context of handling objects (27). The visual salience to humans
of others’ hand movements is evident by the second year of life,
when toddlers shift their visual attention toward another person
handling an object from predominantly toward the face to pre-
dominantly toward the hands (28).
There are sound reasons to propose that nonhuman primates
are also biased to attend to manual actions. Some species of
nonhuman primates in captivity have been shown to attend to
humans’ manual actions with objects (29–31); they are at least as
likely to attend to manual activity of familiar conspecifics, if not
more so. Manual activity is salient to all primates: Using the hands
with visual guidance to collect food and bring it to the mouth is a
primitive characteristic of primates, and all primates use their
hands to explore objects and surfaces, as well as to contact others
during social interactions, such as grooming and play (32). This
fundamental feature of primate behavior is associated with a host
of neuroanatomical, perceptuomotor, and cognitive attributes,
including strong visual and proprioceptive salience to movements
of their own hands and of the hands of others (31).
Through the observer’s bias to attend to others’ manual ac-
tions with objects, others’ actions can support the development
of attentional control by young monkeys during particular
manual activities. In this way, social partners can support young
nonhuman primates that otherwise normally experience brief
sustained attention to others and to their own activities, learning
manual skills that require longer sustained attention. Certainly
using tools qualifies as challenging enough to benefit from sup-
port of this kind. Acknowledging the cognitive dimension of the
constructed niche in nonhuman taxa will strengthen cultural
evolutionary theory. Social influence on the development of
sustained attention is an appropriate early target for research in
this direction. Even if the perceptual biases in primates favoring
attention to actions of others are small, they could nevertheless
powerfully affect learning trajectories, particularly when magni-
fied by shifts in attention and memory (24).
We hypothesize that nonhuman primates and humans share
strong susceptibility to tuning (i.e., extending, strengthening) at-
tention and memory about manual actions and about objects via
interest in others’ manual activity. However, nonhuman primates
face a challenge in sustaining attention to actions with their hands
that humans typically do not, or do not face to the same extent,
which could be particularly important when learning a skill in-
volving handling objects. Nonhuman primates’ attention to their
own activity is typically disrupted every few seconds to scan the
surroundings briefly (surveying the surroundings in this way is
termed “vigilance” in the animal behavior literature) (33, 34).
Vigilance, which functions to inform the perceiver about preda-
tors, conspecifics, and other relevant dynamic features in the
Fragaszy et al.
COLLOQUIUM
PAPER
environment, interferes with sustained attention to one’s own or
another’s activity.
Evaluating Attention to Others’ Actions
Duration of gaze is a common measure of sustained attention to
others’ actions in experimental settings, but this measure is not
useful when the activity of interest is auditory or when the ex-
perimenter cannot see where the subject is looking. Temporally or
spatially synchronized activity between two individuals, commonly
measured in studies with nonhuman animals under the general
heading of “social influence,” is an alternative measure of sus-
tained attention in such cases. We can also measure the decline of
synchronized activity once the “demonstrator” stops performing
the activity. We propose that the onset of synchronized activity is
an indirect measure of the onset of sustained attention and the
decline of synchronized activity (i.e., a return to baseline rates of
performing a particular behavior) is an indirect measure of the
disruption of attention and the subsequent loss of the contents of
working memory (in other words, forgetting).
Theories of social learning predict that an individual is aided in
learning by attending to the actions or products of actions of an-
other, but are silent about how quickly attention is drawn to the
other’s actions or products, how long the influence lasts, or how the
influence declines over time (35). Not surprisingly, contemporary
studies of social learning have not addressed temporal properties
of social influence. To our knowledge, Hoppitt et al. (36) provide
the only published report to quantify the temporal decline of social
influence on behavior in a learning context. In a field experiment,
wild meerkats (Suricata suricatta) were presented with baited boxes
that could be opened in two different ways. Some meerkats
(demonstrators) were trained to open the boxes using a particular
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
method. Subsequently, the boxes were presented to all members of
the group, and the naive meerkats’ interactions with the boxes, as
well as when they watched other meerkats opening the boxes, were
recorded. The researchers found that individuals were more likely
to interact with a box immediately after observing another meerkat
interacting with it; the half-life of the effect was 20 s. Young
meerkats spend much time with adults in the period when they are
learning to forage on the hidden and dangerous scorpions that
these animals capture and eat (37). Adults’ influence on young
individuals in this period is thought to be necessary for meerkats to
master their challenging foraging style (38).
We consider the relation in young monkeys between social
influence on activity and attentional processes associated with
learning. The activity in question relates to using stone hammers
to crack nuts, seeds, or other encased foods, a technical tradition
in several populations of wild bearded capuchin monkeys
(Sapajus libidinosus) (39–42) (Fig. 1). Young capuchin monkeys,
like young meerkats, master finding and feeding on hidden and
sometimes noxious prey, and like meerkats, they are interested in
and affected by others’ actions with objects (43). Thus, they are
good candidates for studies of the temporal dynamics of social
influence on behavior with objects and on the cognitive processes
associated with learning traditional tool-using skills.
Temporal Dynamics of Social Influence as a Window on
Sustained Attention
Our objective was to examine temporal dynamics of social in-
fluence on young monkeys’ behavior in a situation in which the
young monkeys were practicing component actions of a tradi-
tional manual skill. Temporal dynamics of young monkeys’ be-
havioral coordination with others in this context reflect sustained
attention to their own and others’ actions. The skill in question is
cracking palm nuts using stone hammers, a precursor to the
situation facing humans knapping stone (44). Social support in
the form of instruction and demonstration aid in the acquisition
of knapping, but these actions are not sufficient by themselves
for people to learn to knap stone (45). Providing repeated
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7799

Fig. 1. An adult bearded capuchin monkey has cracked a palm nut using a
stone hammer on a log anvil and is removing and eating pieces of the kernel.
A young monkey that cannot crack a nut itself watches closely. Image used
with permission from Luca Antonio Marino, Roma Tre University (Rome, Italy).
occasions for practice and self-discovery of movement solutions
is a crucial dimension of social support for learning this complex
perceptuomotor skill and human traditional skills more generally
(13, 46).
We measured the temporal rise and fall of social influence on
wild young bearded capuchin monkeys (S. libidinosus) in the
company of adults that were cracking palm nuts by striking them
repetitively with stone hammers. These monkeys start to interact
with nuts and stones in the first year of life. They handle nuts,
percuss nuts directly against a hard surface (hereafter, percuss),
and strike nuts or nut shells with stones (hereafter, strike) for
several years before they are able to crack palm nuts themselves
(47, 48). Thus, young monkeys exhibit remarkable persistence in
a foraging activity that they cannot perform effectively. We know
that others’ cracking nuts is partially responsible for the young
monkeys’ persistence. In one recent study, while other group
members were cracking and eating nuts, monkeys 6 y of age or
younger were threefold more likely to be near an anvil, qua-
drupled their rate of interaction with nuts, and doubled their rate
of percussing and striking compared with times when no monkey
in the group was cracking nuts. Interactions with objects other
than nuts showed the opposite pattern (48).
The data for this report are taken from monkeys 6 y of age or
younger belonging to a wild, habituated group of bearded ca-
puchins observed in five periods over 2 y (Table 1). Members of
this group of monkeys routinely crack nuts using stone hammers
at many anvils scattered across their home range (49–51). In this
study, one observer continuously recorded the focal young
monkey’s behavior. A second observer concurrently recorded at
intervals of 1 min the distance from each neighbor (within a 10-m
zone) to the focal monkey, the identity and behavior of each
neighbor, and the occurrence of cracking nuts (i.e., striking a nut
with a stone, producing a sharp cracking noise) by any monkey in
the group. The method allowed us to analyze the focal monkey’s
behavior with nuts and stones and its presence near anvils in
relation to the start, continuation, and end of cracking by other
members of the group. Data collected while the group was in a
frequently visited area with abundant anvils, hammers, and
cracked shells, as detailed below, indicated that the temporal
pattern of others’ influence on young monkeys’ activity with nuts
and stones and their presence near anvils was associated spe-
cifically with the others’ activity with nuts (i.e., synchronization
was not a byproduct of traveling in a cohesive group). Results are
reported for n = 16 monkeys, unless otherwise noted.
Results
Manipulation of Nuts. Young monkeys manipulated nuts at the
highest rate when others were cracking nuts (i.e., striking nuts
with a stone) (median = 8.3 acts per 10 min) and at the lowest
rate (median = 0.9 acts per 10 min) during periods when no
others were cracking nuts, and had not been cracking nuts for at
least 8 min. The difference between these rates was significant
(estimate = 1.98, P < 0.0001). The onset of the effect of others’
cracking nuts was quick: Compared with the minute before
others began to crack, the median probability that a young
monkey would manipulate a nut doubled in the first minute after
others began to crack, and remained doubled or more for at least
5 min when others continued to crack (Fig. 2). Movie S1 provides
a video-clip of a young monkey handling a nut while and after
another monkey is cracking a nut. During the 7 min after the
others stopped cracking, the rate of manipulation of nuts de-
clined exponentially (in At = A0 * e−βt; estimates: A = 9.96, P <
0.0001; β = 0.325, P = 0.0013; Fig. 3), where e is the base of the
natural logarithm, β is the rate by which the dependent variable
declines with time, t is the time since cracking in the group
stopped, and At (the dependent variable) is the rate or percentage

Table 1. Subjects’ date of birth, sex, and body mass at each sampling time point
Name Date of birth (mm/dd/yyyy) Sex Body mass in 2011, kg Body mass in 2012, kg Body mass in 2013, kg

Donzela 01/13/2013 F — — 0.4

Patricia 01/11/2013 F — — 0.5
Titia 01/03/2013 F — — 0.7
Divina 11/07/2012 F — — —
Cachaça 03/15/2012 M — 0.4 1.1
Thais 02/01/2011 F — 1.1 1.3
Presente 02/15/2011* M — 1.0 1.5
Chani 12/15/2010 F — 1.0 1.2
Coco 07/14/2009 M 1.1 1.4 1.7
Paçoca 01/01/2009 F 1.2 1.3 1.6
Pamonha 01/01/2009 F 1.2 1.4 1.6
Doree 11/09/2007 F 1.4 1.6 1.8
Pati 11/02/2007 M 1.7 2.1 2.5
Cangaceiro 09/20/2007 M 1.8 2.1 2.4
Catu 02/05/2007 M 1.8 2.1 2.5
Tomate 12/01/2006 M 1.8 2.0 2.3

F, female; M, male.
*Estimate.

7800 | www.pnas.org/cgi/doi/10.1073/pnas.1621071114 Fragaszy et al.


Fig. 2. Probability that a young monkey (n = 11) manipulated a nut in the
1 min before the onset of striking a nut performed by another monkey (No
cracking) and during each of the 5 min following the onset of striking a nut
by another monkey. The boxes display the median and interquartile range,
and whiskers indicate minimum and maximum values. The solid line within
the box depicts the median. Circles indicate values of outliers.
of time. The output from the model is A0 (the strength of the
effect on the dependent variable) and β. The half-life of the effect
was 2.1 min. The rate of nut manipulation during the first to fifth
minutes after other monkeys stopped cracking nuts was signifi-
cantly higher than the rate during periods 8 min or longer after
others stopped cracking nuts (i.e., the baseline rate) (P <
0.0001 for minutes 1–4 after other monkeys stopped cracking nuts,
P = 0.0146 for minute 5). The rate of manipulation of nuts during
the sixth and seventh minutes after others stopped cracking nuts
and the rate during periods 8 min or longer after others stopped
cracking nuts did not differ significantly.
The temporal pattern for young monkeys’ manipulation of
other objects besides nuts was the opposite of their actions with
nuts. Young monkeys manipulated other objects at the lowest rate
when others were cracking nuts (median = 12.8 per 10 min) and at
the highest rate during periods 8 min or longer after others
stopped cracking nuts (median = 16.7 per 10 min). The difference
between these rates was significant (estimate = 1.29, P < 0.0001).
From the time when others were cracking nuts through 7 min after
they stopped, juveniles’ rate of manipulation of other objects in-
creased exponentially (in At = A0 * e−βt; estimates: A = 14.2, P <
0.0001; β = −0.03, P = 0.015), although the rate of manipulation
of other objects in each of the first 7 min after others stopped
cracking nuts did not differ from the rate in baseline periods (i.e.,
8 min or longer after other monkeys stopped cracking nuts).
Percussing Nuts and Striking Nuts with a Stone. We looked at the
rates of two actions that are related to cracking palm nuts:
percussing a nut on a hard surface (percuss) and striking a nut
placed on a hard surface with a stone (strike). The young mon-
keys’ rate of percussing was highest while others were cracking
nuts (median = 1.3 per 10 min) and lowest during periods 8 min
or longer after others stopped cracking nuts (median = 0.2 per
10 min). These rates were significantly different (estimate = 6.3,
P < 0.0001). As was the case for manipulation of nuts, the onset
of the effect of others’ cracking on young monkeys’ percussion
was nearly immediate. Compared with the minute before others
began to crack nuts, the probability that a young monkey would
percuss a nut more than doubled in the first minute after others
began to crack nuts (from <0.04 to >0.10) and remained at
that level for at least 5 min while others were cracking (Fig. 4).
From the time that others stopped cracking nuts, the immature
Fragaszy et al.
COLLOQUIUM
PAPER
monkeys’ rate of percussion declined exponentially for the next
7 min (in At = A0 * e−βt; estimates: A = 8.16, P = 0.026; β =1.12,
P = 0.014; Fig. 5). The half-life of the effect was 0.6 min, less
than a third as long as the half-life for manipulation of
nuts (2.1 min).
Four monkeys (all <1 y of age) did not strike a nut with a stone
during focal observations, but 12 monkeys (all >1 y of age) did.
For the latter monkeys, the rate of striking was highest when
others were cracking nuts (median = 1.4 per 10 min) and lowest
in periods 8 min or longer after others stopped cracking (me-
dian = 0.3 per 10 min; Fig. 6). The difference between these rates
was significant (estimate = 4.3, P < 0.0001). In the minute after
others stopped cracking, the median rate of striking by the young
monkeys declined to zero, and for all minutes, the rate of young
monkeys’ striking was not significantly different from periods
8 min or longer after others stopped cracking. The data did not
fit an exponential model of decline.
Time Spent Near an Anvil. The percentage of time young monkeys
spent near an anvil was highest when others were cracking nuts
(median = 13.3%) and lowest during periods 8 min or longer
after others stopped cracking (median = 2.6%) (Fig. 7). The
difference between these percentages was significant (estimate =
6.06, P < 0.0001). When others were cracking until 7 min after all
others had stopped cracking, the percentage of time young
monkeys spent near an anvil declined exponentially (in At = A0 *
e−βt; estimates: A = 11.94, P < 0.0001; β = 0.25, P < 0.0001). The
half-life of the effect was 2.8 min.
Discussion
PSYCHOLOGICAL AND
Linking Learning Processes to a Tool-Using Tradition in Monkeys.
COGNITIVE SCIENCES
Culture in a behavioral sense is present when, aided by social
context, individuals consistently learn behaviors exhibited by
others in their community (i.e., they have traditions). Culture can
extend biology in an evolutionary sense when the traditions in
question persist across generations, and when they have selective
consequences. In parallel with natural selection, occasional
spontaneous behavioral variants (arising from developmental
plasticity) that afford some advantage and that are learned by
Fig. 3. Rate per 10 min of manipulation of nuts by young monkeys (n = 16)
when one or more other monkeys struck nuts (Cracking present) during each
of the 7 min after others stopped striking a nut and in periods 8 min or
longer after others stopped striking a nut (8 min or over). The exponential
curve generated by model fitting is overlaid. The boxes display the median
and interquartile range, and whiskers indicate minimum and maximum
values. The solid line within the box depicts the median. Circles indicate
values of outliers. The half-life of the decline occurred at 2.1 min.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7801

Fig. 4. Probability that a young monkey (n = 11) percussed a nut 1 min
before the onset of striking a nut performed by another monkey (No
cracking) and during each of the 5 min following the onset of striking a nut
by another monkey. The boxes display the median and interquartile range,
and whiskers indicate minimum and maximum values. The solid line within
the box depicts the median. Circles indicate values of outliers.
others can become established as new traditions. Learning a
tradition is, by definition, dependent on the social context in
which learning takes place, but it is equally dependent on the
learning processes of the individual. Here, we draw attention to
the power of social partners to influence not just the content of
learning a tradition (i.e., how or when to perform a particular
behavior) but also how learning itself takes place. Working from
findings with humans that sustained attention is developmentally
plastic and susceptible to social influence, we examined behavior
of young monkeys, seeking evidence that social influences shape
attention in young monkeys. Specifically, we examined temporal
dynamics of social influence on young monkeys’ activity with the
materials relevant to the tool-using tradition in their wild group
(cracking nuts with stone hammers) to index their sustained at-
tention to this task. Young monkeys were more likely to be near
an anvil, and to manipulate, percuss, and strike nuts once others
began to strike nuts to crack them (which we call cracking). The
higher probability of these actions in young monkeys persisted
through 5 min while others cracked nuts (5 min was the longest
period for which we had sufficient data for these analyses). They
continued to manipulate nuts at higher rates and were more
likely to be present near an anvil for minutes after others stopped
cracking nuts, with a half-life of more than 2 min for each vari-
able. These findings indicate substantial lingering effects of so-
cial influence on young monkeys’ interest in nuts and anvils and
general exploratory activity directed to nuts. In contrast, these
young monkeys decreased manipulation of other objects while
other monkeys cracked nuts.
Key to our argument that attention (and its related process,
working memory) can be enhanced by participating in socially
supported practice, others’ cracking was a strong facilitator of
young monkeys’ practicing percussion and striking nuts, but only
when others were cracking. The young monkeys percussed a nut
or struck a nut with a stone much less frequently than they
manipulated nuts; moreover, following the end of others’
cracking nuts, young monkeys reduced percussion and striking
quickly (a half-life of about 37 s for percussion and a median of
zero after 1 min for striking).
Building Sustained Attention for the Least Likely Actions. These
findings illustrate how the frequency and temporal duration of
others’ actions can influence how young individuals learn a
7802 | www.pnas.org/cgi/doi/10.1073/pnas.1621071114
challenging perceptuomotor skill. When others act in frequent
bouts, and when these bouts are long-enduring, they support
more frequent performance by young individuals of those actions
for which attention is least well maintained and, subsequently,
working memory is the most fragile. We provisionally identify
this process as social scaffolding for learning attentional skills
that support learning the least familiar component(s) of an ac-
tivity. The socially supported practice of sustained attention in a
given context can powerfully support development of longer
periods of sustained attention that can be marshalled during
other contexts. For example, human infants practice sustained
attention while maintaining joint attention with a caregiver, and
these common and frequent social interactions influence the
development of sustained attention more generally (27). In the
case of young bearded capuchin monkeys, we suggest that
extending sustained attention for percussive actions with stones
and nuts is one outcome of repeated prompting to perform these
actions arising from others striking nuts while cracking them. We
further suggest that the development of longer sustained atten-
tion to their own percussive activity supports the acquisition of
nut-cracking using stone hammers, which is a signature tradition
of tool use in some groups of bearded capuchin monkeys.
Socially Tuned Attention and Learning Traditions in Primates and
Other Orders. This line of reasoning leads us to predict that
tool-using traditions in nonhuman animals include actions that
are infrequent outside of that activity (i.e., traditions are not
composed solely of common actions), and to the related pre-
dictions that (i) the components of tool use traditions that are
infrequent in species-typical activity outside of the traditional
activity will be performed frequently by proficient tool users, and
(ii) frequent performance by adults of uncommon actions will
support practice of these particular actions by young individuals.
Taxonomic variation in social learning may follow from
species-typical variations in attention in combination with tem-
poral dynamics of action. We hypothesize two features of at-
tention and memory distinguish primates from other orders:
(i) Primates are more likely than other orders to have stronger
sustained attention and working memory for actions with objects
compared with other kinds of actions, and (ii) primates have
stronger interest in others’ actions with objects than do other
orders [although there may be other taxa with strong interest in
Fig. 5. Rate of direct percussion of nuts by young monkeys (n = 16) when
one or more other monkeys struck nuts (Cracking present) during the 7 min
after other monkeys stopped striking nuts and in periods 8 min or longer
after others stopped striking a nut (8 min or over). The boxes display the
median and interquartile range; whiskers indicate minimum and maximum
values. The solid line depicts the median. Circles indicate values of outliers.
The exponential curve generated by model fitting is overlaid. The half-life of
the decline occurred at 0.62 min.
Fragaszy et al.

Fig. 6. Rate of striking a nut with a stone by young monkeys (n = 16) when
one or more monkeys in the group struck nuts (Cracking present) during the
7 min after other monkeys stopped striking and in periods 8 min or longer
after others stopped striking a nut (8 min or over). The boxes display the
median and interquartile range, and whiskers indicate minimum and maxi-
mum values. The solid line depicts the median. The exponential curve gen-
erated by model fitting is overlaid. Circles indicate values of outliers.
others’ actions with objects, such as New Caledonian crows,
which also use tools in foraging (52)]. By comparison, some ce-
tacean species (e.g., Tursiops and Orcinus orca) are more likely
than other mammals to have stronger sustained attention to
auditory events and more immediate response to others’ actions,
in keeping with the collaborative nature of their feeding activities
and the efficiency of sound transmission in water (7). In general,
the social dimension of the developmental niche is likely to
contribute to tuning attention, and thus to bias learning, to favor
those skills that are commonly practiced by the young individ-
ual’s social companions.
We hope this work prompts others to develop these ideas and
to test them with data from diverse species. Conceptual and
empirical work along these lines is needed to integrate de-
velopmental and psychological understanding of behavioral
variation into theories of cultural evolution, niche construction,
and evolutionary biology.
Concluding Remarks
Culture potentially extends biology insofar as the setting of de-
velopment supports individuals’ learning traditions, and occasion-
ally learning behavioral variants of these traditions arising in other
individuals that become established as new traditions. Behavioral
traditions are learned in social settings, and the attentional and
memorial processes that underlie that learning are themselves
shaped by social partners. To date, our attention on socially aided
learning, traditions, and cultures in nonhuman species has focused
on the form and function of traditional behaviors (e.g., foraging
skills, social interactional patterns). We argue that we need to in-
clude social influences on the learning process itself in the scope of
cultural inquiry, as cross-cultural educational psychologists have
argued (18). Improving our understanding of the psychological
processes supporting socially biased learning, and thus the tradi-
tions that animals acquire, must be part of advancing theory in
cultural evolution. Working memory and attention are one set of
linked cognitive processes available for study with respect to how
and when learning occurs, and how the social setting of develop-
ment influences learning processes.
We have hypothetically linked temporal dynamics of social
influence to sustained attention and working memory. These
cognitive processes are fundamental to learning, including
learning a traditional skill, cracking nuts with a stone hammer in
the case of the young monkeys that we studied. We suggested
that repeated experiences of performing challenging parts of
the action cycle relating to cracking nuts could lead to ex-
tended sustained attention and working memory for these actions.
Fragaszy et al.
COLLOQUIUM
PAPER
Iterative practice in socially supportive contexts such as we de-
scribe for bearded capuchin monkeys characterizes the learning
context for culturally acquired instrumental skills in humans and
other species (10). Suggestive examples of unlikely actions tightly
synchronized with similar actions by another and of more gradual
decline of more common actions after others have stopped the
activity have been reported in apes. For example, a young chim-
panzee, closely watching another chimpanzee cracking a nut with
a stone, synchronously raised and lowered its arm (a “striking”
action) while the other was striking (53), similar to what we ob-
served with young capuchin monkeys striking nuts in the presence
of others cracking nuts. Young orangutans follow peering at
others at close range feeding or making nests with higher rates of
exploratory actions with the food items handled by others or with
nest materials over the next few minutes (54), exhibiting a tem-
poral persistence of social influence on the order of what we re-
port for capuchin monkeys manipulating nuts.
We further suggested that primates exhibit particular atten-
tional biases toward actions with objects. This bias increases the
likelihood that they learn about such actions from being with
others in concert with practicing themselves. Attentional biases
are likely to vary across taxa in accord with, for example, dominant
modalities of communication and styles of foraging. Differences in
where attentional biases impact learning processes could support
taxonomic variation in the content of what the individual learns
with others (e.g., songs in some species, food preferences and
foraging techniques in other species). The eventual outcome of
taxonomic variation in socially biased learning is that traditions
develop in varied functional domains, and thus that culture is
likely to extend biology in various taxa in different ways.
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
Methods
Study Site. This study was conducted at Fazenda Boa Vista and adjacent lands
in the southern Parnaíba Basin (9°39′S, 45°25′W) in Piauí, Brazil. Palms are
abundant in the area, and many produce fruit at ground level. Two species
of palm nuts in particular were commonly cracked by the monkeys in this
study: tucum (Astrocaryum campestre) and piassava (Orbygnia spp.). A
tucum nut is, on average, 46 mm in length, weighs 15.5 g, and has a 4.1-mm-
thick shell, with a peak-force-at-failure of 5.6 kN. An average piassava nut is
61.3 mm long, weighs 50.6 g, and has a thicker and more resistant shell than
a tucum nut: 6 mm with a peak-force-at-failure of 11.5 kN (55).
The naturally occurring stones used by the monkeys to crack nuts weigh,
on average, 1.1 kg (range: 250 g to 2.5 kg) (56). They are quartz, quartzite,
Fig. 7. Percentage of time spent within an arm’s length of an anvil by
young monkeys (n = 16) when one or more other monkeys struck nuts
(Cracking present) during the 7 min after other monkeys stopped striking
and in periods 8 min or longer after others stopped striking a nut (8 min or
over). The boxes display the median and interquartile range, and whiskers
indicate minimum and maximum values. The solid line depicts the median.
Circles indicate values of outliers. The exponential curve generated by model
fitting is overlaid. The half-life of the decline occurred at 2.8 min.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7803
siltstone, or harder sandstone. The monkeys cracked nuts on naturally oc-
curring anvils (boulders, exposed stone, or horizontal logs with a flat, or
nearly flat, horizontal surface). Anvils are abundant in the region.
Subjects. At the beginning of the study, there were 11 immature monkeys in
the group, aged from 3 mo to 4.5 y (Table 1). Five more infants were born
during the study. At the beginning of the study, none of the subjects could
crack open a whole nut of the more resistant palm species. The two oldest
juveniles and, to some extent, two others mastered this skill during the
study. Apart from the study subjects, the group included three adult males
and five adult females. All but one female habitually cracked palm nuts. The
body mass of each monkey was obtained as monkeys stood individually on a
digital scale to drink from a bowl of water (57).
Data Collection. Data were collected in five discrete collection periods, each
6 to 9 wk, during three dry seasons (May–July) in 2011, 2012, and 2013 and
two wet seasons (January–March) in 2012 and 2013. Observations were
collected using two-person teams. One observer followed a focal subject to
obtain a continuous record of its activities, including manipulation of nuts
and other objects, and locations, specifically if the subject was near an anvil.
Concurrently, the other member of the team recorded, as an instantaneous
observation every minute, the identity, location, and activity of other
monkeys within 10 m of the focal monkey. All observations lasted 20 min, or
until the focal subject went out of view and could not be followed. Obser-
vations lasting <5 min were discarded.
Observers first learned to identify all members of the group and were
subsequently trained in the method with one of the authors (Y.E.). Reliability
for focal observations was calculated using Generalized Sequential Querier
(GSEQ) software (www2.gsu.edu/∼psyrab/gseq/index.html). We used the
time unit method, which compares the codes inserted by two observers and
defines as a match any instant in which both observers used the same code
within a time window of 5 s. For each observer and trainer pair, time unit
kappa was at or above 0.7, which is considered highly reliable (58). Reliability
for instantaneous observations of other monkeys near the focal monkey was
tested separately for each aspect (identity, distance to the focal monkey,
activity, and location) until agreement (sum agreement/agreement plus
disagreement) was over 80% for each variable for 20 consecutive samples.
The protocol was reviewed and approved by the Institutional Animal Care
and Use Committee of the University of Georgia. The study adheres to the code
of best practices for field primatology set by the International Primatological
Society and all applicable Brazilian regulations for the conduct of field research.
Data Analysis. For each subject in each collection period, we collected be-
tween 19 and 53 observations, which lasted cumulatively between 5.3 and
27.1 h. Observations were collated by subject for each season. Ten subjects
appeared in all five collection periods. Data were collected as the monkeys
traveled throughout their home range. The observations were exported from
The Observer to GSEQ software to extract the frequency of different events
(e.g., manipulation of nuts) at times when others in the group cracked (struck)
nuts and at times when they did not.
With respect to change in the young monkeys’ activity following cessation
of others’ cracking, we used general mixed linear models to evaluate the
differences in activity under different conditions and exponential models to
evaluate the temporal pattern of the effects. SAS/STAT14.2 software was
used for the analyses. We examined the rate of manipulation of nuts, ma-
nipulation of objects other than nuts, specific actions with nuts, and time
spent near an anvil. The data are summarized in Dataset S1.
Our independent variables were the presence of nut-cracking activity (which
involved striking nuts with stones) in the group (yes/no) and the time that had
elapsed since this activity stopped (e.g., 0–1 min after the activity stopped,
1–2 min after the activity stopped, 2–3 min after the activity stopped). The
dependent variables were (i) proportion of time the subject spent within an
arm’s length of an anvil, (ii) manipulation of nuts, (iii) manipulation of other
objects, (iv) rate of percussing a nut directly on a surface, and (v) rate of
striking a nut with a stone. “Total time” is defined as all seconds of observa-
tion under a specific condition of the independent variable. For example, the
total time of “3 min after activity stopped” includes all observations from
120 to 180 s after all monkeys in the group stopped cracking nuts. Rates for
1. Fragaszy DM, Perry S (2003) Towards a biology of traditions. Traditions in Nonhuman
Animals: Models and Evidence, eds Fragaszy D, Perry S (Cambridge Univ Press,
Cambridge, UK), pp 1–32.
2. Perry SE, Barrett BJ, Godoy I (2017) Older, sociable capuchins (Cebus capucinus)
invent more social behaviors, but younger monkeys innovate more in other con-
texts. Proc Natl Acad Sci USA 114:7806–7813.
7804 | www.pnas.org/cgi/doi/10.1073/pnas.1621071114
dependent variables in this period were calculated as the number of events
divided by total observation time. The proportion of time near an anvil was
calculated as the number of seconds spent there divided by total observation
time. In the models, we treated the variables as count variables and used total
time as an offset. For variables that did not distribute normally (tested with
the Shapiro–Wilk test), the Poisson distribution was used. Subject identification
was used as a random factor. Randomization of residuals was used to com-
pensate for overdispersion.
We used the exponential model At = A0 * e-βt to describe the dynamics of
the dependent variables with time, where t is the time since cracking in the
group stopped and At (the dependent variable) is the rate or percentage of
time. The output from the model is A0 (the strength of the effect on the
dependent variable) and β, from which the half-life can be calculated as
ln(2)/β. For each variable, we examined the goodness of fit of our data to
this exponential model and determined the estimates for A0 and β, as well as
the half-life of the effect. We used the values of the independent variables
when monkeys were striking nuts in each of the 7 min after this activity
stopped; at times, we also used the values of the independent variables
when monkeys were striking nuts 8 min or longer after this activity stopped
or during observation that did not include striking at all. Data from all ob-
servation periods were used in analyses, except that data from the two wet
seasons on young monkeys’ direct percussion of the nut and striking the nut
with a stone were not used in analyses because these actions occurred very
rarely in the data from these seasons.
We examined the effect of the onset of others’ cracking by tabulating
young monkeys’ actions with nuts in the minute before others began
cracking and in the 5 min following the onset of others’ cracking. The data
are presented in Dataset S2. We present these data descriptively. We used
11 monkeys’ data for these tallies: 2-y-old, 3-y-old, and 4-y-old monkeys,
which our previous analyses had indicated were affected more strongly by
others’ striking than younger or older immature monkeys.
At each data collection period, one-quarter to one-half of observations
were collected while the monkeys visited an area (∼30-m diameter) of their
home range containing several large boulders, fallen logs, and areas of
exposed stone that the monkeys habitually used as anvils to crack nuts. We
use this area as our outdoor laboratory. Several hammer stones were pre-
sent in this area, typically left by the monkeys on or near the anvils. Many
other anvils with hammer stones were present within 200 m in the sur-
rounding area. The monkeys were sometimes provisioned in the outdoor
laboratory with nuts as part of ongoing experiments (59–62).
When in the outdoor laboratory, young monkeys had ample opportunity
to spend time near anvils handling stones and nut shells independent of other
monkeys’ activity. The influence of others in the group cracking nuts in their
vicinity on the rate of manipulating nuts and on proximity to anvils by our
subjects in the outdoor laboratory is approximately the same (manipulation:
P = 0.0406, estimate = 2.33; proximity to anvils: P < 0.0001, estimate = 8.4) as
in the full sample collected over the entire home range (estimate = 1.98 for
manipulation, estimate = 6.06 for proximity to anvil). Young monkeys
approached anvils and handled nuts most often while adults were cracking
nuts, although anvils were available, and nut shells and hammer stones were
equally present and available, when others were not cracking. They showed
the opposite pattern for manipulating other objects. We conclude that the
fine temporal influence of others’ nut-cracking on young monkeys’ activity
with nuts and presence near anvils reported here is not a byproduct of
synchronized travel of a cohesive group.
ACKNOWLEDGMENTS. We thank the assistants who helped collect the data
and Marino Gomes de Oliveira and the Oliveira family for their help and
permission to work on their land. We thank Marcus W. Feldman, Andrew
Whiten, Kevin N. Laland, and Francisco J. Ayala for the opportunity to participate
in the Sackler Colloquium “The Extension of Biology Through Culture.” We
thank the statistical consulting service at the University of Georgia (UGA) and
the UGA Dean’s Award for the funding of this service. This research was funded
by the National Geographic Society, UGA, Coordenadoria de Aperfeiçoamento
de Pessoal de Nível Superior, São Paulo Research Foundation (Grant 08/55684-3),
and Brazilian National Council for Scientific and Technological Development
(CNPq) (Contract 029088). Permission was granted for the research by Instituto
Brasileiro do Meio Ambiente e dos Recursos Renováveis through Permit 28689
and by CNPq/Ministério da Ciência e Tecnologia Permit 0002547/2011.
3. West-Eberhard MJ (2003) Developmental Plasticity and Evolution (Oxford Univ Press,
Oxford).
4. Laland KN, et al. (2015) The extended evolutionary synthesis: Its structure, assump-
tions and predictions. Proc Biol Sci 282:20151019.
5. Fragaszy DM, Perry S (2003) The Biology of Traditions: Models and Evidence (Cambridge
Univ Press, Cambridge, UK).
Fragaszy et al.

6. McGrew W (2004) The Cultured Chimpanzee: Reflections on Cultural Primatology
(Cambridge Univ Press, Cambridge, UK).
7. Whitehead H, Rendell L (2015) The Cultural Lives of Dolphins and Whales (Univ of
Chicago Press, Chicago).
8. Mesoudi A (2011) Cultural Evolution (Univ of Chicago Press, Chicago).
9. Mesoudi A (2017) Pursuing Darwin’s curious parallel: Prospects for a science of cul-
tural evolution. Proc Natl Acad Sci USA 114:7853–7860.
10. Whiten A (2017) Culture extends the scope of evolutionary biology in the great apes.
Proc Natl Acad Sci USA 114:7790–7797.
11. van de Waal E, Bshary R, Whiten A (2014) Field experiments show wild vervet monkey
infants acquire maternal food-processing techniques. Anim Behav 90:41–45.
12. Perry S (2011) Social traditions and social learning in capuchin monkeys (Cebus). Philos
Trans R Soc Lond B Biol Sci 366:988–996.
13. Stout D, Hecht EE (2017) Evolutionary neuroscience of cumulative culture. Proc Natl
Acad Sci USA 114:7861–7868.
14. Lotem A, Halpern JY, Edelman S, Kolodny O (2017) The evolution of cognitive mechanisms
in response to cultural innovations. Proc Natl Acad Sci USA 114:7915–7922.
15. Jablonka E, Lamb M (2014) Evolution in Four Dimensions (MIT Press, Cambridge, MA),
2nd Ed.
16. Flynn EG, Laland KN, Kendal RL, Kendal JR (2013) Target article with commentaries:
Developmental niche construction. Dev Sci 16:296–313.
17. Rogoff B (1991) Apprenticeship in Thinking (Oxford Univ Press, Oxford).
18. Li J (2012) Cultural Foundations of Learning: East and West (Cambridge Univ Press,
Cambridge, UK).
19. Yates FA (1966) The Art of Memory (Chicago Univ Press, Chicago).
20. Carruthers P (2013) Evolution of working memory. Proc Natl Acad Sci USA 110:
10371–10378.
21. Atkinson RC, Schiffrin RM (1968) Human memory: A proposed system and its control
processes. Psychol Learn Motiv 2:89–195.
22. O’Connell RG, et al. (2008) Self-alert training: Volitional modulation of autonomic
arousal improves sustained attention. Neuropsychologia 46:1379–1390.
23. Fragaszy DM, et al. (2013) The fourth dimension of tool use: Temporally enduring
artefacts aid primates learning to use tools. Philos Trans R Soc Lond B Biol Sci 368:
20120410.
24. Thompson B, Kirby S, Smith K (2016) Culture shapes the evolution of cognition. Proc
Natl Acad Sci USA 113:4530–4535.
25. Reynolds GD, Romano AC (2016) The development of attention systems and working
memory in infancy. Front Syst Neurosci 10:15.
26. Pasternak T, Greenlee MW (2005) Working memory in primate sensory systems. Nat
Rev Neurosci 6:97–107.
27. Yu C, Smith LB (2016) The social origins of sustained attention in one-year-old human
infants. Curr Biol 26:1235–1240.
28. Fausey CM, Jayaraman S, Smith LB (2016) From faces to hands: Changing visual input
in the first two years. Cognition 152:101–107.
29. Myowa-Yamakoshi M, Matsuzawa T (2000) Imitation of intentional manipulatory
actions in chimpanzees (Pan troglodytes). J Comp Psychol 114:381–391.
30. Fragaszy DM, Deputte B, Cooper EJ, Colbert-White EN, Hémery C (2011) When and
how well can human-socialized capuchins match actions demonstrated by a familiar
human? Am J Primatol 73:643–654.
31. Rizzolatti G, Sinigaglia C (2008) Mirrors in the Brain: How Our Minds Share Actions,
Emotions and Experience (Oxford Univ Press, Oxford).
32. Fragaszy DM, Crast J (2016) Functions of the hand in primates. The Evolution of the
Primate Hand, Perspectives from Anatomical, Developmental, Functional, and
Paleontological Evidence, eds Kivell T, Schmitt D, Lemelin P (Springer, New York),
Vol 2, pp 313–344.
33. Davis RT (1974) Monkeys as perceivers. Primate Behavior: Developments in Field and
Laboratory Research, ed Rosenblum L (Academic, New York), Vol 3.
34. Treves A (2000) Theory and method in studies of vigilance and aggregation. Anim
Behav 60:711–722.
35. Heyes C (2012) What’s social about social learning? J Comp Psychol 126:193–202.
36. Hoppitt W, Samson J, Laland KN, Thornton A (2012) Identification of learning
mechanisms in a wild meerkat population. PLoS One 7:e42044.
37. Thornton A, Hodge S (2008) The development of foraging microhabitat preferences
in meerkats. Behav Ecol 20:103–110.
Fragaszy et al.
COLLOQUIUM
PAPER
38. Thornton A, McAuliffe K (2006) Teaching in wild meerkats. Science 313:227–229.
39. Visalberghi E, Fragaszy DM (2013) The EthoCebus project. Stone tool use by wild
capuchin monkeys. Tool Use in Animals. Cognition and Ecology, eds Sanz C, Call J,
Boesch C (Cambridge Univ Press, Cambridge, UK), pp 203–222.
40. Ferreira RG, et al. (2009) On the occurrence of Cebus flavius (Schreber 1774) in the
Caatinga, and the use of semi-arid environments by Cebus species in the Brazilian
state of Rio Grande do Norte. Primates 50:357–362.
41. Canale GR, Guidorizzi CE, Kierulff MC, Gatto CA (2009) First record of tool use by wild
populations of the yellow-breasted capuchin monkey (Cebus xanthosternos) and new
records for the bearded capuchin (Cebus libidinosus). Am J Primatol 71:366–372.
42. Falotico T, Ottoni E (2016) The manifold use of pounding stone tools by wild capuchin
monkeys of Serra da Capivara National Park, Brazil. Behav 153:421–442.
43. Fragaszy DM, Visalberghi E, Fedigan L (2004) The Complete Capuchin (Cambridge
Univ Press, Cambridge, UK).
44. Mangalam M, Izar P, Visalberghi E, Fragaszy DM (2016) Task-specific temporal or-
ganization of percussive movements in wild bearded capuchin monkeys. Anim Behav
114:129–137.
45. Stout D, Khreisheh N (2015) Skill learning and human brain evolution: An experi-
mental approach. Camb Archaeol J 25:867–875.
46. Whiten A (2015) Experimental studies illuminate the cultural transmission of per-
cussive technologies in Homo and Pan. Philos Trans R Soc Lond B Biol Sci 370:
20140359.
47. de Resende BD, Ottoni EB, Fragaszy DM (2008) Ontogeny of manipulative behavior
and nut-cracking in young tufted capuchin monkeys (Cebus apella): A perception-
action perspective. Dev Sci 11:828–840.
48. Eshchar Y, Izar P, Visalberghi E, Resende B, Fragaszy D (2016) When and where to
practice: Social influences on the development of nut-cracking in bearded capuchins
(Sapajus libidinosus). Anim Cogn 19:605–618.
49. Spagnoletti N, Visalberghi E, Ottoni E, Izar P, Fragaszy D (2011) Stone tool use by
adult wild bearded capuchin monkeys (Cebus libidinosus). Frequency, efficiency and
tool selectivity. J Hum Evol 61:97–107.
50. Verderane MP, Izar P, Visalberghi E, Fragaszy DM (2013) Socioecology of wild bearded
capuchin monkeys (Sapajus libidinosus): An analysis of social relationships among
female primates that use tools in feeding. Behaviour 150:659–689.
51. Howard AM, Nibbelink N, Bernardes S, Fragaszy DM, Madden M (2015) Remote
sensing and habitat mapping for (Sapajus libidinosus): Landscapes for the use of
stone tools. J Appl Remote Sens 9:096020–096020.
52. Holzhaider J, Hunt G, Gray D (2010) The development of pandanus tool manufacture
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
in wild New Caledonian crows. Behaviour 147:553–586.
53. Fuhrmann D, Ravignani A, Marshall-Pescini S, Whiten A (2014) Synchrony and motor
mimicking in chimpanzee observational learning. Sci Rep 4:5283.
54. Schuppli C, et al. (2016) Observational social learning and socially induced practice of
routine skills in immature wild orangutans. Anim Behav 119:87–98.
55. Visalberghi E, et al. (2008) Physical properties of palm fruits processed with tools by
wild bearded capuchins (Cebus libidinosus). Am J Primatol 70:884–891.
56. Visalberghi E, et al. (2007) Characteristics of hammer stones and anvils used by wild
bearded capuchin monkeys (Cebus libidinosus) to crack open palm nuts. Am J Phys
Anthropol 132:426–444.
57. Fragaszy DM, et al. (2016) Body mass in wild bearded capuchins, (Sapajus libidinosus):
Ontogeny and sexual dimorphism. Am J Primatol 78:389–484.
58. Bakeman R, Deckner DF, Quera V (2005) Analysis of behavioral streams. Handbook of
Research Methods in Developmental Science, ed Teti DM (Wiley, New York), pp
394–420.
59. Massaro L, Liu Q, Visalberghi E, Fragaszy D (2012) Wild bearded capuchin (Sapajus
libidinosus) select hammer tools on the basis of both stone mass and distance from
the anvil. Anim Cogn 15:1065–1074.
60. Fragaszy DM, Liu Q, Wright BW, Allen A, Brown CW (2013) Wild bearded capuchin
monkeys (Sapajus libidinosus) strategically place nuts in a stable position during nut-
cracking. PLoS One 8:E56182.
61. Hanna JB, et al. (2015) Kinetics of bipedal locomotion during load carrying in capu-
chin monkeys. J Hum Evol 85:149–156.
62. Liu Q, Fragaszy DM, Visalberghi E (2016) Wild capuchin monkeys spontaneously ad-
just actions when using hammer stones of different mass to crack nuts of different
resistance. Am J Phys Anthropol 161:53–61, 1.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7805
Older, sociable capuchins (Cebus capucinus) invent
more social behaviors, but younger monkeys
innovate more in other contexts
Susan E. Perrya,b,1, Brendan J. Barrettc,d, and Irene Godoye
a
Department of Anthropology, University of California, Los Angeles, CA 90095-1553; bBehavior, Evolution and Culture Program, University of California, Los
Angeles, CA 90095-1553; cAnimal Behavior Graduate Group, University of California, Davis, CA 95616-8522; dDepartment of Anthropology, University of
California, Davis, CA 95616-8522; and eBehavioural Science Institute, Radboud University Nijmegen, 6500 HE Nijmegen, The Netherlands
Edited by Andrew Whiten, University of St. Andrews, St. Andrews, United Kingdom, and accepted by Editorial Board Member Andrew G. Clark May 16, 2017
(received for review January 18, 2017)
An important extension to our understanding of evolutionary
processes has been the discovery of the roles that individual and
social learning play in creating recurring phenotypes on which
selection can act. Cultural change occurs chiefly through invention
of new behavioral variants combined with social transmission of
the novel behaviors to new practitioners. Therefore, understand-
ing what makes some individuals more likely to innovate and/or
transmit new behaviors is critical for creating realistic models of
culture change. The difficulty in identifying what behaviors qualify
as new in wild animal populations has inhibited researchers from
understanding the characteristics of behavioral innovations and
innovators. Here, we present the findings of a long-term, system-
atic study of innovation (10 y, 10 groups, and 234 individuals) in
wild capuchin monkeys (Cebus capucinus) in Lomas Barbudal,
Costa Rica. Our methodology explicitly seeks novel behaviors, re-
quiring their absence during the first 5 y of the study to qualify as
novel in the second 5 y of the study. Only about 20% of 187 inno-
vations identified were retained in innovators’ individual behavioral
repertoires, and 22% were subsequently seen in other group mem-
bers. Older, more social monkeys were more likely to invent new
forms of social interaction, whereas younger monkeys were more
likely to innovate in other behavioral domains (foraging, investiga-
tive, and self-directed behaviors). Sex and rank had little effect on
innovative tendencies. Relative to apes, capuchins devote more of
their innovations repertoire to investigative behaviors and social
bonding behaviors and less to foraging and comfort behaviors.
innovation | Cebus capucinus | cultural evolution | phenotypic plasticity |
learning
B ehavioral innovation has long been a topic of interest for
researchers dedicated to studying the evolution of culture,
because it is a driver of cultural change (1, 2). The types of be-
havioral traditions that are of greatest interest to evolutionary
modelers are those starting with an innovation that then spreads
via social learning. Understanding the characteristics of (i) be-
havioral innovations (which are roughly analogous to genetic
mutations) and (ii) the individuals who invent these behaviors is
critical to understanding cultural evolution and its relationship to
genetic evolution. Innovation is also of interest to evolutionary
biologists who study the role that learning plays in macroevolu-
tion, because it is a type of phenotypic plasticity that can affect
the direction of natural selection (3). Ability to innovate can
enhance reproductive success (for example, by enabling indi-
viduals to exploit new resources) (4–6). Innovation can generate
the Baldwin effect, in which learned traits create recurring
phenotypes that select for morphological adaptations, eventually
leading to speciation (3, 7). Innovation is also of interest as a
correlate of intelligence more generally, and the ability to solve
novel cognitive problems presented by experimenters can be
positively associated with mating success [e.g., bowerbirds (8)]
and parenting success [e.g., great tits (9)] in wild populations.
7806–7813 | PNAS | July 25, 2017 | vol. 114 | no. 30
Despite the obvious theoretical importance of innovation as a
key element in cultural evolution, there are relatively few studies
of this topic, particularly in wild populations, because of meth-
odological difficulties in stimulating innovation experimentally
or detecting innovations in observational studies (10). This
paucity of information is partly because of the difficulty in cre-
ating operational definitions of innovation that can produce
meaningful datasets for comparative analysis. Here, we loosely
adopt the definitions by Reader and Laland (11) of innovation
(the process) as “a process that results in new or modified
learned behaviour and that introduces novel behavioural variants
into a population’s repertoire” and innovation (the product) as
“a new or modified learned behavior not previously found in the
population” (ref. 11, p.14). Following the definitions of Ramsey
et al. (12) and van Schaik et al. (13), we emphasize that inno-
vations are not part of the innate repertoire and do not arise
predictably in all population members at certain points in the life
history; also, they do not predictably emerge in all population
members in response to particular social or ecological condi-
tions. The definition by Ramsey et al. (12) and van Schaik et al.
(13) differs from the definition by Reader and Laland (11), be-
cause it focuses on the individual rather than the population [i.e.,
Ramsey et al. (12) argue that multiple individuals within the
same population could independently create the same behavior].
We take a compromise position, counting a behavior as an in-
novation if this is the first time that the behavior has been seen in
a particular social group during the putative innovator’s lifetime.
Operational definitions of innovation and invention differ greatly
across fields. Some define innovations as inventions that have
subsequently spread throughout the population via social learning
(14). We use the definition most commonly and currently used in
the animal behavior literature, in which transmission of a new
behavior is not a necessary part of our definition of innovation.
We assume because of the extreme xenophobia exhibited by
white-faced capuchins (15) that widespread social transmission
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: S.E.P. designed research; S.E.P., B.J.B., and I.G. performed research;
S.E.P., B.J.B., and I.G. analyzed data; and S.E.P., B.J.B., and I.G. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. A.W. is a guest editor invited by the Editorial
Board.
Data deposition: The R-code and data used in the statistical analysis reported in this paper
are available at https://github.com/bjbarrett/cebusinnovation2017.
1
To whom correspondence should be addressed. Email: sperry@anthro.ucla.edu.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1620739114/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1620739114

of innovations is inhibited by insufficient exposure to members of
other social groups.
The thorniest challenges in the rapidly growing field of animal
innovation research arise in devising methods for quantifying
innovation rates and determining the characteristics of innova-
tors. Most research on animal innovation thus far has involved
either (i) analysis of anecdotes drawn from published literature
on topics other than innovation (16–18) or (ii) experiments in
either the field or the laboratory, in which experimenters present
individuals with a novel problem to solve (8, 9, 19–22). The most
prominent body of innovation research using observational data
has focused on orangutans (Pongo) using developmental data
from rehabilitants (23) or “geographic contrasts methods” akin
to those used to diagnose probable social traditions (24, 25) to
infer innovation rates by evaluating the patterning of within- and
between-group behavioral variation in short-term studies of both
captive and field populations (12, 13, 26). Few field studies of in-
novation are longitudinal or look at the properties of individuals
that make them more, or less, likely to innovate within their lifetime.
There are advantages and disadvantages to all of these
methods, and none of them seemed entirely suitable for our
goals, which are to (i) document the kinds of behaviors present
in the innovation repertoire in wild white-faced capuchins, (ii) in-
vestigate differences in rates of innovation across behavioral
domains, (iii) determine what features of individuals (rank,
sex, age, and sociality) predict propensity to innovate, and
(iv) determine whether new behaviors become established com-
ponents of individual or group behavioral repertoires. Our ap-
proach differs from previous approaches to documentation
of behavioral innovation, in that it was implemented into the
core data collection protocol over a 10-y period, for which there
was dense behavioral sampling of multiple social groups in the
same ecological context. This approach provides the advantage
of enabling researchers to detect probable innovations in any
behavioral domain and have sufficient time depth to know who
the probable innovators are (details are in Methods).
There are several biologically relevant behavioral contexts, or
domains, that are difficult to study experimentally. Studies have
largely looked at the diffusion of novel foraging tasks for good
reasons. Experimentally seeded innovations or problem-solving
tasks provide controlled contexts, where both the latency of in-
dividuals innovating a solution and the social diffusion of inno-
vations may be studied. Innovating solutions to novel tasks is also
of obvious adaptive value. However, many innovations that are
ecologically or socially relevant are difficult to study experi-
mentally. Some animals have innovative social interaction, i.e.,
“social games,” which may serve as bond testing rituals and can
be socially transmitted (27). Some wild animals display repetitive
self-directed “quirks,” perhaps to self-soothe, akin to the pro-
posed function of some stereotyped coping behaviors in captive
and wild animals (28). Other behavioral innovations have no
apparent immediate biological or ecological function.
Systematic observational study of innovations across a wide
range of behavioral domains permits us to explore whether indi-
vidual propensity to innovate is generalized or whether individuals
will be differentially prone to innovate in different behavioral do-
mains according to their ecology and life history. For example, it has
been suggested that “necessity is the mother of invention” and
hence, that individuals who are young, low-ranking, and/or socially
peripheral will be more prone to inventing new foraging strategies
(29); this hypothesis has yielded mixed results in past literature
reviews (6). In general, there are few strong theoretical expectations
about how age, sex, rank, and sociality affect innovation rates; we
need natural history and observational studies to help guide theory.
Capuchin monkeys (genera Cebus and Sapajus) are expected
to have unusually high innovation rates for myriad reasons.
Comparative studies have shown that brain size covaries with
innovation frequency in primates and birds (30, 31), and capuchins
Perry et al.
COLLOQUIUM
PAPER
are notable for their large relative brain sizes (32). White-faced
capuchins (Cebus capucinus) are also omnivorous extractive for-
agers (33, 34), an ecological trait positively related to innovation in
callitrichids (35), and profit greatly from experimenting with new
foods and feeding techniques. Capuchins have radiated over a
large, ecologically diverse geographic area (34, 36), which suggests
that they are capable of adapting their behavior to novel situa-
tions. They frequently interact with and investigate other species
as predators, prey, and feeding competitors (37, 38). It has been
suggested by Sol et al. (39) that innovative tendencies in birds
have coevolved with life history in generalist species, with slow-
developing, large-brained species prioritizing future over current
reproduction and being more likely to innovate and produce
plastic responses to changing conditions. If this model holds true
in other taxa, capuchins should be excellent candidates for high
innovation rate given their life histories and ecological niche.
White-faced capuchins live in multimale, multifemale social
groups that are highly xenophobic, offering little opportunity to
learn from members of other groups, except by migrating (40).
Capuchins are notable for the high frequency of coalition for-
mation and the creation of quirky social conventions that seem
to serve as means of testing the social bonds that are likely to be
important for enhancing fitness (27, 38, 40). The presence of
social traditions in the foraging and social domains (41) implies
that capuchins innovate as well.
Results
Innovations were classified in four categories or “behavioral
domains”: foraging, self-directed, social, and investigative. The
foraging category included 17 behaviors related to drinking water
or processing food, 4 of which were independently invented in
other groups and only 2 of which persisted in the innovator’s
repertoire. One of these was a form of tool use: the use of leaves
to wrap and scrub the urticating hairs off of Automeris caterpillars.
Another was the use of the tail tip as a sponge to access water
from tree holes too deep to reach with a hand or foot (Movie S1).
The latter was invented only once during the period of 2007–
2011 by a female who did this regularly but did not transmit it to
other group members. This same tail-dipping behavior was in-
ANTHROPOLOGY
dependently invented in the buffer period (2002–2006) by mon-
keys in three other groups and spread to multiple individuals in
one of those groups. (The buffer period was a time period during
which we did not score innovations, but we used this time period
to confirm whether behaviors seen in the subsequent 5-y period
were truly new to that group. Details are in Methods.) It was not
possible to reliably score food choice innovations and assign them
to particular innovators, but the foraging category would have
been much larger had we been able to include them.
The self-directed category included nine behaviors related to
enhancing comfort, dental hygiene, self-soothing, and self-
stimulation. Capuchins are prone to inventing “personal quirks,”
especially involving clutching or poking some part of their own
body for prolonged periods of time. These habits may persist for
years and might be transmitted to other group members. There
were many individuals in the 2007–2011 dataset that were still
practicing postural quirks that they invented during the buffer
period (2002–2006), and those are not evident in SI Appendix,
Table S1. The “body part hold” category lumps together many
different types of postural quirks; if we split this category more
finely, there would be far more innovations in this domain.
The social category includes 47 forms of social interaction that
are not part of the standard species repertoire. Eight of these
were independently invented in multiple groups, and most of
these inventions involved incorporation of behavioral elements
from the foraging repertoire into the exploration or use of the
interaction partner’s body (e.g., explorations of the partner’s
orifices or mouthing parts of the partner). Some behaviors (e.g.,
dental examinations, eye poking, hand sniffing, sucking of body
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7807
parts, toy game, and hair game) have been invented in multiple AA CU FF FL LB
groups over the years and have become well-established tradi-
tions in some groups; hence, many of these behaviors scored as
innovations for this 2007–2011 time period were invented before
2007 in other groups and still in practice (and hence, not counted
as innovations) during this time period. Both past work and the MK NM RF RR SP
patterning of results in SI Appendix, Table S1 suggest that those
social rituals that involve some discomfort or risk (e.g., having an
appendage bitten or damaging an eye) are more prone to remain
in individual repertoires and become established in group rep-
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
ertoires. In addition to the aforementioned behaviors that have estimated annual innovation rate per group
been proposed as bond-testing behaviors (41), the social category
of innovations includes social play, social displays, and some Fig. 1. Posterior predictions of annual innovation rate (number of inno-
creative ways for females to regulate infant behavior. vations per year) for each group. Two-letter names of the social groups are
The investigative category included creative manipulations of at the top of each panel; vertical lines indicate PMs. n = 44 group-years.
other species (e.g., porcupines, howler monkeys, and turtles),
human artifacts, leaves, sand, sticks, water, rocks, and other in-
animate objects as well as innovative ways of locomoting through self-directed domain (PM = 0.018; 89% CI = 0.008–0.037).
the forest. Most of these behaviors had no obvious immediate Much of the variance in probability of innovating can be ex-
purpose and gave the impression that the innovator was engaged plained by individual identity (σid of αp = 1.01) and social group
in recreational creativity, exploring the affordances of the ma- membership (σ group of αp = 1.56). Importantly, differences
terials. To give a few examples, in “cow pie seesaw” (Movie S2), between behavioral domains account for more variation in rates
the young monkey flips over a dried piece of cattle dung the of innovation (Table 2) than between-individual or -group dif-
length of her body, so that the flat side is on top and the rounded ferences for all other varying effect parameters with the excep-
side contacts the ground; then, she stands on top, rocking back tion of βmalep (SI Appendix, Table S2). Because interpreting the
and forth. In “mango hitting game,” a young monkey routinely coefficients of these models can be nonintuitive and because they
finds mangos about one-half the size of her body on the ground interact multiplicatively to estimate innovation rates in each
and applies hard, two-handed strikes to them for 4–5 min at a domain, we instead refer the reader to plots of model predictions
stretch, throwing her body weight into the assault on the mango, (Figs. 2 and 3 and SI Appendix, Figs. S1–S4) for all estimated effects.
with no apparent interest in eating it (SI Appendix). Less than For our model predictions, we display the posterior predic-
15% of investigative behaviors were seen more than once in tions for each individual’s annual innovation rate in each be-
individual repertoires, and only 18% were subsequently observed havioral domain. The y axes may represent (i) the estimated
being practiced by the innovator’s groupmates. number of innovations per individual monkey per year or the
The final dataset, described in detail in SI Appendix, Table S1, joint probability, (1 − p) × λ (Figs. 2 and 3 and SI Appendix, Figs.
included 187 innovations, 127 of which were unique behavior S3A and S4A). In some cases, looking at the individual compo-
types. Of the 187 innovations, 149 (80%) of them we never again nents of a zero-inflated Poisson (ZIP) model can be informative,
saw performed by the innovator, and only 41 (22%) were seen and therefore, we also present (ii) the probability of innovating
performed by other monkeys in the same group (i.e., had the per year, 1 − p, (SI Appendix, Figs. S1 A–D, S2 A–D, S3B, and
potential to have become a tradition during the observation S4B) and (iii) the number of innovations per year estimated to be
period). Of the 127 unique innovations seen, 54 (42.5%) were in observed conditional on being an innovator, λ (SI Appendix, Figs.
the investigative domain, 47 (37.0%) related to social behavior, S1 E–H, S2 E–H, S3, and S4C).
9 (7.1%) were self-directed behaviors, and 17 (13.4%) were
foraging behaviors. At least one innovation, based on our con- Age. Age differentially predicts innovativeness across behavioral
servative criteria, was scored in 117 of 234 individuals included in contexts. Younger individuals innovate at higher rates in the
this dataset. Descriptions of all innovations are included in SI investigative, foraging, and self-directed domains, although the
Appendix, and SI Appendix, Table S1 reports the distribution of effect size is quite small for self-directed and foraging behaviors
these behavioral variants across social groups and individuals. (Fig. 2 A–C). Older individuals are slightly more innovative in
Innovations were rare, being observed, on average, less than the social domain (Fig. 2D and Table 2). This effect was more
once per individual per year in any particular domain. These heavily driven by the probability of being an innovator (SI Appendix,
annual rates may appear low, but our aim is to predict the Fig. S1 A–D and Table S2) than the number of innovations con-
properties of an individual that make him/her more prone to ditional on being an innovator. Younger innovators seem slightly
innovate in particular behavioral domains. If one ignores the more likely to produce more innovations, conditional on being
properties of individuals or behavioral domains, annual group- an innovator, but this effect is small (SI Appendix, Fig. S1 E–H)
wide innovation rates (Fig. 1), which have been the focus of and near zero in most domains.
many field studies of innovation, are much higher; however, we
can learn much about individual innovators by taking this more
detailed approach. Table 1. WAIC estimates for all evaluated innovation models
Our global model (referred to as mASRMG in Table 1) re- Model WAIC dWAIC wWAIC SE
ceived overwhelming support compared with other models
[having a Widely Applicable Information Criterion (WAIC) mASRMG 1,442.02 0 1 81.97
weight of 1.00] and suggests that sociality and age are the most mA 1,478.34 36.32 0 84.38
important predictors of innovation (Table 1). From posterior mS 1,503.03 61.01 0 84.49
median (PM) estimates and 89% credible intervals (89% CIs), m 1,510.11 68.09 0 84.22
our model suggests that marginal rates of innovation per indi- mRG 1,526.1 84.08 0 86.46
vidual per year (innovation rates) are highest in the social do- mM 1,570.25 128.23 0 88.39
main (PM = 0.122; 89% CI = 0.064–0.195) followed by the Capital letters in model names correspond to predictors included: age (A),
investigative domain (PM = 0.085; 89% CI = 0.047–0.147), sociality (S), rank (R), sex (M), and group size (G). dWAIC, difference in WAIC
foraging domain (PM = 0.028; 89% CI = 0.014–0.052), and scores from the highest ranked model; wWAIC, WAIC weight.

7808 | www.pnas.org/cgi/doi/10.1073/pnas.1620739114 Perry et al.

 COLLOQUIUM
PAPER
Table 2. Posterior mean estimates of varying effects of a. foraging b. investigative
behavioral domain

1.5
Behavioral domain

Parameter Foraging Investigative Self-directed Social

1
individual innovation rate
αp −0.12 −0.45 0.01 0.06
αl −0.98 0.07 −1.38 0.49

0.5
βagep 2.37 3.96 2.97 −1.64
βagel 0.03 −0.52 −0.04 0.24
βsocialityp −0.03 −0.31 0.23 −0.13

0
βsocialityl −0.32 −0.09 −0.52 0.51
βrank.highp 0.14 −0.46 −0.02 0.00 c. self-directed d. social

1.5
βrank.highl 0.05 −0.32 0.15 −0.01
βrank.lowp −0.37 −0.42 −0.44 0.19
βrank.lowl −0.21 0.03 −0.21 0.13
βmalep 0.12 −0.57 0.33 −0.09

1
βmalel −0.09 0.08 −0.26 0.20

0.5
Sociality. Sociality also differentially predicts innovation across
behavioral domains. More social individuals showed higher rates

0
of innovation in the social (Fig. 3D) domain. Less social indi-
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
viduals had higher innovation rates in the foraging (Fig. 3A), sociality scores (standardized)
investigative (Fig. 3B), and self-directed (Fig. 3C) domains, al-
though these effects are weak and less certain. More social in- Fig. 3. Joint model predictions for the effect of sociality on the number of
dividuals produced a greater number of innovations per year innovations per individual per year. Dark lines are at the PMs; lighter lines
(conditional on being innovators) in the social domain (SI Ap- are 100 randomly sampled posterior predictions. n = 3,132 individual-years.
pendix, Fig. S2H), whereas less social individuals showed a
greater number of innovations in the self-directed domain (SI
slightly more likely to be innovators (SI Appendix, Fig. S3B).
Appendix, Fig. S2G), and there was little effect of sociality on
Within the social and investigative domains, males showed
foraging (SI Appendix, Fig. S2E) and investigative (SI Appendix,
slightly higher innovation rates both overall and conditional on
Fig. S2F) behaviors.
being innovators, but there were no discernable effects of sex in
Sex. Males (PM = 0.034; 89% CI = 0.014–0.074) have slightly the foraging and self-directed domains (SI Appendix, Fig. S3 A
higher innovation rates than females (PM = 0.024; 89% CI = and C). Where these small differences exist, they are uncertain
0.011–0.045), ignoring between-domain variation. Males were and potentially of no biological significance.

Rank. Ignoring domain-specific effects, we found that innovation

rate is highest in middle-ranking individuals (PM = 0.036; 89%

ANTHROPOLOGY
a. foraging b. investigative CI = 0.017–0.070) followed by high-ranking individuals (PM =
0.026; 89% CI = 0.011–0.058) and lowest in low-ranking indi-
1.5

viduals (PM = 0.021; 89% CI = 0.009–0.48). Rank did not have

consistent or strong effects on overall innovation rates within
domains, although there was a slight tendency for midranking
1
individual innovation rate

individuals to show higher innovation rates in the social and

investigative domains (SI Appendix, Fig. S4A) relative to low- or
high-ranking individuals; this pattern held for the rates of in-
0.5

novation conditional on being an innovator (SI Appendix, Fig.

S4C). However, within each domain, low-rankers had a slightly
higher probability of becoming innovators than mid- or high-
0

c. self-directed d. social ranked individuals (SI Appendix, Fig. S4B). Most of these rank-
related domain-specific effects are relatively small and uncertain,
1.5

and therefore, we hesitate to make strong claims about them.

Discussion
1

Utility of This Method for Identifying Innovations. On the surface, it

would seem that white-faced capuchins have the largest reper-
toire of innovations of any primate species studied thus far, but
0.5

we encourage caution in comparing the sizes of different species’

innovation repertoires or their individual innovation rates be-
cause of differences in methods between studies. The method
0

1 2 3 1 2 3 that we used here differs from previous approaches in the fol-

log(age) lowing ways.

Fig. 2. Joint model predictions for the effect of age on the number of inno- i) Researchers were vigilant for innovations and recorded
vations per individual per year in the domains of (A) foraging, (B) investigative, them throughout the study period, likely resulting in fewer
(C ) self-directed and (D) social behaviors. Dark lines are at the PMs; lighter lines overlooked innovations than in other long-term studies. It
are 100 randomly sampled posterior predictions. n = 3,132 individual-years. also enabled rigorous recording of innovations across a

Perry et al. PNAS | July 25, 2017 | vol. 114 | no. 30 | 7809
 wider range of behavioral domains than would have been the
case had we focused our data collection on a specialized
research topic not specifically related to innovation in multi-
ple behavioral domains. Despite efforts to record all possible
types of innovations, we failed to record food choice innova-
tions with sufficient rigor to include them in this analysis.
ii) This study generated a high density of behavioral observa-
tions compared with most primate field projects because of
the year-round presence of a large number of well-trained
data collectors. This higher sampling density suggests that,
relative to other studies, more true innovations would have
been observed and also, perhaps, that fewer false positives
would have been generated (i.e., fewer rare species-typical
behaviors labeled as innovations simply because they had
not been observed in other practitioners because of low
sampling density).
iii) The large number of groups monitored in essentially iden-
tical ecological circumstances with overlapping home ranges
offered better opportunity than most studies for identifying
innovations via comparison of presence vs. absence in groups
exhibiting similar ecologies. With the exception of five of the
innovations concerned with exploration of human artifacts,
there is no reason to think that monkeys had differential oppor-
tunities to discover particular innovations. Having more groups
in the sample might make it easier to find one that lacked some
rare behavior, although the long-term nature of the study com-
bined with the dense behavioral sampling probably mitigate this
tendency for the large number of groups to produce false pos-
itives of innovation caused by presence/absence contrasts.
iv) The long-term nature of the study and the use of a 5-y buffer
period before the observation period reduce the chance that
behaviors will be falsely termed innovations when they are
actually low-frequency behaviors already in the repertoire. If
our study had been the length of a typical dissertation proj-
ect (i.e., 1 y) and if we had assumed that behaviors were new
to the practitioners if it was the first time that we had seen
them performed in those groups, then we would have had
52% more innovations in our sample than we obtained by
using the buffer period method and requiring each recorded
innovation to be the first sighting by that individual in its
group(s) of residence.
v) Our method is more conservative than the definition used by
Ramsey et al. (12), which defines innovations as being new
to individual repertoires but not necessarily new to group
repertoires. We recognize the possibility of having indepen-
dent inventions within a single social group, and we suspect
that many true innovations in our sample have been discarded
because of the suspicion that they may be the products of
social learning. On the whole, we think that our method pro-
vides a more accurate technique for diagnosing innovations
than alternative observational methods. However, overlook-
ing independent inventions within the same social group is
one way in which we likely underestimate innovation rates.
vi) Our approach looks at the properties that affect individual
propensities to innovate and longitudinally tracks the rate of
innovation over 5 y. Previous studies have looked at group-
level differences or short windows of time. Our hierarchical
statistical approach accounts for unequal sampling effort
among individuals, estimates between- and within-individual
variation, and avoids the potential problem of falsely making
inferences about individual innovation rate from potentially
spurious group-level effects.
Rates and Types of Innovation in Capuchin Monkeys. Evidence that
individuals, on average, (i) innovate in any one of these four
domains less than once per year, (ii) retain only about 20% of
these innovations in their repertoires, (iii) transmit no more than
22% of their innovations to other group members, and (iv)
7810 | www.pnas.org/cgi/doi/10.1073/pnas.1620739114
generally produce innovations with no obvious utility may give an
artificially low impression of the biological importance that in-
novation has for behavioral repertoires overall. Most groups
have a current repertoire of bond-testing signals that have been
steadily practiced by a subset of the group for many years, and
these behaviors are not, for the most part, scored as innovations
in this dataset, because they first appeared in the buffer period or
even earlier. Food choice innovations (not reported here be-
cause of the interobserver reliability challenge of being able to
correctly identify plants the first time that they are eaten) tend to
be adopted quickly and remain in repertoires for long periods of
time, such as is the case for chimpanzees (42) and various species
of monkeys (29). Particularly useful food processing or drinking
techniques, once invented, are likely to persist in repertoires for
many years [or even centuries (43)]. Tail-dipping to access water
in deep tree holes seems to have been independently invented in
four groups at Lomas, originating during the 2002–2006 period
in three of these groups (and therefore, not counting as an in-
novation according to our definition) but persisting during the
2007–2011 period in two of those three groups. Although many
of the innovative behaviors recorded during this 5-y period seem
to be aimless creativity with no obvious utilitarian goal in mind
(aside from the foraging behaviors), it is important to remember
that innovations, like mutations, may not be particularly bene-
ficial on their own but may become exapted (44) and acquire a
benefit when paired with a particular ecological or social context
(even if the initial pairing is accidental). For example, a poten-
tially risky behavior, such as sticking a finger deep into the eye
socket of a partner, might yield benefits if it is incorporated into
a dyadic ritual that tests the quality of an important social bond
(27, 45). Additionally, many of the capuchin innovations in this
dataset might inform the developing monkey about the affor-
dances of objects or how his/her body relates to the environment,
providing useful feedback, even if there is no practical value to
permanently incorporating the new behavior into the behavioral
repertoire.
How the Capuchin Innovation Repertoire Compares with Those of
Chimpanzees and Orangutans. The profound methodological dif-
ferences between studies of these species preclude precise
quantitative comparisons of innovation rates in different be-
havioral domains, but we can at least make some crude quali-
tative comparisons between capuchins and the other two primate
species for which innovations have been systematically cataloged
in the wild. The Mahale chimpanzee researchers (42) present
data on 26 novel behaviors (after excluding food choice to make
their results more comparable with the other datasets) retro-
spectively extracted from their 43-y study of two chimpanzee
communities. Although studying innovation was not an explicit
part of their core data collection protocol, many researchers at
Mahale described behavioral variation and novel behaviors in
detail. They defined innovation as any behavior not seen in the
first 15 y of research. Innovation in orangutans has been ex-
plicitly studied using the geographic contrasts method in short-
term studies by van Schaik et al. (26) at multiple sites, producing
a sample of 44 putative innovations. Comparison of the distri-
bution of behaviors across domains in their datasets with the
composition of the repertoire in the dataset of 127 unique in-
novations from Lomas Barbudal suggests that capuchins, relative
to these ape species, devote a higher proportion of their creative
energy to investigation of their environment and devising new
social behaviors and a lower proportion to comfort-related be-
haviors and foraging. Orangutans are particularly prone to de-
vising new variants on nest-building techniques, and even their
novel acoustic behaviors seem to emerge primarily in a nest-
building context. Both ape species are more innovative with
regard to bodily comfort and hygiene than capuchins. Capuchins
rarely seem to prioritize comfort: they do not build nests, and
Perry et al.

they readily endure stings, bites, urticating hairs, and spiny veg-
etation in the process of acquiring food. Although nearly 4% of
unique innovations may serve to mitigate physical discomfort,
this percentage is low compared with both chimpanzees (15%)
and orangutans (36%). Capuchins devoted about 42% of their
innovation repertoire to the investigative domain compared with
chimpanzees (31%) and orangutans (14%). Orangutans are cu-
riously neophobic relative to both chimpanzees and capuchins,
and the exploration of the environment that they engage in as
young individuals is primarily focused on the mother’s activities
rather than independent discovery (13), whereas capuchins engage
in much object play and independent exploration of the affordances
of the environment at an early age. The preponderance of in-
vestigative sorts of innovations in orangutans seems to be devoted to
solving the problem of how to safely engage in arboreal locomotion
as a large-bodied animal (26, 46)—a problem that is less difficult for
a small-bodied animal with a prehensile tail. Within the category of
social innovations, capuchins were more prone to inventing be-
haviors related to social bonding, whereas chimpanzees placed
more emphasis on aggressive displays. This emphasis is consistent
with the importance of alliances and hence, bond tests in capuchins:
males need allies for parallel dispersal and to acquire and maintain
breeding positions, and females depend on allies to defend food
resources and infants from potentially infanticidal males (40).
What Factors Affect Propensity to Innovate in Other Species? In this
study, younger animals were more innovative in all behavioral
domains except for social interaction, in which the tendency was
reversed. This result is consistent with capuchins’ slow life his-
tory, generalist ecological niche, and complex social relation-
ships. During a long juvenile phase, individuals have much to
gain by experimenting with both new foods and more efficient
ways of acquiring resources. As they age, capuchins increasingly
have reason to form and test the quality of those social rela-
tionships that will prove critical for acquiring and protecting
access to the food and social resources needed for enhancing
reproductive success (40), and therefore, it makes sense that older
animals are more prone to innovation in the social domain.
Studies of other species provide mixed results regarding the effect
of age on innovative tendencies. A review of the published primate
literature suggests that adults innovate more than immatures (16),
and this pattern has been corroborated by experimental studies of
innovation in callitrichids (22) and meerkats (20), in which young
animals were less likely than adults to successfully solve a novel
extractive foraging task, possibly because of insufficient develop-
ment of dexterity. Chimpanzees are a notable exception to this
general pattern; high rates of innovation, particularly of the social
and investigative sort, are observed in immature chimpanzees (16,
29). Among human children, older children are better than
younger children at solving novel problems (47). It is worth noting
that, in this study, we were probably measuring something more
akin to creativity and exploration rather than skill at solving a task,
and it is possible that the innovations created by older capuchins
could be argued to be more sophisticated in some way.
The other variable that had a strong effect on capuchins’
propensity to innovate was sociality: more social capuchins were
more prone to inventing novel social interaction types, and more
social monkeys were also slightly more likely to have their social
innovations picked up by other group members (although we
cannot currently say whether this is because of social learning)
(SI Appendix, Fig. S5B). These results can probably be explained
simply by the fact that more social individuals have more op-
portunities to experiment with novel forms of social interaction.
The existing literature on innovation does not have much to say
about the effects of sociality per se, aside from predicting that
technological innovations will be more common in peripheral
animals who are less distracted by social life (29). However, the
literature does address dominance rank as a predictive variable,
Perry et al.
COLLOQUIUM
PAPER
and subordinates are often less socially central than dominants.
Because there was not a strong rank effect in this dataset, we
remain hesitant about making claims regarding rank mediating
the relationship between sociality and innovative tendencies.
Although sex and rank (16, 20, 29) are often predictive of
innovative tendencies in other species, neither variable had much
explanatory power in the capuchin dataset. Capuchin males were
slightly more likely to innovate in all domains, which is consistent
with observations from other primates (16) and meerkats (20)
but not guppies (48); however, these effects were so small and
uncertain in the capuchin dataset that they are not likely to have
any biological significance.
With a better understanding of innovation, the next logical
questions to address are two questions linking this research to
the social learning literature. What properties of innovations and
what properties of innovators make the innovations most likely
to be socially transmitted to other group members? Adequate an-
swers to these questions are beyond the scope of this paper, but
preliminary analyses suggest that age and group size of the innovator
might be important drivers (SI Appendix, SI Text and Fig. S5).
Methods
Study Site and Subjects. The study was performed at Lomas Barbudal Biological
Reserve in Guanacaste, Costa Rica, a tropical dry forest described in the work by
Frankie et al. (49). This population of C. capucinus has been the subject of long-
term study by Perry et al. (50) since 1990, starting with a single social group.
The number of regularly monitored groups has since expanded by both fission
of research groups and habituation of new groups to include 10 groups by the
end of 2011.
One of the methodological problems in documenting innovations is the
difficulty of knowing whether a particular behavior has ever been performed
or observed by a particular individual. In a short-term study, when the re-
searchers have little experience with the animals, many rare behaviors will be
falsely scored as innovations simply because the researcher has not seen them
previously. Ironically, such sampling biases may result in the misimpression
that innovation rates are higher in short-term studies than long-term studies.
Such problems are mitigated in this study by (i) using data from a long-term
study; (ii) generating a high density of observations by using a large staff to
collect year-round behavioral data, in which new behaviors were explicitly
recorded; (iii) explicitly training observers to watch for and record new be-
haviors; and (iv) having a single observer with 26 y of experience collecting
ANTHROPOLOGY
data on this population (S.E.P.) make the final determinations about which
behaviors are truly new for each group of monkeys, with input from two
additional long-term researchers (B.J.B. and I.G.).
Beginning in January of 2002, all research staff were directed to make
freeform comments about any behaviors that did not neatly fit into the
standard ethogram of species-specific behavior and explicitly mark comment
lines as comment innovation when they thought they were seeing a behavior
that they had never seen before in that group or a behavior that they thought
was a unique behavioral tradition. Naturally, many behaviors seem new to
relatively inexperienced observers, and therefore, not all of the behaviors
initially coded as innovations were true innovations. Also, some behaviors not
coded as innovations in the comments section were, in fact, true innovations.
S.E.P., who personally collected 13,770 h of data on this population from
1990 to 2016, read through all data to determine whether behavioral se-
quences were likely to be true innovations (i.e., behaviors seen for the first
time in that particular social group).
This research was performed in compliance with the laws of Costa Rica,
and the protocol was approved by the University of California, Los Angeles
Animal Care Committee (ARC 1996-122 and 2005-084 plus various renewals).
Buffer Period. We used a 5-y chunk of observational data (35,196 h collected
between January 1, 2007 and December 31, 2011) to look for innovations. We
used the 5 preceding years of data (∼37,514 h of observation collected during
2002–2006) as a “buffer period” (i.e., a period in which we could search for
prior instances of behaviors that appeared to be innovations within the
targeted 2007–2011 time period). If we had not left a large buffer period,
we would have falsely concluded that far more behaviors within the time
period of interest were innovations. SI Appendix, Table S1 reports, for each
innovation, the number of groups in which it occurred, its persistence in the
innovator’s repertoire, and whether it spread to other group members.
Data were collected by a team of 50 highly trained observers (∼12 per
year), each of whom underwent a training period of approximately 3 mo of
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7811
dawn to dusk instruction and interobserver reliability testing before con-
tributing data to the database. Interobserver reliability tests for monkey
identifications and coding were repeated monthly throughout their tenures
at the field site. Obviously, we could not train observers to reliably code
innovations (behaviors that had not yet happened) in the same ways. We
could, however, ensure that the observers recognized and reliably recorded
basic motor movements, gestures, and 205 behaviors considered to be part
of the species-typical behavioral repertoire. This repertoire was based
on >6,000 h of observation invested in studying these monkeys during 1990–
2001 before explicit recording of innovations began, thereby enabling
better detection of the idiosyncratic behaviors.
Although we gave instructions to data collection teams to record any type
of innovation in any behavioral domain, we decided not to include inno-
vations regarding (i) choices of food or medicinal plants and (ii) vocal be-
havior in this analysis. Although we witnessed many instances of apparent
innovation regarding use of novel foods and medicines, we lacked sufficient
confidence in observers’ ability to accurately identify rarely used plants and
insects or accurately identify rarely produced vocalizations.
A behavioral observation was scored as an innovation if it met the fol-
lowing criteria: (i) the behavior was absent in some of the groups that we
regularly monitor, (ii) it was the first time that this behavior had been seen
in this group during 2007–2011, and (iii) this behavior had not been seen in
this group during the buffer period time period preceding 2007 during the
lifetime of the putative innovator. In cases of group fission or migration, we
required that the behavior not have been seen previously in the other
groups of which the potential innovator had been a member. There was one
class of behaviors on which an additional criterion was imposed: postural
habits, such as clutching of one’s own body parts or sniffing one’s own hand
(SI Appendix).
The number of innovations was also counted via two less conservative
methods for methodological comparison. In one version, we eliminated the
buffer period criterion, calling the first observation of a behavior in a par-
ticular group an innovation. This method yielded 263 innovations compared
with 187 produced with the more conservative method described above. In an
even less stringent version that yielded 282 innovations, we termed a be-
havior an innovation if it had not been seen in that group in the past year.
Another challenge in defining innovation is the “grain” problem (i.e.,
determining the descriptive breadth of behavioral categories or in other
words, the extent to which to lump vs. split behaviors) (51). The grain
problem is insoluble; the best that we can do is to use our intuitions about
what the animals themselves seem to consider novel and observations of
how these behaviors cluster in the repertoires of groups and individuals. Are
novel actions used as part of a task already in the behavioral repertoire
novelties? In our view, they usually were. Are the same actions applied to
different objects? Unless the objects and contexts were quite radically dif-
ferent, we opted to lump these together (i.e., we did not designate them as
innovations). This issue was most prominent in our decision-making when
evaluating clutching of different body parts (a self-directed behavior),
sucking of different body parts (a social behavior), and the “toy game,”
which involves passing an object from mouth to mouth. In all of these cases,
we chose to lump rather than split behaviors. In the case of object play (e.g.,
with sand, water, or rocks), different actions used by different individuals in
interacting with these same substances were scored as different behaviors.
Descriptions of the complete list of behaviors that were included in our
analysis are in SI Appendix.
Innovations were classified in four categories or behavioral domains, the
content of which is described in greater detail in Results: foraging, self-
directed behavior, social behaviors, and investigative behaviors (explora-
tion of the environment).
Data Structure. We created a different row for every individual monkey/year/
behavioral domain combination, and a value was scored for the number of
innovations observed for that combination; this number was the output
variable. We measured four main predictor variables to determine what
predicts the number of innovations per individual per year: age, sex, sociality,
and rank. (i) Age. To calculate age, we subtracted the birth year of an indi-
vidual from the year of observation and added one. Individuals born in the
same year as the year of observation were excluded from the dataset. Age was
log-transformed and centered for analysis. (ii) Sex. We coded sex as a dummy
variable (one for males and zero for females). (iii) Sociality. Sociality was cal-
culated using data from group scans, which were taken opportunistically for
all group members, at intervals no closer together than 10 min. For each in-
dividual per calendar year, we calculated the proportion of group scans in
which the individual was in proximity (i.e., within ∼400 cm) to at least one
other group member other than their dependent offspring (i.e., for females,
7812 | www.pnas.org/cgi/doi/10.1073/pnas.1620739114
their offspring that are less than 1 y of age). Individuals with less than 20 scans
per year were excluded from analysis. Sociality scores were standardized, so
that a sociality score of one is 1 SD above the centered mean of zero. (iv) Rank.
For individuals older than 3 y old, rank was calculated using the EloRating
package in R (52). We used outcomes from dyadic interactions involving
avoids, cowers, flees, and supplants to determine dominance. We used default
Elo parameters, with initial Elo scores set to 1,000, and the constant k set to
100. Rank was estimated using average Elo scores calculated for each individual
per calendar year. Young juveniles less than 3 y old were given Elo Scores of
zero to assure a low rank within their social groups. Members of a group were
divided up into tertiles, with the corresponding levels receiving rank categories
of high, middle, and low. High and low ranks were estimated as dummy vari-
ables, with middle rank serving as the intercept-only reference category.
Statistical Methods. Our outcome variable, number of innovations, is a count
variable with many zero values. Each monkey was observed over multiple
years. Membership in a particular group may have affected propensities to
innovate. Therefore, we analyzed these data using a series of hierarchical ZIP
models. ZIP models are mixture models that use two probability distribu-
tions. One component assumes a Bernoulli distribution and estimates p: the
probability of observing a zero. The other component assumes a Poisson
distribution and estimates λ: the estimated mean of a Poisson distribution.
ZIP models permit a mixture of causal factors to be evaluated and help
better predict outcomes when there is a large number of zeros because of
both the rarity of an event and false negatives. The joint likelihood of ob-
serving an innovation can be calculated by multiplying the likelihoods of the
Bernoulli and Poisson outcomes and converting them to the real number
scale using their corresponding link functions. We graphically present joint
posterior predictions here (Figs. 2 and 3) along with model predictions of
p and λ in SI Appendix (SI Appendix, Figs. S1–S4).
We looked at four predictors in this analysis: (i) age, (ii) sex, (iii) sociality,
and (iv) rank. We also estimated unique offsets for each individual, because
they differed in observation time or exposure. We analyzed six models, four
of which corresponded to one of the single aforementioned predictors. The
other two included a global model that looked at all four predictors and an
intercepts-only model. In each model, we used varying intercepts for each
individual (n = 234), social group (n = 10), and behavioral domain (n = 4).
Varying slopes for domains and groups were estimated for all four predic-
tors, and varying slopes for individuals were estimated for sociality and age.
Group size was used as a covariate to control for the numerical likelihood
that, in smaller groups, observed behaviors might be more likely to be
scored as innovations because of our definition of uniqueness and that a
greater number of innovations is more likely in larger groups. In another
analysis, we used experimental year as a varying effect to see if there were
any biases in data collection between field assistant cohorts that would
change our inference. There were none, and therefore, we excluded these
parameters from our final analyses to simplify the presentation of results.
Offsets and Exposure. Because observations of innovations were collected not
only in focal follows but also, ad libitum, and because individuals varied in
their likelihood of being observed because of data collection protocols or
their visibility in the group, they also differed in exposure. To account for
differences in exposure, we calculated an annual offset for each individual in
each calendar year. Offsets for each individual (Oi) were estimated using
 
Gi + Fi
Oi = log ,
365
where Gi is the number of instantaneous group scans calculated per calen-
dar year for each individual i, and Fi is the number of point samples collected
at 2.5-min intervals during focal follows in a calendar year. These offsets
were included alongside linear predictors in each model.
Models were fit using the map2stan function in the R package rethinking
(53). Models were fit using Hamilton Markov Chain Monte Carlo in r-STAN
(v 2.14) (54) in R v. 3.3.2 (55). Models were compared with widely applicable
information criteria (WAIC) using the compare function in rethinking. The
corresponding code and data used for each model and graph production can
be found through a link in SI Appendix.
To estimate group-level difference in annual innovation rate, we summed
the number of innovations observed within each group and across all indi-
viduals and behavioral domains. Exposure rates for individuals within groups
were summed within years. Counts of annual innovations per group were
then fit using a hierarchical Poisson model fit using r-STAN that accounted for
exposure rates using metrics previously described and varying intercepts for
Perry et al.

each group, thus providing an estimate of annual innovation rate per year
per group.
ACKNOWLEDGMENTS. The following field assistants contributed a year or
more of data to the Lomas Barbudal Monkey Project dataset: L. Beaudrot,
M. Bergstrom, R. Berl, A. Bjorkman, L. Blankenship, T. Borcuch, J. Broesch,
J. Butler, F. Campos, C. Carlson, S. Caro, M. Corrales, N. Donati, C. Dillis,
G. Dower, R. Dower, K. Feilen, K. Fisher, A. Fuentes J., M. Fuentes A.,
C. Gault, H. Gilkenson, I. Gottlieb, L. Hack, S. Herbert, C. Hirsch, C. Holman,
S. Hyde, L. Johnson, S. Lee, S. Leinwand, T. Lord, K. Kajokaite, M. Kay,
E. Kennedy, D. Kerhoas-Essens, E. Johnson, S. Kessler, S. MacCarter, J. Manson,
W. Meno, C. Mitchell, Y. Namba, A. Neyer, C. O’Connell, J. C. Ordoñez,
J., N. Parker, B. Pav, R. Popa, K. Potter, K. Ratliff, H. Ruffler, S. Sanford,
M. Saul, I. Schamberg, C. Schmitt, A. Scott, J. Verge, A. Walker-Bolton, E. Wikberg,
and E. Williams. We thank H. Gilkenson, W. Lammers, C. Dillis, M. Corrales,
and R. Popa for helping to manage the field site. E. Wikberg and K. Kajokaite
contributed a year or more of effort to organizing the dataset. D. Cohen
created the MySQL database. R. Hong assisted in the organization and
coding of the innovation data. This paper has benefitted from helpful discus-
sions with D. Caillaud, M. Crofoot, M. Grote, K. Kajokaite, R. McElreath,
1. Imanishi K (1952) Man (Mainichi-Shinbunsha, Tokyo).
2. Kummer H (1971) Primate Societies: Group Techniques of Ecological Adaptation
(AHM Publ Corp, Arlington Heights, IL).
3. West-Eberhard MJ (2003) Developmental Plasticity and Evolution (Oxford Univ Press,
Oxford).
4. Giraldeau L-A, Caraco Y, Valone T (1994) Social foraging: Individual learning and
cultural transmission of innovations. Behav Ecol 5:35–43.
5. Sol D (2003) Behavioural flexibility: A neglected issue in the ecological and evolu-
tionary literature? Animal Innovation, eds Reader SM, Laland KN (Oxford Univ Press,
Oxford), pp 62–82.
6. Reader SM, Morand-Ferron J, Flynn E (2016) Animal and human innovation: Novel
problems and novel solutions. Philos Trans R Soc Lond B Biol Sci 371:371.
7. Wyles JS, Kundel JG, Wilson AC (1983) Birds, behavior, and anatomical evolution. Proc
Natl Acad Sci USA 80:4394–4397.
8. Keagy J, Savard J-F, Borgia G (2009) Male satin bowerbird problem-solving ability
predicts mating success. Anim Behav 78:809–817.
9. Cauchard L, Boogert NJ, Lefebvre L, Dubois F, Doligez B (2013) Problem-solving per-
formance is correlated with reproductive success in a wild bird population. Anim
Behav 85:19–26.
10. Whiten A, Ayala FJ, Feldman MW, Laland KN (2017) The extension of biology through
culture. Proc Natl Acad Sci USA 114:7775–7781.
11. Reader SM, Laland KN (2003) Animal innovation: An introduction. Animal Innovation,
eds Reader SM, Laland KN (Oxford Univ Press, Oxford), pp 3–35.
12. Ramsey G, Bastian ML, van Schaik C (2007) Animal innovation defined and oper-
ationalized. Behav Brain Sci 30:393–407.
13. van Schaik CP, et al. (2016) The reluctant innovator: Orangutans and the phylogeny of
creativity. Philos Trans R Soc Lond B Biol Sci 371:371.
14. Fogarty L, Creanza N, Feldman MW (2015) Cultural evolutionary perspectives on
creativity and human innovation. Trends Ecol Evol 30:736–754.
15. Perry S (1996) Intergroup encounters in wild white-faced capuchins, Cebus capucinus.
Int J Primatol 17:309–330.
16. Reader SM, Laland KN (2001) Primate innovation: Sex, age, and social rank differ-
ences. Int J Primatol 22:787–805.
17. Lefebrvre L, et al. (1998) Feeding innovations and forebrain size in Australasian birds.
Behaviour 135:1077–1097.
18. Lefebrvre L, Whittle P, Lascaris E, Finkelstein A (1997) Feeding innovations and
forebrain size in birds. Anim Behav 53:549–560.
19. Boogert NJ, Reader SM, Hoppitt W, Laland KN (2008) The origin and spread of in-
novations in starlings. Anim Behav 75:1509–1518.
20. Thornton A, Samson J (2012) Innovative problem solving in wild meerkats. Anim
Behav 83:1459–1468.
21. Aplin LM, et al. (2015) Experimentally induced innovations lead to persistent culture
via conformity in wild birds. Nature 518:538–541.
22. Kendal RL, Coe RL, Laland KN (2005) Age differences in neophilia, exploration, and
innovation in family groups of callitrichid monkeys. Am J Primatol 66:167–188.
23. Russon AE (2003) Innovation and creativity in forest-living rehabilitant orangutans.
Animal Innovation, eds Reader SM, Laland KN (Oxford Univ Press, Oxford), pp 279–306.
24. Whiten A, et al. (1999) Cultures in chimpanzees. Nature 399:682–685.
25. van Schaik CP, et al. (2003) Orangutan cultures and the evolution of material culture.
Science 299:102–105.
26. van Schaik CP, van Noordwijk MA, Wich SA (2006) Innovation in wild Bornean
orangutans (Pongo pygmaeus wurmbii). Behaviour 143:839–876.
27. Perry S, et al. (2003) Social conventions in wild white-faced capuchin monkeys: Evi-
dence for traditions in a neotropical primate. Curr Anthropol 44:241–268.
28. Koolhaas JM, et al. (1999) Coping styles in animals: Current status in behavior and
stress-physiology. Neurosci Biobehav Rev 23:925–935.
29. Kummer H, Goodall J (1985) Conditions of innovative behaviour in primates. Philos
Trans R Soc Lond B Biol Sci 308:203–214.
30. Overington S, Morand-Ferron J, Boogert NJ, Lefebvre L (2009) Technical innovations
drive the relationship between innovativeness and residual brain size in birds. Anim
Behav 78:1001–1010.
Perry et al.
COLLOQUIUM
PAPER
J. Manson, K. Perry, B. Scelza, and M. J. West-Eberhard. We thank
Marcus W. Feldman, Andrew Whiten, Kevin N. Laland, and Francisco J.
Ayala for the opportunity to participate in this Sackler Symposium. We
thank the Costa Rican park service (Sistema Nacional de Áreas de Conserva-
ción and Área de Conservación Tempisque), Hacienda Pelon de la Bajura,
Hacienda Brin D’Amor, and the residents of San Ramon de Bagaces for per-
mission to work on their land. This project is based on work supported by the
funding provided to S.E.P. by the Max Planck Institute for Evolutionary
Anthropology, National Science Foundation (NSF) Grants SBR-0613226 and
BCS-0848360, two grants from the Leakey Foundation, two grants from the
National Geographic Society, and multiple Committee on Research grants
from the University of California, Los Angeles (UCLA). B.J.B. was supported
by NSF Graduate Research Fellowship (GRF) Grant 1650042 and two
Achievement Rewards for College Scientists Foundation, Northern Cali-
fornia Chapter fellowships. I.G. was supported by the International Society
for Human Ethology, a Ford Foundation fellowship, an NSF GRF, and multi-
ple grants from UCLA. Any opinions, findings, and conclusions or recommen-
dations expressed in this material are those of the author(s) and do not
necessarily reflect the views of the NSF or other funding agencies.
31. Navarrete AF, Reader SM, Street SE, Whalen A, Laland KN (2016) The coevolution of
innovation and technical intelligence in primates. Philos Trans R Soc Lond B Biol Sci
371:371.
32. Stephan H, Bauchot R, Andy OJ (1970) Data on size of the brain and various brain
parts in insectivores and primates. The Primate Brain, eds Noback CR, Montagna W
(Appleton-Century-Crofts, New York), pp 289–297.
33. Perry S, Ordoñez Jiménez JC (2006) The Effects of Food Size, Rarity, and Processing
Complexity on White-Faced Capuchins’ Visual Attention to Foraging Conspecifics, eds
Hohmann G, Robbins M, Boesch C (Cambridge Univ Press, Cambridge, UK), pp 203–234.
34. Fragaszy DM, Visalberghi E, Fedigan LM (2004) The Complete Capuchin: The Biology
of the Genus Cebus (Cambridge Univ Press, Cambridge, UK).
35. Day RL, Coe RL, Kendal JR, Laland KN (2003) Neophilia, innovation and social learn-
ing: A study of intergeneric differences in callitrichid monkeys. Anim Behav 65:
559–571.
36. Lynch Alfaro JW, et al. (2012) Explosive Pleistocene range expansion leads to wide-
spread Amazonian sympatry between robust and gracile capuchin monkeys. J Biogeogr
39:272–288.
37. Rose L, et al. (2003) Interspecific interactions between white-faced capuchins (Cebus
capucinus) and other species: Preliminary data from three Costa Rican sites. Int J
Primatol 24:759–796.
38. Perry S, Manson JH (2008) Manipulative Monkeys: The Capuchins of Lomas Barbudal
(Harvard Univ Press, Cambridge, MA).
39. Sol D, Sayol F, Ducatez S, Lefebvre L (2016) The life-history basis of behavioural in-
novations. Philos Trans R Soc Lond B Biol Sci 371:371.
40. Perry S (2012) The behavior of wild white-faced capuchins: Demography, life history,
social relationships, and communication. Adv Study Behav 44:135–181.
ANTHROPOLOGY
41. Perry S (2011) Social traditions and social learning in capuchin monkeys (Cebus). Philos
Trans R Soc Lond B Biol Sci 366:988–996.
42. Nishida T, Matsusaka T, McGrew WC (2009) Emergence, propagation or disappear-
ance of novel behavioral patterns in the habituated chimpanzees of Mahale: A re-
view. Primates 50:23–36.
43. Haslam M, et al. (2016) Pre-Columbian monkey tools. Curr Biol 26:R521–R522.
44. Gould SJ, Vrba ES (1982) Exaptation - a missing term in the science of form.
Paleobiology 8:4–15.
45. Zahavi A (1977) The testing of a bond. Anim Behav 25:246–247.
46. Russon AE, Kuncoro P, Ferisa A (2015) Tools for the trees: Orangutan arboreal tool use
and creativity. Animal Creativity and Innovation, eds Kaufman AB, Kaufman JC
(Elsevier, San Diego), pp 419–455.
47. Beck SR, Williams C, Cutting N, Apperly IA, Chappell J (2016) Individual differences in
children’s innovative problem-solving are not predicted by divergent thinking or
executive functions. Philos Trans R Soc Lond B Biol Sci 371:371.
48. Laland KN, van Bergen Y (2003) Experimental studies of innovation in the guppy.
Animal Innovation, eds Reader SM, Laland KN (Oxford Univ Press, New York), pp
155–173.
49. Frankie GW, Vinston SB, Newstrom LE, Barthell JF (1988) Nest site and habitat pref-
erences of Centris bees in the Costa Rican dry forest. Biotropica 20:301–310.
50. Perry S, Godoy I, Lammers W (2012) The Lomas Barbudal Monkey Project: Two de-
cades of research on Cebus capucinus. Long-Term Field Studies of Primates, eds
Kappeler P, Watts D (Springer, New York), pp 141–165.
51. Russon A, Andrews K, Huss B (2007) Innovation and the grain problem. Behav Brain
Sci 30:423–433.
52. Neumann C, Kulik L (2014) EloRating: Animal Dominance Hierarchies by Elo-Rating.
R Package, Version 0.43. Available at https://cran.r-project.org/package=EloRating.
Accessed January 5, 2017.
53. McElreath R (2016) Statistical Rethinking: A Bayesian Course with Examples in R and
Stan (Chapman and Hall/CRC, New York).
54. Stan Development Team (2016) RStan: The R Interface to Stan. R package version
2.14.1. (R Foundation for Statistical Computing, Vienna).
55. R Development Core Team (2013) R: A Language and Environment for Statistical
Computing (R Foundation for Statistical Computing, Vienna).
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7813
Gene–culture coevolution in whales and dolphins
Hal Whiteheada,1
a
Department of Biology, Dalhousie University, Halifax, NS, Canada B3H 4R2
Edited by Marcus W. Feldman, Stanford University, Stanford, CA, and approved May 1, 2017 (received for review January 14, 2017)
Whales and dolphins (Cetacea) have excellent social learning skills
as well as a long and strong mother–calf bond. These features
produce stable cultures, and, in some species, sympatric groups
with different cultures. There is evidence and speculation that this
cultural transmission of behavior has affected gene distributions.
Culture seems to have driven killer whales into distinct ecotypes,
which may be incipient species or subspecies. There are ecotype-
specific signals of selection in functional genes that correspond to
cultural foraging behavior and habitat use by the different eco-
types. The five species of whale with matrilineal social systems
have remarkably low diversity of mtDNA. Cultural hitchhiking,
the transmission of functionally neutral genes in parallel with se-
lective cultural traits, is a plausible hypothesis for this low diver-
sity, especially in sperm whales. In killer whales the ecotype
divisions, together with founding bottlenecks, selection, and cul-
tural hitchhiking, likely explain the low mtDNA diversity. Several
cetacean species show habitat-specific distributions of mtDNA
haplotypes, probably the result of mother–offspring cultural trans-
mission of migration routes or destinations. In bottlenose dolphins,
remarkable small-scale differences in haplotype distribution result
from maternal cultural transmission of foraging methods, and large-
scale redistributions of sperm whale cultural clans in the Pacific have
likely changed mitochondrial genetic geography. With the accelera-
tion of genomics new results should come fast, but understanding
gene–culture coevolution will be hampered by the measured pace
of research on the socio-cultural side of cetacean biology.
gene–culture coevolution | Cetacea | cultural hitchhiking
E volution is dependent on inheritance. Although the trans-
mission of genes is primary, it is not the only mode of in-
heritance. As recently articulated in the extended evolutionary
synthesis, inheritance extends beyond genes to include epigenetic
inheritance, physiological inheritance, ecological inheritance,
social transmission, and cultural inheritance, all of which can
lead to heritability of phenotype (1). From the perspective of
modern humans, cultural inheritance, which directly determines
much of our own behavior as well as indirectly determining how
we affect the global environment, has particular salience (2).
Culture, as an inheritance system, can be defined as behavior
or information shared within a community that is acquired
from conspecifics through some form of social learning (3), i.e.,
learning that is influenced by observation of, or interaction with,
another animal or its products (4). Social learning comes in a
range of forms including imitation, emulation, teaching, and
local enhancement (5), all of which can promote behavioral
similarity between learner and model. Culture may include a
wide range of behavior, including foraging methods, vocaliza-
tions, diet selection, social behavior, movement, habitat use,
social structure, and play (e.g., refs. 3 and 6); potentially higher-
level attributes of culture, e.g., conservative cultures, exploratory
cultures, or pacific cultures (refs. 3, 7, and 8), also can affect a
range of behavior. Cultures may be stable over many generations
or transitory fads. The learning is often from parent to offspring
in the former and from peer to peer in the latter (9).
The different modes of inheritance do not act wholly inde-
pendently. Phenotypic assimilation by one mode of inheritance
may influence the transmission of traits by another mode. Genes
produce the basic organismal phenotype, and this genetic phenotype
7814–7821 | PNAS | July 25, 2017 | vol. 114 | no. 30
places major constraints on the other modes of inheritance.
Circumstances in which other inheritance mechanisms control
the inheritance of genes are much less obvious but can have great
evolutionary significance (10). Thus, there has been particular
interest in gene–culture coevolution (11–13).
Theoreticians started to model potential interactions between
genes and culture in the 1970s (14), and there has been much
empirical and theoretical work since then (11, 13, 15, 16).
However, gene–culture coevolution has rarely been formally
defined. For the perspective of this article, I consider cases in
which culture—i.e., group-specific social learning—affects the
distribution of DNA found in a population. That is, we can
reasonably suppose that the distribution of genes in a population
would be different if individuals were not transmitting cultural
information through social learning. Gene–culture coevolution
includes very specific selective processes in which a particular
cultural practice affects the evolution of a particular gene or
genes. The most famous example is the coevolution of dairy
farming and the lactase gene, allowing adult humans in dairy-
farming cultures to digest milk products (17). However, gene–
culture coevolution can also include general processes. If culture
is driving significant parts of a species’ behavior, then this in-
fluence may constrain genetic evolution. For instance if cultural
processes isolate groups and give them distinctive behavior, then
this isolation may initiate or promote the course of speciation or
reduce the diversity of genes that are transmitted in parallel with
the cultural traits (18–20). Under this concept of gene–culture
coevolution there will be circumstances in which the available
information may not be definitive. Scenarios including or exclud-
ing culture may both be consistent with results.
Gene–culture coevolution has been considered almost entirely
from the perspective of Homo sapiens, and the great majority of
research has been aimed at the hypothesis that there are specific
genes whose distribution has been affected by human cultural
practices (11, 13, 15). Two of the best-documented cases are the
dairy farming–lactase relationship mentioned above and a sce-
nario relating human activities, such as clear-cutting of forests,
with malaria incidence, and sickle cell anemia. Different cultural
practices in different parts of the world that increase suscepti-
bility to malaria are correlated with higher allele frequencies of
the HbS variant of normal adult hemoglobin that in heterozy-
gotes confers protection against malaria infection but in homo-
zygotes results in sickle cell anemia (21, 22).
Population genomics has been able to highlight areas of DNA
subject to selection, give an approximate date for selective
sweeps, and provide links to functionality. The recent expansion
of the genomic enterprise has revealed a great sway of recent
selection on the genomes of modern humans, particularly in
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: H.W. designed research, performed research, analyzed data, and
wrote the paper.
The author declares no conflict of interest.
This article is a PNAS Direct Submission.
1
Email: hwhitehe@dal.ca.
www.pnas.org/cgi/doi/10.1073/pnas.1620736114

populations derived following migration out of Africa (23). It has
been hypothesized that an accelerated rate of adapted substitu-
tions over the last 40,000 y is linked to cultural innovations and
consequent demographic group selection (24). These findings
led authors to propose gene–culture coevolution as the dominant
form of evolution in recent human history, because of the speed
at which genetic change has taken place and correlation with the
occupation of new niches and the establishment of new cultures
(10, 13). However, a temporal correlation, and even a plausible
scenario linking a particular selective gene to a cultural trait,
does not demonstrate culture as causal to genetic change. De-
mographic history can be a vital factor in driving the genetic
variation of contemporary populations. For instance, population
bottlenecks can cause founder effects in both genes and culture,
leading to correlations between specific genes and specific cul-
tures that are not the result of gene–culture coevolution. There
are, in fact, relatively few cases in which human culture has been
irrefutably shown to have caused genetic change (13). That is not
to say that gene–culture coevolution is rare, and indeed the
overall temporal correlation between the recent explosion in the
scope of human culture and an extremely accelerated rate of
selection on genes seems to point toward a major effect. How-
ever, in most instances it is very difficult to provide a compelling
empirical case that the interaction between culture and genetics
is causal.
Even harder to substantiate are the proposals for more general
effects of culture on patterns of human genetics. For instance,
models have shown that the remarkably low levels of human
genetic diversity could result from culturally mediated pop-
ulation structure (25) or from selectively important cultures being
transmitted in parallel with neutral genes (26). However, there is
no strong empirical evidence for either of these processes.
Although culture is clearly present in other species, it is
dwarfed in almost all respects by its impact on modern humans
(27–29). Thus, given the difficulties of demonstrating gene–cul-
ture coevolution in by far the most cultural species, there has
been little effort to look elsewhere. However, gene–culture co-
evolution can theoretically operate in simple cultural systems
(14), and there may be attributes of nonhuman species that could
accentuate the effects of culture on genes or make them more
discernible.
Birds offer one candidate taxon for nonhuman gene–culture
coevolution. In oscine species, birdsong is largely a socially
learned population phenomenon and thus represents culture
(30). For instance, populations of the sharp-beaked Galápagos
ground finch (Geospiza difficilis) have socially learned songs with
an important role in mating but respond fully only to songs from
their own island population (31). Thus, Grant and Grant suggest
that cultural transmission is driving speciation (31). However,
some other studies have found little evidence of links between
cultural and genetic evolution in birds (e.g., ref. 32).
Of all nonhuman species, culture seems particularly prevalent
and significant among Cetacea (3). The cetaceans include about
89 species of whales and dolphins, ranging in size from 1 to 30 m.
They use a range of habitat from large rivers to the deep ocean,
in polar, temperate, and tropical waters. They eat a wide range of
marine animals, from copepods to large whales. Cetaceans are
divided into the mysticetes or baleen whales that filter feed on
schools of prey using baleen, and the odontocetes that use
echolocation to locate and track single prey.
Although cetaceans are not easy to study, there is good evi-
dence for cultural transmission in song, migrations, foraging
behavior, social conventions, cooperative associations with hu-
mans, and play (3). Almost all cetacean species whose behavior
has been much studied show possibly, or likely, culturally ac-
quired behavior (3). For gene–culture coevolution, culture needs
to be quite stable—fads need not apply—and to affect fitness
directly or indirectly. When social groups with distinct cultures
Whitehead
COLLOQUIUM
PAPER
live sympatrically and these groups have differential reproductive
success, then gene–culture coevolution can be driven powerfully
by group selection (3, 27). Stable, sympatric social groups with
distinctive cultures are the hallmark of two large odontocete
species with matrilineal social systems, the killer whale (Orcinus
orca) and the sperm whale (Physeter macrocephalus) (3). In these
species, females, typically, and males, sometimes, spend their
lives grouped with their mothers while both are alive, forming
stable social units of about 10 animals. The social units are
themselves elements of higher-level matrilineally based social
tiers such as clans, communities, and ecotypes. Distinctive be-
havior, which is believed to be generally culturally transmitted,
maps onto the different elements of the social tiers (3). This
socio-cultural relationship sets up conditions in which gene–
culture coevolution could lead to neutral or functional genes
being found in different elements of the social system or to more
general effects such as speciation or reductions in genetic
diversity.
Here I examine four general hypotheses for gene–culture co-
evolution in whales and dolphins: (i) that culture has led to in-
cipient speciation of killer whale ecotypes; (ii) that culture has
driven the evolution of functional genes in killer whales; (iii) that
low levels of mtDNA diversity in five species of matrilineal
odontocete result from cultural hitchhiking; and (iv) that geo-
EVOLUTION
graphical patterns of mtDNA result from cultural behavior. As
with gene–culture coevolution in humans, evidence is not always
conclusive, and I will consider alternative mechanisms for the
observed patterns.
Ecotype Radiations in Killer Whales?
Although killer whales as a species have an extremely diverse
diet, ranging from herring to the largest whales, each killer whale
is typically a member of an ecotype, and most ecotypes are ex-
tremely specialized in what they eat and often in how they catch
it. Ecotypes are distinctive in a range of other behaviors, as well
as morphology (20). In each of the North Pacific, North Atlantic,
and Antarctic two or more ecotypes, each with a few hundred to
a few thousand animals, are sympatric but socially isolated (20).
Best known are the three North Pacific ecotypes: a salmon-
eating specialist ecotype known as “residents,” a mammal-
eating specialist or “transient” ecotype, and an “offshore” eco-
type, which feeds upon sharks (20). All three ecotypes cluster in
distinct mitochondrial and nuclear genome clades (33, 34) that
diverged a few hundred thousand years ago (35). We have less
complete information for the Antarctic, but there appear to
be five ecotypes respectively specializing on minke whales
(Balaenoptera acutorostrata) (ecotype A), seals (ecotype B1),
penguins (ecotype B2), and fish (ecotype C and perhaps eco-
type D) (20). These Antarctic ecotypes seem to have diverged
more recently than those in the North Pacific, and ecotypes B1,
B2, and C, at least, are part of a distinct and relatively shallow
Antarctic clade (34, 35). The distinctions between the ecotypes
have led Morin et al. to suggest that they should be considered
separate species or subspecies (36), but this proposal has not
been formally implemented.
Since the initial discovery of killer whale ecotypes, scientists
have speculated about how they may have arisen. Although some
issues are disputed, a consensus hypothesis about the funda-
mentals of ecotype formation has grown. This culturally driven
ecological speciation scenario has been articulated most clearly
by Riesch et al. (20), with support from more recent genomic
studies (35). Ecological speciation needs (i) an ecological source
of divergent selection; (ii) a form of reproductive isolation; and
(iii) a mechanism linking divergent selection to reproductive
isolation (37). The killer whale scenario is especially interesting
because culture seems to play a key role in all three components.
Within the best-studied killer whale ecotypes there is evidence
that different matrilineal social units can have distinctive ways of
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7815
life. For instance, the AT1 transient matriline of Prince William
Sound, Alaska has particular ways of capturing fast-swimming
Dall’s porpoises, Phocoenoides dalli (38). Such foraging methods
are almost certainly socially learned, primarily within matrilineal
units. As these foraging methodologies become more efficient
through cultural selection, they will tend to become more spe-
cialized, so that matrilines, or groups of socially connected matri-
lines, develop new ways of life. These matrilines are thought to be
the progenitors of the ecotypes (20). In concordance with this
scenario, new genomic studies indicate that ecotypes started with
small populations of related animals (35).
The second and third conditions of ecological speciation, the
emergence of reproductive isolation between ecotypes, are key
to the population structure of killer whales. Riesch et al. consider
four mechanisms through which cultural divergence could lead
to reproductive isolation among ecotypes (20). Most directly,
cultural xenophobia could restrict mating patterns. Vocal dia-
lects, which are socially learned and thus culture, are good
candidates (20). These dialects appear to set barriers to mating
with close associates, and, in accordance with a paradigm of
mating only with animals whose vocalizations are similar but not
too similar, also might prevent mating between proto-ecotypes
(20). Secondly, the cultural specializations that come to define
ecotypes would likely make between-ecotype dispersal difficult
and often functionally impossible. Killer whales are culturally
conservative (7); thus, for instance, adopting the specialized,
collaborative behavior of a seal-eater might be very hard for a
postweaning salmon-eater, who had already absorbed the con-
formist ways of her own lineage. Third, genetic drift within
ecotypes could lead to postzygotic infertility or reduced fitness of
hybrids. Such drift is particularly likely given the generally small
sizes, and even smaller founding bottlenecks, of the ecotypes (20,
35). Riesch et al.’s (20) final mechanism promoting reproductive
isolation among ecotypes is the selection on functional genes by
different cultural practices. New genomic studies have found
strong evidence for this selection (35), and this evidence is de-
scribed in the next section.
Although ecological speciation driven by cultural specializa-
tion is the favored hypothesis for ecotype formation and is
consistent with what is known of the behavior, ecology, genetics,
and genomics of killer whales, the hypothesis is not proven. An
alternative hypothesis is that nonecological cultural divergence
might have predated the ecological specializations (20). For in-
stance, proto-ecotypes could have been based on vocalization
patterns of groups of killer whales that had similar diets, and
then, once isolated, these groups could have begun to specialize
ecologically. It is hard to envisage an alternative ecotype-
generating scenario that does not have a cultural component.
Gene–Culture Coevolution of Functional Genes
We have evidence that culture has influenced the evolution of
functional genes in killer whales, and killer whale ecotypes
provide the substrate for these inferences. In fewer than
10,000 generations, killer whales have radiated from a single
ancestral lineage to colonize regions from the Arctic to the
Antarctic and all waters in between (34, 35). Until humans
reached Antarctica, killer whales were the most widely distrib-
uted mammals and likely were subject to a diverse range of se-
lective pressures associated with different habitats. The stable
cultural traditions of killer whale ecotypes provide a rare and
attractive system for investigating gene–culture coevolution in
functional genes. However, as with most genomically derived
putative linkages between genes and culture in humans (10),
there is no definitive proof that cultural practice has affected
selection pressure on the identified genes.
Genomic studies have looked at two particular contrasts: be-
tween fish-eating and mammal-eating ecotypes and between
temperate and Antarctic ecotypes. Because diet in killer whales
7816 | www.pnas.org/cgi/doi/10.1073/pnas.1620736114
is almost certainly socially learned, and the ecotypes themselves
likely arise from this cultural process (see previous section), the
first contrast can reasonably be deemed cultural; the second is
less directly so, but culture likely underpins almost all behavioral
differences between ecotypes. Niche construction is the modifi-
cation of the environment by an organism, so that the changes
influence natural selection (39). In their review of gene–culture
coevolution, Laland et al. (10) emphasized that the defining
characteristic of niche construction is the modification of selec-
tion pressures and that this modification could be achieved
through processes such as migration, dispersal, and habitat se-
lection. Niche construction then provides a link between the
cultures of killer whale ecotypes and their habitats.
Recent population genomic studies comparing the fish-eating
resident and mammal-eating transient ecotypes of the North
Pacific found a signature consistent with selection on genes as-
sociated with diet (35, 40). The regions of the genome that most
differentiated members of the resident ecotype contained genes
associated with digestive tract formation during developmental
stages and carboxylic ester hydrolase activity, including the hy-
drolysis of long-chain fatty acid esters (35). A signature of se-
lection on the CBS gene, which has a role in the methionine
cycle, was found in the mammal-eating transient ecotype. In-
terestingly, a signature of selection on a different set of genes
that play a key role in the methionine cycle was also found in the
predominantly mammal-eating B1 ecotype from the Antarctic
(35). Methionine is an essential amino acid that cannot be pro-
duced by the body and must be obtained through dietary intake
of protein (41). Foote et al. (35) hypothesized that the more
sporadic intake of large amounts of dietary protein by mammal-
eating killer whales may exert selection pressures on the methi-
onine cycle different from those experienced with a more tem-
porally consistent intake of protein by fish-eating killer whales.
Similar to the parallel examples of lactose tolerance and malaria
resistance in different human populations, the signatures of se-
lection on the same pathway in two independently derived
mammal-eating ecotypes supports the notion that selection may
be associated with culturally acquired ecological niches.
Although the killer whales in the North Pacific inhabit rela-
tively temperate waters, those found in Antarctic waters arguably
inhabit the most extreme environment of this species’ range. The
three Antarctic ecotypes included in a recent population geno-
mics study share a recent common ancestor (34, 35). Targets
of selection along the evolutionary branch to this ancestral
Antarctic population includes genes associated with adipose
tissue development; interestingly, the same process has been
under selection in the polar bear, Ursus maritimus, compared
with the brown bear, Ursus arctos, and are suggestive of a link to
adaptation to a colder climate (42).
The FAM83H gene, which regulates the keratin cytoskeleton,
including that in epithelial cells (43), is one of the genes most
differentiated in the Antarctic and North Pacific ecotypes, con-
taining four fixed nonsynonymous substitutions in the Antarctic
ecotypes (35). Antarctic killer whales are thought to have a slow
regeneration of skin cells because of the thermal constraints of
allowing blood to circulate to the outer epithelial cells (44). This
slow skin regeneration is evidenced from the build-up of dia-
toms, giving Antarctic killer whales a yellowish hue compared
with North Pacific killer whales (45). Genes associated with skin
regeneration, such as FAM83H, therefore may be under selec-
tion because of this thermal constraint. Just as human cultural
practices frequently buffer selective pressures from harsh envi-
ronments (10), Antarctic killer whales seem to use behavioral
adaptation to mitigate the impact of the Antarctic climate upon
skin regeneration. Satellite tagging has revealed that these
whales make regular round trips to warm subtropical waters; when
they return they have lost the yellow hue of their diatom film,
indicating they have regenerated the outer epithelial cells (44).
Whitehead

Several phenotypic distinctions between ecotypes appear
functional, given the ecotypes’ different cultural behaviors, and
are likely genetically based, but the candidate genes for these
distinctions have not yet been identified. Most obviously, eco-
types that eat mammals are generally larger than those that eat
fish (20). This difference in size could be related to the nutri-
tional values and availability of the different foods, but it also
may have a genetic component. In the North Pacific, mammal-
eating transients seem to have more robust mouth parts than
fish-eating residents (20). The type B1 Antarctic killer whales use
a particularly extreme form of coordination, synchronous fluke
beats, to create waves that knock seals off ice flows (46). They
also have relatively larger white eye-patches than other ecotypes.
Are these large eyepatches coincidence or the result of gene–
culture coevolution?
Cultural Hitchhiking in Matrilineal Whales
This section summarizes information in ref. 47. As molecular
genetics was applied to cetacean species during the 1990s, an
unexpected pattern appeared. The four whale species known to
have matrilineal social systems, the killer whale, sperm whale,
and two pilot whale species (Globicephala macrorhynchus and
Globicephala melas), possessed much lower diversities of the
control region of their mtDNA than other cetacean species with
comparable population sizes or latitudinal ranges (19). In the
nearly two decades since then, many additional estimates of
cetacean control region diversity have been published, for other
species (including another presumed matrilineal species, the
false killer whale, Pseudorca crassidens), with larger and more
geographically dispersed samples, and greater coverage of the
genome (47). The pattern still holds (Fig. 1). The range-wide
mtDNA diversities in the matrilineal species are ∼29.8% of
that of nonmatrilineal species with similar latitudinal ranges
(47). For regional estimates the ratio is 16.6%.
Low genetic diversity usually invokes discussion of bottlenecks
and selection. Both these default mechanisms have been used to
explain the low mtDNA diversity of the matrilineal species of
Cetacea (48–53). However, neither bottlenecks nor selection link
the remarkably low diversity of the species to their matrilineal
social systems. Furthermore, contrary to expectations from bot-
tlenecks, the diversity of nuclear microsatellites is not obviously
reduced in the matrilineal whales (although this result may be
partially explained by the greater effective population size and
higher mutation rates of microsatellites compared with mtDNA)
(47). In sperm whales, Alexander et al. (48) found no evidence
for selection in the control region relative to other regions of the
Fig. 1. Mean mtDNA nucleotide diversity (across estimates for a species
with n >100 covering >25% of the species’ range or at least one ocean basin)
against the latitudinal range of cetacean species. Nonmatrilineal species are
indicated by a plus sign, and matrilineal species are indicated by circles. Gma,
short-finned pilot whale; Gme, long-finned pilot whale; Oo, killer whale; Pc,
false killer whale; Pm, sperm whale. Adapted from ref. 47.
Whitehead
COLLOQUIUM
PAPER
mitochondrial genome, but this finding does not exclude a se-
lective sweep originating in these other regions.
More satisfying than bottlenecks or selection are scenarios
postulating important demographic processes operating primar-
ily at the level of the matrilineal group, thus reducing the ef-
fective population size to something like the number of groups
and thus lowering the expected genetic diversity (54–56). How-
ever, agent-based models found no realistic circumstances in
which the demography of matrilineal groups could reduce ge-
netic diversity, unless fitness differentials between groups were
heritable (57). If there is a heritable component to fitness within
matrilineal groups, then the conditions for reduced genetic di-
versity become much relaxed (57). This heritability could happen
with selection on the mitochondrial genome, for instance in re-
sponse to mitochondrial disease or deep diving (47, 48), but
there is no need for a species to have a matrilineal social system
for selection on genes to reduce mtDNA diversity, so this ex-
planation is unsatisfying.
The heritable component to fitness does not need to be ge-
netic; it just needs to be transmitted matrilineally. Killer and
sperm whales are well known for their matrilineally transmitted
cultures (3). Cultural hitchhiking is the hypothesis that these
matrilineally transmitted cultures, with fitness differentials, have
reduced diversity of the mitochondrial genes that are being
EVOLUTION
transmitted in parallel (19). Thus cultural hitchhiking is a form of
gene–culture coevolution. Agent-based models have shown that
cultural hitchhiking can work in circumstances that seem realistic
for the matrilineal whales (57) but do not necessarily imply that
cultural hitchhiking is behind their low mtDNA diversity.
However, it is most parsimonious to assume that a factor
common to the matrilineal species has led to reduced diversity.
Because that factor does not seem to be the direct influence of
matrilineality itself (57), I suspect that it is culture. Matrilineal
social systems are particularly good substrates for the evolution
of stable, group-specific cultures (3), and stable, group-specific
cultures are the prerequisite for cultural hitchhiking (57). How-
ever, culture may affect genetic diversity through different paths
in different species (18), as illustrated by the two best-known
matrilineal species, sperm and killer whales.
Female and immature sperm whales use tropical and sub-
tropical waters, where they live in matrilineally based social units
(58). These units, in turn, are members of coda clans (59). The
coda clans have distinctive dialects, movements, microhabitat
use, and social behavior, as well as differential reproductive
success (59–62). The clans are sympatric but do not associate
with one another; they show no differences in nuclear micro-
satellites but do have distinct mtDNA haplotype distributions
(63, 64). Thus, sperm whales fit the classic cultural hitchhiking
scenario well, with clans being the cultural groups under selec-
tion, and cultural hitchhiking is the most parsimonious expla-
nation for their low mitochondrial diversity. A selective sweep in
the mitochondrial genome is also consistent with available re-
sults on sperm whales (48) but does not obviously provide a link
with matrilineality.
Although killer whales are also matrilineal, their subdivision
into ecotypes adds a layer of additional possible drivers of low
genetic diversity. Population subdivision itself tends to reduce
genetic diversity (65). The highly specialized ways of life of many
of the ecotypes may make them particularly vulnerable to ex-
tirpation, removing characteristic mtDNA haplotypes in the
process, and thus reducing diversity (65). The ecotypes have very
different culturally transmitted behavior—one of the conditions
for cultural hitchhiking—but their lifestyles are so different that
it is hard to see how they would often be in competition (how-
ever, see ref. 66 for a scenario of indirect competition between
ecotypes), so cultural hitchhiking at the ecotype level seems an
incomplete explanation. Processes that reduce diversity within
ecotypes would also affect the overall diversity of killer whales.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7817
Genomic studies give strong support for bottleneck effects dur- grounds (74). The wintering grounds of southern right whales
ing the founding of ecotypes, as well as selection driven by (Eubalaena australis) have distinct mtDNA haplotype distribu-
ecotype-specific cultural traits (35, 67). Cultural hitchhiking also tions, presumably because of maternal migratory fidelity, and
could operate on socio-cultural groups within ecotypes, such as Carroll et al.’s (75) analysis of gene distributions and stable
the communities of resident ecotype killer whales (68). However, isotopes suggests that the migratory cultures also affect genetic
at this time, our evidence for a cultural factor behind the low structure on feeding grounds.
mtDNA diversity of killer whales lies primarily in its role in Such cases of cultural migration routes and habitat use may
setting up the ecotypes and driving selection within them (see not be uncommon among animal species such as birds and un-
previous sections). In this respect, the killer whale situation gulates, although many traditional migrations are insufficiently
corresponds quite closely to Premo and Hublin’s (25) suggestion faithful to set up clear genetic patterning (76). Such patterns can
that culturally transmitted barriers to mating and dispersal may also be found without culture, for instance salmon using chem-
have set up conditions for the reduction in human genetic di- ical traces to home to their native streams, and so set up clear
versity through bottlenecks and selection. geographical genetic distinctions. Using senses alone to home to
We know much less about the three other matrilineal species: native areas consistently is an alternative explanation for the
the short-finned pilot whale, the long-finned pilot whale, and the geographical patterns of mtDNA in cetaceans. However, given
false killer whale. Although the evidence for matrilineal social the social nature of these species, and especially the long
systems is quite good (69–71), we have not identified suitable mother–calf relationship that typically includes several migra-
socio-cultural groups on which cultural hitchhiking might oper- tions, the cultural explanation is more parsimonious.
ate, nor do we know whether such groups are barriers to mating, Geographical patterns of gene distributions can also be driven
as with the killer whale ecotypes, or, alternatively, share nuclear by cultural inheritance of traits that are not themselves about
genes through male dispersal, as with sperm whale coda clans. migration but affect habitat use secondarily. The population of
For these species we also have little idea as to what important bottlenose dolphins (Tursiops spp.) in western Shark Bay, Aus-
matrilineally based cultural traits either provide selection dif- tralia, shows very pronounced distributional differences for three
ferential or define population structure. However, the low main mtDNA haplotypes over scales of a few kilometers (Fig. 2).
mtDNA diversity in these matrilineal species is unlikely to be a There are no barriers to movement between the primary ranges
coincidence; culture is probably involved in some way. of the different haplotypes, and individuals could move between
Culture and the Geographical Distribution of Genes these ranges in less than an hour. Kopps et al. (77) relate the
geographical patterns of mtDNA to foraging methods learned
The geography of the genes of H. sapiens has been thoroughly
from the mother. One particular foraging type, “sponging,” in
redistributed by culture. Both the neutral genes characteristic of
which a sponge is used as a tool to forage in relatively deep waters,
northern Europeans and some functional ones (e.g., those for
is known to be learned almost entirely from the mother (78, 79),
blond hair) are much more widely dispersed than would have
and in Kopps et al.’s (77) sample is restricted to just one haplotype
been expected a millennium ago. This dispersal is a result of their
characteristic of the deeper waters (haplotype E in Fig. 2).
linkage to the knowledge of oceanic navigation and weapons
As illustrated by human colonization, culture can potentially af-
technology, as well as a drive to explore, all culturally acquired. In
fect the dynamics of gene distributions. Sperm whale populations
several cases the geographical distribution of cetacean genes has
a strong cultural imprint. This geographical pattern happens are subdivided into cultural coda clans that are matrilineally based,
most obviously when migration routes are learned matrilineally,
and so migratory destinations may have distinctive mtDNA haplo-
type distributions. However, there are also more complex situations
in which the link between geography and culture is not as direct.
Most beluga whale (Delphinapterus leucas) populations make
seasonal migrations between shallow summering areas, where
the young are born, and deeper-water wintering grounds. The
young belugas are presumed to learn these migration routes
from their mothers and repeat them sufficiently faithfully that
summering grounds that were emptied by intense whaling remain
almost unoccupied several generations later, even though they
border the migration routes of other belugas whose summering
grounds were not intensely whaled. So strong are these matri-
lineal cultural traditions that a population of belugas summering
in western Hudson Bay, Canada, has mtDNA haplotypes traced
to a Pacific refugium during the Wisconsin glaciation, whereas
those summering in eastern Hudson Bay have an almost entirely
different distribution of haplotypes that are presumed to derive
from an Atlantic refugium (72). This distinction has been main-
tained over several thousand generations even though there are no
physical barriers between the summering grounds and the migra-
tion routes and wintering grounds of the two populations, where the
animals mate, overlap so that there is mating between them (73).
Baleen whales also show geographical genetic structure in
mtDNA resulting from maternal learning of migration destina-
tions. For instance, humpback whales (Megaptera novaeangliae) in
the North Atlantic and North Pacific consistently use particular Fig. 2. Segregation of bottlenose dolphin haplotypes by habitat in western
summer feeding grounds in temperate latitudes but breed in Shark Bay, Australia. Survey colors represent haplotypes of dolphins. Each
common tropical wintering areas. Calves use the same sum- sighting of a sampled dolphin was plotted. Green represents land, white
mering grounds as their mothers, but not necessarily their fa- represents shallow water (<10 m), and gray represents deeper water
thers, setting up mtDNA haplotype distinctions among feeding (≥10 m). Adapted from ref. 77.

7818 | www.pnas.org/cgi/doi/10.1073/pnas.1620736114 Whitehead


show distinct mtDNA haplotype distributions, and are often sym-
patric (59, 64). Between 1985 and 1995 there were two principal
cultural coda clans off the Galápagos Islands, but in 2013 and
COLLOQUIUM
PAPER
2014 these had been replaced by two different clans previously
primarily found elsewhere in the Pacific (80). It seems that clan
membership was a primary factor governing the movements of the

Table 1. Potential cases of gene–culture coevolution in whales and dolphins

Strength of evidence
Noncultural for gene–culture
Effect Species Brief description mechanisms/comments coevolution Ref.

Speciation
Ecotype radiations Killer whale A deep division of killer whales Perhaps other cultural Strong (20)
into ecotypes driven by foraging behavior initiated
cultures ecotype formation
Gene-culture coevolution of functional genes
Genes related to Killer whale Differ between mammal-eating Possible (35)
digestive tract and fish-eating ecotypes
Genes related to Killer whale Differ between mammal-eating Independent contrasts in Strong (35)
methionine cycle and fish-eating ecotypes North Pacific and
Antarctic
Genes related to Killer whale Differ between Antarctic and Cultural role not direct Possible (35)
adipose tissue temperate ecotypes
development
Genes related to skin Killer whale Differ between Antarctic and Cultural role not direct Possible (35)

EVOLUTION
regeneration temperate ecotypes
Size Killer whale Differences between mammal- No genes identified; could Weak (20)
eating, fish-eating and bird- be environmental
eating ecotypes
Robustness of mouth Killer whale Differ between mammal-eating No genes identified; Weak (20)
parts and fish-eating ecotypes preliminary study
Coloration Killer whale Most enlarged white eye-patch in No genes identified Weak (20)
ecotype that uses most
coordinated behavior
Cultural hitchhiking in matrilineal whales
Low mtDNA diversity Sperm whale mtDNA has hitchhiked on selective Clans are strong candidate Possible (57)
cultural traits transmitted in for cultural groups; could
parallel be caused by selective
sweep in mtDNA or
bottleneck (less supported)
Low mtDNA diversity Killer whale Ecotype population structure, plus Culture very likely to have Strong (57)
additional cultural effects, role(s); several possible
bottlenecks and/or selection routes; not necessarily
have reduced diversity classic cultural
hitchhiking
Low mtDNA diversity Pilot, false mtDNA has hitchhiked on selective Little evidence other than Weak (57)
killer whales cultural traits transmitted in correlation between
parallel matrilineal social system
and low mtDNA diversity
Culture and gene geography
Distinctive mtDNA Beluga whale Belugas follow their mothers on Born in summer, could be Strong (72)
distributions of first migrations purely environmental
summering areas sensing, but unlikely
Distinctive mtDNA Humpback Humpbacks follow their mothers Born in winter, so, because Strong (74)
distributions of whale on first migrations first visit to summering
summering areas grounds is with mother,
this is social learning
Distinctive mtDNA Southern Right whales follow their mothers As born in winter, could be Strong (75)
distributions of right whale on first migrations purely environmental
wintering areas sensing, but unlikely
Distinctive mtDNA Bottlenose Dolphins learn specialized feeding Social learning of some Strong (77)
distributions at dolphin techniques from their mothers, foraging techniques well
small scales leading to habitat selection established
Change in mtDNA Sperm whale Clans redistributed themselves Genetic change inferred; Possible (80)
distribution caused over 30 y, changing mtDNA redistribution could have
by cultural clan distribution been independent of
redistribution clan membership, but
unlikely

Whitehead PNAS | July 25, 2017 | vol. 114 | no. 30 | 7819

sperms, and so the changes in their mitochondrial gene distributions
(80). Thus, culture can have an important role in structuring the
geographical dynamics of a population and, consequently, its genes.
Discussion
Given the difficulty of studying whale and dolphin behavior,
there is a remarkable amount of evidence for cetacean culture
(3). Some of this culture has driven gene evolution. The extent
and nature of gene–culture coevolution in cetaceans is very un-
certain, as it is in humans (13). Table 1 summarizes the ideas and
evidence. Although the strength of evidence is variable (Table 1),
culture seems to have driven the evolution of both neutral and
functional cetacean genes in several quite different ways. In all
cases, socially learned behavior affects how individuals interact
with their environment or with each other and thus affects the
transmission patterns or selection pressures on genes. Included is
the relatively prosaic mechanism of young animals faithfully
following their mothers on their first migrations and using these
learned migrations for the rest of their lives to set up mtDNA
distinctions between geographical habitats. However, there is
also evidence for unusual processes: culture driving ecological
speciation and cultural hitchhiking reducing genetic diversity.
There are strong parallels with what is known regarding human
gene–culture coevolution and evidence for the evolution of func-
tional genes in different killer whale ecotypes with distinct cultural
behavior (35). However, the clear phylogenetic and cultural divi-
sions between killer whale ecotypes make it easier, in some re-
spects, to pin down gene–culture coevolution in these species than
in the messier trajectory of human evolutionary history.
In 1996 Feldman and Laland (11) suggested that social
learning was rarely stable enough in nonhumans to support long-
standing cultural traditions upon which selection could act.
Documenting cultural stability over generations is not easy in
nonhumans. However, there is archaeological evidence for over
4,000 y of a particular type of nut-cracking by chimpanzees (Pan
troglodytes) at one site in west Africa (81). Faithful transmission
of cultures over many generations is implied by many of the
patterns in cetacean genetics documented in Table 1. Examples
include the strong mtDNA differences between beluga whales on
different summering grounds and the linkage of functional genes
with ecotype behavior in killer whales.
Although evidence for culture is present across the range of
cetacean species whose behavior has been studied (3), there is
considerable variation in how that culture is expressed. One
constant, however, is a strong mother–calf bond, lasting at least
many months and often much longer. This relationship is an
excellent conduit for social learning, and stable cultures (3).
Migration routes, dialects, and foraging strategies are often
learned from the mother and so are transmitted in parallel with
mtDNA, potentially leading to gene–culture coevolution. How-
ever in some of the larger odontocetes the matrilineal influence
is much stronger. Females, and sometimes males, typically spend
their lives with their mothers and other maternal relatives. Not
only does this association increase the potential learning period
and number of models, but the behavior within these matrilineal
social units may become very stereotyped, through processes
such as conformism and symbolic marking (82). These factors
increase the range of ways culture can affect genes, especially the
matrilineally transmitted mitochondrial genes. So there is a
1. Laland KN, et al. (2015) The extended evolutionary synthesis: Its structure, assump-
tions and predictions. Proc Roy Soc B 282:20151019.
2. Maynard Smith J (1989) Evolutionary Genetics (Oxford Univ Press, Oxford, UK).
3. Whitehead H, Rendell L (2015) The Cultural Lives of Whales and Dolphins (Univ of
Chicago Press, Chicago).
4. Heyes CM (1994) Social learning in animals: Categories and mechanisms. Biol Rev
Camb Philos Soc 69:207–231.
5. Hoppitt W, Laland KN (2013) Social Learning: An Introduction to Mechanisms,
Methods, and Models (Princeton Univ Press, Princeton, NJ).
7820 | www.pnas.org/cgi/doi/10.1073/pnas.1620736114
particular emphasis on killer whales, sperm whales, and the other
matrilineal odontocetes. When matrilineal social structures with
distinct and stable cultures cease exchanging mates, then nuclear
genes become targets for gene–culture coevolution. Killer whales
are the only species in which we know that sympatric matrilines
have become reproductively isolated, and it is no coincidence
that killer whales feature so prominently in Table 1. However,
there may be other cetacean species in which mating barriers,
either physical or social, between groups with different cultures
could allow gene–culture coevolution on nuclear genes.
Some of the processes discussed here, particularly the cultural
inheritance of migration routes driving the geography of genes,
may be present in other nonhuman species. In species such as
chimpanzees that have important stable foraging cultures (6, 81)
the evolution of functional genes might be driven by culture; for
instance a group that uses a particular tool might evolve senses
or musculature to make better use of it. In birds, culturally
inherited songs can demarcate social relationships and mating
opportunities, perhaps promoting speciation (31). However, this
process is somewhat different from the culturally driven eco-
logical speciation of killer whales, and the killer whale scenario is
likely rare (20) and maybe unique. Cultural hitchhiking pro-
ducing low genetic diversity may also be rare, because it requires
quite stable groups of relatives with quite stable functionally im-
portant cultures producing consistent fitness differentials; cultural
hitchhiking works much more efficiently if the cultural groups are
sympatric (26, 57).
This review has focused on culture causing differences in genes
within species. However, when cultural transmission becomes
important, there may be species-wide selection for characteristics
that make this transmission more effective. Humans as well as
some species of Cetacea have remarkably large brains, as well as
menopause. There have been suggestions that both traits could be
adaptive in complex cultural systems (3, 83).
The prospects for our understanding of gene–culture co-
evolution in cetaceans, as well as other nonhumans, are good,
but uneven. On the gene side, recent progress in sequencing
technology has allowed the transition from targeting candidate
genes to conducting genome-wide or whole-genome compari-
sons. This advance makes finding gene–culture coevolution
much more likely, because studies are not limited to identifying
genetic variants associated with a particular cultural behavior a
priori. As costs go down and expertise builds, genomic studies
are being used increasingly on cetaceans and are being deployed
with added depth and scope (84). In contrast, studies of cetacean
social structure and culture are naturally slow and painstaking
and are not getting any cheaper. Some shortcuts are being used,
such as inferring diet through stable isotopes and automated
processing of recordings of vocalizations, but we also require
much more detailed observation at sea. For major progress in
understanding gene–culture coevolution, the socio-cultural side
needs innovation, commitment, and investment.
ACKNOWLEDGMENTS. I thank Andrew Foote, who helped write and
validate parts of this article, particularly the parts involving genomics,
contributed alternative explanations, and also reviewed the article; two
anonymous reviewers who provided detailed and constructive comments;
and Francisco J. Ayala, Marcus W. Feldman, Kevin N. Laland, and Andrew
Whiten for organizing such an interesting colloquium.
6. Whiten A, et al. (1999) Cultures in chimpanzees. Nature 399:682–685.
7. Barrett-Lennard L (2011) Killer whale evolution: Populations, ecotypes, species, Oh
my! J Am Cet Soc 40:48–53.
8. Sapolsky RM, Share LJ (2004) A pacific culture among wild baboons: Its emergence
and transmission. PLoS Biol 2:E106.
9. Cavalli-Sforza LL, Feldman MW, Chen KH, Dornbusch SM (1982) Theory and obser-
vation in cultural transmission. Science 218:19–27.
10. Laland KN, Odling-Smee J, Myles S (2010) How culture shaped the human genome:
Bringing genetics and the human sciences together. Nat Rev Genet 11:137–148.
Whitehead

11. Feldman MW, Laland KN (1996) Gene-culture coevolutionary theory. Trends Ecol Evol
11:453–457.
12. Cavalli-Sforza LL, Feldman MW (1981) Cultural Transmission and Evolution: A
Quantitative Approach (Princeton Univ Press, Princeton, NJ).
13. Richerson PJ, Boyd R, Henrich J (2010) Colloquium paper: Gene-culture coevolution in
the age of genomics. Proc Natl Acad Sci USA 107:8985–8992.
14. Feldman MW, Cavalli-Sforza LL (1976) Cultural and biological evolutionary processes,
selection for a trait under complex transmission. Theor Popul Biol 9:238–259.
15. Creanza N, Feldman MW (2016) Worldwide genetic and cultural change in human
evolution. Curr Opin Genet Dev 41:85–92.
16. Creanza N, Kolodny O, Feldman MW (2017) Cultural evolutionary theory: How culture
evolves and why it matters. Proc Natl Acad Sci USA 114:7782–7789.
17. Tishkoff SA, et al. (2007) Convergent adaptation of human lactase persistence in
Africa and Europe. Nat Genet 39:31–40.
18. Premo L (2012) Hitchhiker’s guide to genetic diversity in socially structured pop-
ulations. Curr Zool 58:287–297.
19. Whitehead H (1998) Cultural selection and genetic diversity in matrilineal whales.
Science 282:1708–1711.
20. Riesch R, Barrett-Lennard LG, Ellis GM, Ford JKB, Deecke VB (2012) Cultural traditions
and the evolution of reproductive isolation: Ecological speciation in killer whales?
Biol J Linn Soc Lond 106:1–17.
21. Flint J, et al. (1986) High frequencies of alpha-thalassaemia are the result of natural
selection by malaria. Nature 321:744–750.
22. Piel FB, et al. (2010) Global distribution of the sickle cell gene and geographical
confirmation of the malaria hypothesis. Nat Commun 1:104.
23. Fan S, Hansen ME, Lo Y, Tishkoff SA (2016) Going global by adapting local: A review
of recent human adaptation. Science 354:54–59.
24. Hawks J, Wang ET, Cochran GM, Harpending HC, Moyzis RK (2007) Recent accelera-
tion of human adaptive evolution. Proc Natl Acad Sci USA 104:20753–20758.
25. Premo LS, Hublin JJ (2009) Culture, population structure, and low genetic diversity in
Pleistocene hominins. Proc Natl Acad Sci USA 106:33–37.
26. Whitehead H, Richerson PJ, Boyd R (2002) Cultural selection and genetic diversity in
humans. Selection 3:115–125.
27. Richerson PJ, Boyd R (2005) Not by Genes Alone: How Culture Transformed Human
Evolution (Univ of Chicago Press, Chicago).
28. Whiten A (2011) The scope of culture in chimpanzees, humans and ancestral apes.
Philos Trans R Soc Lond B Biol Sci 366:997–1007.
29. Whiten A, Ayala FJ, Feldman MW, Laland KN (2017) The extension of biology through
culture. Proc Natl Acad Sci USA 114:7775–7781.
30. Catchpole CK, Slater PJB (2008) Bird Song: Biological Themes and Variations (Cam-
bridge Univ Press, Cambridge, UK).
31. Grant BR, Grant PR (2002) Simulating secondary contact in allopatric speciation: An
empirical test of premating isolation. Biol J Linn Soc Lond 76:545–556.
32. Wright TF, Wilkinson GS (2001) Population genetic structure and vocal dialects in an
amazon parrot. Proc Biol Sci 268:609–616.
33. Moura AE, et al. (2015) Phylogenomics of the killer whale indicates ecotype di-
vergence in sympatry. Heredity (Edinb) 114:48–55.
34. Morin PA, et al. (2015) Geographic and temporal dynamics of a global radiation and
diversification in the killer whale. Mol Ecol 24:3964–3979.
35. Foote AD, et al. (2016) Genome-culture coevolution promotes rapid divergence of
killer whale ecotypes. Nat Commun 7:11693.
36. Morin PA, et al. (2010) Complete mitochondrial genome phylogeographic analysis of
killer whales (Orcinus orca) indicates multiple species. Genome Res 20:908–916.
37. Rundle HD, Nosil P (2005) Ecological speciation. Ecol Lett 8:336–352.
38. Matkin C, Durban J (2011) Killer whales in Alaskan waters. J Am Cet Soc 40:24–29.
39. Laland KN, Odling-Smee J, Feldman MW (2000) Niche construction, biological evo-
lution, and cultural change. Behav Brain Sci 23:131–146, discussion 146–175.
40. Moura AE, et al. (2014) Population genomics of the killer whale indicates ecotype
evolution in sympatry involving both selection and drift. Mol Ecol 23:5179–5192.
41. Finkelstein JD (1990) Methionine metabolism in mammals. J Nutr Biochem 1:228–237.
42. Liu S, et al. (2014) Population genomics reveal recent speciation and rapid evolu-
tionary adaptation in polar bears. Cell 157:785–794.
43. Forman OP, et al. (2012) Parallel mapping and simultaneous sequencing reveals de-
letions in BCAN and FAM83H associated with discrete inherited disorders in a do-
mestic dog breed. PLoS Genet 8:e1002462.
44. Durban JW, Pitman RL (2012) Antarctic killer whales make rapid, round-trip move-
ments to subtropical waters: Evidence for physiological maintenance migrations? Biol
Lett 8:274–277.
45. Pitman RL, Ensor P (2003) Three forms of killer whales (Orcinus orca) in Antarctic
waters. J Cetacean Res Manag 5:131–139.
46. Pitman RL, Durban JW (2012) Cooperative hunting behavior, prey selectivity and prey
handling by pack ice killer whales (Orcinus orca), type B, in Antarctic Peninsula waters.
Mar Mamm Sci 28:16–36.
47. Whitehead H, Vachon F, Frasier TR (2017) Cultural hitchhiking in the matrilineal
whales. Behav Genet 47:324–334.
48. Alexander A, et al. (2013) Low diversity in the mitogenome of sperm whales revealed
by next-generation sequencing. Genome Biol Evol 5:113–129.
49. Lyrholm T, Leimar O, Gyllensten U (1996) Low diversity and biased substitution pat-
terns in the mitochondrial DNA control region of sperm whales: Implications for es-
timates of time since common ancestry. Mol Biol Evol 13:1318–1326.
Whitehead
COLLOQUIUM
PAPER
50. Moura AE, et al. (2014) Killer whale nuclear genome and mtDNA reveal widespread
population bottleneck during the last glacial maximum. Mol Biol Evol 31:1121–1131.
51. Hoelzel AR, et al. (2002) Low worldwide genetic diversity in the killer whale (Orcinus
orca): Implications for demographic history. Proc Biol Sci 269:1467–1473.
52. Janik VM (2001) Is cetacean social learning unique? Behav Brain Sci 24:337–338.
53. Mesnick SL, et al. (1999) Culture and genetic evolution in whales. Science 284:2055a.
54. Amos W (1999) Culture and genetic evolution in whales. Science 284:2055a.
55. Tiedemann R, Milinkovitch M (1999) Culture and genetic evolution in whales. Science
284:2055a.
56. Siemann LA (1994) Mitochondrial DNA Sequence Variation in North Atlantic Long-
Finned Pilot Whales, Globicephala melas. PhD thesis (Massachusetts Institute of
Technology, Cambridge, MA).
57. Whitehead H (2005) Genetic diversity in the matrilineal whales: Models of cultural
hitchhiking and group-specific non-heritable demographic variation. Mar Mamm Sci
21:58–79.
58. Christal J, Whitehead H, Lettevall E (1998) Sperm whale social units: Variation and
change. Can J Zool 76:1431–1440.
59. Rendell LE, Whitehead H (2003) Vocal clans in sperm whales (Physeter macro-
cephalus). Proc Biol Sci 270:225–231.
60. Marcoux M, Rendell L, Whitehead H (2007) Indications of fitness differences among
vocal clans of sperm whales. Behav Ecol Sociobiol 61:1093–1098.
61. Cantor M, Whitehead H (2015) How does social behavior differ among sperm whale
clans? Mar Mamm Sci 31:1275–1290.
62. Whitehead H, Rendell L (2004) Movements, habitat use and feeding success of cul-
tural clans of South Pacific sperm whales. J Anim Ecol 73:190–196.
63. Whitehead H (2003) Sperm Whales: Social Evolution in the Ocean (Univ of Chicago
Press, Chicago).
64. Rendell L, Mesnick SL, Dalebout ML, Burtenshaw J, Whitehead H (2012) Can genetic
differences explain vocal dialect variation in sperm whales, Physeter macrocephalus?
EVOLUTION
Behav Genet 42:332–343.
65. Whitlock MC, Barton NH (1997) The effective size of a subdivided population.
Genetics 146:427–441.
66. Baird RW, Abrams PA, Dill LM (1992) Possible indirect interactions between transient
and resident killer whales: Implications for the evolution of foraging specializations in
the genus Orcinus. Oecologia 89:125–132.
67. Foote AD, et al. (2011) Positive selection on the killer whale mitogenome. Biol Lett 7:
116–118.
68. Ford JKB, Ellis GM, Balcomb KC (2000) Killer Whales (Univ of British Columbia,
Vancouver).
69. Amos B, Schlötterer C, Tautz D (1993) Social structure of pilot whales revealed by
analytical DNA profiling. Science 260:670–672.
70. Heimlich-Boran JR (1993) Social Organization of the Short-Finned Pilot Whale Glo-
bicephala macrorhynchus, with Special Reference to the Comparative Social Ecology
of Delphinids. PhD thesis (Cambridge University, Cambridge, UK).
71. Baird RW, et al. (2008) False killer whales (Pseudorca crassidens) around the main
Hawaiian Islands: Long-term site fidelity, inter-island movements, and association
patterns. Mar Mamm Sci 24:591–612.
72. Brown Gladden JG, Ferguson MM, Clayton JW (1997) Matriarchal genetic population
structure of North American beluga whales Delphinapterus leucas (Cetacea: Mono-
dontidae). Mol Ecol 6:1033–1046.
73. Turgeon J, Duchesne P, Colbeck GJ, Postma LD, Hammill MO (2012) Spatiotemporal
segregation among summer stocks of beluga (Delphinapterus leucas) despite nuclear
gene flow: Implication for the endangered belugas in eastern Hudson Bay (Canada).
Conserv Genet 13:419–433.
74. Baker CS, et al. (1990) Influence of seasonal migration on geographic distribution of
mitochondrial DNA haplotypes in humpback whales. Nature 344:238–240.
75. Carroll EL, et al. (2015) Cultural traditions across a migratory network shape the
genetic structure of southern right whales around Australia and New Zealand. Sci Rep
5:16182.
76. Harrison XA, et al. (2010) Cultural inheritance drives site fidelity and migratory con-
nectivity in a long-distance migrant. Mol Ecol 19:5484–5496.
77. Kopps AM, et al. (2014) Cultural transmission of tool use combined with habitat
specializations leads to fine-scale genetic structure in bottlenose dolphins. Proc Roy
Soc B 281:20133245.
78. Mann J, Sargeant B (2003) The Biology of Traditions; Models and Evidence, eds
Fragaszy DM, Perry S (Cambridge Univ Press, Cambridge, UK), pp 236–266.
79. Krützen M, et al. (2005) Cultural transmission of tool use in bottlenose dolphins. Proc
Natl Acad Sci USA 102:8939–8943.
80. Cantor M, Whitehead H, Gero S, Rendell L (2016) Cultural turnover among Galápagos
sperm whales. R Soc Open Sci 3:160615.
81. Mercader J, et al. (2007) 4,300-year-old chimpanzee sites and the origins of percussive
stone technology. Proc Natl Acad Sci USA 104:3043–3048.
82. Cantor M, et al. (2015) Multilevel animal societies can emerge from cultural trans-
mission. Nat Commun 6:8091.
83. van Schaik CP, Isler K, Burkart JM (2012) Explaining brain size variation: From social to
cultural brain. Trends Cogn Sci 16:277–284.
84. Cammen KM, et al. (2016) Genomic methods take the plunge: Recent advances in
high-throughput sequencing of marine mammals. J Hered 107:481–495.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7821
Song hybridization events during revolutionary song
change provide insights into cultural transmission
in humpback whales
Ellen C. Garlanda,b,1, Luke Rendella,b, Luca Lamonia,b, M. Michael Poolec, and Michael J. Noadd
a
Centre for Social Learning and Cognitive Evolution, School of Biology, University of St. Andrews, St. Andrews, KY16 9TH, United Kingdom; bSea Mammal
Research Unit, School of Biology, University of St. Andrews, St. Andrews, KY16 8LB, United Kingdom; cMarine Mammal Research Program, Maharepa,
Moorea 98728, French Polynesia; and dCetacean Ecology and Acoustics Laboratory, School of Veterinary Science, University of Queensland, Gatton, QLD
4343, Australia
Edited by Kevin N. Laland, University of St. Andrews, St. Andrews, United Kingdom, and accepted by Editorial Board Member Andrew G. Clark June 16, 2017
(received for review January 14, 2017)
Cultural processes occur in a wide variety of animal taxa, from
insects to cetaceans. The songs of humpback whales are one of the
most striking examples of the transmission of a cultural trait and
social learning in any nonhuman animal. To understand how songs
are learned, we investigate rare cases of song hybridization, where
parts of an existing song are spliced with a new one, likely before
an individual totally adopts the new song. Song unit sequences
were extracted from over 9,300 phrases recorded during two song
revolutions across the South Pacific Ocean, allowing fine-scale
analysis of composition and sequencing. In hybrid songs the current
and new songs were spliced together in two specific ways: (i) sing-
ers placed a single hybrid phrase, in which content from both songs
were combined, between the two song types when transitioning
from one to the other, and/or (ii) singers spliced complete themes
from the revolutionary song into the current song. Sequence anal-
ysis indicated that both processes were governed by structural sim-
ilarity rules. Hybrid phrases or theme substitutions occurred at
points in the songs where both songs contained “similar sounds
arranged in a similar pattern.” Songs appear to be learned as seg-
ments (themes/phrase types), akin to birdsong and human language
acquisition, and these can be combined in predictable ways if the
underlying structural pattern is similar. These snapshots of song
change provide insights into the mechanisms underlying song learn-
ing in humpback whales, and comparative perspectives on the evo-
lution of human language and culture.
vocal learning | cultural transmission | song | cetacean | humpback whale
C ultural transmission has been shown in a wide variety of taxa,
spanning birds, fish, insects, cetaceans, and nonhuman pri-
mates (1, 2). We define culture in the broad sense as shared in-
formation or behavior acquired through some form of social
learning from conspecifics (3–5). Each of these studies has pro-
vided examples demonstrating a behavioral trait being passed
from one individual to another, and on occasion entire pop-
ulations, through some form of social learning. Cetaceans show
some of the most sophisticated and complex vocal and cultural
behavior outside of humans (6, 7), including vocal learning, shared
traditions, and gene–culture coevolution. For example, southern
right whales (Eubalaena australis) demonstrate strong migratory
culture (8), whereas bottlenose dolphins (Tursiops truncatus and
Tursiops aduncus) demonstrate the cultural transmission of tool
use (9, 10). Both sperm whales (Physeter macrocephalus) and killer
whales (Orcinus orca) have culturally transmitted group vocaliza-
tions that are maintained over decades (11, 12), and also appear to
undergo gene–culture coevolution (13–15).
Humpback whales (Megaptera novaeangliae) possess multiple,
independently evolving cultural traditions, including maternally
directed site fidelity to breeding and feeding grounds (16), so-
cially learned feeding tactics (17), and song displays that are
subject to cultural evolution and revolution (18–20). Humpback
7822–7829 | PNAS | July 25, 2017 | vol. 114 | no. 30
whale song is one of the most elaborate acoustic displays in the
animal kingdom (21). The song is produced solely by adult males
(22) and is therefore considered a product of sexual selection,
even though the details of how it functions as a signal are still
debated (23).
Song is organized in a nested hierarchy: single sounds are
termed “units,” a sequence of units is grouped into a “phrase,”
phrases are repeated to form a “theme,” and a number of different
themes are usually sung in a set order to form the “song” (24). To
move from one theme into another, a single “transitional phrase”
is sometimes sung that contains content from the preceding and
following themes (20). Different versions of the display (contain-
ing different themes) are termed “song types” (18). Within each
population, there is usually strong conformity to a single song type
at any point in time (25). However, the song is constantly changing
(20), and all males must continuously incorporate these alterations
to maintain the observed conformity. This slow and gradual
change is a process of cultural evolution in which subtle changes
occur over time at a population scale (20, 26).
Populations within an ocean basin sing similar songs, but the
similarity depends on both geographic (27, 28) and temporal dis-
tances, as transmission of song changes across a region may take
several years (18, 29, 30). In the western and central South Pacific
region, song also undergoes dramatic cultural “revolutions,” where
the song type from a neighboring population is rapidly adopted by
all of the males in an adjacent population (18, 19). We have pre-
viously described the rapid, repeated, and regular horizontal cul-
tural transmission of multiple song types, creating multiple song
revolutions across the western and central South Pacific region (18,
29, 30). Among populations in any nonhuman animal, this is a very
rare, possibly unique, example of population-wide horizontal cul-
tural transmission where behavioral variants are transmitted rapidly
and repeatedly (18). However, we know little regarding the un-
derlying vocal and sequence learning mechanisms governing this
extraordinary cultural phenomenon.
Mechanisms of vocal learning are far better understood for
human language acquisition and birdsong than for cetacean
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: E.C.G., L.R., and M.J.N. designed research; E.C.G., M.M.P., and
M.J.N. performed research; E.C.G. contributed new reagents/analytic tools; E.C.G., L.L.,
and M.J.N. analyzed data; and E.C.G., L.R., L.L., M.M.P., and M.J.N. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. K.N.L. is a guest editor invited by the
Editorial Board.
1
To whom correspondence should be addressed. Email: ecg5@st-andrews.ac.uk.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1621072114/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1621072114

vocalizations (7). Humpback whales are “vocal production learn-
ers,” as they are able to modify the form of their own vocal signals
after experience with signals from other individuals (7). Vocal
production learning is not widespread; thus far, only a few mam-
malian groups, including cetaceans, pinnipeds, bats, elephants,
and humans have been shown to be capable of it (7). An impor-
tant limitation when studying large cetacean species is the inability
to conduct controlled laboratory experiments. Although learning
and song production can be mapped to different pathways in the
brain for songbirds, mice, and humans (to name a few) (31), this is
not yet possible with large, free-roaming cetaceans. However, we
can explore the learning mechanisms involved by examining the
structure and arrangement of the song displays in detail, and
comparing any rules we uncover governing the arrangement and
learning of song to those currently known in human language
acquisition and birdsong learning.
Statistical learning, where patterns and structure are identified
based on the statistical information present in sensory stimuli, is a
common human learning mechanism present in all sensory mo-
dalities (32). From a very young age human infants are able to
detect, extract, and generalize statistical regularities (i.e., simple
algebraic rules) from their auditory environment (32, 33), and
understanding how they use this statistical information to learn
language is a major research focus (32). The ability to detect
transition probabilities—the probability that a given sound or
syllable follows another one (32, 33)—is important in under-
standing word segmentation or grammar learning tasks. From a
comparative perspective, recent work has demonstrated that zebra
finches (Taeniopygia guttata) generate phonological categories
that result in the song being easier for others to learn (34).
Understanding how humpback whales learn their extended
song sequences is therefore of interest in the comparative study
of mechanisms for learning sequences and patterns in cultural
vocal signals.
Segmentation, the chunking of sequences into smaller com-
ponents (phrases or words) that can later be recombined, is
another important mechanism in human language acquisition
(32, 33, 35, 36). Songbirds have been shown to segment when
learning their song displays (37–40). Segments are typically
separated by longer pauses (silence), and these pauses may
provide an emphasis that aids in memorization of segment
chunks (39). In a recent review of human language and non-
human animal communication, Birchenall (33) suggests that the
process of segmentation may also be present in humpback whale
song learning. Given the importance of segmentation to lan-
guage acquisition and the presence of this mechanism in the
learning of birdsong, this is a logical starting place to study
humpback whale song learning.
Here, we present evidence that humpback whales use seg-
mentation in song learning by examining recordings made during
the process of learning a new song in the context of a song rev-
olution event. Recording a whale in the act of changing his song is
challenging; they are highly mobile and one cannot simply record
all of an individual’s song during a 2- to 3-mo breeding season
and >6,000-km migration. We therefore investigate some rare
cases of song hybridization recorded during song revolution events
to understand how individual whales transition between two dif-
ferent songs. These hybrid songs, which contain themes and ele-
ments from both the previous song and the new, revolutionary
song, presumably represent a transition phase in the process by
which singers change their song display to a new, completely dif-
ferent arrangement. We aim to identify if there are any underlying
structural rules governing song change (e.g., segmentation, tran-
sition probabilities) that can provide insight into how new songs
can be learned so rapidly. We hypothesize that new songs will be
learned as segments if segmentation is a taxon-general mechanism
(hypothesis 1). Identifying the level in the song hierarchy (phrase,
theme, or song) that comprises a segment will provide important
Garland et al.
COLLOQUIUM
PAPER
information as to how the song is memorized. Alternatively, parts
of both song types may be spliced together in a random ar-
rangement of new and old units. This would indicate that the
structural arrangement of an individual’s song disintegrates to a
babbling/subsong phase (41) before learning the new song ar-
rangement, and that segmentation is not occurring. Additionally,
we hypothesize that if segmentation occurs, then the combination
of these segments from both song types by an individual will not be
random (hypothesis 2). That is, the insertion of new song segments
into the existing song will be at locations in the existing song where
there is some structural similarity in the sound units, phrases, or
themes of the old and new songs, rather than at random positions.
This “similar sounds in similar arrangements” mechanism would
be akin to word substitutions in humans, such as malapropisms,
where an incorrect word with a similar sound is used in place of
the correct word (42). To test these hypotheses, we first investi-
gated how each singer displaying a hybrid song transitioned be-
tween song types, and second we quantified the similarity in
arrangement between the themes from each song using sequence
analysis metrics. We analyzed four hybrid songs recorded during
two different song revolutions from two geographic locations
(eastern Australia and French Polynesia). Thus far, these are the
only examples of hybrid songs in over 20 y of fieldwork from five
populations where song revolutions are known to occur regularly,
and from which ∼1,500 song sequences representing at least
100,000 phrases have been analyzed. [A fifth hybrid song has been
identified. This recording is of a very poor quality (low signal-to-
noise ratio). Themes can be sporadically identified but the clear
transitions between themes required for the current analysis is
lacking. We therefore excluded this recording from analysis. The
recording was from eastern Australia in 1997 as part of the pink/
black song revolution presented in ref. 19.]
Results
Three separate datasets were included in the analysis, as each
contained one or more hybrid songs. These spanned two geo-
graphic locations: Peregian Beach, eastern Australia (1996–1997
and 2002–2003), and Mo’orea, French Polynesia (2005); and two
song revolutions: from pink to black (Australia 1996–1997), the
“original” song revolution (19), and from blue to dark red
[which occurred in Australia in 2002–2003 and French Polynesia
in 2005 (20)]. Over 46 h of song from 50 singers and 4 song types
(each given an arbitrary color label—blue, dark red, pink, and
black—to be consistent with published analyses of these song
types) were analyzed from French Polynesia (2005: 18 singers,
1 hybrid) and eastern Australia (1996–1997: 2 singers each
based on the highest quality singer for each song type from
ECOLOGY
249 singers presented in ref. 19, 2 hybrids; and 2002–2003:
26 singers, 1 hybrid).
To identify if new songs were learned as segments (hypothesis
1), we first needed to classify each potential segment. Because
there are multiple levels in the humpback song hierarchy, each
being a potential basis for segmentation, we analyzed each level.
First, individual sounds were classified into categories (i.e., unit
types) (SI Methods and Tables S1 and S2). Then the stereotyped
sequences of units that made phrases were established and fur-
ther grouped into themes (SI Methods and Table S1). Themes
from each song type were labeled 1 through 37 (Table 1; also see
SI Methods and Table S1) following previous classification of
these song types (18, 19, 29, 43, 44). The song type of origin (pink
or black) for theme 11 was uncertain and thus remained un-
resolved, as it was not heard in any nonhybrid songs (Fig. 1,
Table 1, and Table S1). The sequence of themes for each hybrid
singer was established (Table 1). It is immediately obvious that
the hybrid songs examined here comprised complete themes
from the two different song types combined into a single song;
segmentation occurred at the theme level.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7823
Table 1. Theme sequence of hybrid songs from French Polynesia (2005) and eastern Australia (1997 and 2003)
/ Represents a transitional phrase between the two labeled themes. // Represents a break in recording for <1 min. SLA, surface level attenuation, where the
whale is breathing at the surface and the song content is difficult to hear and therefore uncertain. Themes are color-coded by song type. Note the song type
of origin for theme 11 was uncertain (colored gray). See Table S1 for description of theme content.
*Themes 31a and 31b repeated multiple times. No break in recording.
†
Themes 8a and 9b repeated multiple times. No break in recording.
‡
Themes 8b and 9b repeated multiple times. No break in recording.
Given that hybrid songs contained theme segments from each
song type, we investigated if there were any patterns to the
arrangement of themes (hypothesis 2). To do this, we: (i) estab-
lished the location of hybrid transitions in the song, (ii) in-
vestigated how each singer transitioned between the song types,
and (iii) quantified the similarity of theme content using sequence
analysis metrics to understand why a singer might switch at that
particular location in the song.
To understand the location of theme transitions, the full se-
quence of themes from all singers was used to construct a first-
order Markov model based on the frequencies of transition be-
tween phrases (Fig. 2). Transitions occurred between the pink and
black song types at multiple locations in the song (Fig. 2A and
Table 1) but, in contrast, transitions between the blue and dark red
song types occurred only at two locations in the song (Fig. 1 and
Fig. 2B). At these transition locations, singers often placed a tran-
sitional phrase between the two song types to mediate the transition
(Tables 1 and 2). This single phrase combined the starting units
from the preceding phrase with units from the following phrase
(typically the ending units) (Fig. 1, Table 2, and Table S1).
We characterized the structural similarity, that is the similarity
in the sequence of units that comprised each theme/phrase type
(laid out in Table S1), between each pair of songs (e.g., blue vs.
dark red) using the Levenshtein distance (LD), a common simi-
larity metric in linguistic and humpback song comparisons (29, 43,
45, 46). In songs from the 2005 French Polynesia blue/dark red
revolution, hierarchical clustering of themes showed a single lo-
cation on the dendrogram where themes from both song types
grouped together on a branch (Fig. 3A). This was where the singer
of the hybrid song in the French Polynesian dataset switched
7824 | www.pnas.org/cgi/doi/10.1073/pnas.1621072114
between song types (Fig. 2 and Tables 1 and 2). In songs from the
eastern Australia 2002 revolution involving the same song types,
this pattern was not as clear because theme transitions did not
occur at the most similar themes (Fig. 3B). Instead, theme tran-
sitions were mediated by a transitional phrase (Tables 1 and 2).
Finally, in songs from the 1996 eastern Australia pink/black rev-
olution, the dendrogram showed a single location where themes
from both song types grouped together on a branch (Fig. 3C and
SI Results). This was where the majority of transitions in hybrid
songs occurred between the song types (Table 2). The hybrid
singers replaced the next theme in the song sequence with a
similarly arranged theme from the other song type (Fig. 1 and
Table 2). The remaining theme transitions were either mediated
by a transitional phrase or the mechanism of transition between
the song types was unclear (Tables 1 and 2). Regardless, in ad-
dition to transitional phrases this final analysis strongly indicates
that transitions between song types are not random and occur
more often at locations where theme content is most similar.
Discussion
Hybrid songs are recorded extremely rarely but are of interest
because they capture some part of the process by which singers
change their song display from an older version (type) to a new,
completely different arrangement. The hybrid songs presented
here were all captured during song revolution events, when
singers using both the old and new song types were in the same
population. It is clear that new songs are learned as segments,
confirming hypothesis 1 (see also ref. 33), indicating that seg-
mentation is a learning mechanism found in the cetacean
lineage. The way singers move between song types during singing
Garland et al.
 COLLOQUIUM
PAPER
Fig. 1. Example spectrograms of hybrid transitional phrases, corresponding parent themes, and substituted themes from the blue and dark red song types
(A and B), and the pink and black song types (C and D). A shows the theme progression (from left to right) of the transition from blue theme 24, through the
hybrid 24/37a phrase into dark red theme 37a and then theme 37b (singer HYB1) (Table 1). B shows the theme progression from dark red theme 31a to blue
theme 27, mediated by hybrid transition phrases 27/31a and 31a/27 [note the difference in arrangement depending on the direction of transition (singer
HYB2)]. C shows the theme progression (from left to right) from pink theme 1, through hybrid phrase 1/7a into black theme 7a (singers HYB3 and HYB4)

ECOLOGY
(Table 1), and the substituted pink theme 2. D shows the theme progression (from left to right) from black theme 9b, through hybrid phrase 9b/4 into pink
theme 4 (singers HYB3 and HYB4) (Table 1). It also shows pink theme 3 and the unresolved theme 11. Spectrograms were 2,048-point fast Fourier transform
(FFT), Hanning window and 75% overlap, generated in RAVEN PRO 1.4 (see also Audios S1–S4 for corresponding audio files).

bouts suggests that these displays are unlikely to be learned as a
whole. Instead, songs are split into theme segments, and the fact
that transitions between song types occur at specific points
in the theme sequence suggests that each theme is learned as
a separate entity. Segmentation or chunking of sequences is
an important mechanism in human language acquisition (35),
where a stream of utterances is segmented into smaller com-
ponents (phrases or words) and later recombined (36). Song-
birds have also been shown to segment their song displays (37–
40) and statistically learn sound categories (34). Juvenile male
songbirds may learn their song from one or more tutors as a
sequence of syllable segments, which they recombine to form
their own song (37–40). In humpback whales, our results sug-
gest that a male learns the new song as theme segments, which
he combines with older themes as he progressively learns
the new song. The novelty-threshold hypothesis suggests that
novelties in the song are adopted by singers once reaching a
threshold prevalence (47), and therefore an individual male
would need to hear a new song from multiple individuals before
adopting the change. The male therefore has multiple potential
models for each theme and a general overview of the “correct”
sequence of the themes. The highly stereotyped nature of theme
and phrase sequences, both of which we quantified as transition
probabilities (e.g., Fig. 2 and ref. 48), strongly suggests hump-
back whales, like songbirds, use statistical learning in learning
their song display (34).
In songbirds, segments are typically separated by longer pauses
(silence), and these pauses may provide an emphasis that aids in
memorization of segment chunks (39). This feature of pauses
between segments of zebra finch song is also a feature of
humpback whale song, as a phrase is delineated from the start of
another phrase by a longer pause (24, 49). Given that a single

Garland et al. PNAS | July 25, 2017 | vol. 114 | no. 30 | 7825
Fig. 2. First-order Markov model of theme transitions to understand hybridization between (A) pink/black song types (n = 2,222 phrase transitions, n =
4 individuals), and (B) blue/dark red song types (n = 8,852 phrase transitions, n = 46 individuals). Each node represents a theme or phrase type, color-coded by
song type. White nodes represent transitional phrases and dashed lines indicate transitions between song types. Arrows represent the direction of movement
and thicker lines indicate higher transition probabilities. Transitions between the pink and black song types (A) occurred at multiple locations (themes 1 to 7b,
9b to 4, 10a to 1, 4 to 10a, 10a to 5b, 8b to 4, and 8a to 5b). In contrast, transitions between the blue and dark red song types (B) occurred only at two specific
locations in the song: blue theme 27–dark red theme 31a (both directions), and blue theme 24–dark red theme 37a (one-way). Phrase repetitions are removed
from the figure for ease of display.

humpback whale song can last anywhere from 5 to 30 min
(24), any aid in memorization of such a long display would be
under strong selection. The repetition of phrases within themes
introduces redundancy in the song, and likely aids memorization
through repetition and reduced content. Furthermore, rhyme-like
patterns in humpback song (50) appear similar to rhyme patterns
in human poems or prose, which also aid recall (51). The question
of how humpback whales remember their song display (they rarely
sing the wrong thing) is still open. From playback studies we know
humpback whales react more strongly to novel songs than to the
song of the current year (see ref. 52). The whales can identify
“same” from “different.” It would be interesting to explore how
long their song memory lasts, as bottlenose dolphins have been
shown to remember vocalizations (signature whistles of conspe-
cifics) for over 20 y (53). Such a song memory could drive the
directional change in song revolutions (to stop whales reverting
back to the previous song type), leading to the broad-scale cultural
phenomenon we observe (18).
Hybrid songs from both song revolutions contained themes from
one song type that were spliced into the middle of the other song
type (Table 1). There are multiple examples of such hybrid song
production in songbirds at the boundary of two song dialect areas
or the boundary between two closely related species (41). For ex-
ample, orange-tufted sunbirds (Nectarinia osea) have sharp dialect
boundaries, but a small number of birds along these boundaries
sing songs from both dialects (i.e., hybrids) (54). Similarly, in the
village indigobird (Vidua chalybeata), a species that undergoes
continuous population-wide song evolution in some ways similar to
humpback whale songs, males along dialect boundaries have been
recorded singing hybrid songs that combined songs from each

7826 | www.pnas.org/cgi/doi/10.1073/pnas.1621072114 Garland et al.


Table 2. Theme transitions between two different song types in hybrid songs
The direction of transition (i.e., old song to new song, or vice versa), and the number of times this transition occurred as a percentage of the total number
of hybrid transitions for each pair of song types (taken from Table 1) are noted. The similarity in sound units or their arrangement is described along with
whether this similarity was supported in the dendrograms (i.e., both themes present on a branch). The presence of a transitional phrase is noted, and a description
of the potential mechanism assisting the transition is suggested.
dialect (55). In yellow-rumped caciques (Cacicus cela vitellinus),
another species with continuous population-wide song evolution,
males in a colony may occasionally incorporate a foreign song type
as part of their yearly population dialect if the two colonies are
closely situated (56). In another example, at the range interface of
black-capped chickadees (Poecile atricapillus) and Carolina chick-
adees (Poecile carolinensis), birds from both species displayed bi-
lingual or atypical repertoires (57). Clearly, segmentation is an
important general mechanism in vocal learning present in multiple
independent lineages.
Transitions between humpback whale song types were often
mediated by a transitional phrase containing individual sound
units from the previous and following phrases that were common
to both song types (Figs. 1 and 2 and Tables 1 and 2). Transi-
tional phrases are a neglected component of the song in general,
as they are often excluded from analyses focused on delineating
song types (49). The variable structure of transitional phrases can
make them difficult to categorize, particularly if they are not
routinely used in all transitions between themes. Nevertheless, it
is clear this normal component of song organization is important
to allow an ordered progression from one theme into another,
regardless of the song types.
Transitions between song types were partially governed by
structural similarity, based on the Markov model and sequence
analysis (Figs. 2 and 3), rejecting random combinations of seg-
ments (hypothesis 2). The sequence analysis indicated that
transitions or theme substitutions occurred more often in loca-
tions that contained “similar sounds arranged in a similar pat-
tern” in old and new songs (Fig. 3). Themes either progressed
into a similarly sounding theme of the other song type or
replaced that similarly sounding theme altogether (Table 2). In
addition to segmenting, song learning and change are partially
governed by structural similarity rules where transitions or theme
substitutions occur in locations that contain similar sounds
Garland et al.
COLLOQUIUM
PAPER
arranged in a similar pattern (i.e., a “switch-when-similar” rule).
Word substitutions in humans, such as malapropisms—the use of
an incorrect word in place of a word with a similar sound (42)—is
highly suggestive for a general mechanism. These transition
points based on similarity could act as a point of reference or
cue, allowing the singer to switch from the old into the new song
at this position in the song. Such anchors are present in human
vocal performances [e.g., oral traditions (51)], and single sounds
or words and similar note arrangements are used to transition
among songs in human music performances. Finally, the ability
to jump from one song into another is also a feature of birdsong;
for example, counter-singing allows a male to select a matching
song of a rival male and switch to singing that song in an ag-
gressive context (41). This skill strongly suggests the presence of
ECOLOGY
an underlying mechanism allowing plasticity in vocal output
shared among vocal learning species.
We suggest the switch-when-similar rule may be stronger and
thus more important in one direction (i.e., old-to-new themes)
(Table 2), assisting singers in learning new themes sequentially
and in the “correct” order. The whale is attempting to learn the
new display; this is very directional. The location in the song
where old themes encroach back into the song display may be
less important and is unlikely to be governed by this similarity
rule (explaining the majority of unsimilar transitions backward).
These new-to-old song transitions appear to be mediated more
often by transitional phrases (Table 2).
The process of vocal production learning (7) of a completely
new song type could occur through a number of structural
changes to the song, as new themes must be learned and old
themes removed. Multiple studies indicate that male humpback
whales adhere to the current arrangement of the song (e.g., refs.
20 and 25). Importantly, once a new song is recorded in a pop-
ulation, all males switch to this new song (18, 19). Clearly, the
song is learned as theme segments to aid in the learning of this
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7827

Fig. 3. Dendrogram of bootstrapped (1,000) similarity matrices of average-
linkage hierarchical clustered median unit sequences for each theme for
(A) French Polynesia 2005 blue and dark red song types (CCC = 0.93),
(B) eastern Australia 2002–2003 blue and dark red song types (CCC = 0.88), and
(C) eastern Australia 1996–1997 pink and black song types (CCC = 0.95).
Multiscale bootstrap resampling (AU, Left, red dot indicates P > 95%) and
normal bootstrap probabilities (Right, green dot indicates P > 70%) are dis-
played. Branches with high AU values are strongly supported by the data.
Dashed boxes indicate where themes from different song types appear to-
gether on a branch.
complex display. In male village indigobirds, immigrant males
add song types from their new dialect and then drop their old,
foreign song types within a year (55). We suggest humpback
whales may undertake a similar process by adding in new themes
starting at similar locations and then progressively deleting the
old themes. Intense cultural conformity is likely influencing these
vocal displays, which are in turn also driven by sexual selection.
The presence of an innate template likely governs the underlying
processes and rules of song learning (58), overlaid with a more
flexible cultural component that governs what variant of the song
display to sing, regardless of the species. The details of how songs
change when there is a general conformity to a population song,
and how this process interacts with sexual selection that un-
derlies the humpback song display, are important questions for
future research.
Conclusions
Humpback whales provide a unique perspective for understanding
of animal culture. Their mammalian heritage also makes them
7828 | www.pnas.org/cgi/doi/10.1073/pnas.1621072114
particularly important to our understanding of structurally arranged
vocal communication and the potential origins of human language.
Here, by investigating rare cases of song hybridization, where parts
of an existing song are spliced with a novel, revolutionary song, we
have unearthed a number of underlying structural rules governing
song change, including segmentation and transition/substitution of
themes based on the similarity in sound sequences. These rules
likely assist humpback whales in rapidly learning their complex and
ever-changing songs, and provide insights into the evolution of
human language and culture.
Methods
Song Recordings. All recordings covered the frequency range of humpback
whale song (see SI Methods for detailed recording settings). The units in each
recording were transcribed by a human classifier (E.C.G. or L.L.), and a subset
of units measured for a suite of acoustic parameters to ensure consistent
naming (45). As humpback whale song is highly stereotyped (24), units were
grouped into phrases, phrases into themes, and themes into song types. Pre-
vious studies have identified and quantified these four song types (pink, black,
blue, and dark red), the themes (labeled 1–37), and unit types within each, and
their cultural transmission across the western and central South Pacific (18, 19,
29, 43, 59).
Theme Transitions to Understand Song Sequences. For each recording, the
sequences of themes, including phrase repetitions, transitional phrases, and
hybrid phrases, were noted. Transition tables were calculated and a first-
order Markov model of phrase transition probabilities was constructed for
each song revolution using these data: pink to black and blue to dark red. The
2002–2003 eastern Australian and 2005 French Polynesian data were com-
bined, given that they represented the same song types (18, 29), and the aim
of this higher-level analysis was to identify positions within a song where a
singer may transition between two song types.
Structural Similarity of Themes. The LD or string edit distance is a powerful
metric for comparing humpback whale song sequences, which we and others
have used extensively to understand song similarity at all levels within the
song hierarchy (29, 43, 45, 46, 59–61). The LD similarity index produces a
measure of similarity (between 0 and 1) among multiple sequences of
varying lengths, and provides an overall understanding of the similarity of all
sequences (see ref. 45). Here, we compared the sequence of units (i.e., a
phrase) to establish the most representative phrase for each theme based on
the similarity in the sequence of units (see SI Methods for further in-
formation, and Table S1) (29, 43, 45, 46). These representative phrases for
each theme (laid out in Table S1) were then compared between the two
song types (pink vs. black or blue vs. dark red) to quantify the structural
similarity among themes in an attempt to identify any underlying structural
rules for the transitions highlighted in the Markov models. Similarity scores
were hierarchically clustered and bootstrapped in R using the hclust, pvclust,
and pvrect packages to ensure the resulting structure was stable and likely to
occur (43, 45, 62). Branches with high bootstrap values (AU significance P >
95% and bootstrap probability significance P > 70%) are strongly supported
by the data, whereas lower values suggest variability in their division (45). As
a further test of how well each dendrogram represented the data, the
Cophenetic Correlation Coefficient (CCC) was also calculated. A CCC score of
over 0.8 is considered a good representation of the associations within the
data (63).
ACKNOWLEDGMENTS. This report is based on a presentation at the Sackler
Colloquium on “The Extension of Biology Through Culture”; we thank the or-
ganizers of the Colloquium and Emma Carroll, Elena Miu, and two anonymous
reviewers for providing helpful comments on previous versions of this manu-
script. E.C.G. and this study were supported by a Newton International Fellow-
ship from the Royal Society of London; L.L. was supported by Leverhulme
Trust Research Project Grant RPG-2013-367; L.R. was supported by the Marine
Alliance for Science and Technology for Scotland (MASTS) pooling initiative.
MASTS is funded by the Scottish Funding Council (Grant HR09011) and con-
tributing institutions. Song recordings in eastern Australia were funded by the
Scott Foundation, the US Office of Naval Research, and the Australian Defence
Science and Technology Organization. We thank everyone involved with this
project. Some funding and logistical support was provided to M.M.P. by the
US National Oceanic Society, Dolphin & Whale Watching Expeditions (French
Polynesia), Vista Press, and the International Fund for Animal Welfare (via the
South Pacific Whale Research Consortium).
Garland et al.

1. Laland KN, Janik VM (2006) The animal cultures debate. Trends Ecol Evol 21:542–547.
2. Whiten A, Ayala FJ, Feldman MW, Laland KN (2017) The extension of biology through
culture. Proc Natl Acad Sci USA 114:7775–7781.
3. Rendell L, Whitehead H (2001) Culture in whales and dolphins. Behav Brain Sci 24:
309–324, discussion 324–382.
4. Fragaszy DM, Perry S (2003) Preface. The Biology of Traditions: Models and Evidence,
eds Fragaszy DM, Perry S (Cambridge Univ Press, Cambridge, UK), pp xiii–xvi.
5. Whiten A (2009) The identification and differentiation of culture in chimpanzees and
other animals: From natural history to diffusion experiments. The Question of Animal
Culture, eds Laland KN, Galef BG (Harvard Univ Press, Cambridge, MA), pp 99–124.
6. Whitehead H, Rendell L (2015) The Cultural Lives of Whales and Dolphins (Univ of
Chicago Press, Chicago).
7. Janik VM (2014) Cetacean vocal learning and communication. Curr Opin Neurobiol 28:
60–65.
8. Carroll EL, et al. (2015) Cultural traditions across a migratory network shape the ge-
netic structure of southern right whales around Australia and New Zealand. Sci Rep 5:
16182.
9. Krützen M, et al. (2005) Cultural transmission of tool use in bottlenose dolphins. Proc
Natl Acad Sci USA 102:8939–8943.
10. Kopps AM, Krützen M, Allen SJ, Bacher K, Sherwin WB (2014) Characterizing the
socially transmitted foraging tactic “sponging” by bottlenose dolphins (Tursiops sp.)
in the western gulf of Shark Bay, Western Australia. Mar Mamm Sci 30:847–863.
11. Rendell LE, Whitehead H (2003) Vocal clans in sperm whales (Physeter macrocephalus).
Proc Biol Sci 270:225–231.
12. Deecke VB, Ford JKB, Spong P (2000) Dialect change in resident killer whales: Impli-
cations for vocal learning and cultural transmission. Anim Behav 60:629–638.
13. Foote AD, et al. (2016) Genome-culture coevolution promotes rapid divergence of
killer whale ecotypes. Nat Commun 7:11693.
14. Whitehead H (1998) Cultural selection and genetic diversity in matrilineal whales.
Science 282:1708–1711.
15. Whitehead H (2017) Gene–culture coevolution in whales and dolphins. Proc Natl Acad
Sci USA 114:7814–7821.
16. Baker CS, et al. (1990) Influence of seasonal migration on geographic distribution of
mitochondrial DNA haplotypes in humpback whales. Nature 344:238–240.
17. Allen J, Weinrich M, Hoppitt W, Rendell L (2013) Network-based diffusion analysis
reveals cultural transmission of lobtail feeding in humpback whales. Science 340:
485–488.
18. Garland EC, et al. (2011) Dynamic horizontal cultural transmission of humpback whale
song at the ocean basin scale. Curr Biol 21:687–691.
19. Noad MJ, Cato DH, Bryden MM, Jenner MN, Jenner KCS (2000) Cultural revolution in
whale songs. Nature 408:537.
20. Payne K, Payne RS (1985) Large scale changes over 19 years in songs of humpback
whales in Bermuda. Z Tierpsychol 68:89–114.
21. Wilson EO (2000) Sociobiology: The New Synthesis (Harvard Univ Press, Cambridge,
MA), 25th Ed.
22. Glockner DA (1983) Determining the sex of humpback whales (Megaptera no-
vaeangliae) in their natural environment. Communication and Behavior of Whales,
ed Payne R (AAAS Selected Symposia Series, Boulder, CO), pp 447–464.
23. Herman LM (November 6, 2016) The multiple functions of male song within the
humpback whale (Megaptera novaeangliae) mating system: Review, evaluation, and
synthesis. Biol Rev Camb Philos Soc, 10.1111/brv.12309.
24. Payne RS, McVay S (1971) Songs of humpback whales. Science 173:585–597.
25. Payne K, Tyack P, Payne R (1983) Progressive changes in the songs of humpback
whales (Megaptera novaeangliae): A detailed analysis of two seasons in Hawaii.
Communication and Behavior of Whales, ed Payne R (AAAS Selected Symposia Series,
Boulder, CO), pp 9–57.
26. Winn H, Winn L (1978) The song of the humpback whale Megaptera novaeangliae in
the West Indies. Mar Biol 47:97–114.
27. Payne R, Guinee LN (1983) Humpback whale (Megaptera novaeangliae) songs as an
indicator of “stocks.” Communication and Behavior of Whales, ed Payne R (AAAS
Selected Symposia Series, Boulder, CO), pp 333–358.
28. Darling JD, Acebes JMV, Yamaguchi M (2014) Similarity yet a range of differences
between humpback whale songs recorded in the Philippines, Japan and Hawaii in
2006. Aquat Biol 21:93–107.
29. Garland EC, et al. (2013) Quantifying humpback whale song sequences to understand
the dynamics of song exchange at the ocean basin scale. J Acoust Soc Am 133:
560–569.
30. Garland EC, et al. (2013) Humpback whale song on the Southern Ocean feeding
grounds: Implications for cultural transmission. PLoS One 8:e79422.
31. Arriaga G, Zhou EP, Jarvis ED (2012) Of mice, birds, and men: The mouse ultrasonic
song system has some features similar to humans and song-learning birds. PLoS One
7:e46610.
32. Romberg AR, Saffran JR (2010) Statistical learning and language acquisition. Wiley
Interdiscip Rev Cogn Sci 1:906–914.
33. Birchenall LB (2016) Animal communication and human language: An overview. Int J
Comp Psychol 29:1–27.
Garland et al.
COLLOQUIUM
PAPER
34. Fehér O, Ljubičic I, Suzuki K, Okanoya K, Tchernichovski O (2016) Statistical learning in
songbirds: From self-tutoring to song culture. Phil Trans R Soc Lond B Biol Sci 372:
20160053.
35. Jusczyk PW (1999) How infants begin to extract words from speech. Trends Cogn Sci 3:
323–328.
36. Doupe AJ, Kuhl PK (1999) Birdsong and human speech: Common themes and mecha-
nisms. Annu Rev Neurosci 22:567–631.
37. Williams H, Staples K (1992) Syllable chunking in zebra finch (Taeniopygia guttata)
song. J Comp Psychol 106:278–286.
38. Takahasi M, Yamada H, Okanoya K (2010) Statistical and prosodic cues for song seg-
mentation learning by Bengalese finches (Lonchura striata var. domestica). Ethology
116:481–489.
39. Spierings M, de Weger A, Ten Cate C (2015) Pauses enhance chunk recognition in
song element strings by zebra finches. Anim Cogn 18:867–874.
40. Slabbekoorn H, And AJ, Bell DA (2013) Microgeographic song variation in island
populations of the white-crowned sparrow (Zonotrichia leucophrys nutalli): In-
novation through recombination. Behaviour 140:947–963.
41. Catchpole C, Slater PJB (2008) Bird Song: Biological Themes and Variations (Cam-
bridge Univ Press, Cambridge, UK), 2nd Ed.
42. Fay D, Cutler A (1977) Malapropisms and the structure of the mental lexicon. Linguist
Inq 8:505–520.
43. Garland EC, et al. (2012) Improved versions of the Levenshtein distance method for
comparing sequence information in animals’ vocalisations: Tests using humpback
whale song. Behaviour 149:1413–1441.
44. Smith JN, Goldizen AW, Dunlop RA, Noad MJ (2008) Songs of male humpback whales,
Megaptera novaeangliae, are involved in intersexual interactions. Anim Behav 76:
467–477.
45. Garland EC, Rendell L, Lilley MS, Poole MM, Noad MJ (2017) The devil is in the detail:
Quantifying vocal variation in a complex, multi-levelled, and rapidly evolving display.
Acoustical Soc Am, 10.1121/1.4991320.
46. Kershenbaum A, Garland EC (2015) Quantifying similarity in animal vocal sequences:
Which metric performs best? Methods Ecol Evol 6:1452–1461.
47. Noad MJ (2002) The use of song by humpback whales (Megaptera novaeangliae)
during migration off the east coast of Australia. PhD dissertation (University of
Sydney, NSW, Australia).
48. Helweg DA, Herman LM, Yamamoto S, Forestell PH (1990) Comparison of songs of
humpback whales (Megaptera novaeangliae) recorded in Japan, Hawaii, and Mexico
during the winter of 1989. Sci Reports Cetacean Res 1:1–20.
49. Cholewiak DM, Sousa-Lima RS, Cerchio S (2013) Humpback whale song hierarchical
structure: Historical context and discussion of current classification issues. Mar Mamm
Sci 29:1–21.
50. Guinee LN, Payne KB (1986) Rhyme-like repetitions in songs of humpback whales.
Ethology 79:295–306.
51. Rubin DC (1997) Memory in Oral Traditions: The Cognitive Psychology of Epic, Ballads,
and Counting-Out Rhymes (Oxford Univ Press, Oxford).
52. Helweg DA, Frankel AS, Mobley JR, Jr, Herman LM (1992) Humpback whale song: Our
current understanding. Marine Mammal Sensory Systems, eds Thomas JA,
Kastelein RA, Supin AY (Plenum Press, New York), pp 459–483.
53. Bruck JN (2013) Decades-long social memory in bottlenose dolphins. Proc R Soc B 280:
20131726.
54. Leader N, Wright J, Yom-Tov Y (2000) Microgeographic song dialects in the orange-
tufted sunbird (Nectarinia Osea). Behaviour 137:1613–1627.
55. Payne RB (1985) Behavioral continuity and change in local song populations of village
indigobirds Vidua chalybeata. Z Tierpsychol 70:1–44.
56. Trainer JM (1989) Cultural evolution in song dialects of Yellow-rumped Caciques in
Panama. Ethology 80:190–204.
57. Sattler GD, Sawaya P, Braun MJ (2017) An assessment of song admixture as an in-
dicator of hybridization in black-capped chickadees (Poecile atricapillus) and Carolina
chickadees (P. carolinensis). Auk 124:926–944.
58. Cerchio S, Jacobsen JK, Norris TF (2001) Temporal and geographical variation in songs
ECOLOGY
of humpback whales, Megaptera novaeangliae: Synchronous change in Hawaiian and
Mexican breeding assemblages. Anim Behav 62:313–329.
59. Garland EC, et al. (2015) Population structure of humpback whales in the western and
central South Pacific Ocean as determined by vocal exchange among populations.
Conserv Biol 29:1198–1207.
60. Eriksen N, Tougaard J (2006) Analysing differences among animal songs quantita-
tively by means of the Levenshtein distance measure. Behaviour 143:239–252.
61. Helweg DA, Cato DH, Jenkins PF, Garrigue C, McCauley RD (1998) Geographic vari-
ation in South Pacific humpback whale songs. Behaviour 135:1–27.
62. Suzuki R, Shimodaira H (2004) An application of multiscale bootstrap resampling to
hierarchical clustering of microarray data: How accurate are these clusters. 15th
Annual International Conference of Genome Informatics, Posters and Software
Demonstrations. Available at http://stat.sys.i.kyoto-u.ac.jp/prog/pvclust/. Accessed October
29, 2015.
63. Sokal RR, Rohlf FJ (1962) The comparison of dendrograms by objective methods.
Taxon 11:33–40.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7829
Conformity does not perpetuate suboptimal traditions
in a wild population of songbirds
Lucy M. Aplina,1 , Ben C. Sheldona , and Richard McElreathb,c
a
Edward Grey Institute, Department of Zoology, University of Oxford, Oxford OX1 3PS, United Kingdom; b Department of Human Behavior, Ecology, and
Culture, Max Planck Institute for Evolutionary Anthropology, 04103 Leipzig, Germany; and c Department of Anthropology, University of California, Davis,
CA 95616
Edited by Kevin N. Laland, University of St. Andrews, St. Andrews, United Kingdom, and accepted by Editorial Board Member Andrew G. Clark May 30,
2017 (received for review January 17, 2017)
Social learning is important to the life history of many ani-
mals, helping individuals to acquire new adaptive behavior. How-
ever despite long-running debate, it remains an open question
whether a reliance on social learning can also lead to mismatched
or maladaptive behavior. In a previous study, we experimentally
induced traditions for opening a bidirectional door puzzle box in
replicate subpopulations of the great tit Parus major. Individuals
were conformist social learners, resulting in stable cultural behav-
iors. Here, we vary the rewards gained by these techniques to
ask to what extent established behaviors are flexible to changing
conditions. When subpopulations with established foraging tra-
ditions for one technique were subjected to a reduced foraging
payoff, 49% of birds switched their behavior to a higher-payoff
foraging technique after only 14 days, with younger individu-
als showing a faster rate of change. We elucidated the decision-
making process for each individual, using a mechanistic learning
model to demonstrate that, perhaps surprisingly, this population-
level change was achieved without significant asocial exploration
and without any evidence for payoff-biased copying. Rather, by
combining conformist social learning with payoff-sensitive indi-
vidual reinforcement (updating of experience), individuals and
populations could both acquire adaptive behavior and track envi-
ronmental change.
social learning | animal culture | conformity | Parus major
S ocial learning, the acquisition of behavior by observation of,
or interaction with, other individuals, is common to many
animal species. It provides a relatively cheap way of acquiring
valuable information and shields naive individuals from the risks
of engaging in trial and error learning. A range of studies have
further highlighted the crucial role of social learning in promot-
ing cultural behavior and shared traditions (1–4) and suggested
that the cultural inheritance of information across generations
may be an important component of the behavioral ecology of
some animals (1, 5, 6). However, social learning may also be dis-
advantageous, if copied information is outdated or mismatched
to the observing individual. How individuals balance costs and
benefits of social learning has therefore been the focus of much
recent research aiming to understand how natural selection has
shaped learning (7).
The use of social learning “strategies” is one possible route
by which animals can combine and filter different kinds of infor-
mation to optimize learning outcomes (8–10). Here, individuals
use social cues, often from multiple conspecifics, to bias learn-
ing in favor of better quality information. These include pref-
erences to copy kin or more prestigious or older individuals,
as well as conformist and payoff-biased social learning (11, 12).
Such learning strategies also have implications for the spread and
persistence of information in populations and for cultural evolu-
tion more broadly (13, 14). For example, conformist transmis-
sion, here defined as the disproportionate tendency to copy the
most common behavioral variant (15, 16), may evolve as a means
of providing naive individuals with a quick way of ascertaining
7830–7837 | PNAS | July 25, 2017 | vol. 114 | no. 30
locally adaptive information. However, it may also have the out-
come of maintaining group differences in behavior, with within-
group traditions resilient to invasion by alternative variants.
Empirical evidence for conformity in nonhuman animals is
currently limited, but hints at a wide taxonomic occurrence, with
proposed cases in fish (17), birds (18), and primates (4, 19). Fur-
thermore, theoretical modeling has suggested that conformist
transmission should evolve under a wide range of conditions and
be particularly favored when environments are spatially hetero-
geneous (15, 20). Yet if individuals are exclusively conformist,
then any new environmental change may result in a mismatch
with the majority behavior, leading to a perpetuation of subop-
timal or maladaptive traditions over time (21) and exaggerating
the disadvantages of social information use. Evolutionary mod-
eling has gone as far as to suggest that in socially learning ani-
mals, coupling of conformist learning with environmental change
could lead to population collapse (22).
This apparent paradox of nonadaptive culture has thus been
the subject of much debate (23–28), with two individual-level
strategies proposed as a potential means of evading this evolu-
tionary trap. First, individuals could switch from socially learned
behavior to engaging in asocial learning (individual innovation)
when the rewards gained for performing the established tradi-
tion is smaller than previously (7, 12, 27, 29). Second, individu-
als could combine conformist tendencies with payoff-biased social
learning, using a “behavioral toolbox” of social learning strate-
gies when choosing what behavior to adopt, thus integrating infor-
mation about both the frequency of behavior and the relative
rewards gained by demonstrators (8, 30). Such a strategy has been
observed in laboratory experiments in humans (8). However,
empirical tests for the occurrence or emergence of suboptimal
or maladaptive traditions in animals have been limited (31, 32).
In a previous study, we demonstrated an influence of con-
formist transmission on the social learning of novel foraging
techniques in wild great tits (Parus major) (18). There, we trained
two demonstrators in each of five subpopulations to one of two
equal alternative solutions to a novel task, where food could
be gained from a puzzle box by pushing a bidirectional door to
the right (technique A) or the left (technique B). Using auto-
mated tracking of individuals and their choices, we then mapped
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, ”The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension of Biology Through Culture.
Author contributions: L.M.A. and B.C.S. designed research; L.M.A. performed research;
L.M.A. and R.M. analyzed data; and L.M.A., B.C.S., and R.M. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. K.N.L. is a guest editor invited by the Editorial
Board.
1
To whom correspondence should be addressed. Email: lucy.aplin@zoo.ox.ac.uk.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1621067114/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1621067114

the spread and establishment of these seeded behaviors through
each subpopulation. Results showed a sigmoidal relationship
between a naive individual’s probability of adopting one of the
two techniques and its frequency in their foraging group, with
birds disproportionately likely to learn the majority technique.
Across each subpopulation, the most common behavior there-
fore became increasingly prevalent over time (18). These strong
preferences at the subpopulation for one technique persisted
over the two generations measured, suggesting that technique
choice had become established as a stable cultural behavior.
Here, we vary the rewards gained by these techniques to inves-
tigate whether a reliance on conformist social learning will result
in mismatched behavior after conditions change. First, in four
subpopulations, two with established traditions for technique A
(“left”) and two for technique B (“right”), we stocked all puzzle
boxes with a less preferred food reward over 2 days, so that solv-
ing using either technique was rewarded with a lower payoff. This
“equal low payoffs” condition was designed to test whether indi-
viduals would explore alternative foraging behaviors when con-
fronted with a lower payoff from their previously learned behav-
ior. Second, and immediately following this condition, all puzzle
boxes were stocked with unequal rewards for a period of 14 days.
Solving using the established tradition was thus rewarded with
this same less preferred food, but solving with the uncommon
behavioral variant now resulted in a high payoff. Whereas most
individuals had no experience of the uncommon solution, a small
minority of individuals already preferred it and so provided avail-
able social information for the new difference between the two
techniques. Finally, all visits and behaviors at the puzzle boxes
were monitored using automated tracking. By quantifying the
decision-making process for each individual’s visit to the puzzle
box we then examined (i) whether individuals switched from con-
formist to payoff-biased copying when observing others receiv-
ing variable rewards, (ii) whether individuals flexibly adjusted
their behavior in response to learning about and gaining variable
rewards, and (iii) whether this resulted in a population-level shift
in behavior.
Results
Condition 1: Equal High Payoffs. A tradition for pushing a bidirec-
tional door either to the right (variant A, T1–T2) or to the left
(variant B, T3–T4) was experimentally induced in four subpop-
1
A B
Proportion of each variant in all solves
0.8
0.6
0.4
0.2
0
T1
Equal High Payoffs
Fig. 1. (A) A puzzle box where visiting individuals can slide the door open from either the blue/left side (variant A) or the red/right side (variant B) to
access a reward in a concealed feeder behind the door. The individual pictured is solving using variant A. Solving using either option can give the same
(equal condition) or different (unequal condition) rewards. Puzzle boxes record identity, contact duration, and solution choice and reset after each visit.
(B) Proportion of variant A or B used in each replicate (T1–T4) in three sequential conditions after variant A (T1–T2) or variant B (T3–T4) was initially
introduced by a trained demonstrator: (i) equally high payoffs for each solving option, proportion for last 5 days shown; (ii) equally low payoffs for each
solving option (2 days); and (iii) unequal payoffs, with the established tradition leading to a lower reward (14 days). Solid circles and error bars show mean
and 95% CI of the probability of individuals’ first solve in each condition being the uncommon variant.
Aplin et al.
COLLOQUIUM
PAPER
ulations of great tits [see Aplin et al. (18) for detailed methods]
(Fig. 1A). Either solving technique was rewarded with the same
highly preferred food, a live mealworm. In all subpopulations,
the large majority (87–98%) of solutions used the technique that
was introduced by the trained demonstrator at the beginning of
each experiment, with the population-level bias to this technique
becoming increasingly entrenched over time (18) (Fig. 1B).
Condition 2: Equal Low Payoffs. For a 2-day period following con-
dition 1, three puzzle boxes were distributed in each subpopula-
tion that rewarded both variants A and B with a less preferred
food, sunflower seed (Fig. S1) (18). Compared with condition 1,
there was no change in the overall proportion of each solution
performed in any subpopulation (Welch two-sample test: T1, t =
0.82, P = 0.42; T2, t = −0.16, P = 0.87; T3, t = −0.57, P =
0.57; T4, t = 1.42, P = 0.16) (Fig. 1B). Whereas these patterns
may have differed over a longer time frame, these results suggest
that, at least over 2 days [average number of solves per individual
across replicates = 22(14–30)] birds did not change their sam-
pling behavior in response to experiencing or observing lower
rewards.
Condition 3: Unequal Payoffs. Immediately following condition
2, puzzle boxes with unequal rewards were installed at all
sites/subpopulations over a period of 14 days. Here, solving using
the established tradition was rewarded with this same less pre-
ferred food of sunflower seeds, whereas solving with the uncom-
mon variant was rewarded with live mealworms. Overall, there
were significantly more solutions of the alternative variant per-
formance in condition 3 than in the previous conditions (Welch
two-sample test, first vs. third condition: T1, t = −5.28, P < 0.001;
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
T2, t = −3.87, P < 0.001; T3, t = −3.75, P < 0.001; T4, t = −5.86,
P < 0.001) (Fig. 1B). In the last part of condition 1, an average
of 8(2–16)% of individuals either showed no preference or pre-
ferred the alternative variant. For these individuals, their prefer-
ence did not change in condition 3. By contrast, 49(33–71)% of
other individuals switched to change their variant preference by
the end of the experimental period (proportion of all solves for
each individual over last 2 days).
Similarly to Aplin et al. (18), we analyzed the change in
individual and population preferences over time. First, as the
data were clearly bimodal (Fig. S2), a longitudinal clustering
1
Variant A
Probability of first using uncommon variant
Variant B
0.8
0.6
0.4
0.2
0
T2 T3 T4 T1 T2 T3 T4 T1 T2 T3 T4
Equal Low Payoffs Unequal Payoffs
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7831
 1
vidual choosing either option was a combination of both social
cues and accumulated experience. Social cues included the fre-
quencies of each behavior and the relative value of demonstrated
rewards and were calculated from activity immediately before a
Proportion of solves are seeded variant

given solve at a given puzzle box (18). Code sufficient to repeat

0.8

our results is available as an R package, wythamewa, that con-

tains the data, models, and simulation code, as well as code for
reproducing each figure to follow (Figs. 3–5).
Four model parameters were considered: s, g, λ, and y, rep-
0.6

resenting, respectively, the influence of social cues on choice,

the updating rate (how quickly an individual’s attraction to an
option responds to newly experienced payoffs), an individual’s
conformist exponent, and its payoff social learning bias (popula-
0.4

tion averages presented in Table S1). There was little evidence

for a payoff social learning bias in individuals’ behavior, with no
individuals weighting this parameter higher than 0.1 (Fig. S2). All
other parameters were more important, yet there was consider-
0.2

able individual variation in each (Table S1 and Fig. S4). Solving

T1 individuals ranged in their behavior from little use of social infor-
T2 mation to placing more than half of their decision-making weight
T3 on social cues (s). Similarly, the updating rate (g) ranged from
T4
0.0

2 4 6 8 10
Time (1-14 days)
Fig. 2. The proportion of solutions using the seeded technique decreased
over time in each replicate, with individuals moving toward preferring the
previously uncommon technique. Each replicate is shown in a different
color/shape, and solid and open symbols represents the two distinct clus-
ters of individuals identified in the longitudinal clustering algorithm (solid
symbols, cluster 1; open symbols, cluster 2). Lines show the generalized esti-
mating equation model fit for each cluster/replicate.
nearly zero to over 0.8. Notably, these two parameters were corre-
lated across individuals, indicating that birds that tended to learn
12 14
socially also updated their personal information more quickly.
In contrast to payoff social learning biases, the large majority
of individuals had a conformity exponent (λ) above 1, indicating
at least mild conformity in their use of social cues (above dotted
line, Fig. 3A). However, there was a strong negative correla-
tion with overall reliance on social learning (s), such that birds
A B

0.2 0.4 0.6 0.8 1.0

(b) (d)
conformity exponent (lambda)
4

probability of high option

algorithm was used to group individuals into two behavioral tra-
3

jectories, with 48% (0.33–0.67%) of individuals falling into clus-

ter 1. Clusters were then analyzed separately, using a generalized
estimating equation (GEE) model where the dependent variable
2

was the proportion of solves using the established technique on

each of 15 days and the explanatory variables were day, individ-
1

uals weighted by their total number of solves, and subpopulation

0.0

(Fig. 2). There was strong evidence in cluster 1 that the prefer- 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6 0.8 1.0
ence for the established tradition decreased over time (pooled social learning weight (s) observed freq high option
replicate data; coefficient ± SEM = −0.27 ± 0.02, P < 0.001).
Cluster 2 showed a significant but much lower decreasing prefer- C D
ence for the established tradition (pooled replicate data; coeffi-
0.8
0.6

cient ± SEM = −0.13 ± 0.02, P < 0.001) (Fig. 2). This bimodality
social learning weight

in the rate of change over time was related to age, with younger
0.6
updating rate
0.4

individuals more likely to fall into cluster 1 [general linear mixed

model (GLMM): z140 = −2.94, P = 0.003]. There was no rela-
0.4

tionship between cluster membership and any other measured

0.2

0.2

variables (sex GLMM, z116 = 0.26, P = 0.79; prior preference

strength GLMM, z116 = −0.89, P = 0.37; number of solves in
0.0

condition 1 GLMM, z116 = −0.89, P = 0.37). Finally, and in

0.0

support of this result, there was a negative correlation between -1 0 1 2 3 -1 0 1 2 3

age and likelihood of preferring the high-payoff variant in the last
2 days of the experimental period [linear mixed model (LMM):
−0.07 ± 0.02, t = −3.83, P < 0.001] (Fig. S3).
Analysis of Learning Mechanisms. We next investigated the relative
contribution of different learning mechanisms to the observed
change in behavior during condition 3. To address these ques-
tions statistically, we used a sequential learning model of a form
previously used to study the interaction of social and asocial
learning (8, 33). This framework allows for individual choices
to be modeled as products of time-varying interactions of dif-
ferent modes of learning. Specifically, each solution decision was
modeled as a binary outcome in which the probability of an indi-
age (standardized) age (standardized)
Fig. 3. Individual parameter estimates for the mechanistic learning model.
Each circle represents the posterior mean for an individual bird. (A) Strength
of conformity (λ) is negatively correlated with reliance on social learning
(s), but most individuals show some conformist bias (exponent above 1).
(B) Implied social learning influence functions (expression 5). The diagonal
line represents unbiased social learning. S-shaped curves are conformist indi-
viduals. The weak influence of payoff bias shifts these curves upward in
the lower left corner. (C) Reliance on social cues tends to decline with age,
explained mainly by the presence of large values of s in the youngest indi-
viduals. Individuals with low values are present at all ages. (D) Updating rate
g tends to decline with age. The youngest individuals can be highly respon-
sive to individual experience, whereas the oldest individuals change their
attraction scores more slowly.

7832 | www.pnas.org/cgi/doi/10.1073/pnas.1621067114 Aplin et al.


who put lower weights on social learning were more strongly
conformist (Fig. 3A). It is easier to understand the impact of
these individual differences by translating the estimates for each
individual into an implied social learning function, as defined by
expression 5. Fig. 3B shows these implied functions, plotted for
the posterior mean of each individual. Here, an s-shaped curve
corresponds to conformist learning. Whereas a few birds appear
slightly anticonformist, it must be noted that Fig. 3B shows poste-
rior means, with considerable uncertainty about the exact func-
tion of any one individual. Inspecting individual functions with
full uncertainty envelopes confirms that there is no strong evi-
dence for anticonformity in this population, only for a minority
of individuals whose behavior is consistent with both weak anti-
conformity and weak conformity.
There was a consistent negative impact of age on both weight
given to social cues (s) and updating rate (g). In contrast, there
were no consistent or strong effects of age on conformity (λ) or
on payoff bias (y). These relationships are shown in Fig. 3 C
and D and suggest that older individuals were much less influ-
enced by social cues and also changed their behavior more slowly
in response to changes in personal experience. This is in agree-
ment with the descriptive results: Older individuals switched to
the high-payoff variant more slowly. Younger individuals, in con-
trast, adapted more quickly; these “adaptor” individuals showed
the strongest use of social information and simultaneously tended
to be less conformist, yet also tended to update their own attrac-
tion scores more quickly in response to new personal experience.
Modeling the Link Between Individual Behavior and Population-Level
Patterns. The results in the previous section suggest a combina-
tion of conformist social learning and payoff-sensitive individual
reinforcement (updating) in the population, with individual vari-
ation in all measures. Surprisingly, given our initial hypotheses,
we found no evidence of individual exploration or payoff-biased
social learning of sufficient strength to explain the patterns of
behavior change. Therefore, does the conformity present in the
population slow the rate of switching? Or does it instead help
A Enough conformity
9
7
bird
5
3
1
0 10 20 30 40 50
turn
C Too little conformity
9
7
bird
5
3
1
0 10 20 30 40 50
turn
Fig. 4. Simulations of the population consequences of mixes of conformist social learning and individual reinforcement. In each plot, each row is an
individual agent and each column is a time period. Open and solid circles represent alternative behavior. Before the vertical dashed line at turn 30, solid
is adaptive. After turn 30, open is adaptive. All groups of learners initialized with nonadaptive attraction scores, s = 0.5, g = 0.6, and y = 0. (A) λ = 1, no
conformity. (B) λ = 10, high conformity. (C) λ = 5, intermediate conformity. (D) Ten birds sampled from posterior distribution of the fitted model.
Aplin et al.
COLLOQUIUM
PAPER
both individuals and the population show adaptive responses to
environmental change?
To test this, we used the same learning model as for data anal-
ysis and used it as a forward evolutionary simulation. We simu-
lated groups of individuals learning together, with parameters for
the weight of social cues, strength of conformity, and updating
rate. We first used these simulations to validate our data analy-
sis code and then explored the population dynamics arising from
different parameter settings. Finally, we computed selection gra-
dients on the parameters.
First, we consider three examples to show the role of con-
formist social learning (Fig. 4). Each plot shows the time series
of a simulated group of 10 learners, where payoffs switch for the
variants at turn 1 and again at turn 30, with groups starting by
preferring the low variant. Only conformist strength λ is varied,
with other parameters held at s = 0.5, g = 0.6, and y = 0 (repre-
sentative of the posterior distribution in the fitted model). When
conformity is turned off (λ = 1), individuals take a long time to
stabilize on the high-payoff variant and are slow to switch back
when the environment changes again (Fig. 4A). Similarly, when
conformity is set very high (λ = 10), adaptive learning is very
slow, and most individuals fail to establish on the high-payoff
variant at all before the environment switches (Fig. 4B). How-
ever, when conformity is set to a moderately high value (λ = 5),
all individuals stabilize on the high-payoff variant before turn 30
and quickly switch back after the variants switch again (Fig. 4C).
Second, we ran a simulation using parameter values for 10
birds sampled from the posterior distribution deduced from the
experimental data (Fig. 4D). The simulation shown here is repre-
sentative, and if anything, the birds showed less conformity than
PSYCHOLOGICAL AND
is optimal in this setting. Nevertheless, the amount of conformity
COGNITIVE SCIENCES
that is present, combined with payoff-sensitive updating, allows
individuals to track changes in behavioral variants.
Finally, selection gradients for social learning weight (s) and
conformity strength (λ) were used to determine whether selec-
tion favors larger or smaller values of each parameter, condi-
tional on the value of the other. We calculated the selection
Too much conformity
B
9
7
bird
5
3
1
60 0 10 20 30 40 50 60
turn
D Individuals sampled from posterior distribution
9
7
bird
5
3
60 0 10 20 30 40 50 60
turn
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7833
gradients by conducting 20,000 simulations at each of 63 combi-
nations of s and λ (for a total of 1,260,000 simulations) to com-
pute the selection differential of a mutant (Fig. 5). In each sim-
ulation, we considered the difference in total payoffs between a
mutant individual with parameters s + δs or λ + δλ and an aver-
age common-type individual with parameters s and λ. This dif-
ference defines a numerical estimate of the selection gradient
for an invader. The parameters g and y were again fixed at 0.6
and 0. We display the results in Fig. 5 as a vector field. Selec-
tion adjusts combinations on s and λ in the directions indicated
by the arrows, with longer arrows indicating stronger selection.
The red dashed contour is the combinations of s and λ at which
selection on conformity is neutral. Selection increases conformity
below this contour. The blue dashed contour is the combinations
where selection on s is neutral. Selection increases s above this
contour. Therefore, selection favors more conformity for most
of the gradient space, becoming disadvantageous only at high
weights of social learning. Social learning weight increases above
and to the left of the blue contour. Here, social learning weight
cannot increase from zero unless social learning is slightly con-
formist (above the central blue contour); however, once confor-
mity is above a threshold value, higher social learning weight is
favored only up to a point. These processes combine to produce
evolutionary dynamics that favor conformity combined with an
intermediate weighting of social learning (Fig. 5).
In summary, the birds in the experiment obviously did not
evolve in the experiment, and we do not expect them to be pre-
cisely adapted to it. Nevertheless, these simulations help us to
understand why conformist social learning, in combination with
payoff-sensitive individual reinforcement, facilitated the ability
of individuals and groups to track environmental change.
Discussion
Our experiment reveals that socially learned foraging traditions
in great tits are flexible in response to environmental change.
3.0
2.5
conformity strength
2.0
1.5
1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6
social learning
Fig. 5. Selection gradient on social learning weight and conformity, rep-
resented as a vector field for social learning weight (horizontal axis) and
conformity (vertical axis). Selection increases conformity below the outer
red contours. Selection increases social learning above and to the left of the
central blue contour. For these parameter values, selection does not favor
much social learning, unless social learning is also conformist.
7834 | www.pnas.org/cgi/doi/10.1073/pnas.1621067114
Indeed when subpopulations with strongly established forag-
ing traditions for a single behavioral variant were subjected to
a change in foraging payoffs, after just 14 days 49% of birds
switched their behavior to prefer an alternative higher-payoff
variant, with almost all individuals sampling this option. By
modeling the decision-making process for each individual we
show that, perhaps surprisingly, this population-level flexibility
was achieved without significant asocial sampling and despite
an ongoing bias for conformity at the individual level. Instead,
switching depended on two factors. First, there was an interac-
tion between social information and personal experience, with
individuals that experienced the higher-payoff behavior having
a strong preference for that variant in future solves. Second,
there was extensive individual variation, with those individuals
that relied more on social information showing a weaker con-
formist bias. These factors allowed some individuals to switch
once fortuitously exposed to, and experiencing, the high-payoff
variant. These individuals then provided the correct social infor-
mation for others, leading to a positive feedback loop and even-
tual population-level turnover.
There has been extensive speculation about whether a reliance
on social learning can lead to mismatched or out-of-date behav-
ior (23, 25). Conformity has been thought to potentially exacer-
bate this process, as conformist individuals rely on an indirect
cue of information quality (the proportion of individuals exhibit-
ing a behavior) rather than assessing the value of the information
itself (21, 22). However, there has been a paucity of empirical
evidence in nonhuman animals. In the only prior study, guppies
(Poceilia reticulata) were trained on a longer, suboptimal route
to reach a feeding station. This route preference transmitted and
persisted over several days before eroding toward a faster route
(31). It was assumed that this erosion was associated with aso-
cial learning, but this was not tested. The learning mechanisms
that individuals may be using to optimally exploit variable envi-
ronments were more explicitly tested in Rendell et al. (7), where
a computer tournament was used to compete different sets of
learning strategies. Winning strategies invested in social learning
over asocial learning learned most when individuals experienced
a reduced payoff and relied on recently acquired information.
Most interestingly, strategies did not benefit in variable environ-
ments either by using conformist learning or by preferentially
copying high payoffs.
Our experiment supports the findings from Rendell et al. (7)
in two main ways. First, we found no evidence that individu-
als used asocial sampling to change established behavior. This
is contrary to previous theory, which suggested that individu-
als should use “critical social learning,” switching to individual
innovation under reduced payoffs (27, 34). In Rendell et al. (7),
indiscriminate social learning was adaptive because other agents
were “rational”; that is, they reliably demonstrated the highest-
payoff behavior in their repertoire. Again, our results reflect this
finding; individuals exhibited a clear payoff bias in their per-
sonal experience, preferring to use the high-payoff technique
once experiencing both possible options. Second, we found no
evidence that individuals were incorporating information about
the differences in payoffs achieved by different observed demon-
strators. Thus, our results, along with those of ref. 7, suggest that
the evolution of adaptive social learning strategies like “copy the
high payoff” may not be necessary for individuals and popula-
tions to cope with temporally variable environments.
Our results differ in one major respect from those of ref. 7;
individuals exhibited a conformist bias in their social information
use. This agrees with our previous experiment conducted under
stable payoff conditions, where individuals were also conformist
in their social learning (18). This conformity did not prevent
the population from tracking environmental change. Rather, our
simulations demonstrated that the combination of conformist
social learning with the payoff-sensitive individual reinforcement
Aplin et al.

we observed allowed naive individuals to learn adaptive behav-
ior, but then actually promoted their ability to learn new infor-
mation and switch behavior once conditions changed. However,
our simulations found that evolutionary dynamics favored con-
formity only under intermediate social learning. This reflects the
model in Kandler and Laland (35), which also suggested that
conformity bias should be associated with a weaker influence for
social learning (20, 21).
Interestingly, we also found considerable individual varia-
tion in these parameters, with individuals that used the most
social information also generally being less conformist. The role
that this this between-individual variation plays in mediating
population-level outcomes merits more investigation. Addition-
ally, future research should investigate whether these underlying
individual differences in learning behavior are consistent across
contexts and whether they relate to other correlates of behavior,
for example differences in personality or in developmental con-
ditions (36, 37).
Our simulations of model parameters therefore suggest that a
mix of conformist social learning and individual reinforcement
is sufficient to result in population-level switches in groups of
free-mixing individuals. However, it is interesting to consider how
social structure might have additionally influenced this process in
the wild population. Great tits show a fission–fusion social struc-
ture, with extensive mixing and remixing of small foraging flocks
(38, 39) and with social information moving between individuals
via these foraging associations (18, 40). It seems likely that this
social system might have acted to increase the rate at which the
population could flexibly adjust. That is, if individuals occur in
small groups that frequently fission, then if there is any behav-
ioral heterogeneity within the population, then even as conformist
learners, they will likely have some opportunity to acquire this
alternative information. In species with highly modular networks,
by contrast, social structure could instead act to slow the rate
at which individuals and populations could flexibly adjust their
socially learned behavior, with individuals repeatedly exposed to
the same mix of potential demonstrators when copying.
In addition to social structure, population demography could
also influence the speed at which populations can flexibly adjust
socially learned behavior in a variable environment. Whereas we
found no sex differences in learning, unlike in refs. 41 and 42,
younger individuals in our population tended to show a faster
move away from the established low-payoff technique than older
individuals and had a higher probability of preferring the high-
payoff technique by the end of the experiment. All individuals
had equal opportunity in condition 1 to learn and practice the
established behavior, and this result was unrelated to previous
experience. Rather, it appears that younger birds were generally
more likely than older individuals to use social information and,
once having experienced the high-payoff technique, were also
more able to flexibly adjust their behavior. As younger individu-
als are often also more likely to disperse, such flexibility in behav-
ior could be advantageous when moving between new habitats
(41). More broadly, future work should model the effects of pop-
ulation demography and social network structure on the ability
of socially learning populations to track environmental change.
Indeed, both population demography and social structure
could also potentially be manipulated, and their effect experi-
mentally tested.
In conclusion, we show that socially learned traditions in wild
populations of great tits will track environmental change. We
further find that populations can track payoffs while individu-
als remain conformist social learners and use simulations to elu-
cidate the mechanisms by which this counterintuitive outcome
occurs. Indeed, our results suggest that conformist social learn-
ing actually helps the population adapt to and retain high-payoff
behavior, provided it is not too strong. This adds further weight
to arguments that social learning will be adaptive in a wide range
Aplin et al.
COLLOQUIUM
PAPER
of environments and contexts (7, 12). It is intriguing to consider
what circumstances might therefore promote the perpetuation
of maladaptive traditions. One possibility is that some kinds of
socially learned information might be more vulnerable to this,
for example, where matching group patterns is more important
than the absolute adaptive value of a behavior or where the adap-
tive value of a behavior is obtuse or delayed. Future work should
continue to investigate how general these findings are to other
species, including humans, and explore their possible implica-
tions for the adaptive significance of animal culture.
Materials and Methods
Study System. This study was conducted in a population of great tits
(P. major) at Wytham Woods, near Oxford (51◦ 46’ N, 01◦ 20’ W; 385 ha).
This population has been the subject of a long-term study; all resident
great tits were caught as chicks or adults and fitted with a British Trust
for Ornithology metal leg ring and a plastic leg ring encasing a passive
integrated transponder (PIT) tag (IB Technology). In addition to this main
marking scheme, regular mist netting targeted individuals immigrating into
the population and was also used to age and sex birds (by plumage). Immi-
grants could be classed only as first year or older on plumage; however, as
most individuals disperse as relatively young individuals, in all analyses they
were assigned as their youngest possible age based on first capture date.
From autumn to winter, birds form loose flocks of unrelated individuals (38,
39) with groups aggregating to exploit patchy food sources. In spring and
summer, great tits prefer insect prey, but switch to a seed-based diet in win-
ter when insects are less available [e.g., beech mast, Fagus sylvatica (40)].
All experiments mimicked this diet, using live mealworms, unhusked sun-
flower seeds, and peanut granules; previous work has established that meal-
worms are a highly preferred food type, and sunflower seed is preferred to
peanut granules (18). All work was conducted with relevant ethics approval
from the University of Oxford, and by license holders from the British Trust
PSYCHOLOGICAL AND
for Ornithology and the Natural England (Natural England license numbers
COGNITIVE SCIENCES
20123075, 20131205, 20145171).
Experimental Apparatus. The social learning task consisted of a plastic box
containing a feeder that was accessed by sliding a bidirectional door either
to left or to right. The left side of this door was colored blue and the right
side red, and it had a raised front section to allow an easier grip. A perch
in front in the door functioned as a radio-frequency identification (RFID)
antenna registering the identity, visit duration, and action of each visiting
individual; these were recorded and controlled by a printed circuit board
(Stickman Technology) inside the box. One second after a bird was recorded
as departed from the antenna, the door reset back to the middle. When
installed in the woodland, each puzzle box was surrounded by a 1 × 1-
m cage with a 5 × 5-cm mesh to prevent access by larger species, and a
freely accessible bird feeder providing peanut granules was provided at 1 m
distance.
In experimental condition 1, the puzzle-box feeder contained live meal-
worms. However, for experimental conditions 2 and 3, the puzzle box was
modified to provide two different rewards, depending on the solving tech-
nique. This modification involved widening the door by ∼1.5 cm; however,
no other changes were made to the puzzle-box interface.
Experimental Design. A social learning and foraging experiment was con-
ducted in four relatively isolated subpopulations across the woodland, in
4-wk periods between December 2013 and January 2014 (treatment 1 and
treatments 3 and 4) and in a 4-wk period in January 2013 for treatment 2.
First, two males were caught from each subpopulation and trained in cap-
tivity to solve a novel puzzle box: In two subpopulations (T1 and T2), they
were trained to solve using variant A (solving pushing right from the blue
side), whereas in T3 and T4 they were trained to solve using variant B (solv-
ing pushing left from the red side) (Fig. 1A). All birds were then released
to act as the initial demonstrators for this behavior, and three such puzzle
boxes were installed 250 m apart in each subpopulation. These then con-
tinuously operated from dawn on Monday to dusk on Friday for a total of
20 days (18). In all areas the solving behavior spread rapidly, with 68–83%
(n = 37–96 per subpopulation) of resident individuals solving either vari-
ant at least once. Puzzle boxes were used frequently, with 7,945–12,411
rewarded visits per subpopulation; for more detail see ref. 18.
In January and February 2014, these four replicates were then exposed to
a modified puzzle box providing changed rewards. In condition 2 (2 days),
this modified puzzle box provided sunflower seeds as a reward for solving
using either technique. This was followed by condition 3 (14 days): Here
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7835

the behavioral variant introduced in the initial experiment was rewarded
with sunflower seed, whereas the alternative technique was rewarded with
live mealworms. In three replicates, these conditions occurred immediately
following the initial experiment. In T2, the experiment occurred 1 y later;
however, the population was given 15 days of exposure to the puzzle box
immediately before condition 2, with 73% of resident individuals solving
the task (either variant) in this period. The results from this replicate were
similar to those from the three other replicates.
Statistical Analysis. To analyze the change over time in individual- and
population-level preferences in condition 3, we used a GEE model where
the dependent variable was the proportion of solves using the established
tradition and explanatory variables were day, replicate, and individuals,
weighted by their total number of solves per day. As the data distribution
was bimodal, we first divided the data into two clusters, using a longitudi-
nal clustering algorithm that fitted the data for the relative proportion of
events that were variant A for individuals over cumulative 2-h time periods.
This method was implemented using the R package kml3d.
Learning mechanisms underlying individual changes in behavior were
analyzed using a sequential learning model that modeled individual choices
as products of time-varying interactions of different modes of learning. The
foundation of this framework is the experience-weighted attraction learn-
ing model (43), but with additional terms that allow behavior to be guided
by social cues. Specifically, we assume that the probability of observing a
choice k at time t by individual i is given by
pkit = (1 − si ) Ikit + si Skit , [1]
where si is the influence of social cues on choice, Ikit the probability of choice
k according only to accumulative individual attraction, and Skit the probabil-
ity of k according only to social cues. The individual attractions are modeled
as ordinary experience-weighted attractions with a simple reinforcement
model, such that the attraction score for an option k at time t and individ-
ual i is given by
Aki,t = (1 − gi ) Aki, t−1 + gi πk , [2]
where gi is the importance of newly experienced payoff πk . Therefore,
gi = 1 when there is no influence of past experience. Here we estimate both
gi for each individual i and the unobservable payoff πk to each option k.
Attraction scores at time t imply choice probability by means of a softmax
choice rule:
exp (Akit )
Ikit = . [3]
sumn exp (Anit )
In fitting the model, we set the initial attraction scores at time t = 0 for
each individual to the empirical preferences from the first condition. This
accounts for the fact that most individuals begin the second condition with
strong preferences for the formerly high option.
Social cues at time t can influence choice by changing the probability Skit .
In the simplest example, conformist learning is modeled as
1. Muller CA, Cant MA (2010) Imitation and traditions in wild banded mongooses. Curr
Biol 20:1171–1175.
2. Whiten A, et al. (1999) Cultures in chimpanzees. Nature 399:682–685.
3. van Schaik CP, et al. (2003) Orangutan cultures and the evolution of material culture.
Science 299:102–105.
4. van de Waal E, Borgeaud C, Whiten A (2013) Potent social learning and conformity
shape a wild primate’s foraging decisions. Science 340:483–485.
5. Slagsvold T, Wiebe KL (2007) Learning the ecological niche. Proc Biol Sci 274:
19–23.
6. Slagsvold T, Wiebe KL (2011) Social learning in birds and its role in shaping a foraging
niche. Philos Trans R Soc Lond B Biol Sci 366:969–977.
7. Rendell L, et al. (2010) Why copy others? Insights from the social learning strategies
tournament. Science 328:208–213.
8. McElreath R, et al. (2008) Beyond existence and aiming outside the laboratory: Esti-
mating frequency-dependent and pay-off-biased social learning strategies. Philos
Trans R Soc Lond B Biol Sci 363:3515–3528.
9. Boyd R, Richerson P (1985) Culture and the Evolutionary Process (Univ of Chicago
Press, Chicago).
10. Henrich J, McElreath R (2003) The evolution of cultural evolution. Evol Anthropol
12:123–135.
11. Laland K (2004) Social learning strategies. Learn Behav 32:4–14.
12. Rendell L, et al. (2011) Cognitive culture: Theoretical and empirical insights into social
learning strategies. Trends Cogn Sci 15:68–76.
13. Cantor M, Whitehead H (2013) The interplay between social networks and culture:
Theoretically and among whales and dolphins. Philos Trans R Soc Lond B Biol Sci
368:20120340.
7836 | www.pnas.org/cgi/doi/10.1073/pnas.1621067114
nλ
Skit = P kt λ , [4]
m nmt
where nkt is the frequency of choice k among social cues at time t and λ = λi
is individual i’s conformist exponent. When this exponent is 1, social learning
is unbiased by frequency and behavior is sampled merely in proportion to
its occurrence among cues. When, however, the exponent exceeds 1, social
learning is conformist. In this study, we consider also payoff-biased social
learning, which favors the highest-payoff choice among choices observed at
time t. Specifically, we construct a convex combination of conformist and
payoff-bias terms
nλ
Skit = (1 − yi ) P kt λ + yi (1), [5]
m nmt
where yi is individual i’s reliance on payoff bias. When there is no variation
in social cues, we assume that yi = 0, which means that payoff bias is active
only when observed payoffs vary.
We allow learning strategy to vary at the individual level, estimating si , gi ,
λi , and yi for each individual i in the sample. In each case, we construct each
parameter such that its log-odds are a linear combination of an individual
random effect and an age-specific offset. For example, the submodel for si is
logit (si ) = ai, 1 + bs xi , [6]
where xi is the standardized age of individual i and ai is a vector of individ-
ual random effects, one for each parameter s, g, λ, and y. The conformity
exponent λ is given a log rather than a logit link.
Model fitting was performed using Hamiltonian Monte Carlo, as imple-
mented in version 2.12 of Stan (44), to draw samples from the posterior
distribution. We assessed convergence by inspection of the trace plots,
Gelman–Rubin R̂, and an estimate of the effective number of samples.
Finally, model priors were defined to be weakly informative and conserva-
tive, so that estimated effects and correlations were shrunk slightly toward
zero. Specifically, the averages for s, g, λ, and y were assigned Normal(0,1)
priors on the latent scale. The standard deviations of each random effect
were assigned Exponential(2) priors, also on the latent scale. For the corre-
lation matrix of random effects, we used the LKJ family of distributions of
matrices and assigned η = 3, which shrinks correlations away from extreme
values near −1 or +1 and toward zero. For the unobserved payoff advan-
tage of the high-payoff option, we assigned a Cauchy(0,1) prior, which is
essentially uninformative. Code sufficient to repeat our results is available
as an R package, wythamewa, that contains the data, models, and simula-
tion code.
ACKNOWLEDGMENTS. We thank Keith McMahon, Stephen Lang, and other
members of the Edward Gray Institute for help with various aspects of
fieldwork and data collection and Damien Farine for discussions leading
to the formation of the project and for developing the software for the
puzzle boxes. This research was supported by a grant from the Biotechnol-
ogy and Biosciences Research Council (BB/L006081/1) (to B.C.S.). L.M.A. was
supported by a junior research fellowship at St. John’s College, University
of Oxford.
14. Rendell L, et al. (2011) How copying affects the amount, evenness and persistence of
cultural knowledge: Insights from the social learning strategies tournament. Philos
Trans R Soc Lond B Biol Sci 366:1118–1128.
15. Morgan TJH, Laland K (2012) The biological bases of conformity. Front Neurosci 6:1–7.
16. Aplin LM, et al. (2015) Counting conformity: Evaluating the units of information in
frequency-dependent social learning. Anim Behav 110:e5–e8.
17. Pike TW, Laland K (2010) Conformist learning in nine-spined stickleback’s foraging
decisions. Biol Lett 6:466–468.
18. Aplin LM, et al. (2015) Experimentally induced innovations lead to persistent culture
via conformity in wild birds. Nature 518:539–541.
19. Whiten A, Horner V, de waal FB (2005) Conformity to cultural norms of tool use in
chimpanzees. Nature 437:737–740.
20. Nakahashi W, Wakano JY, Henrich J (2012) Adaptive social learning strategies in tem-
porally and spatially varying environments : How temporal vs. spatial variation, num-
ber of cultural traits, and costs of learning influence the evolution of conformist-
biased transmission, payoff-biased transmission, and individual learning. Hum Nat
23:386–418.
21. Henrich J, Boyd R (1998) The evolution of conformist transmission and the emergence
of between-group differences. Evol Hum Behav 19:215–241.
22. Whitehead H, Richerson PJ (2009) The evolution of conformist social learning can
cause population collapse in realistically variable environments. Evol Hum Behav
30:261–273.
23. Galef BG (1995) Why behaviour patterns that animals learn socially are locally adap-
tive. Anim Behav 49:1325–1334.
24. Galef BG (1995) A new model system for studying behavioural traditions in animals.
Anim Behav 50:705–717.
Aplin et al.

25. Laland K (1996) Is social learning always locally adaptive? Anim Behav 52:637–640.
26. Franz M, Matthews LJ (2010) Social enhancement can create adaptive, arbitrary and
maladaptive cultural traditions. Proc Biol Sci 277:3363–3372.
27. Enquist M, Eriksson K, Ghirlanda S (2007) Critical social learning: A solution to Rogers’s
paradox of nonadaptive culture. Am Anthropol 109:727–734.
28. Giraldeau LA, Valone TJ, Templeton JJ (2002) Potential disadvantages of using socially
acquired information. Philos Trans R Soc Lond B Biol Sci 357:1559–1566.
29. Kendal RL, Coolen I, van Bergen Y, Laland KN (2005) Trade-offs in the adaptive use of
social and asocial learning. Adv Stud Behav 35:333–379.
30. Kendal J, Giraldeau LA, Laland K (2009) The evolution of social learning rules: Payoff-
biased and frequency-dependent biased transmission. J Theor Biol 260:210–219.
31. Laland K, Williams K (1998) Social transmission of maladaptive information in the
guppy. Behav Ecol 9:495–499.
32. Bates L, Chappell J (2002) Inhibition of optimal behaviour by social transmission in
the guppy depends on shoaling. Behav Ecol 13:827–831.
33. McElreath R, et al. (2005) Applying evolutionary models to the laboratory study of
social learning. Evol Hum Behav 26:483–508.
34. Rendell L, Fogarty L, Laland K (2010) Rogers’ paradox recast and resolved: Pop-
ulation structure and the evolution of social learning strategies. Evolution 64:
534–548.
35. Kandler A, Laland K (2013) Tradeoffs between the strength of conformity and number
of conformists in variable environments. J Theor Biol 332:191–202.
Aplin et al.
COLLOQUIUM
PAPER
36. Mesoudi A, Chang L, Dall SR, Thornton A (2016) The evolution of individual and cul-
tural variation in social learning. Trends Ecol Evol 31:215–225.
37. Farine DR, Spencer KA, Boogert NJ (2015) Early-life stress triggers juvenile zebra
finches to switch social learning strategies. Curr Biol 25:2184–2188.
38. Farine DR, et al. (2015) The role of social and ecological processes in structuring ani-
mal populations: A case study from automated tracking of wild birds. R Soc Open Sci
2:150057.
39. Aplin LM, et al. (2015) Consistent individual differences in the social phenotypes of
wild great tits, Parus major. Anim Behav 108:117–127.
40. Aplin LM, Farine DR, Morand-Ferron J, Sheldon BC (2012) Social networks predict
patch discovery in a wild population of songbirds. Proc Biol Sci 279:4199–4205.
41. Aplin LM, Sheldon B, Morand-Ferron J (2013) Milk-bottles revisited: Social learn-
ing and individual variation in the blue tit (Cyanistes caeruleus). Anim Behav 85:
1225–1232.
42. Brodin A, Urhan AU (2015) Sex differences in learning ability in a common songbird,
the great tit - females are better observational learners than males. Behav Ecol Socio-
biol 69:237–241.
43. Camerer C, Ho TH (1999) Experience-weighted attraction learning in normal form
games. Econometrica 67:827–874.
44. Stan Development Team (2016) Stan Modeling Language Users Guide and Reference
Manual, Version 2.12.0. Available at mc-stan.org/users/documentation/. Accessed
June 19, 2017.
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7837
A social insect perspective on the evolution of social
learning mechanisms
Ellouise Leadbeatera,1 and Erika H. Dawsonb,2
a
School of Biological Sciences, Royal Holloway University of London, Egham TW20 0EX, United Kingdom; and bLaboratoire Evolution, Génomes,
Comportement et Ecologie, CNRS, 91198 Gif-sur-Yvette, France
Edited by Kevin N. Laland, University of St. Andrews, St. Andrews, United Kingdom, and accepted by Editorial Board Member Andrew G. Clark May 29, 2017
(received for review January 14, 2017)
The social world offers a wealth of opportunities to learn from
others, and across the animal kingdom individuals capitalize on
those opportunities. Here, we explore the role of natural selection
in shaping the processes that underlie social information use, using
a suite of experiments on social insects as case studies. We illustrate
how an associative framework can encompass complex, context-
specific social learning in the insect world and beyond, and based on
the hypothesis that evolution acts to modify the associative process,
suggest potential pathways by which social information use could
evolve to become more efficient and effective. Social insects are
distant relatives of vertebrate social learners, but the research we
describe highlights routes by which natural selection could coopt
similar cognitive raw material across the animal kingdom.
|
social learning associative learning | observational conditioning |
|
social insects Bombus
A n expanding body of work now shows that social learning
(1), once considered the preserve of vertebrate species, is a
feature of insect behavioral repertoires (2, 3). Insects not only
learn about foraging skills, food preferences, brood hosts, and
potential mates by responding to information provided inad-
vertently by others (4–11), but also transmit these behaviors
further, such that they propagate through groups (8, 12) and
possibly even through wild populations (13). Some of these
phenomena appear similar to socially learned behavior patterns
that have been described in vertebrates (11, 14) at least outside
the context of imitation, and as such they are interesting exten-
sions to the taxonomic distribution of social learning. However,
there is more to be gained from these findings than the claim
that “insects are smarter than we thought.” An insect perspective
encourages questions about the basic cognitive raw material that
underlies social learning.
Associative learning theory describes the rules that govern the
strength of connections between neural representations of stimuli,
and there is little doubt that these rules are taxonomically widespread
(15–18). We now know that associative learning is a means to build a
complex, representation-based picture of the world that is far from
the simple stimulus–response/reinforcement paradigms through
which it was originally characterized (19–22). Correspondingly, it has
been repeatedly argued that associative explanations are powerful
enough to encompass many cognitive phenomena (23–29), and social
learning is no exception. Heyes in particular has highlighted that
social learning mechanisms may be fundamentally associative (28–30;
see also ref. 31). But the contention that no major qualitative leaps
are required to distinguish at least some social learning processes
from asocial learning is not an argument against evolutionary change.
It has been argued that a likely pathway for cognitive evolution
may be the accumulation of small quantitative changes that render
domain-general processes a closer fit to the specific cognitive de-
mands of a particular niche (32, 33). Because almost all animals in-
teract with others at some point during their lifetime, these demands
most likely include effective processing of social information. In other
words, whereas there may be little evidence to support a polarized
view of asocial- and social learning processes, it does not follow that
7838–7845 | PNAS | July 25, 2017 | vol. 114 | no. 30
the ability to learn socially is simply a useful byproduct of the fun-
damental ability to learn asocially that has remained untouched by
natural selection. There is clear empirical evidence that natural se-
lection can fine-tune associative processes to particular tasks (34–36).
For certain social learning processes that typify human behavior,
such as imitation or learning through language or instruction, this
mechanistic middle ground between adaptation- and preadaptation-
based explanations for social learning is the subject of substantial
empirical attention (37–40). However, this is not the case for those
social learning mechanisms that underlie the majority of social in-
formation use outside the primates, such as local or stimulus en-
hancement, social facilitation, or goal emulation (41). The effects
produced by these “simpler” processes are well-labeled (1), but
these definitions often tell us little about how the social stimuli upon
which they are based come to control behavior. For example,
imagine that an individual observes a demonstrator using a tool to
get food, and then uses the same tool to extract food for him or
herself, without using the same exact actions. This fits a classic
definition of “emulation” (1), because observation of a demon-
strator interacting with objects in its environment has rendered the
observer more likely to perform any actions that have a similar
effect on those objects. An associative hypothesis might put forward
that perhaps, through the demonstrator’s actions, the observer came
to associate the tool with food, and thus manipulated the tool in its
own manner (1). This is not an alternative option, but rather a
different level of explanation: the label describes the effect, and
associative learning theory puts forward a testable hypothesis as
to why that effect occurs, on a mechanistic level (39). Under-
standing those mechanisms is critical to forming a picture of how
natural selection has acted upon cognitive raw material, but as yet,
such an understanding is lacking. As Galef (41) puts it, “an entire
field of local enhancement awaits exploration.”
Addressing this issue first requires exploration of the basic
mechanisms through which associative learning could produce
adaptive social information use, and here we review a series of
case studies carried out by ourselves and by other researchers
that specifically illustrate how social learning effects can sit
within an associative framework. In the light of claims that stu-
dents of animal cognition might sometimes be “association-
blind” (24), we hope that this approach proves useful not only in
highlighting how associative learning theory can be applied to
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: E.L. and E.H.D. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. K.N.L. is a guest editor invited by the
Editorial Board.
1
To whom correspondence should be addressed. Email: Elli.Leadbeater@rhul.ac.uk.
2
Present address: Unité de Modélisation Mathématique et Informatique des Systèmes
Complexes, Unité Mixtes Internationales, Institut de Recherche pour le Développe-
ment/Univesité Pierre et Marie Curie 209, 93143 Bondy, France.
www.pnas.org/cgi/doi/10.1073/pnas.1620744114

empirical work on animal social learning, but also in stimulating
consideration of how natural selection might modify associative
mechanisms. Social learning involves upstream input mechanisms
that define how social information is received, and downstream
parameters that define how associations are weighted, processed,
stored, retrieved, and applied, and all may be potential targets for
selection (29, 32, 33). Given the phylogenetic distance between
insects and vertebrates (and even between the diverse vertebrate
groups within which social learning abilities are manifest), we do
not infer that any evolutionary modifications would necessarily be
conserved across lineages through common descent; such inference
would have little foundation. Rather, our aim is to illustrate the
cognitive toolbox from which social learning can be built. We hope
to provide some food for thought for those researchers from both
the invertebrate and the vertebrate world that are interested in the
question of how associative learning could be shaped by natural
selection to create something specifically, if not exclusively, social.
An Associative Framework
Bumblebees (Bombus spp.) and honey bees (Apis spp.) forage for
nectar and pollen in rapidly changing floral landscapes, where
floral reward levels vary not only between flower species but also
with season, time of day, local pollinator abundance, and even
shade patterns (42). Foragers visit thousands of flowers per day,
and quickly learn about the stimuli that predict where to find
rewards in the particular flower species and patches on which
they forage. Information provided inadvertently by other forag-
ing bees influences this learning process, and a simple example
whereby social bees learn from their conspecifics about rewarding
flower types provides a good introduction to an associative
framework for studying social learning.
Worden and Papaj (5) allowed bumblebees (Bombus impatiens)
to observe their foraging conspecifics through a screen. Those
conspecifics foraged on only one of two available flower colors,
and when the observers were later permitted to forage alone, they
“copied” the color preferences of the demonstrators (5, 43, 44).
This type of learning initially appears conceptually opaque, be-
cause learning theory is based upon the fundamental concept of
prediction error, which requires a difference between predicted
and experienced outcomes (45). In this observational paradigm,
observers do not directly experience any rewarding outcome, nor
had these laboratory bees ever had the chance to learn that
matching the color choices of conspecifics was rewarding. Thus,
the bees in this situation seem to have learned about flower color
simply by observing the behavior of their conspecifics.
Second-order conditioning is an associative phenomenon that can
potentially explain why animals respond to certain stimuli as though
they have been directly associated with food rewards, when they have
not. This process is best known for its use in psychological research
to study the contents of learning (46), but perhaps the most illus-
trative example derives from the work of Pavlov (47), who famously
trained dogs that the tick of a metronome (a conditioned stimulus,
CS) predicted the arrival of food (an appetitive unconditioned
stimulus, US+). When Pavlov later trained the same dogs that
presentation of a black square (a second conditioned stimulus, CS2)
predicted the sound of the metronome in the absence of food, he
found that subsequent presentation of the square alone evoked
salivation. In other words, the black square (CS2) had come to elicit
the same response as the food (US+), despite the two never having
been experienced together. An association had formed between the
black square and either the food itself or the appetitive state induced
by the conditioned state of the bell (it remains unclear which) (20,
48), and this association constitutes a second-order conditioned re-
lationship. Whereas in classic conditioning paradigms, it is an un-
conditioned response to a stimulus that comes to be elicited by a new
stimulus, in second-order conditioning paradigms, it is a conditioned
response that undergoes what is functionally the same effect.
Leadbeater and Dawson
COLLOQUIUM
PAPER
Second-order conditioning is relevant to social learning be-
cause it provides a mechanism by which a learned response to a
social stimulus, such as conspecific behavior or its products,
could come to be elicited by a new asocial stimulus (49). Hence,
any biologically important stimulus (US) that an animal has
learned to associate with conspecific presence, or with a partic-
ular conspecific behavior pattern, or a vocalization—indeed, with
any social stimulus at all—could become secondarily conditioned
to a new asocial stimulus (a CS2). In Worden and Papaj’s bum-
blebee example (5), observers might “copy” the color preferences
of demonstrators because they have previously learned to associate
conspecifics (a CS) with food rewards (US+), perhaps because
other bees are found at rewarding feeders or flowers. On sub-
sequent observation of conspecifics on a particular flower color (a
CS2), that flower color should be rendered attractive through
second-order conditioning, even though it has not been directly
paired with food.
We recently carried out an experiment to test this hypothesis
(Fig. 1) (43), in which we controlled the previous social foraging
experience of observer bees (Bombus terrestris). Individuals that
had learned to associate conspecifics (CS) with sucrose (US+)
behaved similarly to bees in previous experiments (5), “copying”
the color preferences of the demonstrators that they had ob-
served through a screen. However, bees that had learned to as-
EVOLUTION
sociate conspecifics (CS) with an aversive substance (US−)
actively avoided those same colors, and bees that had never
foraged with conspecifics were not influenced by conspecific
choices. In other words, the results support that observing con-
specifics through a screen influenced forage preferences through
second-order conditioning.
Second-order conditioning is an associative mechanism, and the
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
use of an associative framework to explain empirical results in
social learning research is not new. Perhaps the best-known ex-
ample comes from the work of Cook and Mineka (50–52), who
demonstrated almost 30 years ago that juvenile rhesus monkeys
(Macaca mulatta) can acquire a fear of snakes through observation
of a frightened conspecific interacting with a snake, a phenome-
non that they termed “observational conditioning.” As the name
implies, these results are considered to reflect a classic condi-
tioning process, whereby the sight of a frightened conspecific (US)
elicits an unconditioned fear response in observers. This affective
state becomes conditioned to stimuli that are experienced at the
same time, in this case a snake model (CS), which thus also ac-
quire the ability to elicit fear (28, 52). In other words, the snake is
simply a conditioned stimulus that comes to elicit the fear re-
sponse. Cook and Mineka (50) found that fear could be condi-
tioned to snakes in this way, but not to flowers, implying that fear
cannot be socially conditioned to any randomly chosen stimulus.
However, this is not an argument against an associative explana-
tion, because in primates snake stimuli are generally particularly
easily conditioned to fear responses, whereas flower stimuli are
not (53–56). For example, in humans, pairing of snake pictures
with electric shock leads those pictures to later elicit heart-rate
acceleration (indicating fear), whereas the same protocol involving
flower pictures leads to heart-rate deceleration (indicating arousal
of attention) (54, 56). It is thus not surprising that a fear response
invoked by a social stimulus can be easily conditioned to snakes
but was not detectable for flowers.
What is the link between the second-order conditioning and the
observational conditioning process that we describe above? In both
cases, a response to a social stimulus becomes conditioned to a new,
asocial stimulus. In observational conditioning, an unconditioned
response (here fear, elicited by seeing a frightened conspecific)
becomes conditioned to a new stimulus. In second-order condition-
ing, a conditioned appetitive response becomes conditioned to a new
stimulus. The two mechanisms are functionally analogous, and the
difference lies in whether the initial response to the social stimulus is
acquired through learning (a conditioned response) or unlearned
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7839
 A B C

Fig. 1. Second-order conditioning of flower color preferences in bumblebees (43). (A) Observer bees were initially allowed to forage on a floral array in
which the presence of conspecifics (CS1) predicted either sucrose (US+; Upper) or quinine (US−; Lower). A third group completed this training in the absence
of any conspecifics (not pictured). (B) All bees then observed conspecifics (CS1) foraging on one flower color (CS2) and ignoring an alternative, through a glass
screen. (C) Finally, each observer was permitted to forage alone on the colored array.

predisposition [an unconditioned response; although Heyes (29)
highlights that presumed predispositions may in fact reflect learning].
Note that although we have framed this argument in terms of
classic conditioning, the same case can be made for operant
conditioning (57). Matched-dependent behavior (58) describes
instances whereby matching of another animal’s behavior (i.e., a
response to a social stimulus) is rewarded (e.g., by finding food; a
reinforcer) and thus increases in frequency, and Church (59) has
shown that this response can become conditioned to the asocial
stimulus that initially elicited the demonstrator’s behavior. The
key point is that responses to social stimuli—be they acquired
through classic conditioning, operant conditioning, or a history
of natural selection—are conditioned to new asocial stimuli.
As an illustration, consider the following example (60). Social
information from injured conspecifics, in the form of volatiles
from stressed conspecifics, typically elicits aversion responses in
foraging honey bees (Apis mellifera) (61). We recently found that
bees not only avoid areas where they currently detect such vol-
atiles, but also later avoid colored lights that were experienced at
the same time (60) (Fig. 2). Thus, the avoidance response, which
is initially elicited by the social stimulus, becomes conditioned to
a new asocial stimulus. In the laboratory, we used colored lights
as asocial stimuli; in the wild, floral features that predict the
presence of sit-and-wait predators, such as crab spiders (Thomisidae
spp.) could fulfill the same role.
It is important to reiterate that this associative framework
complements, rather than adds to, the collection of processes
that are labeled as social learning mechanisms, such as local or
stimulus enhancement, or social facilitation (1). These labels
describe effects, and learning theory offers an explanation for
why these effects occur, rather than an additional alternative
effect (39). In some cases, an associative framework is already

A B

Fig. 2. Honey bees learn to respond to colored lights through exposure to alarm volatiles (60). (A) We used an assay whereby highly phototactic subjects
walked up a dark tube toward a colored light (balanced blue/green design; only blue is pictured). Warning triangle indicates the presence of volatiles from a
stressed conspecific. (B) In the training phase, bees in groups E1 (experimental) and E2 (control for sensitization/habituation effects) were slower to approach
the light, but only group E1 were slower in the testing phase. Thus, responses were conditioned to the specific stimuli that had been contiguous with stress
volatiles in the training phase.

7840 | www.pnas.org/cgi/doi/10.1073/pnas.1620744114 Leadbeater and Dawson


part of the definition of a particular process (e.g., observational
conditioning or matched-dependent behavior) and for others (e.g.,
local enhancement, stimulus enhancement, emulation, or imitation)
it is not. Furthermore, our focus here is on mechanisms that are
taxonomically widespread within the animal kingdom, but similar
arguments have been made for processes that are typically associ-
ated with the primate lineage [32, 39, 40; see also Lotem et al. (33),
who make such an argument for imitation within this issue]. More-
over, this associative framework is relevant to any form of social
stimulus, irrespective of whether it is visual, olfactory, or auditory,
or whether it is a specific behavior pattern, bias, or expression, or
simply physical presence. The key concept is that social stimuli elicit
affective states, neural representations, or motor patterns, which
can then be associated with contiguous asocial stimuli.
Salience of Social Stimuli
Above, we argued that responses to social stimuli can produce a
functionally analogous outcome irrespective of whether they
arise through unlearned predisposition or through learning, be-
cause they can be conditioned to new, asocial stimuli. However,
the distinction between such unconditioned and conditioned
responses to social stimuli is important for questions about evolved
traits. A clear pathway for natural selection might be to implement
small quantitative changes that render social associations more likely
to be learned, such that information deriving from informative cues is
acquired particularly rapidly. One means (but not the only means, as
we discuss later) to do so could involve changes to upstream mech-
anisms that determine whether animals notice or pay attention to
social stimuli. We begin with a focus on the question of whether social
stimuli are more salient than less-ecologically informative alternatives.
The salience of a stimulus—the property that renders it con-
spicuous or noticeable, and thus likely to be attended to—is a
key determinant of the speed with which it can be associated with
other stimuli (62). Salience depends on the species-specific
characteristics of the receiver’s sensory system, such that a
stimulus that is salient for one taxonomic group may be less so
for another (63). For example, carbon disulphide (a component
of rat breath) is a particularly salient stimulus for young rats, and
is quickly associated with any food flavors that are encountered
at the same time (64). Experimental evolution paradigms pro-
vide evidence that environments where particular cue types are
reliably useful for decision-making select for changes to the at-
tentional or perceptual mechanisms that determine stimulus
salience. For example, Drosophila lines for which learning about
olfactory stimuli is the best method of identifying a good brood
host develop enhanced sensitivity to such cues and ignore visual
alternatives, and vice versa (36). In a social context, the question
of whether social cues are especially salient has rarely been di-
rectly addressed [with the exception of Galef et al. (64)], but a
number of bee studies touch upon the topic, with mixed results.
When two conditioned stimuli (e.g., a social and an asocial
stimulus) are simultaneously paired with an unconditioned stimu-
lus, the more salient stimulus overshadows learning about the less
salient stimulus (47). Accordingly, an experiment by Dunlap and
colleagues initially appears to suggest that conspecifics present
might be a particularly salient stimulus for bumblebees (65). In this
study, subjects (B. impatiens) were trained to find sucrose
rewards in floral arrays where both asocial and social cues (floral
color and pinned conspecifics, respectively) provided some in-
dication of which flowers were rewarding, before assessing which
of the two cue types the bees preferentially used on a test array.
When both cue types had been equally reliable at predicting
sucrose, the bees disproportionately favored social cues in the
tests, and most surprisingly, they also used social cues even when
floral cues had been more reliable. Bees only resorted to using
floral cues if social cues had been entirely useless predictors of
reward. These results would seem to indicate that the social
stimulus is the more salient alternative, overshadowing any
Leadbeater and Dawson
COLLOQUIUM
PAPER
learning about the floral cue, and correspondingly, the authors
discuss why the ecological niche of pollinators might favor
reliance on social cues. However, given that floral color is also
likely a very useful cue for bees, it seems surprising that social
cues should be more salient, and consideration of these findings
in the context of blocking (66)—another mainstay of associative
learning theory—raises a testable alternative explanation.
Association between a CS (e.g., flower color) and a US (e.g.,
sucrose) is typically impaired if that CS is presented together with
another CS that has previously been associated with the US. This
phenomenon could potentially explain Dunlap et al.’s (65) results,
because the bees have prior experience of social foraging but not
of the flower colors in question. As members of laboratory colo-
nies that forage ad libitum from feeders placed in flight arenas
when not participating in experiments, individuals would have had
an opportunity to associate conspecific presence with reward, but
no similar experience of flower colors. During the training, bees
should thus learn little about the reliability of the floral CS, be-
cause the simultaneous presence of the social CS blocks learning
about the asocial cue. The only situation where bees should learn
about the floral CS would occur if the social CS became unreli-
able. In this situation, the association between conspecific pres-
ence and sucrose rewards should move toward extinction, allowing
for new learning of the association between floral color and sucrose,
EVOLUTION
exactly as Dunlap et al.’s results illustrate.
It may well be the case that bees treat asocial and social cues
differently, and post hoc associative explanations should gener-
ate testable predictions to be useful. Here, one simple means to
rule out a blocking explanation would be to closely control the
previous foraging experience of the bees. In this study system,
where individuals never leave the flight arena and the colony’s
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
foraging needs can be met by providing sucrose and pollen di-
rectly into the nest, it is entirely possible to create individual
foragers with no experience of social foraging (43, 67, 68).
Finding the same effect under these circumstances would render
the case for enhanced salience of social stimuli compelling, and a
further experiment by Dawson and Chittka (68) employs such an
approach. These authors compared the speed at which bum-
blebees (B. terrestris) learned to associate either a social CS (the
presence of dead, pinned conspecifics) or an asocial CS (a coin/
plastic disk/black wooden cuboid) with sucrose, in a free-flying
choice paradigm. Bees were more likely to associate the social
than asocial CS with reward, and were also more likely to use the
social CS to identify rewarding flowers in a transfer test involving
a novel target flower color. On realizing that their results could
reflect the fact that subjects had previous experience of social
foraging from laboratory feeders, Dawson and Chittka repeated
their experiment using bees that had never foraged with others,
and found that the effect was maintained. Interestingly, pinned
honey bee (A. mellifera) demonstrators, which visit comparable
but not identical floral resources, elicited similar results, suggesting
that natural selection might lead to increased salience of cues that
derive from useful heterospecifics, as well as conspecifics.
Social Associations
Salience is an important determinant of the ease with which an
association is acquired, but it is by no means the only contrib-
uting factor. Whereas salience affects the extent to which a
stimulus is made available for learning, learning itself requires
that associations between neural representations of stimuli, af-
fective states, or motor patterns, are formed around that stim-
ulus. Natural selection might act upon the parameters that
determine the number and timing of exposures that are required
before a particular association is committed to memory (69). The
key feature of such prepared learning is that certain combi-
nations of stimuli, rather than any particular stimulus alone,
elicit rapid learning. For example, consider the oft-cited “Garcia
effect,” whereby rats rapidly learn to associate tastes but not
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7841
audible tones with gastrointestinal illness. The combination of
taste and illness is critical here, and the same effect is not ob-
served when tastes are associated with shock (35, 70), ruling out
the suggestion that tastes are simply more salient stimuli than
tones. In fact, Garcia and Koelling (35) found that tones were
more easily associated with shock than taste was, potentially
because loud noises are likely to be a relevant cue to imminent
pain, whereas taste is not.
In a social learning context, there are clear candidate hypoth-
eses concerning stimulus combinations that await investigation.
For example, above we discussed evidence that bees might acquire
associations between nectar rewards and social stimuli more rap-
idly than associations between nectar rewards and asocial stimuli
(65, 68). A prepared learning hypothesis would predict that social
stimuli might be particularly easy to associate with sucrose reward
levels, but not with an aversive stimulus. Or alternatively, perhaps
the sign of the CS–US relationship is also important, such that
bees learn positive sucrose/social cues relationships easily but not
negative ones. These latter two alternatives—that bees are
particularly sensitive to social CSs when learning about where to
find food, or that they are particularly likely to learn positive social
CS-sucrose combinations—are qualitatively different traits. The-
ory correspondingly predicts that they should evolve under dif-
ferent circumstances (36, 71). To visualize the difference, consider
again the Garcia effect (35). Although many stimuli might precede
a feeling of illness, the true cause will often be related to recently
eaten food, so an a priori prioritization of associations between
taste and illness, rather than sound and illness, makes evolutionary
sense. Now consider the situation where a particular flavor always
predicts illness. Here, theory would predict the fixation of an
aversion to that flavor, rather than prepared learning (71). Thus, a
priori expectations and prepared learning about social stimuli are
both means by which natural selection could facilitate social
learning processes, but the ecological conditions that favor their
evolution are distinct. Teasing apart the roles of stimulus salience,
prepared learning, and a priori expectation is a task that invites
empirical exploration in our social insect system and in others.
Retrieval and Implementation of Learned Associations
We have discussed potential pathways by which natural selection
could modify stimulus salience, or the downstream learning pa-
rameters that influence memory formation, and suggested means by
which such hypotheses could be explored. However, we have not yet
touched upon the possibility that selection could produce modifi-
cations to the final retrieval or implementation of learned in-
formation. This is particularly important in light of the large volume
of literature on “social learning strategies” that describes how ani-
mals use social information most often in those situations in which it
is most beneficial (72, 73). We begin with an example that illustrates
how associative processes could bring about such context-specificity.
As we have alluded to in an earlier example in this paper,
foraging bees suffer predation by camouflaged sit-and-wait pred-
ators that ambush individuals as they land on flowers to feed.
Dawson and Chittka (74) allowed bees to forage in environments
that mimicked high or low risk of such predation, and found that
bumblebees (B. terrestris) appear to use social information to
identify safe flowers specifically in dangerous environments. Their
subjects were initially trained on an array where landing on flowers
of one color morph led to brief capture in a pressure trap (sim-
ulating spider attack), whereas an alternative color morph was
safe. Note that the color morphs simulate flower species with
different levels of spider occupancy, rather than spiders them-
selves, which are typically cryptic. When subsequently tested on an
array containing only flowers of the dangerous morph, bees
strongly preferred to land on the single flower where a live dem-
onstrator could be seen feeding, an effect that was entirely absent
when tested on flowers of the safe morph (Fig. 3). In other words,
bees seemed to use social information adaptively when foraging
7842 | www.pnas.org/cgi/doi/10.1073/pnas.1620744114
on flower types where predation was a real threat. Bees could not
have simply learned that the social cue was useful on dangerous
flowers but irrelevant on safe ones because individuals were
trained on the mixed array entirely alone.
How might associations between stimuli produce such context-
specific and potentially adaptive behavior? Conditioned suppres-
sion is an associative phenomenon widely used to study the ac-
quisition and extinction of fear, and it predicts exactly the effect
that Dawson and Chittka (74) demonstrated in bumblebees (30).
In the presence of a cue predicting an aversive stimulus that an
animal would usually avoid, learned food-seeking behaviors are
typically repressed, although they do not disappear altogether
(75). Thus, a rat that has learned to press a lever for food typically
reduces the frequency of pressing when a light that predicts foot
shock is turned on (76). Similarly, when under predation pressure
from a threat that is hard to directly detect, bumblebees increase
both their latency to probe and rejection rate of all flowers of the
color morph associated with danger (77, 78). With this in mind,
picture a bee that enters a flight arena filled with flowers of the
safe color morph, and happens to first come across an unoccupied
flower. Because the bee has learned that the holes in the flowers
reliably contain sucrose rewards, it is very likely to land and feed.
If it instead first comes across an occupied flower where a dem-
onstrator is foraging, it is also very likely to land and feed. Even if
the presence of the demonstrator renders flowers much more
attractive, the difference in acceptance rates between unoccupied
(very attractive) and occupied (extremely attractive) flowers will
be hard to detect experimentally, because all flowers are very
likely to be accepted. Now consider an environment where the
floral color morph predicts danger. All conditioned responses will
be suppressed, such that flowers in general are less likely to be
accepted. In this situation, if the presence of a demonstrator bee is
attractive, the effect is much more likely to be detected experi-
mentally (Fig. 4), because it is no longer the case that a bee almost
invariably accepts the first flower that it encounters.
This associative hypothesis does not require that selection
deriving from risky contexts has influenced the weight that bees
ascribe to social information. It is simply an alternative to the
suggestion that individuals strategically respond to the circum-
stances in which they find themselves by computing the likely
pay-offs of to social information use. However, our hypothesis
again generates a testable prediction. If dangerous environments
simply render the effect of an attractive social stimulus more
detectable, the same should be true for an attractive asocial
stimulus. Thus, replacing conspecific demonstrators with asocial
stimuli that have previously been conditioned to sucrose should
produce analogous results.
A study by Smolla et al. (79) employs exactly this approach in a
different social learning context. Based on the premise that bees
should use a “copy-when-uncertain” strategy, which follows from an
agent-based model that they develop, these authors pretrained bees
(B. terrestris) that either a social (model bee) or an asocial (green
rectangle) cue predicted reward in a floral array. Half of the bees in
each group were then trained that the floral array contained highly
variable rewards and half learned that rewards were constant, in the
absence of both cue types. When subsequently presented with a
nonrewarding test array, those bees that had learned that rewards
were variable used the social cue to find food, but those that had
experienced the constant array did not. Their results thus support a
“copy-when-uncertain” interpretation, but as pointed out above, the
fact that more variable rewards render flowers less attractive (80, 81)
means that the use of any cue, not just social ones, should be ren-
dered more detectable. However, crucially, the difference between
contexts was much less evident for the asocial cue and did not reach
statistical significance. This difference seems unlikely to be attributed
to greater salience or associability of the social cue, because Smolla
et al. (79) state that pretraining was similarly successful for both cue
types. Smolla et al.’s results invite further exploration that has not yet
Leadbeater and Dawson
 COLLOQUIUM
PAPER
Fig. 3. Social information use varied with context (74). Bees were trained that sucrose could be obtained by extending the proboscis through holes demarked

EVOLUTION
by colored squares. During training, one color (here yellow) predicted brief capture in a pressure trap. Bees were then tested in either a “dangerous” or a
“safe” environment, where one live demonstrator was foraging at a single location.

been carried out, but their approach of comparing context-specific learning should be accredited with the status of a “default” ex-
responses to social and asocial stimuli is a promising one that seems planation for social learning phenomena, to be assumed true

a useful way to evaluate the specific characteristics that govern
learning about social stimuli.
Competing Hypotheses
As a whole, the insect-based studies that we have discussed il-
lustrate the power of associative learning to generate adaptive
behavior. There is nothing new about associative explanations
for social learning phenomena; Heyes in particular has long
championed this approach (24, 28–30, 39, 40). However, the fact
that a hypothesis has explanatory power is not evidence of its
truth, and consequently, debate arises over whether associative
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
unless proven false (24). Arguments that favor this approach in
animal cognition are typically based on Morgan’s canon, which
states that “no animal activity should be interpreted in terms of
higher psychological processes if it can be fairly interpreted in
terms of processes that stand lower in the scale of psychological
evolution and development” (82), but this has led to debate over
whether Morgan’s canon itself it up to the task (24).
Perhaps such a polarized perspective is less productive than it
could be. If there were clear evidence that social learning pro-
cesses were more efficient in species where social information
repeatedly presents itself than in more solitary species (for

A Safe B Dangerous
1.5

1.5
Probability of Acceptance

Probability of Acceptance
1.0

1.0
0.5

0.5
0.0

0.0

Unoccupied Occupied Unoccupied Occupied

Fig. 4. Conditioned suppression should render social effects more detectable when aversive stimuli are present. (A) Safe environment. We assume that bees have
learned to associate the feeding holes in artificial flowers with sucrose rewards. Therefore, in the absence of aversive stimuli, the chances of acceptance on en-
countering a flower are high (here set at 0.9, for illustration purposes). The presence of a demonstrator is also attractive (here set at 0.45), perhaps because bees have
previously learned to associate conspecifics with sucrose. If two appetitive conditioned stimuli are presented together, the subject’s expectation is equal to their
combined strength (62), so occupied flowers are very attractive, but a probability of acceptance cannot exceed 1. Thus, the detectable effect of demonstrator presence
is small (arrow). (B) Dangerous environment. The presence of an aversive stimulus (the dangerous flower color) reduces all responses for food (here, suppression ratio
has been set at 0.5). Thus, the difference in the probability of acceptance between the unoccupied and occupied flowers is now relatively more detectable (arrow).

Leadbeater and Dawson PNAS | July 25, 2017 | vol. 114 | no. 30 | 7843
example, in the social vs. solitary insects, which span multiple
origins of sociality), it might be argued that the two are suffi-
ciently different traits to justify discussion of evolutionary par-
simony. However, such evidence is sparse. Most studies that
compare learning efficiency across social and asocial contexts
conclude that the two are highly correlated (83–85; but see also
ref. 86). Here, we have suggested that a productive way forward
might be to search for small quantitative differences between
associative processes in asocial and social contexts, rather than
qualitative leaps. Heyes (29) has highlighted that the pathway of
least resistance for natural selection might be to modify the input
mechanisms that determine which aspects of the world are made
available for learning, contrasting such processes with learning
mechanisms that determine how stimuli are linked and com-
mitted to memory. Increased salience of social stimuli is one
such input mechanism, but increased salience of social stimuli
could be evolutionarily advantageous for many reasons other
than acquiring information, such as recognizing kin, selecting
mates, or defending a territory (87). Perhaps more convincing
evidence that natural selection has shaped responses to social
information would be evidence of prepared learning about social
stimuli, of unconditioned responses to social stimuli that were
specific to a social learning context, or of social learning strate-
gies that cannot be accounted for by associative learning theory.
Our choice of study taxon—the social insects—has meant that
we have made little mention of processes that characterize mainly
primate behavior, such as imitation and (to a lesser extent) em-
ulation. Nonetheless, very recent work has begun to focus on
potentially emulative behavior even in bees (8, 9), and it is not our
intention to imply that such processes follow a different evolu-
tionary pathway to other forms of social learning. In fact, asso-
ciative explanations for imitation are prominent in the psychology
literature [40, 88; see also Lotem et al. (33)]. Explaining how in-
dividuals copy a novel sequence of actions through imitation in-
vokes a “correspondence problem” (89, 90) because the seen
movements of others must somehow be matched to motor rep-
resentations of self-movements. Associative models of imitation
propose that such links could arise through previous experience of
contingency between performing an action and seeing it per-
formed (91). For example, an infant might often observe others
smiling when she or he smiles, leading to association between the
visual and motor representations of smiling. She or he will also
typically observe a raised arm each time that she raises her own
arm, or react to an unexpected surprise with the same expressions
as others nearby. This elegant idea generates both theoretical
debate and extensive empirical exploration. However, imitation is
but one social learning mechanism, and this intensive exploration
1. Hoppitt W, Laland K (2013) Social Learning: An Introduction to Mechanisms, Methods
and Models (Princeton Univ Press, Princeton, NJ).
2. Grüter C, Leadbeater E (2014) Insights from insects about adaptive social information
use. Trends Ecol Evol 29:177–184.
3. Leadbeater E, Chittka L (2007) Social learning in insects—From miniature brains to
consensus building. Curr Biol 17:R703–R713.
4. Mery F, et al. (2009) Public versus personal information for mate copying in an in-
vertebrate. Curr Biol 19:730–734.
5. Worden BD, Papaj DR (2005) Flower choice copying in bumblebees. Biol Lett 1:
504–507.
6. Leadbeater E, Chittka L (2007) The dynamics of social learning in an insect model, the
bumblebee (Bombus terrestris). Behav Ecol Sociobiol 61:1789–1796.
7. Leadbeater E, Chittka L (2008) Social transmission of nectar-robbing behaviour in
bumble-bees. Proc Biol Sci 275:1669–1674.
8. Alem S, et al. (2016) Associative mechanisms allow for social learning and cultural
transmission of string pulling in an insect. PLoS Biol 14:e1008529; erratum in 14(12):
e1008529.
9. Loukola OJ, Perry CJ, Coscos L, Chittka L (2017) Bumblebees show cognitive flexibility
by improving on an observed complex behavior. Science 355:833–836.
10. Battesti M, Moreno C, Joly D, Mery F (2015) Biased social transmission in Drosophila
oviposition choice. Behav Ecol Sociobiol 69:83–87.
11. Dagaeff A-C, et al. (2016) Drosophila mate copying correlates with atmospheric
pressure in a speed-learning situation. Anim Behav 121:163–173.
7844 | www.pnas.org/cgi/doi/10.1073/pnas.1620744114
of mechanism is not the norm. It is surprising that for the vast
majority of cases in which animals respond to social information,
we understand very little about underlying mechanisms at all (41).
To return to Galef’s statement that “an entire field of local en-
hancement awaits exploration” (41), we suggest that this explo-
ration begins with a full understanding of how reported “simpler”
social learning phenomena fit into an associative framework.
Conclusion
We became interested in insect social learning based on a
thought-provoking anecdotal observation that the acceptance of
a laboratory feeder by a single bumblebee seemed to render that
feeder popular, rather than through careful choice of a study
system. It was surprising to subsequently find that insects seemed
capable of what we considered a relatively sophisticated cogni-
tive process. However, questions about cognition in invertebrates
inevitably involve detailed dissection of mechanisms, and a
critical feature of this system has been the ability to closely
control the previous social foraging experience of experimental
subjects in a way that would not be possible for many species. We
hope that the body of research presented above shows that there
was value in the novel perspective offered by a rather unlikely
model. The studies that we have described are based on a small-
brained study organism, but one that excels at accomplishing the
cognitive tasks that are relevant to its own ecological niche (92),
that is highly social, and that offers exceptional experimental
tractability. Other insect systems offer similar advantages for
investigating social learning contexts that we have not mentioned
in detail here; for example, a large literature now documents the
existence of remarkably rapid and generalizable mate-choice
copying in Drosophila fruit flies (3, 11, 12). It may well be the
case that social learning abilities have traveled further along
some evolutionary routes in other lineages, and less far in others,
but the rules that govern associative learning are taxonomically
widespread within the animal kingdom, and natural selection
may well coopt the same cognitive raw material across multiple
evolutionary lineages.
ACKNOWLEDGMENTS. We thank the organizers and funders of the Arthur M.
Sackler Colloquium on “The Extension of Biology Through Culture,” from
which this paper derives, the many participants at the colloquium who pro-
vided informative feedback and discussion, and Simon Reader for comments
on an earlier draft of the manuscript. Several of the empirical studies pre-
sented here were coauthored by Lars Chittka, and the ideas that we have
discussed owe much to his direction and guidance. E.L. is funded by Euro-
pean Research Council Starting Grant BeeDanceGap, and the empirical
work described here also derives from an Early Career Fellowship from
The Leverhulme Trust. E.H.D.’s current position is funded by a Fyssen foun-
dation postdoctoral fellowship.
12. Battesti M, Moreno C, Joly D, Mery F (2012) Spread of social information and dy-
namics of social transmission within Drosophila groups. Curr Biol 22:309–313.
13. Goulson D, Park KJ, Tinsley MC, Bussiere LF, Vallejo-Marin M (2013) Social learning
drives handedness in nectar-robbing bumblebees. Behav Ecol Sociobiol 67:1141–1150.
14. Danchin E, Giraldeau LA, Valone TJ, Wagner RH (2004) Public information: From nosy
neighbors to cultural evolution. Science 305:487–491.
15. Thompson R, McCONNELL J (1955) Classical conditioning in the planarian, Dugesia
dorotocephala. J Comp Physiol Psychol 48:65–68.
16. Kemenes G, Benjamin PR (1989) Appetitive learning in snails shows characteristics of
conditioning in vertebrates. Brain Res 489:163–166.
17. Walters ET, Carew TJ, Kandel ER (1981) Associative learning in Aplysia: Evidence for
conditioned fear in an invertebrate. Science 211:504–506.
18. Spatz HC, Emanns A, Reichert H (1974) Associative learning of Drosophila mela-
nogaster. Nature 248:359–361.
19. Rescorla RA (1988) Pavlovian conditioning. It’s not what you think it is. Am Psychol 43:
151–160.
20. Holland PC, Sherwood A (2008) Formation of excitatory and inhibitory associations
between absent events. J Exp Psychol Anim Behav Process 34:324–335.
21. Timberlake W (1994) Behavior systems, associationism, and Pavlovian conditioning.
Psychon Bull Rev 1:405–420.
22. Pearce JM, Bouton ME (2001) Theories of associative learning in animals. Annu Rev
Psychol 52:111–139.
23. Dickinson A (2012) Associative learning and animal cognition. Philos Trans R Soc Lond
B Biol Sci 367:2733–2742.
Leadbeater and Dawson

24. Heyes C (2012) Simple minds: A qualified defence of associative learning. Philos Trans
R Soc Lond B Biol Sci 367:2695–2703.
25. Dwyer D, Burgess KV (2011) Rational accounts of animal behaviour? Lessons from
C. Lloyd Morgan’s canon. Int J Comp Psychol 24:349–364.
26. Guilford T, Burt De Perera T (2016) An associative account of avian navigation. J Avian
Biol 48:191–195.
27. Shettleworth SJ (2010) Clever animals and killjoy explanations in comparative psy-
chology. Trends Cogn Sci 14:477–481.
28. Heyes CM (1994) Social learning in animals: Categories and mechanisms. Biol Rev
Camb Philos Soc 69:207–231.
29. Heyes C (2012) What’s social about social learning? J Comp Psychol 126:193–202.
30. Heyes C, Pearce JM (2015) Not-so-social learning strategies. Proc Biol Sci 282:7.
31. Galef BG (1992) The question of animal culture. Hum Nat 3:157–178.
32. Lotem A, Halpern JY (2012) Coevolution of learning and data-acquisition mecha-
nisms: A model for cognitive evolution. Philos Trans R Soc Lond B Biol Sci 367:
2686–2694.
33. Lotem A, Halpern JY, Edelman S, Kolodny O (2017) The evolution of cognitive mechanisms
in response to cultural innovations. Proc Natl Acad Sci USA 114:7915–7922.
34. Mery F, et al. (2007) Natural polymorphism affecting learning and memory in Dro-
sophila. Proc Natl Acad Sci USA 104:13051–13055.
35. Garcia JK, Koelling RA (1966) Relation of cue to consequence in avoidance learning.
Psychon Sci 4:123–124.
36. Dunlap AS, Stephens DW (2014) Experimental evolution of prepared learning. Proc
Natl Acad Sci USA 111:11750–11755.
37. Truskanov N, Lotem A (2017) Trial-and-error copying of demonstrated actions reveals
how fledglings learn to ‘imitate’ their mothers. Proc Biol Sci 284:20162744.
38. Ho MK, MacGlashan J, Littman ML, Cushman F (March 21, 2017) Social is special: A nor-
mative framework for teaching with and learning from evaluative feedback. Cognition,
10.1016/j.cognition.2017.03.006.
39. Heyes C (2017) When does social learning become cultural learning? Dev Sci 20:
e12350.
40. Cook R, Bird G, Catmur C, Press C, Heyes C (2014) Mirror neurons: From origin to
function. Behav Brain Sci 37:177–192.
41. Galef BG (2013) Imitation and local enhancement: Detrimental effects of consensus
definitions on analyses of social learning in animals. Behav Processes 100:123–130.
42. Heinrich B (1979) Bumblebee Economics. (Harvard Univ Press, Cambridge, MA).
43. Dawson EH, Avarguès-Weber A, Chittka L, Leadbeater E (2013) Learning by obser-
vation emerges from simple associations in an insect model. Curr Biol 23:727–730.
44. Avarguès-Weber A, Chittka L (2014) Observational conditioning in flower choice
copying by bumblebees (Bombus terrestris): Influence of observer distance and
demonstrator movement. PLoS One 9:e88415.
45. Schultz W, Dickinson A (2000) Neuronal coding of prediction errors. Annu Rev
Neurosci 23:473–500.
46. Rescorla R (1980) Pavlovian Second-Order Conditioning: Studies in Associative
Learning (Laurence Erlbaum Associates, Hillsdale, NJ).
47. Pavlov IP (1927) Conditioned Reflexes (Oxford Univ Press, Oxford).
48. Winterbauer NE, Balleine BW (2005) Motivational control of second-order condi-
tioning. J Exp Psychol Anim Behav Process 31:334–340.
49. Chittka L, Leadbeater E (2005) Social learning: Public information in insects. Curr Biol
15:R869–R871.
50. Cook M, Mineka S (1989) Observational conditioning of fear to fear-relevant versus
fear-irrelevant stimuli in rhesus monkeys. J Abnorm Psychol 98:448–459.
51. Mineka S, Cook M (1988) Social learning and the acquisition of snake fear in monkeys.
Social Learning: Pyschological and Biological Perspectives, eds Zentall TR, Galef BG
(Laurence Erlbaum Associates, Hillsdale, NJ), pp 51–75.
52. Mineka S, Cook M (1993) Mechanisms involved in the observational conditioning of
fear. J Exp Psychol Gen 122:23–38.
53. Ohman A (2009) Of snakes and faces: An evolutionary perspective on the psychology
of fear. Scand J Psychol 50:543–552.
54. Cook EW, 3rd, Hodes RL, Lang PJ (1986) Preparedness and phobia: Effects of stimulus
content on human visceral conditioning. J Abnorm Psychol 95:195–207.
55. Tomarken AJ, Sutton SK, Mineka S (1995) Fear-relevant illusory correlations: What
types of associations promote judgmental bias? J Abnorm Psychol 104:312–326.
56. Ohman A, Mineka S (2003) The malicious serpent: Snakes as a prototypical stimulus
for an evolved module of fear. Curr Dir Psychol Sci 12:5–9.
57. Zentall TR, Galef BG, eds (1988) Social Learning: Pyschological and Biological
Perspectives (Laurence Erlbaum Associates, Hillsdale, NJ).
58. Miller NE, Dollard J (1941) Social Learning and Imitation (Yale Univ Press, New Haven, CT).
59. Church RM (1968) Applications of behaviour theory to social psychology. Social
Facilitation and Imitative Behavior, eds Simmel EC, Hoppe RA, Milton GD (Allyn &
Bacon, Boston).
Leadbeater and Dawson
COLLOQUIUM
PAPER
60. Dawson EH, Chittka L, Leadbeater E (2016) Alarm substances induce associative social
learning in honeybees, Apis mellifera. Anim Behav 122:17–22.
61. Balderrama N, et al. (1996) A deterrent response in honeybee (Apis mellifera) for-
agers: Dependence on disturbance and season. J Insect Physiol 42:463–470.
62. Rescorla RA, Wagner AR (1972) A theory of Pavlovian conditioning: Variations in the
effectiveness of reinforcement and nonreinforcement. Classical Conditioning II:
Current Research and Theory, eds Black AH, Prokasy WF (Appleton-Century-Crofts,
NY), pp 64–99.
63. Shettleworth S (2010) Cognition, Evolution and Behavior (Oxford Univ Press, Oxford).
64. Galef BG, Jr, Mason JR, Preti G, Bean NJ (1988) Carbon disulfide: A semiochemical
mediating socially-induced diet choice in rats. Physiol Behav 42:119–124.
65. Dunlap AS, Nielsen ME, Dornhaus A, Papaj DR (2016) Foraging bumble bees weigh
the reliability of personal and social information. Curr Biol 26:1195–1199.
66. Kamin LJ (1969) Predictability, surprise, attention, and conditioning. Punishment and
Aversive Behavior, eds Campbell BA, Church RM (Appleton-Century-Crofts, New
York), pp 279–296.
67. Leadbeater E, Chittka L (2011) Do inexperienced bumblebee foragers use scent marks
as social information? Anim Cogn 14:915–919.
68. Dawson EH, Chittka L (2012) Conspecific and heterospecific information use in
bumblebees. PLoS One 7:e31444.
69. Domjan M, Galef BG (1983) Biological constraints on instrumental and classical-con-
ditioning—Retrospect and prospect. Anim Learn Behav 11:151–161.
70. Dwyer DM (2015) Experimental evolution of sensitivity to a stimulus domain alone is
not an example of prepared learning. Proc Natl Acad Sci USA 112:E385.
71. Dukas R (1998) Evolutionary ecology of learning. Cognitive Ecology, ed Dukas R (Univ
of Chicago Press, Chicago), pp 129–174.
72. Laland KN (2004) Social learning strategies. Learn Behav 32:4–14.
73. Kendal RL, Coolen I, Laland KN (2009) Adaptive trade-offs in the use of social and
personal information. Cognitive Ecology II, eds Dukas R, Ratcliffe JM, pp 249–271.
74. Dawson EH, Chittka L (2014) Bumblebees (Bombus terrestris) use social information as
EVOLUTION
an indicator of safety in dangerous environments. Proc Biol Sci 281:20133174.
75. Annau Z, Kamin LJ (1961) Conditioned emotional response as a function of intensity
of US. J Comp Physiol Psych 54:428–432.
76. Estes WK, Skinner BF (1941) Some quantitative properties of anxiety. J Exp Psychol 29:
390–400.
77. Ings TC, Chittka L (2008) Speed-accuracy tradeoffs and false alarms in bee responses
to cryptic predators. Curr Biol 18:1520–1524.
78. Lenz F, Ings TC, Chittka L, Chechkin AV, Klages R (2012) Spatio-temporal dynamics of
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
bumblebees foraging under predation risk. Phys Rev Lett 108:098103.
79. Smolla M, Alem S, Chittka L, Shultz S (2016) Copy-when-uncertain: Bumblebees rely
on social information when rewards are highly variable. Biol Lett 12:20160188.
80. Shafir S, Wiegmann DD, Smith BH, Real LA (1999) Risk-sensitive foraging: Choice
behaviour of honeybees in response to variability in volume of reward. Anim Behav
57:1055–1061.
81. Seefeldt S, De Marco RJ (2008) The response of the honeybee dance to uncertain
rewards. J Exp Biol 211:3392–3400.
82. Lloyd Morgan C (1909) An Introduction to Comparative Psychology (Walter Scott
Publishing, London).
83. Lefebvre L, Giraldeau L-A (1996) Is social learning an adaptive specialization? Social
Learning in Animals: The Roots of Culture, eds Heyes CM, Galef BG, Jr (Academic,
London), pp 107–128.
84. Reader SM, Hager Y, Laland KN (2011) The evolution of primate general and cultural
intelligence. Philos Trans R Soc Lond B Biol Sci 366:1017–1027.
85. Reader SM (2003) Innovation and social learning: Individual variation and brain
evolution. Animal Biology 53:147–158.
86. Templeton JJ, Kamil AC, Balda RP (1999) Sociality and social learning in two species of
corvids: the pinyon jay (Gymnorhinus cyanocephalus) and the Clark’s nutcracker
(Nucifraga columbiana). J Comp Psychol 113:450–455.
87. Leadbeater E (2015) What evolves in the evolution of social learning? J Zool (Lond)
295:4–11.
88. Heyes C (2016) Homo imitans? Seven reasons why imitation couldn’t possibly be as-
sociative. Philos Trans R Soc Lond B Biol Sci 371:20150069.
89. Brass M, Heyes C (2005) Imitation: Is cognitive neuroscience solving the correspon-
dence problem? Trends Cogn Sci 9:489–495.
90. Nehaniv CL, Dautenhahn K (2002) The correspondence problem. Imitation in Animals
and Artifacts, eds Dautenhahn K, Nehaniv CL (MIT Press, Cambridge, MA).
91. Catmur C, Walsh V, Heyes C (2009) Associative sequence learning: The role of expe-
rience in the development of imitation and the mirror system. Philos Trans R Soc Lond
B Biol Sci 364:2369–2380.
92. Chittka L, Thomson JD (2001) Cognitive Ecology of Pollination (Cambridge Univ Press,
Cambridge, UK).
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7845
Cultural macroevolution matters
Russell D. Graya,b,c,1 and Joseph Wattsa,d
a
Department of Linguistic and Cultural Evolution, Max Planck Institute for the Science of Human History, Jena D-07745, Germany; bSchool of Psychology,
University of Auckland, Auckland 1142, New Zealand; cResearch School of the Social Sciences, Australian National University, Canberra, ACT 2601, Australia;
and dDepartment of Experimental Psychology, University of Oxford, Oxford OX1 3PH, United Kingdom
Edited by Andrew Whiten, University of St. Andrews, St. Andrews, United Kingdom, and accepted by Editorial Board Member Andrew G. Clark May 29, 2017
(received for review January 16, 2017)
Evolutionary thinking can be applied to both cultural microevolu-
tion and macroevolution. However, much of the current literature
focuses on cultural microevolution. In this article, we argue that
the growing availability of large cross-cultural datasets facilitates
the use of computational methods derived from evolutionary
biology to answer broad-scale questions about the major transi-
tions in human social organization. Biological methods can be
extended to human cultural evolution. We illustrate this argument
with examples drawn from our recent work on the roles of Big
Gods and ritual human sacrifice in the evolution of large, stratified
societies. These analyses show that, although the presence of Big
Gods is correlated with the evolution of political complexity, in
Austronesian cultures at least, they do not play a causal role in
ratcheting up political complexity. In contrast, ritual human
sacrifice does play a causal role in promoting and sustaining the
evolution of stratified societies by maintaining and legitimizing
the power of elites. We briefly discuss some common objections to
the application of phylogenetic modeling to cultural evolution and
argue that the use of these methods does not require a commit-
ment to either gene-like cultural inheritance or to the view that
cultures are like vertebrate species. We conclude that the careful
application of these methods can substantially enhance the
prospects of an evolutionary science of human history.
cultural evolution | macroevolution | phylogenetics | religion | Big Gods
arwin’s On the Origin of the Species ends with the poetic
D phrase, “From so simple a beginning endless forms most
beautiful and most wonderful have been, and are being, evolved”
(1). The central challenge for evolutionary biology is to explain
this diversity of endless forms. Evolutionary biologists tackle this
task by studying both microevolution (changes in gene frequency
within a population) and macroevolution (changes between
species over much longer time periods). The aim is to have a
mechanistic understanding of the evolution of biological diversity
that integrates microlevel processes and macrolevel patterns.
This work examines ways in which evolutionary thinking and
methods can be extended into the realm of culture, extending the
scope of biology to include questions that have traditionally been
restricted to the humanities and social sciences. Human cultures
also display a vast variety of most beautiful and most wonderful
forms. We speak ∼7,000 different languages, engage in hundreds
of different religious practices, build many different types of
houses, exploit different resources for subsistence, use numerous
different kinship systems, and abide by a striking array of marital,
sexual, and child-rearing norms (2). The cultural processes that
produce such striking cultural diversity must be explained. The
field of cultural evolution is currently beginning to blossom (Fig.
1). There is a new cultural evolution society, a proposed journal,
and an inaugural conference (3). However, with a few notable
exceptions (4), much of the current work on cultural evolution
focuses on microevolutionary processes. For example, in Dan
Sperber’s influential book Explaining Culture: A Naturalistic Ap-
proach (5), cultural macroevolution rates only a passing mention
on p. 2. More recently, in Lewens’ (6) otherwise masterful anal-
ysis of current work on cultural evolution, macroevolutionary
phenomena again fail to feature. This article is a plea—a plea for
7846–7852 | PNAS | July 25, 2017 | vol. 114 | no. 30
the importance of studying cultural macroevolution. Although we
completely understand the need for elegant empirical work and
appropriate models of cultural change within populations, we
should never forget that the large-scale patterns of diversity be-
tween cultures also cry out for evolutionary analyses and expla-
nation. The macro really matters.
Big(ish) Data and Need for Computational Methods
It is a cliché these days to talk about big data transforming the
social sciences. However, clichés can be true. Certainly, there are
a growing number of global comparative cultural and linguistic
databases, such as D-PLACE (2), DRH (7), WALS (8), ASJP
(9), and Phoible (10), as well as relatively large regional data-
bases, such as the Austronesian Basic Vocabulary Database (11),
SAILS (12), Chirilla (13), and Pulotu (14). Although these da-
tabases might not technically qualify as “big data,” they are large
enough to afford the application of the type of sophisticated
computational methods that are often used in the biological
sciences such as network analysis of reticulate evolution, epide-
miological models, and phylogenetic comparative methods.
These methods can be used to compare the relative importance
of different factors in the distribution of traits, model the un-
derlying dynamics of evolutionary change, and infer the history
of traits. The combination of big(ish) data and computational
methods has the potential to transform the social sciences and
humanities by enabling powerful quantitative tests of hypotheses
that would have previously only been analyzable in much more
limited ways.
To illustrate the promise of this approach, we present a recent
study by Botero et al. (15) titled, “The Ecology of Religious Be-
liefs,” in which the authors examined the global distribution of
moralizing high gods (MHGs)—supernatural beings who are
claimed to have created or govern all reality, intervene in human
affairs, and enforce or support human morality (sometimes re-
ferred to as “Big Gods”). These gods are central to the Abrahamic
religions, which includes the two largest religious families in the
world today, Christianity and Islam. Scholars have debated the
social and physical environments in which MHGs most readily
spread, and previous studies found rather contradictory results,
with resource scarcity both positively and negatively associated
with a belief in a MHG (16–18). These studies were limited by
the use of crude metrics of ecology or indirect measures of
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: R.D.G. and J.W. designed research, performed research, analyzed
data, and wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. A.W. is a guest editor invited by the Editorial
Board.
Data deposition: The data reported in this paper have been deposited in the D-PLACE
(https://d-place.org/home), Pulotu (https://pulotu.shh.mpg.de), and ABVD (https://abvd.
shh.mpg.de/austronesian/) databases.
1
To whom correspondence should be addressed. Email: gray@shh.mpg.de.
www.pnas.org/cgi/doi/10.1073/pnas.1620746114

2.5
Percentage of search results
2.0
1.5
1.0
0.5
1950 1960 1970 1980 1990 2000
Year
Fig. 1. Percentage of Google Scholar search results containing the term
“cultural evolution” from 1950 to 2010.
agricultural potential, relatively small sample sizes, and a failure
to account for the nonindependence of societies as a result of
spatial proximity and common ancestry (19). Botero et al. (15)
extracted fine-grained environmental data, as well as cultural,
linguistic, and geographic information from the open-access
D-PLACE database (https://d-place.org) (2). They used multimodel
inference to simultaneously evaluate the effects of environ-
mental variables, shared ancestry, geographic proximity, and so-
cial structures on belief in MHGs in 583 societies from around
the globe. Generalized linear models and generalized linear
mixed models were fitted in R (20) by using the lme4 (21) and
MuMin (22) packages. The best-fitting models included spatial
proximity, political complexity, animal husbandry, resource abun-
dance, and resource stability. Belief in MHGs was more prevalent in
societies from harsher environments and more likely in politically
complex societies that had animal husbandry. Strikingly, this mul-
timodel inference approach was able to predict the global distri-
bution of belief in MHGs in a separate sample of cultures with an
accuracy of 91%.
Major Transitions: Big Questions for Big(ish) Data
John Maynard Smith and Eörs Szathmáry’s 1995 book, The
Major Transitions in Evolution (23), is perhaps one of the most
important and insightful contributions to evolutionary theory
in the last 50 years. In this book, Maynard Smith and Szathmáry
not only document fundamental changes in biological organi-
zation, such as the emergence of the genetic code, the origins
of cells, the evolution of the eukaryotic cell, and multicellular-
ity, they also show how these changes in biological organiza-
tion change the way in which biological systems can evolve.
The major transitions create entirely new evolutionary possibil-
ities built upon new and more powerful ways of storing and
transmitting information (24). According to Maynard Smith
and Szathmáry (25), these transitions have at least five general
properties:
1. Smaller entities form larger entities;
2. Smaller entities become differentiated as part of the
larger entity;
3. The smaller entities are often unable to replicate in the ab-
sence of the larger entity;
4. The smaller entities can sometimes disrupt the development
of the larger entity; and
5. New ways of transmitting information arise.
Gray and Watts
COLLOQUIUM
PAPER
By themselves, large cross-cultural datasets and sophisticated
computational methods are insufficient to create an exciting
macroevolutionary science of human history. What is needed are
big hypotheses and a powerful synthetic framework. We would
like to suggest that much of the evolutionary thinking behind
The Major Transitions in Evolution could be applied to cultural
macroevolution (23).
Ten thousand years ago, most humans lived in small, kin-
based, and relatively egalitarian groups (26). Today, we live in
colossal nation states with distantly related members, complex
hierarchical organization, and huge social inequality. Although
kin selection and reciprocity explain a great deal of cooperation
in the animal kingdom, these mechanisms break down in modern
societies because the sheer scale of modern societies means that
people can be anonymous and only distantly biologically related
(27). The challenge is to explain the cultural forces that enabled
2010
this major transition in the size and complexity of human social
groups to occur.
The potential of religious beliefs and practices to bind to-
gether social groups has long been recognized (28), although
these functions have only recently been considered from an ex-
plicitly evolutionary perspective (29–31). One prominent theory
is that belief in supernatural punishment, particularly by pow-
erful and omnipotent Big Gods, inhibits selfishness and increases
cooperation among adherents (32–34). The intuitive idea is that
if people believe a punishing and moral supernatural agent is
monitoring them, they are more likely to behave themselves. By
facilitating cooperation in large groups of nonkin, beliefs in su-
pernatural punishment are thought to have played a causal role
in the emergence of large, complex human societies (33, 35, 36).
It is crucial to emphasize that, at least as it was initially formu-
lated, this hypothesis was both causal and directional. Big Gods
were needed to make big societies. For example:
It is no coincidence that the world is now dominated by a few great
monotheisms, and that much human behaviour is influenced by the
belief in a few high gods. To achieve a civilization of this scale, it was
necessary to invent them (36);
and
ANTHROPOLOGY
One reason societies were able to develop cultural complexity in the
first place is partly on account of the cooperative benefits attained
through a belief in moralizing gods (35).
In support of the Supernatural Punishment Hypothesis, a
number of cross-cultural studies have shown that belief in MHGs
is positively correlated with a range of measures of social com-
plexity, such as political hierarchy, agriculture, and taxation
systems (18, 35). On the face of it, the cross-cultural evidence for
EVOLUTION
the Supernatural Punishment Hypothesis appears compelling.
However, these studies have a number of important limitations
(37). First, these studies do not actually get at the direction of
causality. Although one possible explanation for these results is
that MHGs facilitate social complexity, another is that social
complexity makes cultures more likely to adopt MHGs. Second,
these studies are ether based on a single dataset called the
Ethnographic Atlas or a subset of this dataset known as the
Standard Cross-Cultural Sample (17, 18, 35, 38). The MHGs in
these datasets are almost all derived from the closely related
family of Abrahamic religions—Christianity, Judaism, and Islam
(37). These religions share a wide range of features, such as
providing a universal rather than ethnocentric doctrine and en-
couraging fertility, and it is not clear whether it is an MHG
specifically or some other part of these religions that is related
to social complexity (37, 39). Third, cultures often inherit traits
such as language, customs, oral traditions, and social norms from
their ancestors (19). These relationships between cultures mean
that cultures cannot be treated as statistically independent–a
problem famously first pointed out by Francis Galton (40, 41).
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7847
The studies mentioned above do not adequately account for
Galton’s Problem, so the correlation observed between the
presence of MHGs and social complexity might merely arise
because of the historical relationships between cultures (42).
Thus, to rigorously test hypotheses about the role of MHGs in
driving the major transitions in human history, we need data
from cultures with non-Abrahamic religions, as well as methods
that avoid Galton’s Problem and can explicitly test causal
predictions.
Phylogenetic methods have revolutionized the field of evolu-
tionary biology (43). These methods solve Galton’s Problem by
explicitly estimating ancestral state changes on phylogenetic
trees (19, 41). Thus, there is no overcounting or undercounting
of evolutionary events. Phylogenetic methods have recently been
used to make inferences about things such as the ancestral state
of postmarital residence patterns in Austronesian cultures (44),
the evolution of political complexity (45), the effects of cultural
ancestry on deforestation (46), and the links between cattle and
matrilinity (47). Mark Pagel and Andrew Meade introduced a
method called “Discrete” in the program BayesTraits that
models the evolution of two binary traits and tests between de-
pendent and independent models of evolution (48, 49). In an
independent model, the gains and losses of each trait are mod-
eled separately from each other (Fig. 2A). In the dependent
model, the rate at which a trait is gained or lost depends on the
state of the other trait, as would be expected if there is a causal
relationship between traits (Fig. 2 B and C). This approach gets
at the direction of causality by inferring the temporal order that
traits tend to arise and the effects they have on one another (49).
Using this approach and data from the Pulotu database (14,
49), we recently tested a series of hypotheses about the role of
religion in the emergence of social complexity (14, 50, 51). The
Pulotu database contains quantitative variables documenting the
traditional religious beliefs and practices as well as the social
organization of 106 Austronesian cultures. Special care was
taken in the coding of these data to ensure that, as far as pos-
sible, the coding reflected the state of the culture before con-
version to major world religions and colonization (14). Previously
published language-based phylogenies were used as a proxy for
the population history of these cultures (52). These trees fit re-
markably well with archaeological evidence that shows Austronesian-
speaking cultures were some of the greatest ocean voyagers in
human history, sailing from their homeland in Taiwan to settle
on islands ranging in size from the 0.4-km2 island of Anuta up to
the 785,000-km2 continental island of New Guinea (14, 53, 54).
The archaeological, genetic, and linguistic evidence suggests that
this expansion started ∼5,000 y ago and spread in a series of
expansion pulses and pauses through Island South East Asia and
the Pacific (52–55). The cultures that evolved on these islands
ranged from small kin-based groups, such as the Berawan (56),
up to federated kingdoms, such as Southern Toraja (57). Pop-
ulation sizes ranged from ∼200 people on Anuta (58) to ap-
proximately half a million people in the case of the Merina of
Madagascar (59). No less diverse were their religious systems,
with supernatural beliefs including anthropomorphic, animistic,
and nature deities, and religious rituals ranging in scale from
humble personal offerings to multiday community-wide festivals
(14). Because Austronesian cultures were some of the last cul-
tures in the world to have contact with major world religions, and
their traditional beliefs were well documented, they provide an
ideal sample for testing theories about the role of religion in the
emergence of social complexity.
We ran two series of analyses to test the Supernatural Pun-
ishment Hypothesis (50). In the first, we tested the effect of
Broad Supernatural Punishment on the evolution of political
complexity. Agents counted in this test included a wide range of
punishing and morally concerned supernatural agents, such as
ancestral spirits, natural spirits (e.g., forest and sky gods), and
mythical heroes, in addition to MHGs (14). Belief in Broad
Supernatural Punishment was found in just over two-thirds of the
cultures sampled. We found modest support for the coevolution
of Broad Supernatural Punishment and political complexity, with
Broad Supernatural Punishment facilitating the rise of political
complexity, but not helping to sustain it. In the second series
of analyses, we tested whether the specific belief in MHGs
coevolved with political complexity. We were surprised to find
evidence of MHGs in just 6 of the 96 traditional Austronesian
cultures we studied. Although our analyses suggested that MHGs
coevolved with political complexity, instead of MHGs driving
political complexity, our results indicated that MHGs tended to be
gained after political complexity had already emerged (Fig. 2C).
Our analyses suggested that these MHGs had been gained only
recently, and most of these MHGs occurred in regions where
there had been early contact with Muslim traders. Although we

Fig. 2. An independent model (A) of evolution alongside the dependent model predicted by the Supernatural Punishment Hypothesis (B) and the dependent
model resulting from analyses of traditional Austronesian cultures (C). The red figure represents the presence of a MHG, and the black figure represents the
presence of political complexity (PC). Arrows indicate the rates of change between states, and the width of the arrows are proportional to the size of the
transition rates. (A) In independent models of evolution, the rate at which each trait is gained or lost is independent of the state of the other trait. In this
example, cultures are more likely to gain PC than to lose it (rate c is lower than rate d). (B) In dependent models of evolution, the rate at which each trait is
gained and lost can be dependent on the state of the other. In the model predicted by the Supernatural Punishment Hypothesis, the rate at which PC
is gained is higher when a MHG is present (rate d) than when it is absent (rate b), and the rate at which PC is lost is lower when an MHG is present (rate g) than
when an MHG is absent (rate e). (C) The resulting models from our analyses suggested that MHGs had little effect on the gain and loss of PC, but that MHGs
were rarely gained in cultures without PC (rate a is lower than rate f).

7848 | www.pnas.org/cgi/doi/10.1073/pnas.1620746114 Gray and Watts


excluded all clear cases of direct borrowings from Abrahamic re-
ligions, it is likely that the concept of a MHG was subtly borrowed
and transferred to the names of indigenous deities (60).
Defenders of the Big Gods hypothesis might argue that, for
some reason, the Austronesian cultures do not reflect the role
MHGs played in the emergence of social complexity in other
regions of the world. However, a closer examination of previous
cross-cultural studies suggests otherwise. Of the 40 MHGs in the
Standard Cross Cultural Sample, 32 are of Christian or Islamic
origin, and the remaining 8 are either other Abrahamic religions
or plausibly influenced by them (37). The Abrahamic religions
arose ∼3,000 y ago, long after humans had begun forming large,
sedentary, and complexly organized societies (26). Although
there is a substantial body of experimental research showing that
Abrahamic MHGs can increase cooperation within groups (61),
the timing of their origin means that they cannot explain at least
the initial emergence of social complexity in human history. This
finding tells us that microlevel processes observed in contem-
porary cultures do not necessarily explain the macroevolutionary
patterns observed in human history.
An alternative vein of scholarship has focused on the darker
role of religion in human social life (62, 63). Archaeological,
historical, and ethnographic records reveal that in early societies
religious and political authority often overlapped (26), providing
ample opportunities for elites to use religious systems toward
their own ends. As a result, religious narratives in early human
societies often legitimize the authority of those in power and
involve rituals that benefit the elite at the expense of under-
classes (64). A particularly gruesome example is the practice of
ritualized human sacrifice that occurred in early human societies
throughout the world (64–68). According to the Social Control
Hypothesis (64, 66, 68), ritualized human sacrifice was used by
social elites as a religiously sanctioned means of terrifying un-
derclasses into obedience.
To test the Social Control Hypothesis, we went back to the
Pulotu database (14), coded variables on human sacrifice and
social stratification, and tested for their coevolution (51). The
term “social stratification” refers to inherited differences in
wealth and status and is thought to have been one of the earliest
forms of hierarchical structuring to emerge in human history
(26). We found human sacrifice to have been remarkably com-
mon in traditional cultures, occurring in almost half of those
sampled (51). Typically, social elites orchestrated the sacrifices,
with social underclasses becoming the victims. The results of our
analyses showed that human sacrifice coevolved with social strati-
fication and functioned to stabilize social inequality in general, as
well as facilitated the emergence of rigid class systems (Fig. 3). This
result does not imply that human sacrifice was necessarily functional
for the whole group, nor that it would have these effects in modern
societies, which have developed more sophisticated methods of
sustaining social inequality. What our results do show is that ritual
human sacrifice was used by social elites as a tool to maintain their
social standing in the early stages of social complexity.
Overextension of Biological Metaphors and Methods?
The famous evolutionary biologist Richard Lewontin often liked
to cite Rosenblueth and Wiener’s quip that, “The price of met-
aphor is eternal vigilance” (69). One of the things that Lewontin
is particularly skeptical about is the metaphorical extension of
evolutionary ideas to cultural history (70, 71). Part of this
skepticism is driven by his opposition to Dawkins’ meme concept
(72). Fracchia and Lewontin write (70):
But, unlike genes, memes are not entities with an existence inde-
pendent of the theory. They are a mental construct whose only de-
fined property is to fill in the gap in an elaborate metaphor.
However, Lewontin’s own three central principles for systems
to evolve by natural selection (phenotypic variation, differential
Gray and Watts
COLLOQUIUM
PAPER
fitness, and inheritance) are satisfied by cultural systems (4, 73).
As Henrich and Boyd have shown, adaptive cultural evolution
does not require replicator-like inheritance systems (74).
There is, however, another line of skepticism that is sometimes
directed against attempts to apply phylogenetic methods to cul-
tural evolution. Lewontin’s former colleague, Steven Jay Gould
(75), put this view with characteristic vigor:
Human cultural evolution proceeds along paths outstandingly dif-
ferent from the ways of genetic change. . . Biological evolution is
constantly diverging; once lineages become separate, they cannot
amalgamate (except in producing news species by hybridization—a
process that occurs very rarely in animals). Trees are correct topol-
ogies of biological evolution. . . In human cultural evolution, on the
other hand, transmission and anastomosis are rampant. Five minutes
with a wheel, a snowshoe, a bobbin, or a bow and arrow may allow an
artisan of one culture to capture a major achievement of another.
Although Gould may have hugely overestimated how easy it is
to reverse-engineer the manufacture of these items, the critique
of cultural phylogenetics has not gone away. Recently, Norenzayan
et al. (76) stated:
We caution against rushing to embrace analytical techniques im-
ported from genetic evolution – used to reconstruct species phylog-
enies – to cultural evolution. Cultural evolution is in some crucial
respects unlike genetic evolution. . . Species, for example, are not
subject to intergroup competition that creates massive and directed
horizontal transmission of only some traits. Therefore, we think the
first step should be to benchmark phylogenetic techniques to cultural
history using known historical cases.
For the sake of clarity, we should be clear that we are not
advocating the blanket adoption of phylogenetics to all cultural
phenomena. So, let us look more closely at what the legitimate
concerns may or may not be. The statements above could be
boiled down to four linked, but logically separate, claims:
1. Culture evolves differently from biology. Biological evolution
is treelike, but in culture reticulation rules.
2. Cultures are not (vertebrate) species. Different aspects of
culture will have quite different histories.
3. The estimation of phylogenetic trees will be biased by
ANTHROPOLOGY
horizontal transmission.
4. The accuracy of cultural phylogenies has not been validated.
The first claim displays a shocking lack of knowledge of bi-
ology and human culture. There is a great deal of biology that
does not fit tidily on the “tree of life.” Indeed, the tree of life has
been mocked as the “tree of 1%” (77). A very significant amount
of cross-lineage transfer occurs in biological evolution, especially
in microbes (78). Mallet (79) estimated that there is hybridiza-
EVOLUTION
tion in ∼10% of animal and 25% of plant species. Dagan and
Martin’s (80) analysis of 190 prokaryotic genomes suggests that
horizontal gene transfer has affected at least two-thirds of >57,000
gene families.
In the literature on cultural microevolution, there is evidence
that the majority of social learning occurs between members of
the same population, but the relative importance of parent-to-
offspring and peer-to-peer social learning is debated (81–84).
What matters for the application of phylogenetic methods are the
resulting macroevolutionary patterns. Given that social learning
occurs predominantly within a population, both peer-to-peer and
parent-to-offspring learning can result in vertical transmission at
the macroevolutionary level. The relative importance of vertical
and horizontal transmission between populations is likely to vary
across domains of culture, world regions, and periods of history.
For example, the design of the internal combustion engine has
been borrowed between cultural lineages. Conversely, basic vo-
cabulary items, such as terms for hand and eye, lower numerals,
and kinship terms show clear evidence of vertical transmission
down cultural lineages (85).
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7849
 A Atayal
Tsou
Ami
Bunun
Puyuma
Paiwan
Sama Dilaut
Tagbanwa
Palawan Batak
Bukidnon
Subanun
Isneg
Gaddang
Bontok
Ifugao
Kalinga
Tinguian
Ngaju
Merina
Bidayuh
Kayan
Berawan
Kelabit
Moken
Iban
Toba Batak
Southern Toraja
Minahasa
Eastern Toraja
Nias
Palau
Chamorro
Ata Tana ‘Ai
Kedang
Manggarai
Savu
Laboya
East Sumba
Tanimbar
Tetum
Atoni
Roti
Biak
Waropen
Lakalai
Wogeo
Manam
Dobu
Trobriand Islands
Mekeo
Motu
Nendo
Bughotu
Lau
Kwaio
To’abaita
Kwara’ae
Tolai
Buka
Cheke Holo
Choiseul
Simbo
Roviana
Lifou
Mare
Eromanga
Tanna
Mota
Southern Malekula
Kiribati
Marshall Islands
Pohnpei
Woleai
Chuuk
East Fiji
Tonga
a = gain of HS
Samoa
Anuta
Rennell and Bellona
East Futuna
B c = loss of HS
Niue

b = gain of SS

d = gain of SS
g = loss of SS
West Futuna

e = loss of SS
Tikopia
Kapingamarangi
Ontong Java
Tokelau
Rapa Nui
Marquesas
Mangareva f = gain of HS
Rarotonga
Maori
Hawaii
Tahiti h = loss of HS

Fig. 3. (A) Ancestral state reconstruction of human sacrifice and social stratification on a maximum clade credibility consensus tree of 93 Austronesian
languages. The circles at the tips of the tree represent the known traditional states of cultures, and the circles found across the nodes of the tree represent the
state of prehistoric cultures inferred by a Markov chain Monte Carlo analysis in BayesTraits. In the analysis, 4,200 of the most likely possible trees were used,
and the consensus tree is a summary of these trees for illustrative purposes. The gray at each of the internal nodes represents the proportion of trees sampled
without this node and provides an indication of phylogenetic uncertainty. (B) The resulting dependent model shows that cultures with ritualized human
sacrifice were less likely to lose social stratification than those that lacked human sacrifice (rate g is lower than rate e). Adapted from ref. 51.

The second claim is more sensible, but does not undermine
the use of phylogenetic methods. If anything, it points out an
important role for their use—to assess the coherence of differ-
ence aspects of culture. In their book chapter, “Are Cultural
Phylogenies Possible?”, Boyd et al. (86) describe a range of
positions along a continuum on the question of how integrated
cultural histories are: (i) Cultures are tightly integrated like
vertebrate species; (ii) cultures contain a core of traditions that
are tightly linked and vertically transmitted, with peripheral as-
pects that are less cohesive and marked by frequent borrowing;
(iii) cultures contain some aspects that are bound together, but
there are no core traditions; and (iv) cultures are collections of
ephemeral entities. Just as biologists talk about “every gene
having its own history,” and have developed methods to map
these gene genealogies on to a species phylogeny, so cultural
phylogenetists could construct trees for different aspects of cul-
ture and evaluate their fit with population history (87, 88). For
example, genealogies of religious beliefs, material culture, kinship
systems, music genres, and styles of art could be mapped and
compared with language-based cultural histories. Phylogenetic
methods make the traditional social science debate about the
extent to which a culture is an integrated whole testable.
The third objection—that the estimation of cultural phylog-
enies will be biased by horizontal transmission—is a quantitative
issue that can be evaluated by simulation modeling. Greenhill
et al. (89) simulated language phylogenies with different tree
topologies, different borrowing scenarios, and different levels
of borrowing. The results show that tree topologies constructed
with Bayesian phylogenetic methods are robust to realistic
levels of borrowing. Inferences about divergence dates were
slightly less robust and showed a tendency to underestimate
dates.
The final objection—have inferences from cultural phyloge-
netics been validated?—is a fair enough concern, but one that
applies to much of computational biology, and indeed the ex-
trapolation of laboratory studies to the field. In brief, we will
point out that the Austronesian languages phylogenies built from
basic vocabulary fit strikingly well with both archaeological (55)
and recent genetic data (90, 91), both in terms of the sequence
and the timing of the Austronesian expansion.
Conclusion
In the coming years, more quantitative phylogenies for the major
language families will be published, and the number and richness
of comparative cultural databases will undoubtedly grow (7, 92).

7850 | www.pnas.org/cgi/doi/10.1073/pnas.1620746114 Gray and Watts


The series of studies we have discussed in this work illustrate
how causal theories about the emergence of major transitions in
human social organization can be tested with the combination of
large quantitative cross-cultural data and computational phylo-
genetic methods. We do not claim that these methods are ap-
propriate for all questions and for all spatial and temporal time
scales in cultural evolution. Instead, we suggest that, when they
are used carefully in cases where there is clear historical signal,
such as the Austronesian or Bantu expansions (52, 93), and
where the inferences are triangulated with other lines of evidence
(94), then they can make an important contribution to our
1. Darwin C (1872) On the Origin of Species by Means of Natural Selection (John Murray,
London), 6th Ed.
2. Kirby KR, et al. (2016) D-PLACE: A global database of cultural, linguistic and envi-
ronmental diversity. PLoS One 11:e0158391.
3. The Evolution Institute (2016) A New Society for the Study of Cultural Evolution.
Available at https://evolution-institute.org/project/society-for-the-study-of-cultural-
evolution/. Accessed January 3, 2017.
4. Mesoudi A (2011) Cultural Evolution: How Darwinian Theory Can Explain Human
Culture and Synthesize the Social Sciences (Univ of Chicago Press, Chicago).
5. Sperber D (1996) Explaining Culture: A Naturalistic Approach (Blackwell, Oxford).
6. Lewens T (2015) Cultural Evolution: Conceptual Challenges (Oxford Univ Press,
Oxford).
7. Slingerland E, Sullivan B (2017) Durkheim with data: The Database of Religious His-
tory. J Am Acad Relig 85:312–347.
8. Haspelmath M (2005) The World Atlas of Language Structures (Oxford Univ Press,
Oxford).
9. Wichmann S, Holman EW, Brown CH (2016) The ASJP Database. Version 17. Available
at asjp.clld.org/. Accessed January 3, 2017.
10. Moran S, McCloy D, Wright R (2014) PHOIBLE Online (Max Planck Institute for Evo-
lutionary Anthropology, Leipzig).
11. Greenhill SJ, Blust R, Gray RD (2008) The Austronesian Basic Vocabulary Database:
From bioinformatics to lexomics. Evol Bioinform Online 4:271–283.
12. Muysken P, et al. (2016) South American Indigenous Language Structures (SAILS)
Online (Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany).
Available at sails.clld.org. Accessed January 3, 2017.
13. Bowern C (2016) Chirila: Contemporary and Historical Resources for the Indigenous
Languages of Australia. Lang Doc Conserv 10:1–44.
14. Watts J, et al. (2015) Pulotu: Database of Austronesian supernatural beliefs and
practices. PLoS One 10:e0136783.
15. Botero CA, et al. (2014) The ecology of religious beliefs. Proc Natl Acad Sci USA 111:
16784–16789.
16. Snarey J (1996) The natural environment’s impact upon religious ethics: A cross-
cultural study. J Sci Study Relig 35:85–96.
17. Brown C, Eff EA (2010) The state and the supernatural: Support for prosocial be-
havior. Struct Dyn 4:1–21.
18. Roes FL, Raymond M (2003) Belief in moralizing gods. Evol Hum Behav 24:126–135.
19. Mace R, Pagel M (1994) The comparative method in anthropology. Curr Anthropol 35:
549–564.
20. R Core Team (2015) R: A Language and Environment for Statistical Computing (R
Foundation for Statistical Computing, Vienna).
21. Bates D, Maechler M, Bolker B, Walker S (2015) Package lme4. J Stat Softw 67:1–91.
22. Barton K (2015) MuMIn: Multi-model inference. R package, Version 1.15.1. Available
at r-forge.r-project.org/projects/mumin/. Accessed January 3, 2017.
23. Maynard Smith J, Szathmáry E (1995) The Major Transitions in Evolution (Oxford Univ
Press, Oxford).
24. Calcott B, Sterelny K (2011) A big picture of big pictures of life’s history. The Major
Transitions in Evolution Revisited, eds Calcott B, Sterelny K (MIT Press, Cambridge,
MA).
25. Szathmáry E, Smith JM (1995) The major evolutionary transitions. Nature 374:
227–232.
26. Flannery K, Marcus J (2012) The Creation of Inequality: How our Prehistroic Ancestors
Set the Stage for Monarchy, Slavery, and Empire (Harvard Univ Press, Cambridge,
MA).
27. Gintis H, Bowles S, Boyd R, Fehr E (2003) Explaining alturistic behavior in humans. Evol
Hum Behav 24:153–172.
28. Durkheim E (1915) The Elementary Forms of the Religious Life (Allen & Unwin,
London).
29. Sosis R (2009) The adaptationist-byproduct debate on the evolution of religion: Five
misunderstandings of the adaptationist program. J Cogn Cult 9:315–332.
30. Bulbulia J (2004) The cognitive and evolutionary psychology of religion. Biol Philos 18:
655–686.
31. Wiebe D (2008) Does talk about the evolution of religion make sense? Evolution of
Religion: Studies, Theories and Critiques, eds Bulbulia J, et al. (Collins Foundation,
Santa Margarita, CA), pp 339–346.
32. Johnson DD, Krüger O (2004) The good of wrath: Supernatural punishment and the
evolution of cooperation. Polit Theol 5:159–176.
33. Norenzayan A (2013) Big Gods: How Religion Transformed Cooperation and Conflict
(Princeton Univ Press, Princeton).
Gray and Watts
COLLOQUIUM
PAPER
understanding of cultural macroevolution. They can even be used
to predict political and economic changes (95). Although there is
much still to be done to integrate microlevel processes and mac-
rolevel patterns, the macro not only matters, it is tractable.
ACKNOWLEDGMENTS. We thank colleagues Quentin Atkinson, Carlos Botero,
Joseph Bulbulia, Michael Gavin, Simon Greenhill, and Oliver Sheehan for their
important contributions to the joint work on the cultural evolution of religion
discussed here. Olivier Morin and Kim Sterelny made useful comments on the
manuscript. This work was supported by John Templeton Foundation Grant
28745; a PhD scholarship from the University of Auckland; and Marsden Fund
Grant UOA1104.
34. Schloss JP, Murray MJ (2011) Evolutionary accounts of belief in supernatural pun-
ishment: A critical review. Religion Brain Behav 1:46–99.
35. Johnson DDP (2005) God’s punishment and public goods: A test of the supernatural
punishment hypothesis in 186 world cultures. Hum Nat 16:410–446.
36. Shariff AF, Norenzayan A, Henrich J (2011) The birth of high gods: How the cultural
evolution of supernatural policing influenced the emergence of complex, cooperative
human societies, paving the way for civilization. Evolution, Culture, and the Human
Mind, eds Schaller M, Norenzayan A, Heine SJ, Yamagishi T, Kameda T (Psychology,
New York), pp 119–136.
37. Atkinson Q, Latham A, Watts J (2015) Are Big Gods a big deal in the emergence of big
groups? Religion Brain Behav 5:266–274.
38. Peoples HC, Marlowe FW (2012) Subsistence and the evolution of religion. Hum Nat
23:253–269.
39. Watts J, Bulbulia J, Gray RD, Atkinson QD (2016) Clarity and causality needed in claims
about Big Gods. Behav Brain Sci 39:41–42.
40. Jordan FM (2013) Comparative phylogenetic methods and the study of pattern and
process in kinship. Kinship Systems: Change and Reconstruction, eds McConvell P,
Keen I, Hendery R (Univ of Utah Press, Salt Lake City), pp 43–58.
41. Mace R, Jordan F, Holden C (2003) Testing evolutionary hypotheses about human
biological adaptation using cross-cultural comparison. Comp Biochem Physiol A Mol
Integr Physiol 136:85–94.
42. Dow M, Eff E (2008) Global, regional, and local network autocorrelation in the
standard cross-cultural sample. Cross-Cultural Res 42:148–171.
43. Freckleton RP, Harvey PH, Pagel M (2002) Phylogenetic analysis and comparative data:
A test and review of evidence. Am Naturalist 160:712–726.
44. Fortunato L, Jordan F (2010) Your place or mine? A phylogenetic comparative analysis
of marital residence in Indo-European and Austronesian societies. Philos Trans R Soc
Lond B Biol Sci 365:3913–3922.
45. Currie TE, Greenhill SJ, Gray RD, Hasegawa T, Mace R (2010) Rise and fall of political
complexity in island South-East Asia and the Pacific. Nature 467:801–804.
46. Atkinson QD, Coomber T, Passmore S, Greenhill SJ, Kushnick G (2016) Cultural and
environmental predictors of pre-European deforestation on Pacific Islands. PLoS One
11:e0156340.
ANTHROPOLOGY
47. Holden CJ, Mace R (2003) Spread of cattle led to the loss of matrilineal descent in
Africa: A coevolutionary analysis. Proc Biol Sci 270:2425–2433.
48. Pagel M (1994) Detecting correlated evolution on phylogenies: A general method for
the comparative analysis of discrete characters. Proc Biol Sci 255:37–45.
49. Pagel M, Meade A (2006) Bayesian analysis of correlated evolution of discrete char-
acters by reversible-jump Markov chain Monte Carlo. Am Nat 167:808–825.
50. Watts J, et al. (2015) Broad supernatural punishment but not moralising high gods
precede the evolution of political complexity in Austronesia. Proc R Soc B Biol Sci
282:20142556.
51. Watts J, Sheehan O, Atkinson QD, Bulbulia J, Gray RD (2016) Ritual human sacrifice
EVOLUTION
promoted and sustained the evolution of stratified societies. Nature 532:228–231.
52. Gray RD, Drummond AJ, Greenhill SJ (2009) Language phylogenies reveal expansion
pulses and pauses in Pacific settlement. Science 323:479–483.
53. Kirch PV, Green RC (2001) Hawaiki, Ancestral Polynesia: An Essay in Historical
Anthropology (Cambridge Univ Press, Cambridge, UK).
54. Ko AM, et al. (2014) Early Austronesians: Into and out of Taiwan. Am J Hum Genet 94:
426–436.
55. Wilmshurst JM, Hunt TL, Lipo CP, Anderson AJ (2011) High-precision radiocarbon
dating shows recent and rapid initial human colonization of East Polynesia. Proc Natl
Acad Sci USA 108:1815–1820.
56. Huntington R, Metcalf P (1979) Celebrations of Death: The Anthropology of Mortuary
Ritual (Cambridge Univ Press, Cambridge, UK).
57. Nooy-Palm H (1979) The Sa’dan-Toraja: A Study of Their Social Life and Religion
(Martinus Nijhoff, The Hague).
58. Feinberg R (1991) Anuta. Oceania, Encyclopedia of World Cultures, ed Hays TE (G. K. Hall,
New York), Vol II, pp 13–16.
59. Campbell G (1991) The state and pre-colonial demographic history: The case of late
Nineteenth-Century Madagascar. J Afr Hist 32:425–445.
60. Buck PH (1952) The Coming of the Maori (Human Relations Area Files Press, New
Haven, CT).
61. Shariff AF, Willard AK, Andersen T, Norenzayan A (2016) Religious priming: A meta-
analysis with a focus on prosociality. Pers Soc Psychol Rev 20:27–48.
62. Marx K, Engels F (1975) Karl Marx and Friedrich Engels: Collected Works (In-
ternational, New York).
63. Cronk L (1994) Evolutionary theories of morality and the manipulative use of signals.
Zygon 29:81–101.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7851
64. Carrasco D (1999) City of Sacrifice (Beacon, Boston).
65. Bremmer JN (2007) The Strange World of Human Sacrifice (Peeters, Leuven, Belgium).
66. Turner CG, Turner JA (1999) Man Corn: Cannibalism and Violence in the Prehistoric
American Southwest (Univ of Utah Press, Salt Lake City).
67. Girard R (1987) Violent origins: Ritual killing and cultural formation. Violent Origins,
eds Hamerton-Kelly R, Burkert W, Girard R, Smith J (Stanford Univ Press, Stanford,
CA), pp 73–105.
68. Winkelman M (2014) Political and demograpic-ecological determinants of in-
stitutionalised human sacrifice. Anthropol Forum 24:47–70.
69. Lewontin RC (2001) In the beginning was the word. Science 291:1263–1264.
70. Fracchia J, Lewontin RC (2005) The price of metaphor. Hist Theory 44:14–29.
71. Fracchia J, Lewontin RC (1999) Does culture evolve? Hist Theory 38:52–78.
72. Dawkins R (1976) The Selfish Gene (Oxford Univ Press, Oxford).
73. Lewontin RC (1970) The units of selection. Annu Rev Ecol Syst 1:1–18.
74. Henrich J, Boyd R (2002) On modeling cognition and culture: Why cultural evolution
does not require replication of representations. J Cogn Cult 2:87–112.
75. Gould SJ (2010) An Urchin in the Storm: Essays About Books and Ideas (W. W. Norton,
New York).
76. Norenzayan A, et al. (2016) The cultural evolution of prosocial religions. Behav Brain
Sci 39:e1.
77. Dagan T, Martin W (2006) The tree of one percent. Genome Biol 7:118.
78. Shapiro JA (2016) Nothing in evolution makes sense except in the light of genomics:
Read-write genome evolution as an active biological process. Biology (Basel) 5:E27.
79. Mallet J (2005) Hybridization as an invasion of the genome. Trends Ecol Evol 20:
229–237.
80. Dagan T, Martin W (2007) Ancestral genome sizes specify the minimum rate of lateral
gene transfer during prokaryote evolution. Proc Natl Acad Sci USA 104:870–875.
81. Tehrani JJ, Collard M (2009) On the relationship between interindividual cultural
transmission and population-level cultural diversity: A case study of weaving in Ira-
nian tribal population. Evol Hum Behav 30:286–300.
7852 | www.pnas.org/cgi/doi/10.1073/pnas.1620746114
82. Hewlett BS, Cavalli-Sforza LL (1986) Cultural transmission among Aka Pygmies. Am
Anthropol 88:922–934.
83. Henrich J, Broesch J (2011) On the nature of cultural transmission networks: Evidence from
Fijian villages for adaptive learning biases. Philos Trans R Soc Lond B Biol Sci 366:1139–1148.
84. Aunger R (2000) The life history of culture learning in a face-to-face society. Ethos 28:
445–481.
85. Haspelmath M, Tadmor U (2009) World Loanword Database (WOLD) (Max Planck
Digital Library, Leipzig, Germany).
86. Boyd R, Borgerhoff-Mulder M, Durham WH, Richerson PJ (1997) Are cultural phy-
logenies possible? Human by Nature: Between Biology and the Social Sciences, eds
Weingart P, Richerson P, Mitchell S, Maasen S (Lawrence Erlbaum Associates, Mahwah,
NJ), pp 355–386.
87. Gray RD, Greenhill SJ, Ross RM (2007) The pleasures and perils of Darwinizing culture
(with phylogenies). Biol Theory 2:360–375.
88. Gray RD, Bryant D, Greenhill SJ (2010) On the shape and fabric of human history.
Philos Trans R Soc Lond B Biol Sci 365:3923–3933.
89. Greenhill SJ, Currie TE, Gray RD (2009) Does horizontal transmission invalidate cul-
tural phylogenies? Proc R Soc B Biol Sci 276:2299–2306.
90. Lipson M, et al. (2014) Reconstructing Austronesian population history in Island
Southeast Asia. Nat Commun 5:4689.
91. Lind J, Lindenfors P, Ghirlanda S, Lidén K, Enquist M (2013) Dating human cultural
capacity using phylogenetic principles. Sci Rep 3:1785.
92. Turchin P, et al. (2015) Seshat: The global history databank. Cliodynamics J Quant Hist
Cult Evol 6(1).
93. Currie TE, Meade A, Guillon M, Mace R (2013) Cultural phylogeography of the Bantu
Languages of sub-Saharan Africa. Proc R Soc London B Biol Sci 280:20130695.
94. Gray RD, Atkinson QD, Greenhill SJ (2011) Language evolution and human history:
What a difference a date makes. Philos Trans R Soc Lond B Biol Sci 366:1090–1100.
95. Matthews LJ, Passmore S, Richard PM, Gray RD, Atkinson QD (2016) Shared cultural
history as a predictor of political and economic changes among nation states.
PLoS One 11:e0152979.
Gray and Watts

Pursuing Darwin’s curious parallel: Prospects for a
science of cultural evolution
Alex Mesoudia,1
a
Human Biological and Cultural Evolution Group, Department of Biosciences, University of Exeter, Penryn, Cornwall TR10 9FE, United Kingdom
Edited by Marcus W. Feldman, Stanford University, Stanford, CA, and approved April 10, 2017 (received for review January 11, 2017)
In the past few decades, scholars from several disciplines have
pursued the curious parallel noted by Darwin between the genetic
evolution of species and the cultural evolution of beliefs, skills,
knowledge, languages, institutions, and other forms of socially
transmitted information. Here, I review current progress in the
pursuit of an evolutionary science of culture that is grounded in
both biological and evolutionary theory, but also treats culture as
more than a proximate mechanism that is directly controlled by
genes. Both genetic and cultural evolution can be described as
systems of inherited variation that change over time in response
to processes such as selection, migration, and drift. Appropriate
differences between genetic and cultural change are taken seriously,
such as the possibility in the latter of nonrandomly guided variation
or transformation, blending inheritance, and one-to-many trans-
mission. The foundation of cultural evolution was laid in the late
20th century with population-genetic style models of cultural mi-
croevolution, and the use of phylogenetic methods to reconstruct
cultural macroevolution. Since then, there have been major efforts
to understand the sociocognitive mechanisms underlying cumulative
cultural evolution, the consequences of demography on cultural
evolution, the empirical validity of assumed social learning biases,
the relative role of transformative and selective processes, and the
use of quantitative phylogenetic and multilevel selection models to
understand past and present dynamics of society-level change.
I conclude by highlighting the interdisciplinary challenges of study-
ing cultural evolution, including its relation to the traditional social
sciences and humanities.
|
cultural evolution cumulative culture | gene–culture coevolution |
|
human evolution social learning
The formation of different languages and of distinct species, and the
proofs that both have been developed through a gradual process, are
curiously parallel. . .
Charles Darwin, The Descent of Man, p 90
T his quote from Charles Darwin (1) draws a parallel between,
on the one hand, the genetic evolution of species, and on the
other, cultural change (i.e., changes in socially learned infor-
mation, such as beliefs, knowledge, tools, technology, attitudes,
norms, and, as Darwin mentions, languages). This idea is the
basic premise of cultural evolution: Cultural change constitutes a
Darwinian evolutionary process that shares key characteristics
with the genetic evolution of species. The emergence of this second
evolutionary process saw an unprecedented extension of genetic
evolution by allowing organisms to adapt more rapidly to, and
more powerfully create and shape, their environments.
Since the 1980s, this parallel between genetic and cultural
evolution has been pursued by scholars from a range of disci-
plines across the social, behavioral, and biological sciences. In
this article, I review the current state of this interdisciplinary
effort, focusing on topics of major recent research interest. No
new theories or findings are presented, but in presenting dispa-
rate strands of work alongside each other, I hope to identify links
between strands and foster a synthetic evolutionary science of
culture (2, 3) paralleling the interdisciplinary synthesis of the
biological sciences in the early 20th century.
www.pnas.org/cgi/doi/10.1073/pnas.1620741114
COLLOQUIUM
PAPER
Historical Context
Darwin’s comment above was inspired by historical linguists of
his time, who, even before publication of On the Origin of Species
(4), were constructing tree-like schemas of extant languages ex-
plicitly based on the assumption of common descent (5). Although
“evolutionary” ideas became popular ways of describing cultural
change in the late 1800s, such ideas were confused. Many scholars
erroneously saw evolution as inevitable progress along fixed stages
of increasing complexity (e.g., from savagery to barbarism to civi-
lization), drawing more from Herbert Spencer than from Darwin
(6). There was also much confusion, in the absence of a clear
understanding of genetics, about genetic and cultural inheritance.
Theories were often literally Lamarckian, with ideas, artifacts, and
words somehow thought to become part of the germ line through
repeated use (7). Due to this confusion, as well as the misuse of
pseudoevolutionary racial theories for distasteful political ends,
early 20th century social scientists declared culture to be separate
from “the organic” (8); the biological and social sciences went
separate ways; and the notion of cultural evolution, or indeed any
evolutionary basis for human behavior, fell from favor.
Cultural Microevolution
It was not until the 1970s and 1980s that a properly Darwinian
theory of cultural change was formulated, first by Cavalli-Sforza
and Feldman (9, 10) and then by Boyd and Richerson (11). This
theory comprised quantitative models of cultural microevolution,
describing the mechanisms by which cultural variation is trans-
mitted from person to person, and the processes that change this
variation over time within populations (Table 1), thus embodying
the “population thinking” that characterizes Darwin’s approach.
ANTHROPOLOGY
Here, “culture” is defined as “information capable of affecting
individuals’ behavior that they acquire from other members of their
species through teaching, imitation, and other forms of social trans-
mission” (12). “Social transmission,” “social learning,” and
“cultural transmission” are used interchangeably to denote the
nongenetic transfer of learned information from one individual
to another. “Cultural trait,” “cultural variant,” and sometimes
PSYCHOLOGICAL AND
“meme” are used to refer to the information (e.g., ideas, attitudes,
COGNITIVE SCIENCES
skills) that is transmitted. All of these terms hide huge complexity
and caveats. Such simplification is typical of a modeling approach.
This approach follows population genetics, which makes simplify-
ing assumptions (e.g., infinitely large populations) to understand
similarly complex genetic evolutionary processes. The simplification
in both cases is tactical, aiming to understand complex processes in
a piecemeal fashion and to formalize verbal arguments (13).
Some of the processes in Table 1 have parallels in genetic
evolution. Selection-like “content” or “direct” biases favor the
acquisition and transmission of some cultural variants over others
due to their memorability or effectiveness (14, 15), just as some
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: A.M. wrote the paper.
The author declares no conflict of interest.
This article is a PNAS Direct Submission.
1
Email: a.mesoudi@exeter.ac.uk.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7853–7860
 Table 1. Comparison of genetic and cultural evolution as originally modeled by Cavalli-Sforza and Feldman (10)
and Boyd and Richerson (11), with citations to recent empirical tests or examples
Process Genetic evolution Cultural evolution

Variation Genetic mutation Undirected cultural mutation/invention (16)

Recombination Recombination (133)
— Guided variation or transformation (109, 112)
Inheritance Particulate Particulate or “meme-like” (134)
— Blending of continuous cultural variation (60)
Vertical (parent to offspring) Vertical cultural transmission (75)
Horizontal gene transfer Horizontal/oblique cultural transmission (76, 77)
Selection Natural selection Cultural selection or direct/content biases (14, 15)
— Frequency-dependent biases (e.g., conformity) (88, 89)
— Indirect (e.g., success, prestige) biases (92, 99)
Migration Gene flow Demic diffusion (movement of people with their cultural traits) (17)
— Cultural diffusion (movement of traits without people) (135)
Drift Genetic drift Cultural drift (18)

alleles have higher fitness than others. Random cultural “mutation”
occurs where new variation arises randomly, such as via perceptual
error (16), akin to random genetic mutation. Migration allows indi-
viduals to introduce novel variants to a population as they move (17),
just as gene flow spreads alleles. Some cultural traits, such as first
names, fit the expectations of neutral drift (18), just like some alleles.
However, this enterprise is not simply the unthinking transfer
of models from genetic to cultural evolution. In many cases,
cultural variation is generated, inherited, and changed in very dif-
ferent ways from genetic variation, and models have addressed these
differences (10, 11). Examples include the blending of continuous,
nonparticulate cultural variation; the systematic, nonrandom
generation of cultural variation, or “guided variation”; frequency-
dependent biases, such as conformity, where variants are adopted
based on their commonness in the population; and indirect biases,
where traits are adopted based on the characteristics of their
bearers, such as success or prestige. Where possible, evidence
from psychology, anthropology, sociology, and other fields was
used to justify these processes (10, 11). However, the use of
quantitative models went beyond typical theory in the social and
behavioral sciences by (i) precisely and explicitly defining these
processes rather than relying on imprecise verbal descriptions of
phenomena (e.g., “conformity”) and (ii) exploring the population-
level consequences of such processes, such as the consequences
of frequency-dependent biases for between- and within-group
cultural variation (19).
Cultural Macroevolution
In the 1990s, the study of cultural microevolution was supple-
mented by the study of cultural macroevolution, defined as long-
term cultural change at or above the level of the society. Mace
and Pagel (20) introduced the phylogenetic comparative method
as a means to (i) reconstruct the cultural evolutionary history of
a particular trait or set of traits and (ii) test functional hypoth-
eses concerning the spread or distribution of cultural variation
across societies while controlling for evolutionary history. The
latter had been a problem within anthropology for over a cen-
tury. In 1889, Francis Galton (21) pointed out that even if two
traits (e.g., cattle-keeping and patriliny) often co-occur across
many societies, this co-occurrence does not necessarily provide
evidence that they are functionally associated (e.g., cattle-keeping
causes patriliny), because all these societies may have culturally
inherited this combination from a common ancestral society. So-
cieties are not necessarily statistically independent data points, due
to shared history. This problem is the same one facing biologists
when comparing across species, and, in the meantime, biologists
had developed methods for controlling for nonindependence due
to common descent (22). Mace, Pagel, and others imported these
methods to test functional evolutionary hypotheses in the same
way, showing, for example, that cattle-keeping did likely cause a
switch from matriliny to patriliny even after controlling for
descent (23).
Concurrently, archaeologists began using phylogenetic meth-
ods to reconstruct the history of artifacts, such as projectile
points (24), and, following Darwin’s original insight, others be-
gan reconstructing the history of language families (25). Like for
microevolution, the advantage here was in the use of quantitative
methods borrowed from biology that were explicit in their as-
sumptions about how to reconstruct historical relationships (e.g.,
maximum parsimony, maximum likelihood), repeatable and ex-
tendable by others, and easily scaled up to large datasets (26), in
contrast to the informal, idiosyncratic, and subjective schemas of
historical linguistics or archaeology.
Is It Evolution?
A common question is whether culture really evolves. This
question comes from both social scientists skeptical of any kind
of parallel with evolution, and biologists insistent that Darwin’s
theory applies only to genetic evolution (27). Importantly, no
one argues that genetic evolution and cultural evolution operate
identically. From the outset, microevolutionary modelers in-
corporated processes unique to cultural change, such as one-to-
many transmission (10) or nonrandomly guided variation (11).
However, an examination of Table 1 indicates that the parallels
are numerous enough to warrant an evolutionary theory of culture,
as long as these differences are taken seriously. At its heart, cultural
change is a process of inherited variation that changes due to se-
lection, drift, migration, and other processes, which, in their details,
may operate similar to or different from the genetic case.
Similarly, at the macroevolutionary level, it is sometimes ar-
gued that human culture is so riven with cross-lineage diffusion
that it is not tree-like, and thus not amenable to phylogenetic
methods (26). Although this argument may be true for some
cultural domains, many, such as languages or some artifacts,
have been shown to be tree-like due to strong intergenerational
cultural descent (28). Moreover, cross-lineage blending is a
common feature of genetic evolution when we look beyond our
own kingdom to, say, prokaryotes, where horizontal gene trans-
fer is rife (29). Indeed, network-based methods exist for dealing
with non–tree-like data (30).
One indirect, but perhaps most important, test of the parallel
between genetic and cultural change is whether methods bor-
rowed from evolutionary biology, suitably modified, actually
prove useful in explaining cultural change in a manner that adds
to the findings of nonevolutionary methods. Table 2 lists such
methods, which are further discussed throughout this article.
Evolution of Cultural Evolution
In parallel to the study of cultural change itself, that is, changes
in the contents of culture, modelers have also examined when
and why the capacity for cultural evolution evolved. Models

7854 | www.pnas.org/cgi/doi/10.1073/pnas.1620741114 Mesoudi


suggest social learning evolves when environments change at
intermediate rates: too quickly for genes to track directly, but not
so fast that socially learned information is outdated (11, 31). An
influential model by Rogers (32) showed that whereas social
learning readily evolves in such conditions, it does not increase
mean population fitness, contrary to claims that culture is responsible
for our species’ evolutionary success. Further models showed that
social learning does enhance population fitness when it is cumulative;
that is, individuals can learn via social learning what they could not
possibly learn alone (33). This finding has led to extensive study of the
factors that permit cumulative culture (discussed below). Finally,
gene–culture coevolution models and data have examined how
cultural evolution interacts with and affects genetic evolution (34).
There is extensive evidence for culture-driven genetic change in
humans, including agriculture-induced genetic adaptations for
digesting starch, dairy products, and alcohol (34, 35). Overall, this
research shows the extent to which culture extends genetic evolu-
tion by independently tracking environmental change that is too
rapid for genes to track, by generating diverse cultural adaptations
to those environmental challenges, and by driving genetic
evolution.
Recent Research Trends
Cumulative Culture. Just 20 y ago, little was known about social
learning and culture in nonhuman species. Many definitions of
culture stated that it was unique to humans, making the notion of
“nonhuman culture” nonsensical. Now, it is established that a
range of species from diverse taxa exhibit social learning (36), as
well as cultural traditions, where social learning generates long-
lasting behavioral differences between groups (37).
However, there is still a gulf between the cultural achieve-
ments of humans and other species. Recent work has focused on
cumulative culture, where knowledge is built up over successive
generations to exceed anything that a single individual could
invent alone (38, 39). This ability appears to be unique to hu-
mans: Although chimpanzees’ nut-cracking (36) or dolphins’ use
of protective nose-sponges (40) does not seem to exceed what a
lone chimpanzee or dolphin could invent alone, it is surely im-
possible for a single human to have discovered quantum me-
chanics, invented smartphones, visited the moon, or achieved
any of the other feats that require standing on the shoulders of
previous generations. As noted above, models of the evolution of
culture show that cumulative culture is particularly effective at in-
creasing mean population fitness beyond the population fitness of
noncumulative cultural species (33). A major research question is
therefore “What allows human culture to be uniquely cumulative?”
There has been much focus on high-fidelity social learning,
which is needed to preserve modifications over successive gen-
erations such that they can accumulate (41, 42). It was initially
suggested that imitation (i.e., the copying of bodily actions rather
than products) was key to this high fidelity (43). This claim seems
doubtful, given that chimpanzees can imitate tool use techniques
with high enough fidelity for alternate techniques to stabilize in
different groups (44). Rather than an “imitation vs. no-imitation”
dichotomy, perhaps humans are more effective, spontaneous, or
compulsive imitators (45). Other comparative work has sug-
gested roles for prosociality and language-mediated teaching
(46). As well as individual cognitive abilities, cumulative culture
Table 2. Methods and concepts that have been adapted from evolutionary biology to study cultural change
Evolutionary biology
Population genetic models
Gene-based phylogenetics
Comparative (cross-species) method
Population dynamic models
Multilevel selection
Genetic drift
Multigeneration breeding experiments
Mesoudi
COLLOQUIUM
PAPER
may also require favorable demography, such as larger pop-
ulations (42, 47) or populations partially connected via migration
(48, 49), as is typical of human societies (50). Currently, there is
no consensus on which of these factors is key to explaining
uniquely human cumulative culture. A combination of more than
one factor is probably necessary, perhaps explaining why cumu-
lative culture is confined to just one extant species.
Is “cumulative culture” synonymous with “cultural evolution”?
In principle, evolutionary change can be noncumulative, in-
volving changes in trait frequencies over time such as occurs with
genetic drift or local adaptive changes in gene frequencies. In
this sense, nonhuman, noncumulative cultural change can justi-
fiably be called cultural evolution. However, genetic evolution is
clearly also cumulative, involving the gradual accumulation of
beneficial genetic modifications over time to produce complex ad-
aptations, such as eyes or wings. Human cumulative cultural evo-
lution bears a clear parallel with this form of cumulative genetic
evolution. Indeed, the gradual accumulation of cultural innovations
results in complex cultural adaptations, such as telescopes or air-
planes, that resemble and rival complex genetic adaptations (12).
Ideas regarding the origin of cumulative culture can inform
thinking about factors that might affect recent and ongoing cu-
mulative cultural evolution. The invention of writing, followed by
digital media, surely greatly increased the fidelity of social
learning and, potentially, the speed of cumulative culture (51).
Demographically, the threefold increase in the world population
and increased global mobility in the past century should also
have accelerated cumulative culture. However, there are also
constraints. For example, as the amount of knowledge that is
accumulated increases (which, by definition, it must), it should
take longer for each new individual to acquire that knowledge.
This increased acquisition cost may result in extended educational
periods, and the eventual slowing down of cumulative culture as
innovation becomes harder (52).
Demography. Biologists have long recognized that demographic
factors, such as population size, structure, and interconnectedness,
are crucial for understanding trajectories of evolutionary change
(53). Although the effect of demography on cultural evolution was
modeled in the 1980s (10), the past decade has seen a major focus,
ANTHROPOLOGY
mostly in archaeology, on the way in which population structure
affects patterns of cultural variation and the gain and loss of
cultural complexity. Shennan (54) and Henrich (47) argued that
population size has been a major determinant of cultural com-
plexity in hunter-gatherers, often measured as the number of tools
in a toolkit or the number of components per tool. Henrich (47)
argued that the loss of toolkit complexity in Tasmania following
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
isolation from Australia 12 kya was due to reduced effective
population size, because the isolated population was too small to
maintain complexity, given imperfect social learning. In his model
of this process, each member of each new generation acquires the
skill of the most skillful member of the previous generation with
systematic loss due to copying error and some chance of im-
provement. Larger populations make the loss of skills due to im-
perfect copying less likely and improvements more likely. Shennan
and coworkers (48, 54) argued that increasing population densities
in Upper Paleolithic Europe around 45 kya caused the major increase
Cultural evolution
Cultural evolution (or gene–culture coevolution) models (10, 11)
Cultural phylogenetics (24, 26)
Comparative (cross-cultural) method (20)
Historical dynamic models (120)
Multilevel cultural selection (122, 123)
Cultural drift (15, 18)
Multigeneration transmission chain experiments (111, 136)
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7855
in symbolic and technological complexity seen in the archaeological
record, rather than a genetic mutation that enhanced cognitive ability.
Subsequent tests of the link between demography and cultural
complexity have been mixed. Analyses of tools (55) and lan-
guages (56) on islands in the Pacific Ocean are supportive, with
toolkit complexity and word gain rate increasing with island size.
Others have found environmental risk, not population size, to
predict toolkit complexity (57). These findings remain to be
reconciled. A recent critique arguing that “population size does
not explain past changes in cultural complexity” (58) is surely too
strong (59). Although not even the strongest advocates of de-
mography would argue that population size should always pre-
dict cultural complexity, the notion that demography has no
effect at all is surely also incorrect in light of the positive evi-
dence cited above. Recent models integrating population density
and mobility (not simply population size) as proxies of cul-
tural transmission (49) offer promise for more robust tests of
demographic hypotheses.
Experiments have also examined how population size affects
cultural complexity (60–63). Experiments cannot prove whether
demography affects cultural evolution in the real world, but they
can test the validity of behavioral assumptions of demographic
models and manipulate factors in a way we cannot in real life.
For example, the one study not to find a link between group size
and cultural complexity used a simple task that is unlikely to
benefit from a large pool of demonstrators (61), highlighting the
importance of task difficulty. Another showed that blending in-
heritance, where learners combine information from multiple
demonstrators, also leads to a group size effect (60), providing an
alternative mechanism to the assumption that people solely copy
the most skilled group member (47).
Cultural Phylogenetics. The use of phylogenetic methods to re-
construct human history has been greatly extended in both
methodology and subject matter. Methodologically, maximum
parsimony has been superseded by Bayesian Markov chain
Monte Carlo (MCMC) techniques that allow the testing of ex-
plicit evolutionary hypotheses such as whether artifacts exhibit
core packages of traits that are inherited together (64); network-
based methods to handle non–tree-like reticulation (30); and/or
phylogeographic methods that explicitly model the spread of
cultural traits in space, making assumptions about geographical
constraints [e.g., water bodies as potential barriers to language
diffusion (65)].
Such methods have been used to test hypotheses about human
history with more rigor than traditional nonquantitative, non-
evolutionary methods, shedding light on, for example, the origin
of the Indo-European language family (55), whether similarities
between hand-axe assemblages are caused by shared descent or
convergence (66), the historical links between folktales from
different regions (30), and even the recent evolution of computer
programming languages (67). These analyses do not ignore
existing work in traditional historical disciplines. They often test
existing hypotheses, but using larger samples and more powerful
statistical techniques. For example, Currie et al. (68) found
support in South-East Asia and the Pacific for the hypothesis
that political complexity increases in a unilinear sequence (69),
moving from acephalous to simple chiefdoms, to complex
chiefdoms, to states, without skipping stages, but with possible
collapses down to any earlier stage (note that this latter point is
evidence against the aforementioned Spencerian “progress”
theories of social evolution, which posited that progress toward
increasing complexity is an inevitable law). Similarly, Haynie and
Bowern (70) used phylogenetic comparative analyses of Australian
languages to test Berlin and Kay’s sequence model of color term
acquisition (71), where languages first acquire terms for black/
white, then red, then green/yellow, then blue, then brown. This
model was generally supported, albeit with many exceptions to
the sequence.
7856 | www.pnas.org/cgi/doi/10.1073/pnas.1620741114
Empirical Tests of Social Learning Biases. Cultural evolution models
contain assumptions about how people learn from one another.
Some posit biases such as conformity, where people are dispro-
portionately more likely to copy common cultural traits, or
success or prestige biases, where people preferentially copy traits
of successful or prestigious individuals (11). Others explore the
consequences of learning from parents (vertical transmission),
elders (oblique transmission), and peers (horizontal trans-
mission) (10). Where possible, these assumptions are made using
evidence from the social and behavioral sciences, such as the
social learning literature in psychology (72) or the diffusion of
innovations literature in sociology (73). However, such evidence
was not collected with these models in mind, and is often un-
suitable. For example, classic psychology studies of conformity
confound social influence and personal judgment in a way that
cannot reveal whether conformity in the sense modeled in the
cultural evolution literature is present (74). Such differences may
seem trivial, but they have significant population-level implica-
tions: Only conformity in the sense of disproportionately copying
the majority can lead to within-group homogeneity (11). Con-
sequently, the past decade has seen the use of field studies and
laboratory experiments that test the assumptions and predictions
of cultural microevolution models.
Field studies, typically conducted in small-scale societies where
paths of social transmission are easier to trace, have examined
whether and when people use vertical, oblique, or horizontal
transmission to acquire key skills, beliefs, or knowledge. This
research shows a cumulative refinement of findings. An initial
study on the Aka, relying solely on retrospective self-report
(asking people from whom they learned X), found substantial
vertical transmission for most traits, including hunting skills, food
gathering/preparation techniques, infant care methods, and sharing
norms, with dancing and singing the only domains with substantial
nonparental influence (75).
However, retrospective self-report is vulnerable to recall bias.
Later studies used nonretrospective methods (e.g., asking to
whom people would go to learn X) or non–self-report regres-
sions to assess pairwise similarity across respondents, on the
assumption that higher similarity indicates transmission between
those individuals. These studies support a two-stage model of
skill acquisition (76–79): People initially learn from their parents,
and then update this knowledge by learning from older adults or
peers later in life. Moreover, the updating stage typically targets
highly skilled or knowledgeable individuals (76). Interestingly,
this pattern resembles evidence regarding childhood learning in
Western societies, where peers are more important than parents
for the acquisition of key skills and knowledge (80). Such findings
are also consistent with models of age-based learning schedules,
which find that cumulative cultural evolution is facilitated when
learning becomes increasingly oblique and horizontal with age
(81). Nevertheless, these generalizations betray many exceptions,
and effects may vary with domain (82).
Laboratory experiments have explored similar questions re-
garding from whom people learn. Experiments offer more con-
trol than field studies, albeit with reduced external validity.
Participants face an unfamiliar task designed to resemble a real-
life task faced by people past or present. They can solve this task
through asocial learning and/or various forms of social learning
(e.g., conformity, success bias). Such studies paint a consistent
pattern despite using different tasks and protocols (74, 83–87).
At least some people behave adaptively as predicted by models,
by learning socially when appropriate (e.g., when asocial learning
is costly) and in an appropriate manner (e.g., using success bias
rather than copying at random). Moreover, some people exhibit
adaptive flexibility, such as using payoff bias in a task to determine
which of two choices yields higher payoffs, but frequency-based
biases in a coordination task where it pays to match others’
choices (85).
However, across these experiments, there are unexplained
individual differences in social learning use and often an un-
derutilization of social information (74, 83–87). There is some
Mesoudi

evidence linking this individual variation to other individual
differences, such as personality (87) or intelligence (88, 89), but
these correlations are weak and exploratory. There is also evi-
dence of cross-cultural variation in social learning, specifically
higher social learning in collectivistic East Asian societies than in
individualistic Western societies (90, 91). Similar individual and
cultural variation in nonexperimental data suggests that this
finding is not just a laboratory artifact (92).
The causes and consequences of this individual and cultural
variation are unclear. Individual variation in social learning is
found in many species. It can be viewed on a continuum of
phenotypic plasticity, from genetically polymorphic and de-
velopmentally fixed individual differences, to developmentally
determined facultative responses to external environments or
physiological state, to the associative learning of learning strat-
egies (93). Whether individual variation in human social learning
is nonadaptive noise, reflects frequency-dependent equilibria
between information producers and scroungers with no group-
level benefit (32), or is adaptive at the group level by maintaining
a mix of innovation and tradition is unknown.
Cultural variation in human social learning suggests that we
acquire norms via social learning that, in turn, affect our degree
of social learning (93). This process may generate cultural dy-
namics entailing the “social learning of social learning” (94),
rather than the typical modeling assumption that learning
strategies are genetically inherited. The origin of this cultural
variation is unknown, but it might be an historical and societal
response to different rates of environmental change (95). An-
other line of work has identified intentional, institution-based
mechanisms for generating innovation (96), suggesting similar
flexibility in asocial learning. There is great scope to use the
methods of cultural macroevolution outlined above to test these
hypotheses and explain how cultural macroevolution, as well as
genetic evolution, has shaped the behavioral responses measured
in experiments, rather than assuming that participants are
coming into the laboratory as “blank slates.”
Development. The brief mentions above of work with children
belie rapid growth in the study of the developmental basis of
cultural evolution (97). Such studies are crucial for un-
derstanding how people developmentally acquire the learning
biases noted above to be individually and culturally variable, and
how these learning biases interact with developing abilities, such
as language and theory of mind. Experiments have shown that
children are sophisticated social learners and exhibit biases
predicted by models to be adaptive, such as preferentially
learning from accurate over inaccurate individuals (98) and
prestigious over nonprestigious individuals (99). There is work
combining biases, finding that children copy groups over indi-
viduals when both are equally successful but copy successful in-
dividuals over unsuccessful groups, thus adaptively switching
between frequency and success information (100). Other work
has addressed the motivation for copying, with children more
likely to imitate indiscriminately when tasks are presented as
conventional rather than instrumental, such that the motiva-
tion is to affiliate with one’s group rather than acquire effective
skills (101).
There are exceptions to these impressive skills, however. One
study found that children copy adults over peers even when peers
are more knowledgeable (102). In addition, there is similar in-
dividual (103) and cultural (90) variation as seen in adults, which
has yet to be explained. Further work is needed to link the study
of social learning in childhood and adulthood, ideally using
models to link developmentally changing learning schedules to
macroevolutionary patterns of cumulative culture (81).
Cultural Attraction. An ongoing debate has been over the relative
role of preservative, selection-like processes and nonselective,
transformative processes in explaining cultural change (104–
107). Many of the cultural evolution models described earlier
assume high-fidelity transmission plus random copying error or
Mesoudi
COLLOQUIUM
PAPER
mutation, with selection-like processes, such as content or con-
formist biases, altering the frequency of cultural traits over time.
Sperber and coworkers (105, 106) have argued that many in-
stances of cultural change do not take this form. Instead, they
argue, cultural transmission is transformative: People reconstruct
what they learn from others according to their preexisting
knowledge, cognitive or perceptual biases, or other factors. This
process of transformation is known as cultural attraction, and the
points at which representations converge are called cultural
attractors (105, 106) [although ambiguities in these definitions
are discussed elsewhere (108)].
For example, recall the phylogenetic analysis showing that
Australian languages typically acquire color terms in the specific
Berlin–Kay sequence, explaining cross-cultural regularities in
color terminology (70). Cultural attraction offers a plausible
microevolutionary explanation for this finding: People share
perceptual systems that lead them to invent and transform color
terms independently in the same way. This explanation is sup-
ported by experiments in which initially random artificial color
terms were passed along chains of people (109). Each participant
learned unfamiliar terms for each color, with these labels, in-
cluding errors, passed to the next person as his or her learning
set. In each of 30 independent chains, the artificial terminology
converged on the predicted Berlin–Kay scheme, as each person
transformed the labels in a systematic manner. Similar experi-
ments have shown convergence toward universal patterns of
grammatical structure (110), category learning (111) and
bloodletting as a medical practice (112).
The cultural attraction approach moves the explanatory focus
from the population to the individual level. Rather than explaining
patterns of cultural diversity, stability, and change in terms of the
differential selection of certain cultural variants (e.g., content
biases) or differential copying of certain individuals (e.g., success
bias, prestige bias), cultural attraction focuses on how individuals
systematically transform representations. The latter is similar
[but not identical (107)] to evolutionary psychologists’ notion of
“evoked culture” (113), where genetically evolved cognitive
biases cause the independent recurrence of genetically adap-
tive behavior, and the process of iterated learning, where re-
peated learning and transmission cause convergence on inductive
biases or priors (111).
ANTHROPOLOGY
Although cultural attraction is sometimes presented as an al-
ternative to cultural evolution, the two approaches are compat-
ible (104). Many “standard” cultural evolution models, in fact,
do not model transmission as high fidelity, and allow for trans-
formation (47, 114). The notion of guided variation (11) is
similar to the individual, nonrandom transformation described as
cultural attraction, and can operate in parallel to the more
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
selection-like transmission biases in Table 1. However, cultural
attraction proponents have a valid point that, in practice, such
transformative processes have not received adequate attention.
The relative influence of each likely varies with domain (104).
Where there are clear inductive biases favoring certain repre-
sentations, such as bloodletting or color terms, then explanations
in terms of transformation/attraction will be useful. Where there
are no clear intuitions or inductive biases, selection-like pro-
cesses will be more important. Bloodletting, for example, has
been replaced in many societies with surgical techniques that are
the product of a long refinement and accumulation of unintuitive
knowledge and skills. Many medical, scientific, and technological
practices are the product of accidental invention followed by
payoff-biased selection in the face of resistance due to confor-
mity to prior practices or transformation back to intuitive
attractors (115). Examples include glassmaking (116); musical
instruments (12]); and the theory of evolution, an unintuitive
idea that needs conscious effort to understand (118).
Even where there is clear evidence for inductive biases, as in
the case of color terms (109), the prediction of cross-cultural
universals is only partially upheld in real-world data, as shown
by the many exceptions to the Berlin–Kay scheme identified by
phylogenetic analyses (70). Further work might show these
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7857
exceptions to be determined by processes like migration or
prestige bias. A Bayesian framework can be useful here, having
been used to model both attraction-like inductive biases (111) and
selection-like biases (119). This integration of selective and
transformative processes, and use of both microevolutionary evi-
dence regarding cognitive biases and macroevolutionary analyses
of actual cultural diversity and change, holds great promise.
Historical and Contemporary Social Dynamics. As well as cultural
phylogenetics, another approach to understanding historical
change uses population dynamic models of the kind used within
ecology to model changes in population size over time in re-
sponse to births, deaths, predation, or migration (120). This
approach is more useful when detailed temporal data are avail-
able, such as on the rise and fall of empires. Again, models and
theories are not unthinkingly imported from ecology and applied
to human societies. They are adapted to take into account the
unique aspects of human culture, often drawing on traditional
theories from history and sociology. However, the advantages of
this approach are that (i) theories can be precisely quantified in a
way that generates clear predictions, unlike verbal arguments,
and (ii) these predictions can be empirically tested often across
multiple societies, regions, and time periods, in a way that his-
torians seldom do (120).
For example, Turchin (120) used population dynamic models
to test competing theories for the rise and fall of empires in
Europe during the period from AD 0–1900. One idea, proposed
by Ibn Khaldun in the 14th century, can be interpreted as a
theory of multilevel selection. Here, societies grow by solving
collective action problems, such as building irrigation systems or
organizing collective defense against enemies. Societies that
more effectively solve such problems grow in size and defeat
other less internally cooperative societies, eventually becoming
empires. When societies are very large, however, there is over-
production of elites who fight among themselves for power, as
well as a disconnect between the majority and the squabbling
elites. This internal conflict allows another society to invade,
typically one in a border region with higher internal cooperation
due to its smaller size and common enemy (the larger empire).
This new empire grows larger, internal cooperation breaks down,
the new empire is itself eventually invaded, and the cycle
continues.
Turchin (120) converted this verbal theory into a quantitative
model, incorporating within-group cooperation as well as factors
traditionally considered important, such as access to resources
and geographical overreach (121). This model provided a better
fit to historical data on the rise and fall of actual empires than
models without cooperation. This work shows not only the value
of quantitative cultural evolution models and empirical tests as
applied to history but also support for the idea of multilevel
cultural selection, where societies grow as a result of superior
within-group cooperation that provides an advantage in between-
group competition (11, 122). Subsequent work has found further
support for this theory in a spatially explicit model tested with data
beyond Europe (123).
More recently, Turchin (124) has applied these ideas to current
societies, particularly the United States, extending structural-
demographic theories from political science (125). Worryingly,
elite overproduction, interelite conflict, and social inequality, which
were key markers of low within-group cooperation and impending
empire collapse in the past, have been increasing in the United
States in recent decades. Examples of interelite conflict include
decreased Republican-Democrat cooperation in government and
the Tea Party challenge within the Republican Party. The success
of self-styled antiestablishment figures, such as Donald Trump,
is arguably due to rising social inequality and a disconnect
between the majority of voters and the increasingly conflict-
ridden political elites.
Predicting the future can never be done with complete certainty,
for either genetic or cultural evolution. However, the work of
Turchin (124) and others makes events like the unexpected election
7858 | www.pnas.org/cgi/doi/10.1073/pnas.1620741114
of Donald Trump more understandable, and provides early warning
signs of societal collapse based on the long view of cultural evolu-
tion that we ignore at our peril. Continual progress toward more
stable forms of political organization is far from inevitable (68), and
institutions are fragile balancing acts between individual and group
interests (122). Better understanding of this balance, within the long
view of cultural evolution, is surely crucial for creating and main-
taining sustainable societies.
Conclusions
Science itself is a cumulative cultural evolutionary process (126).
Is the science of cultural evolution accumulating an increasingly
reliable body of knowledge concerning human culture? I hope I
have shown that it is. Initial claims derived from self-reports that
cultural transmission is largely vertical have been replaced with a
two-stage model in which parental knowledge is updated via
horizontal or oblique transmission, often targeted in age- or skill-
appropriate ways (76, 77). The common modeling assumption
that social learning is under fixed genetic control is incompatible
with experimental evidence of substantial individual and cultural
variation, and is being revised (93). The “demographic turn”
within archaeology, itself an improvement on unrealistic “single
genetic mutation” explanations for increases in past cultural
complexity, has been refined to focus on population density and
migration rather than just population size (49). Long-standing
hypotheses regarding the acquisition of color terms, origin of
Indo-European languages, and trajectory of sociopolitical com-
plexity have been tested using phylogenetic methods that are
more powerful than informal comparisons using cherry-picked
examples (26).
However, major questions remain, which is also the normal
course of science. Ongoing work seeks to explain why only hu-
mans exhibit cumulative cultural evolution, the origins of indi-
vidual and cultural variation in social learning and innovation,
the relative influence of selection and attraction across domains,
and the balance between individual and group interests that
shape societal cohesion and stability. A welcome trend is to apply
cultural evolution theory to real-world problems, including en-
vironmental sustainability (122), the social effects of new digital
media (51) and society-level cooperation (124). The work reviewed
here is, moreover, a small selection of cultural evolution research.
There is no space to cover, say, economics (127), neuroscience
(128), literature (129), or religion (130).
As well as the use of quantitative methods, often borrowed
from biology (Table 2), a major strength of cultural evolution
work is its interdisciplinarity. Findings from, say, experimental
psychology can be applied to problems in archaeology (16). Con-
versely, a consideration of the population-level consequences of
psychological constructs, such as conformity, highlights their lim-
itations (74). This interdisciplinarity should be pushed further,
especially by integrating microevolution and macroevolution. For
example, phylogenetic analyses of actual color terminologies (70)
reveal broad support but key exceptions to the universality pre-
dicted by experiments (109). Are these exceptions also predictable
from individual cognition, or are other factors needed? More
broadly, interdisciplinarity is facilitated by open science. Although
data and analytical techniques have traditionally been kept within
disciplines as protected knowledge, the public release of data and
use of reproducible analyses (e.g., R scripts) encourage scholars
from other disciplines to explore that data and familiarize them-
selves first-hand with key findings.
However, there is still an epistemological gulf between the
scientific, evolutionary approach outlined here and much of the
social sciences and humanities, which, after all, study the same
cultural phenomena (131, 132). Most cultural anthropologists,
sociologists, historians, and the like are reluctant to use the sim-
plifying assumptions and reductionism inherent in quantitative
models and methods, instead highlighting complexity and contradic-
tions. There is also a reluctance to generalize across societies, re-
gions, or time periods and, instead, a focus on specificity and
uniqueness. There is also a reluctance to consider continuities
Mesoudi

between human behavior and the behavior of other species, with
the latter often perceived to be “instinctive” or genetically de-
termined in a way that human behavior is not [an overview of
these issues from both sides is provided elsewhere (131)].
This reluctance is, I think, misguided. The methods and ap-
proaches described here add to, rather than detract from, tra-
ditional methods and knowledge. Phylogenetic analyses typically
use existing linguistic or ethnographic data to reconstruct his-
torical relationships and test causal historical hypotheses, but
with greater rigor than is possible by considering single case
studies or ignoring Galton’s problem (20). Quantitative models
are simply verbal arguments expressed more precisely and un-
ambiguously, and in a way that affords easier empirical testing.
Those verbal arguments often come from the social sciences and
humanities (120, 121, 124, 125). There are also excellent exam-
ples of traditional ethnography providing important corrections
to cultural evolution theory, such as Wilf’s (96) demonstration
that cultural innovation can be institutionally driven. Further-
more, cultural evolution offers a link to the wider evolutionary
1. Darwin C (1871) The Descent of Man (John Murray, London); reprinted (2003)
(Gibson Square, London).
2. Mesoudi A (2011) Cultural Evolution (Univ of Chicago Press, Chicago).
3. Mesoudi A, Whiten A, Laland KN (2006) Towards a unified science of cultural evo-
lution. Behav Brain Sci 29:329–347, discussion 347–383.
4. Darwin C (1859) On the Origin of Species (John Murray, London); reprinted
(1968) (Penguin, London).
5. van Wyhe J (2005) The descent of words: Evolutionary thinking 1780-1880.
Endeavour 29:94–100.
6. Freeman D (1974) The evolutionary theories of Charles Darwin and Herbert Spencer.
Curr Anthropol 15:211–221.
7. Stocking GW (1962) Lamarckianism in American social science: 1890-1915. J Hist Ideas
23:239–256.
8. Kroeber AL (1917) The superorganic. Am Anthropol 19:163–213.
9. Cavalli-Sforza LL, Feldman MW (1973) Cultural versus biological inheritance: Phe-
notypic transmission from parents to children. (A theory of the effect of parental
phenotypes on children’s phenotypes). Am J Hum Genet 25:618–637.
10. Cavalli-Sforza LL, Feldman MW (1981) Cultural Transmission and Evolution (Prince-
ton Univ Press, Princeton).
11. Boyd R, Richerson PJ (1985) Culture and the Evolutionary Process (Univ of Chicago
Press, Chicago).
12. Richerson PJ, Boyd R (2005) Not by Genes Alone (Univ of Chicago Press, Chicago).
13. Servedio MR, et al. (2014) Not just a theory–the utility of mathematical models in
evolutionary biology. PLoS Biol 12:e1002017.
14. Stubbersfield JM, Tehrani JJ, Flynn EG (2015) Serial killers, spiders and cybersex:
Social and survival information bias in the transmission of urban legends. Br J Psychol
106:288–307.
15. Sindi SS, Dale R (2016) Culturomics as a data playground for tests of selection:
Mathematical approaches to detecting selection in word use. J Theor Biol 405:
140–149.
16. Kempe M, Lycett S, Mesoudi A (2012) An experimental test of the accumulated
copying error model of cultural mutation for Acheulean handaxe size. PLoS One
7:e48333.
17. Hamilton MJ, Buchanan B (2007) Spatial gradients in Clovis-age radiocarbon dates
across North America suggest rapid colonization from the north. Proc Natl Acad Sci
USA 104:15625–15630.
18. Bentley RA, Hahn MW, Shennan SJ (2004) Random drift and culture change. Proc
Biol Sci 271:1443–1450.
19. Mesoudi A (2009) How cultural evolutionary theory can inform social psychology
and vice versa. Psychol Rev 116:929–952.
20. Mace R, Pagel M (1994) The comparative method in anthropology. Curr Anthropol
35:549–564.
21. Galton F (1889) Comment on Tylor, E. B., On a method of investigating the devel-
opment of institutions, applied to laws of marriage and descent. J R Anthropol Inst
18:270.
22. Harvey P, Pagel M (1991) The Comparative Method in Evolutionary Biology (Oxford
Univ Press, Oxford).
23. Holden CJ, Mace R (2003) Spread of cattle led to the loss of matrilineal descent in
Africa: A coevolutionary analysis. Proc Biol Sci 270:2425–2433.
24. O’Brien MJ, Darwent J, Lyman RL (2001) Cladistics is useful for reconstructing ar-
chaeological phylogenies. J Archaeol Sci 28:1115–1136.
25. Gray RD, Jordan FM (2000) Language trees support the express-train sequence of
Austronesian expansion. Nature 405:1052–1055.
26. Gray RD, Watts J (2017) Cultural macroevolution matters. Proc Natl Acad Sci USA
114:7846–7852.
27. Fracchia J, Lewontin RC (1999) Does culture evolve? Hist Theory 38:52–78.
28. Collard M, Shennan S, Tehrani J (2006) Branching, blending, and the evolution of
cultural similarities and differences among human populations. Evol Hum Behav 27:
169–184.
Mesoudi
COLLOQUIUM
PAPER
sciences without inappropriate genetic determinism. Unlike so-
ciobiology and evolutionary psychology approaches that downplay
transmitted culture (113), the work outlined here assumes that
cultural evolution is semiautonomous from genetic evolution,
allowing rapid cultural adaptation to novel physical and social
environments without genetic change. Although some aspects of
cultural diversity may reflect genetically evolved, content-rich
cognitive biases, many others reflect the accumulation of modifi-
cations via content-independent learning rules such as success
bias, leading to diverse, historically contingent pathways of
culturally inherited knowledge. The theory of cultural evolution
offers a means of taking culture seriously within a scientific,
evolutionary framework.
ACKNOWLEDGMENTS. I thank Francisco Ayala, Marc Feldman, Kevin Laland,
and Andrew Whiten for inviting me to participate in the Sackler Colloquium
upon which this special issue is based, and two anonymous reviewers for
helpful comments.
29. Doolittle WF, Bapteste E (2007) Pattern pluralism and the Tree of Life hypothesis.
Proc Natl Acad Sci USA 104:2043–2049.
30. Tehrani JJ (2013) The phylogeny of Little Red Riding Hood. PLoS One 8:e78871.
31. Aoki K, Wakano JY, Feldman MW (2005) The emergence of social learning in a
temporally changing environment. Curr Anthropol 46:334–340.
32. Rogers A (1988) Does biology constrain culture? Am Anthropol 90:819–831.
33. Boyd R, Richerson PJ (1995) Why does culture increase human adaptability? Ethol
Sociobiol 16:125–143.
34. Laland KN, Odling-Smee J, Myles S (2010) How culture shaped the human genome:
Bringing genetics and the human sciences together. Nat Rev Genet 11:137–148.
35. Henrich J (2015) The Secret of Our Success (Princeton Univ Press, Princeton).
36. Hoppitt W, Laland KN (2013) Social Learning (Princeton University Press, Princeton,
NJ).
37. Whiten A (2017) Culture extends the scope of evolutionary biology in the great apes.
Proc Natl Acad Sci USA 114:7790–7797.
38. Dean LG, Vale GL, Laland KN, Flynn E, Kendal RL (2014) Human cumulative culture: A
comparative perspective. Biol Rev Camb Philos Soc 89:284–301.
39. Tennie C, Call J, Tomasello M (2009) Ratcheting up the ratchet: On the evolution of
cumulative culture. Philos Trans R Soc Lond B Biol Sci 364:2405–2415.
40. Krützen M, et al. (2005) Cultural transmission of tool use in bottlenose dolphins. Proc
Natl Acad Sci USA 102:8939–8943.
41. Lewis HM, Laland KN (2012) Transmission fidelity is the key to the build-up of cu-
mulative culture. Philos Trans R Soc Lond B Biol Sci 367:2171–2180.
42. Kempe M, Lycett SJ, Mesoudi A (2014) From cultural traditions to cumulative culture:
ANTHROPOLOGY
Parameterizing the differences between human and nonhuman culture. J Theor Biol
359:29–36.
43. Tomasello M (1999) The Cultural Origins of Human Cognition (Harvard Univ Press,
Cambridge, MA).
44. Whiten A, McGuigan N, Marshall-Pescini S, Hopper LM (2009) Emulation, imitation,
over-imitation and the scope of culture for child and chimpanzee. Philos Trans R Soc
Lond B Biol Sci 364:2417–2428.
45. Lyons DE, Young AG, Keil FC (2007) The hidden structure of overimitation. Proc Natl
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
Acad Sci USA 104:19751–19756.
46. Dean LG, Kendal RL, Schapiro SJ, Thierry B, Laland KN (2012) Identification of the
social and cognitive processes underlying human cumulative culture. Science 335:
1114–1118.
47. Henrich J (2004) Demography and cultural evolution. Am Antiq 69:197–214.
48. Powell A, Shennan S, Thomas MG (2009) Late Pleistocene demography and the
appearance of modern human behavior. Science 324:1298–1301.
49. Grove M (2016) Population density, mobility, and cultural transmission. J Archaeol
Sci 74:75–84.
50. Salali GD, et al. (2016) Knowledge-sharing networks in hunter-gatherers and the
evolution of cumulative culture. Curr Biol 26:2516–2521.
51. Acerbi A (2016) A cultural evolution approach to digital media. Front Hum Neurosci
10:636.
52. Mesoudi A (2011) Variable cultural acquisition costs constrain cumulative cultural
evolution. PLoS One 6:e18239.
53. Charlesworth B (2009) Fundamental concepts in genetics: Effective population size
and patterns of molecular evolution and variation. Nat Rev Genet 10:195–205.
54. Shennan S (2001) Demography and cultural innovation. Camb Archaeol J 11:5–16.
55. Kline MA, Boyd R (2010) Population size predicts technological complexity in Oce-
ania. Proc Biol Sci 277:2559–2564.
56. Bromham L, Hua X, Fitzpatrick TG, Greenhill SJ (2015) Rate of language evolution is
affected by population size. Proc Natl Acad Sci USA 112:2097–2102.
57. Collard M, Buchanan B, O’Brien MJ (2013) Population size as an explanation for
patterns in the Paleolithic archaeological record. Curr Anthropol 54:S388–S396.
58. Vaesen K, Collard M, Cosgrove R, Roebroeks W (2016) Population size does not
explain past changes in cultural complexity. Proc Natl Acad Sci USA 113:
E2241–E2247.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7859
59. Henrich J, et al. (2016) Understanding cumulative cultural evolution. Proc Natl Acad
Sci USA 113:E6724–E6725.
60. Kempe M, Mesoudi A (2014) An experimental demonstration of the effect of group
size on cultural accumulation. Evol Hum Behav 35:285–290.
61. Caldwell CA, Millen AE (2010) Human cumulative culture in the laboratory: Effects of
(micro) population size. Learn Behav 38:310–318.
62. Derex M, Beugin M-P, Godelle B, Raymond M (2013) Experimental evidence for the
influence of group size on cultural complexity. Nature 503:389–391.
63. Muthukrishna M, Shulman BW, Vasilescu V, Henrich J (2014) Sociality influences
cultural complexity. Proc Biol Sci 281:20132511.
64. Matthews LJ, Tehrani JJ, Jordan FM, Collard M, Nunn CL (2011) Testing for divergent
transmission histories among cultural characters: A study using Bayesian phyloge-
netic methods and Iranian tribal textile data. PLoS One 6:e14810.
65. Bouckaert R, et al. (2012) Mapping the origins and expansion of the Indo-European
language family. Science 337:957–960.
66. Lycett SJ (2009) Are Victoria West cores “proto-Levallois”? A phylogenetic assess-
ment. J Hum Evol 56:175–191.
67. Valverde S, Solé RV (2015) Punctuated equilibrium in the large-scale evolution of
programming languages. J R Soc Interface 12:20150249.
68. Currie TE, Greenhill SJ, Gray RD, Hasegawa T, Mace R (2010) Rise and fall of political
complexity in island South-East Asia and the Pacific. Nature 467:801–804.
69. Johnson AW, Earle TK (2000) The Evolution of Human Societies (Stanford Univ Press,
Stanford, CA).
70. Haynie HJ, Bowern C (2016) Phylogenetic approach to the evolution of color term
systems. Proc Natl Acad Sci USA 113:13666–13671.
71. Berlin B, Kay P (1991) Basic Color Terms (Univ of California Press, Berkeley, CA).
72. Bandura A (1977) Social Learning Theory (Prentice Hall, Oxford).
73. Rogers E (1995) The Diffusion of Innovations (Free Press, New York).
74. Efferson C, Lalive R, Richerson PJ, McElreath R, Lubell M (2008) Conformists and
mavericks. Evol Hum Behav 29:56–64.
75. Hewlett BS, Cavalli-Sforza LL (1986) Cultural transmission among Aka pygmies. Am
Anthropol 88:922–934.
76. Henrich J, Broesch J (2011) On the nature of cultural transmission networks: Evidence
from Fijian villages for adaptive learning biases. Philos Trans R Soc Lond B Biol Sci
366:1139–1148.
77. Hewlett BS, Fouts HN, Boyette AH, Hewlett BL (2011) Social learning among Congo
Basin hunter-gatherers. Philos Trans R Soc Lond B Biol Sci 366:1168–1178.
78. Reyes-Garcia V, et al. (2009) Cultural transmission of ethnobotanical knowledge and
skills. Evol Hum Behav 30:274–285.
79. Kline MA, Boyd R, Henrich J (2013) Teaching and the life history of cultural trans-
mission in Fijian villages. Hum Nat 24:351–374.
80. Harris J (1995) Where is the child’s environment? Psychol Rev 102:458–489.
81. Lehmann L, Wakano JY, Aoki K (2013) On optimal learning schedules and the
marginal value of cumulative cultural evolution. Evolution 67:1435–1445.
82. McElreath R, Strimling P (2008) When natural selection favors imitation of parents.
Curr Anthropol 49:307–316.
83. McElreath R, et al. (2008) Beyond existence and aiming outside the laboratory: Es-
timating frequency-dependent and pay-off-biased social learning strategies. Philos
Trans R Soc Lond B Biol Sci 363:3515–3528.
84. Mesoudi A (2011) An experimental comparison of human social learning strategies.
Evol Hum Behav 32:334–342.
85. Molleman L, van den Berg P, Weissing FJ (2014) Consistent individual differences in
human social learning strategies. Nature Commun 5:3570.
86. Morgan TJH, Rendell LE, Ehn M, Hoppitt W, Laland KN (2012) The evolutionary basis
of human social learning. Proc Biol Sci 279:653–662.
87. Toelch U, Bruce MJ, Newson L, Richerson PJ, Reader SM (2014) Individual consistency
and flexibility in human social information use. Proc Biol Sci 281:20132864.
88. Muthukrishna M, Morgan TJH, Henrich J (2016) The when and who of social learning
and conformist transmission. Evol Hum Behav 37:10–20.
89. Shennan S, Wilkinson J (2001) Ceramic style change and neutral evolution. Am Antiq
66:577–593.
90. DiYanni CJ, Corriveau KH, Kurkul K, Nasrini J, Nini D (2015) The role of consensus and
culture in children’s imitation of inefficient actions. J Exp Child Psychol 137:99–110.
91. Mesoudi A, Chang L, Murray K, Lu HJ (2015) Higher frequency of social learning in
China than in the West shows cultural variation in the dynamics of cultural evolu-
tion. Proc Biol Sci 282:20142209.
92. Beheim BA, Thigpen C, McElreath R (2014) Strategic social learning and the pop-
ulation dynamics of human behavior. Evol Hum Behav 35:351–357.
93. Mesoudi A, Chang L, Dall SRX, Thornton A (2016) The evolution of individual and
cultural variation in social learning. Trends Ecol Evol 31:215–225.
94. Acerbi A, Enquist M, Ghirlanda S (2009) Cultural evolution and individual develop-
ment of openness and conservatism. Proc Natl Acad Sci USA 106:18931–18935.
95. Chang L, et al. (2011) Cultural adaptations to environmental variability. Educ Psychol
Rev 23:99–129.
96. Wilf E (2015) Routinized business innovation: An undertheorized engine of cultural
evolution. Am Anthropol 117:679–692.
97. Legare CH (2017) Cumulative cultural learning: Development and diversity. Proc Natl
Acad Sci USA 114:7877–7883.
7860 | www.pnas.org/cgi/doi/10.1073/pnas.1620741114
98. Birch SAJ, Vauthier SA, Bloom P (2008) Three- and four-year-olds spontaneously use
others’ past performance to guide their learning. Cognition 107:1018–1034.
99. Chudek M, Heller S, Birch S, Henrich J (2012) Prestige-biased cultural learning. Evol
Hum Behav 33:46–56.
100. Wilks M, Collier-Baker E, Nielsen M (2015) Preschool children favor copying a suc-
cessful individual over an unsuccessful group. Dev Sci 18:1014–1024.
101. Herrmann PA, Legare CH, Harris PL, Whitehouse H (2013) Stick to the script: The
effect of witnessing multiple actors on children’s imitation. Cognition 129:536–543.
102. Wood LA, Kendal RL, Flynn EG (2012) Context-dependent model-based biases in
cultural transmission. Evol Hum Behav 33:387–394.
103. Burdett ERR, et al. (2016) Do children copy an expert or a majority? PLoS One 11:
e0164698.
104. Acerbi A, Mesoudi A (2015) If we are all cultural Darwinians what’s the fuss about?
Clarifying recent disagreements in the field of cultural evolution. Biol Philos 30:
481–503.
105. Claidière N, Scott-Phillips TC, Sperber D (2014) How Darwinian is cultural evolution?
Philos Trans R Soc Lond B Biol Sci 369:20130368.
106. Sperber D (1996) Explaining Culture: A Naturalistic Approach (Oxford Univ Press,
Oxford).
107. Morin O (2015) How Traditions Live and Die (Oxford Univ Press, Oxford).
108. Buskell A (March 17, 2017) What are cultural attractors? Biol Philos, 10.1007/s10539-
017-9570-6.
109. Xu J, Dowman M, Griffiths TL (2013) Cultural transmission results in convergence
towards colour term universals. Proc Biol Sci 280:20123073.
110. Kirby S, Cornish H, Smith K (2008) Cumulative cultural evolution in the laboratory:
An experimental approach to the origins of structure in human language. Proc Natl
Acad Sci USA 105:10681–10686.
111. Griffiths TL, Kalish ML, Lewandowsky S (2008) Review. Theoretical and empirical
evidence for the impact of inductive biases on cultural evolution. Philos Trans R Soc
Lond B Biol Sci 363:3503–3514.
112. Miton H, Claidière N, Mercier H (2015) Universal cognitive mechanisms explain the
cultural success of bloodletting. Evol Hum Behav 36:303–312.
113. Tooby J, Cosmides L (1992) The psychological foundations of culture. The Adapted
Mind, eds Barkow JH, Cosmides L, Tooby J (Oxford Univ Press, London), pp 19–136.
114. Henrich J, Boyd R, Richerson PJ (2008) Five misunderstandings about cultural evo-
lution. Hum Nat 19:119–137.
115. Muthukrishna M, Henrich J (2016) Innovation in the collective brain. Phil Trans R Soc
B 371:20150192.
116. Macfarlane A, Martin G (2002) Glass: A World History (Univ of Chicago Press, Chi-
cago).
117. Nia HT, et al. (2015) The evolution of air resonance power efficiency in the violin and
its ancestors. Proc Math Phys Eng Sci 471:20140905.
118. Shtulman A (2006) Qualitative differences between naïve and scientific theories of
evolution. Cognit Psychol 52:170–194.
119. Perreault C, Moya C, Boyd R (2012) A Bayesian approach to the evolution of social
learning. Evol Hum Behav 33:449–459.
120. Turchin P (2003) Historical Dynamics (Princeton Univ Press, Princeton).
121. Collins R (1995) Prediction in macrosociology: The case of the Soviet collapse. Am J
Sociol 100:1552–1593.
122. Waring TM, et al. (2015) A multilevel evolutionary framework for sustainability
analysis. Ecol Soc 20:34.
123. Turchin P, Currie TE, Turner EAL, Gavrilets S (2013) War, space, and the evolution of
Old World complex societies. Proc Natl Acad Sci USA 110:16384–16389.
124. Turchin P (2016) Ages of Discord (Beresta Books, Chaplin, CT).
125. Goldstone JA, et al. (2010) A global model for forecasting political instability. Am J
Pol Sci 54:190–208.
126. Hull DL (1988) Science as a Process (Chicago Univ Press, Chicago).
127. Nelson RR, Winter SG (2002) Evolutionary theorizing in economics. J Econ Perspect
16:23–46.
128. Stout D (2013) Neuroscience of technology. Cultural Evolution: Society, Technology,
Language, and Religion, ed Richerson PJ (MIT Press, Cambridge, MA), pp 157–174.
129. Moretti F (2013) Distant Reading (Verso Books, London).
130. Norenzayan A, et al. (2016) The cultural evolution of prosocial religions. Behav Brain
Sci 39:e1.
131. Slingerland E, Collard M (2011) Creating Consilience: Integrating the Sciences and
the Humanities (Oxford Univ Press, Oxford).
132. Ingold T, Palsson G (2013) Biosocial Becomings: Integrating Social and Biological
Anthropology (Cambridge Univ Press, Cambridge, UK).
133. Youn H, Strumsky D, Bettencourt LMA, Lobo J (2015) Invention as a combinatorial
process: Evidence from US patents. J R Soc Interface 12:20150272.
134. Howe CJ, et al. (2001) Manuscript evolution. Trends Genet 17:147–152.
135. Henrich J (2001) Cultural transmission and the diffusion of innovations. Am
Anthropol 103:992–1013.
136. Mesoudi A, Whiten A (2008) Review. The multiple roles of cultural transmission
experiments in understanding human cultural evolution. Philos Trans R Soc Lond
B Biol Sci 363:3489–3501.
Mesoudi

Evolutionary neuroscience of cumulative culture
Dietrich Stouta,1 and Erin E. Hechtb,c
a
Department of Anthropology, Emory University, Atlanta, GA 30322; bCenter for Behavioral Neuroscience, Georgia State University, Atlanta, GA 30303;
and cYerkes National Primate Research Center, Emory University, Atlanta, GA 30302-5090
Edited by Marcus W. Feldman, Stanford University, Stanford, CA, and approved May 1, 2017 (received for review January 12, 2017)
Culture suffuses all aspects of human life. It shapes our minds and
bodies and has provided a cumulative inheritance of knowledge,
skills, institutions, and artifacts that allows us to truly stand on the
shoulders of giants. No other species approaches the extent,
diversity, and complexity of human culture, but we remain unsure
how this came to be. The very uniqueness of human culture is both
a puzzle and a problem. It is puzzling as to why more species have
not adopted this manifestly beneficial strategy and problematic
because the comparative methods of evolutionary biology are
ill suited to explain unique events. Here, we develop a more
particularistic and mechanistic evolutionary neuroscience approach
to cumulative culture, taking into account experimental, develop-
mental, comparative, and archaeological evidence. This approach
reconciles currently competing accounts of the origins of human
culture and develops the concept of a uniquely human technolog-
ical niche rooted in a shared primate heritage of visuomotor
coordination and dexterous manipulation.
brain evolution | cultural evolution | archaeology | imitation
M odern humans live in a culturally constructed niche of
artificial landscapes, structures, artifacts, skills, practices,
and beliefs accumulated over generations and beyond the ability
of any one individual to recreate in a lifetime (1, 2). Like the air
we breathe, this cumulative cultural matrix is so immersive that it
is easy to forget it is there. However, this is the medium through
which we grow, act, and think, and it exerts profound influences
on human life across a range of behavioral (2), developmental
(3), and evolutionary (4) scales. How did our species find itself in
this remarkable situation?
Niche construction is not unique to humans (5), and many
animals reliably transmit behavioral traditions across generations
(1). In contrast, it is controversial whether any examples of
nonhuman cumulative culture exist and all can agree that no
other species approaches the extent, diversity, and complexity of
human culture (6, 7). Two kinds of explanations have been
proposed for this disjunction. “Individual cognition” accounts
(IC) propose that humans accumulate more complex cultures
primarily because the biological evolution of greater intelligence
in human individuals has promoted innovation and allowed
mastery of more complex concepts and skills (7, 8). Alternatively,
“cultural evolution” accounts (CE) propose that the difference
arises from uniquely human psychological specializations for
“high-fidelity” social learning [e.g., theory of mind (ToM), imi-
tation], which have enabled the lossless “ratchet-effect” of cul-
tural accumulation to supplant biology as humanity’s primary
mode of adaptation (2, 9). Both views recognize a role for social
learning in reducing the costs of knowledge and skill acquisition,
but they differ on the phylogenetic uniqueness (e.g., ref. 7 vs. ref.
9) and transformative power (e.g., ref. 2 vs. ref. 8) of human
high-fidelity social transmission. This plays out in starkly differ-
ent visions of cumulative culture as either a fundamental evo-
lutionary transition (cf. ref. 10) that altered the very medium of
human adaption (2), or just another “unique or extreme” bi-
ological trait comparable to “the elephant’s trunk, the narwhal’s
tusk, the whale’s baleen, the platypus’s duckbill, and the arma-
dillo’s armor” (8). Neither option, however, makes it immediately
www.pnas.org/cgi/doi/10.1073/pnas.1620738114
COLLOQUIUM
PAPER
clear why this exceptional capacity for culture has evolved in hu-
mans and humans alone.
If there is one thing on which IC and CE agree, it is that
cultural capacity is a good thing: It has “undeniable practical
advantages” (8) that have allowed our species to have “expanded
across the globe and. . .occupy a wider range than any other ter-
restrial species” (2). Indeed, the benefits are so substantial that
even small “initial increments” (8) in this direction are expected to
generate powerful biocultural feedback leading to further brain
and cognitive evolution (2, 8). Proponents of a highly modular view
of IC nevertheless argue that this feedback will lead to coordinated
enhancement across multiple domains (8). CE advocates similarly
suggest that evolved cognitive mechanisms (i.e., modules in a loose
sense) for social learning will lead to more general brain size and
intelligence increases to deal with increased amounts of “valuable
cultural information” (2). So why are humans the only species to
have fallen into this virtuous cycle?
Arguing from a CE perspective, Boyd and Richerson (11)
suggest that cumulative culture is rare because of the evolu-
tionary costs of requisite social-learning mechanisms. According
to this argument, accumulation must begin with simple skills that
are within the inventive potential of individuals. Insofar as such
simple skills could be socially learned using existing “low-fidelity”
mechanisms, any small increases in learning efficiency provided by
new high-fidelity social-learning mechanisms would be unlikely to
pay for the (presumably high) metabolic and developmental costs
of those mechanisms. In this case, it would only be after accu-
mulation had already generated a sufficient body of complex,
difficult-to-learn, and useful cultural content that these expen-
sive mechanisms would begin to pay for themselves. Boyd and
Richerson (11) thus suggest that high-fidelity social-learning
capacities initially arose as a side effect (exaptation) of some
other adaptation, such as behavior prediction for Machiavellian
social strategizing (12, 13). Although reasonable, this hypothesis
is weakened by its reliance on assumptions regarding the cost of
social-learning mechanisms and the rarity of social-cognitive pre-
adaptations. In fact, the hypothesis does not so much explain the
rarity of cumulative culture as shift the puzzle to explaining the
rarity of preadaptations, like behavior prediction, which might also
be assumed to be quite generally beneficial.
An alternative, IC-compatible proposal is that it is large brains
in general that are expensive, rather than social-learning capacities
in particular. Thus, Pinker (8) argues that human cognitive ex-
ceptionalism arises from the fortuitous fact that “hominid ances-
tors, more so than any other species, had a collection of traits that
had tilted the payoffs toward further investment in intelligence.”
NEUROSCIENCE
Whether such “intelligence” is thought to be composed of discrete
but tightly coevolving innate modules (8) or a general-purpose
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
ANTHROPOLOGY
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: D.S. and E.E.H. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1
To whom correspondence should be addressed. Email: dwstout@emory.edu.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7861–7868
capacity only “secondarily” specialized by experience (14), the
premise is that bigger brains are generally advantageous but can
only evolve under a narrow set of conditions. Lifespans must be
long enough (i.e., mortality low enough) to reward early invest-
ments in growth and additional energetic costs must be accom-
modated through increased intake or reallocation (4). For many
species, constraints such as small body size, unstable environ-
ments, and unavoidable mortality from predation and disease may
simply make large brains impractical. In this framework, high-
fidelity social learning can increase the payoffs of large brains
(8, 15) but is not associated with its own unique costs, as it is
thought to rely on cognitive abilities that overlap (15) or evolved in
tandem with (8) asocial learning and problem solving. This ap-
proach actually parallels the suggestion of Boyd and Richerson
(11) that initial enhancements of social-learning capacities were a
byproduct of selection on other capacities, and similarly shifts the
question back to identifying the unique trait (4) or combination of
traits (8) that pushed humans, and no others, into an auto-catalytic
coevolutionary feedback loop.
The strength of both IC and CE accounts is that they explain
a wide range of distinctive human traits in terms of a single co-
evolutionary process. However, this parsimony is offset by the
perceived need to posit a uniquely exceptional [and possibly even
random (16)] initial cause inaccessible to more general evolu-
tionary explanation (8, 17, 18). An alternative is to consider cu-
mulative culture, not as a unitary capacity that is either present or
absent, but as a complex trait with a correspondingly complex
history of gradual or piecemeal emergence. Although any event
that altered the evolutionary cost/benefit analysis of either brain
expansion generally (4, 8) or high-fidelity social learning specifi-
cally (2, 16) could theoretically have initiated runaway biocultural
coevolution in our lineage, there is no reason to assume this
feedback would be indefinitely self-sustaining once initiated, nor
that that it would necessarily produce constant increase as op-
posed to more complex dynamics. In fact, both comparative bi-
ological evidence (4) and cultural evolutionary models (19)
indicate the potential for just such interactions and dynamics and
this is entirely consistent with the emerging paleoanthropological
picture of multilineal, intermittent, asynchronous change over
human evolution (20, 21). This indication suggests a more con-
tingent evolutionary history, likely involving multiple inflection
points in response to perturbations both intrinsic (19) and ex-
trinsic (20) to hominin behavior systems. If this is the case, un-
derstanding coevolutionary feedback dynamics (4, 8) and CE
processes (2, 19) will be necessary but not sufficient to explain the
actual path of human evolution, which will additionally require
the application of these general principles to explain particular
historical contingencies (17).
In our view, this situation calls for a particularistic and mecha-
nistic approach to the study of cumulative culture. As exemplified
below, such an “evolutionary neuroscience” perspective evaluates
comparative evidence of brain and behavioral variation in light of
(i) evolutionary and developmental processes, (ii) primary archae-
ological and paleontological evidence of evolutionary timing and
context, and (iii) the ethnographic, ethological, and experimental
analogies needed to interpret this primary evidence.
High-Fidelity Social Reproduction
To begin with, it is necessary to further dissect “cumulative cul-
ture” as a complex trait with heterogeneous cognitive and be-
havioral prerequisites (cf. ref. 6) and capable of various degrees of
expression. Although there is broad agreement (2, 8, 9, 15, 22)
that accurate transmission is necessary for cultural accumulation,
the actual fidelity (i) required to accumulate particular behaviors,
(ii) associated with different social-learning mechanisms, and (iii)
typically exhibited by different species are all largely unknown
(22). Experimental studies of artifact reproduction in humans
suggest that the required fidelity and relevant mechanisms depend
7862 | www.pnas.org/cgi/doi/10.1073/pnas.1620738114
on the complexity and difficulty of the production process.
Transmission chains building spaghetti towers and paper air-
planes achieve sufficient fidelity for cumulative improvement
even in purely end-state emulative (i.e., reverse engineering)
conditions (23), whereas more challenging tasks—such as de-
signing virtual “fishing nets” (24), building real weight-bearing
devices (25), and reproducing particular artifact forms (26)—
may require imitative copying of specific actions or processes.
Interactions between particular tasks demands, required fidelity,
and sufficient mechanisms are critical to the interpretation of
comparative and evolutionary evidence, yet remain underex-
plored and undertheorized. The objective is obviously to move
from particularistic analyses of specific behaviors to identifica-
tion of general principles, yet it is not obvious how to design or
select experimental tasks to efficiently advance this goal.
One solution is to seek inspiration from the archaeological
record of human evolution (e.g., refs. 26 and 27). As the name
implies, this early “Paleolithic” evidence is dominated by stone
tools. These artifacts are valuable, not only because they endure
but because they provide prolific and fine-grained evidence of
behavioral changes across a critical evolutionary interval during
which hominin brains tripled in volume to assume their modern
proportions. Stone tools were key components of premodern
subsistence and survival strategies and likely helped to shape the
very course of this evolution. Experimental, comparative, and
ethnographic evidence indicate that stone tool-making (“knap-
ping”) is a complex skill integrating demands for planning, problem
solving, and perceptual-motor coordination within a collaborative
social context (28–31). It encompasses an evolutionary continuum
from early Paleolithic skills at or just beyond the limits of modern
apes (32) to the virtuoso craftsmanship of later prehistory (33). The
study of knapping skill acquisition and transmission is thus a
promising avenue for evolutionarily grounded investigation into the
foundations of human cumulative culture, including the copying
fidelity needed to explain empirical patterns in the archaeological
record (34, 35). To be clear, knapping is but one of many evolu-
tionarily relevant skills that might be studied, but it is one of which
we have a good archaeological record and which may reasonably be
hoped to be representative of broader trends.
Knapping is a “reductive” technology involving the sequential
detachment of flakes from a stone core using precise ballistic
strikes with a handheld hammer (typically stone, bone, or antler)
to initiate controlled and predictable fracture. This means that
small errors in strike execution can have catastrophic, unrever-
sible effects. Experiments by Bril and colleagues have shown that
fracture prediction and control is a demanding perceptual-motor
skill reliably expressed only in expert knappers (28, 29). Building
on this work, Stout and colleagues (31, 36, 37) found that even
22 mo (x ̄ = 167 h) of knapping training produced relatively little
evidence of perceptual-motor improvement, in contrast to clear
gains in conceptual understanding.
The key bottleneck in the social reproduction of knapping is
thus the extended practice required to achieve perceptual-motor
competence. This requires mastery of relationships, for example
between the force and location of the strike and the morphology,
positioning, and support of the core (29, 38, 39), that are not
perceptually available to naïve observers and cannot be directly
communicated as semantic knowledge. Attempts to implement
semantic knowledge of knapping strategies before perceptual-
motor skill development are ineffective at best (40, 41), and
such knowledge decays rapidly along knapping transmission chains
when practice time is limited, even if explicit verbal teaching is
allowed (27). For observational learning, the challenge is to
translate visual and auditory information of another’s actions to
appropriate motor commands for one’s own body. This may be
accomplished by linking the observed behavior with preexisting
internal models of one’s own body and actions through associative
learning and stimulus generalization (42, 43). Novel behaviors are
Stout and Hecht

copied by breaking them down into familiar action elements (e.g.,
lift, turn, twist), matching these, and reassembling (44). Such
matching always requires a degree of approximation, if only be-
cause no two bodies are identical, but where low-level differences
are not critical it can enable fast imitation learning (13). Unfor-
tunately, such low-level details do matter for knapping. Learners
must thus proceed by incorrectly executing an observed gesture or
communicated strategy, checking the actual vs. predicted outcome,
and embarking on a lengthy process of behavioral exploration to
(re)discover relevant task constraints and develop corresponding
internal models through a self-conscious process of deliberate
practice (cf. ref. 45). These learning challenges call for an iter-
ative approach that alternates social-learning opportunities
(observation, instruction) with motivated individual practice (46),
as commonly seen in coaching or apprenticeship learning. Whiten
(47) describes this as a “helical curriculum.”
This complexity is not easily reconciled with the dichotomy of
high-fidelity process copying (imitation) vs. low-fidelity product
copying (emulation) prevalent in discussions of cumulative cul-
ture (e.g., ref. 48). For knapping and similar crafts, what is ac-
tually reproduced across individuals is a flexible skill rather than
an invariant formula, and what must be learned is a structured
set of relations [compare with, for example, “affordances” (49)]
between effective means and appropriate goals at multiple levels
of task organization (50). This is achieved through a diverse
helical curriculum in which many varieties of social learning may
scaffold individual development of perceptual-motor skill and
strategic competence (Fig. 1). Whiten and colleagues (e.g., see
ref. 12) have identified a “portfolio” of such processes, ranging
from copying bodily postures and gestures to object movement
reenactment and end-state emulation. In humans, these may
even extend to copying cultural norms of affect, behavior, and
social affiliation (30). Although such copying may be a goal in
itself when learning conventional behaviors, like human dance or
the “do-as-I-do” experimental paradigm used with apes (12), in
instrumental skill acquisition these processes are means to the
end of learning subtle affordances and complex causal relations
(51). Such learning can be further facilitated by (passive or ac-
tive) niche construction, increasing exposure to relevant mate-
rials and situations (30, 52) and by teaching, potentially including
the provision of practice opportunities, the direction of attention,
Stout and Hecht
COLLOQUIUM
PAPER
and affective feedback, as well as intentional demonstration and
explicit instruction (53).
Underlying this diversity of processes, are a small number of
fundamental psychological capacities required for efficient high-
fidelity skill learning in a helical curriculum. These include:
(i) the ability to relate fine details of actions and objects to
complex goals during behavior observation and execution (46)
and (ii) the prosocial motivation and ToM needed to support
apprenticeship learning (9, 54, 55). In keeping with comparative
(12) and archaeological (50) evidence, variation in these capac-
ities is expected to be continuous rather than binary, and to be
reflected in the complexity and difficulty of skills that can be
accurately reproduced. The latter point suggests that the ar-
chaeological record of increasingly complex and difficult stone
tool-making (31, 36, 50) could help adjudicate between (or rec-
oncile) IC and CE accounts of human evolution, but only in the
context of a solid inferential framework linking brains, behaviors,
and evolutionary processes. Key questions include the extent and
nature of overlap between processes supporting behavior exe-
cution, observation, and interpretation (e.g., ToM), and the
relevance of evolutionary processes other than natural selection
(e.g., CE). An emerging extended evolutionary synthesis (EES)
effectively addresses both topics through its core concepts of
constructive development and reciprocal causation (56).
An Extended Evolutionary Framework
As Deacon (57) provocatively put it, “brain evolution should be
impossible.” How could random mutational changes to such a
complex integrated system be anything other than catastrophic?
The answer is that brain development is itself an evolutionary
process of remarkable flexibility and adaptability. Basic patterning
mechanisms and developmental selection (58) can produce func-
tional systems even in the face of quite significant environmental
or genetic perturbation. Such plasticity may be essential to the
evolvability of larger brains, as developmental programs and net-
work architectures are forced to accommodate shifting geometries
together with massive increases in the number of potential neu-
ronal connections to be specified. It also creates a medium for the
multilevel interactions between genes, developmental mecha-
nisms, environments both internal and external, and organismal
behavior that the EES describes as “constructive development”
NEUROSCIENCE
Fig. 1. A learning cycle in the helical curriculum.
Social resources both passive and pedagogical (53),
together with constructed learning contexts (52),
provide opportunities and structure for individual
ANTHROPOLOGY
practice, which can include a portfolio (12) of pro-
cesses ranging from “emulative” end-state copying to
the imitation of specific body movements. This be-
havioral continuum maps roughly onto the differing
contributions of dorsal and ventral processing streams
in the primate brain.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7863
(56). Any viable evolutionary account of cumulative culture must
address these dynamics.
The primate neocortex is divided into large-scale functional
networks characterized by high within-network functional and
anatomical connectivity (59, 60). In humans, these networks are
organized in a processing hierarchy from concrete perceptual
and motor functions to abstract, domain-general processing. This
arrangement is realized anatomically as a cortex-wide gradient of
topological distance and connectivity patterns extending from
distinct peripheral sensorimotor cortices at one end to highly
interconnected central association cortices at the other (61).
Along this gradient, seven widely recognized networks can be
arrayed into four organizational tiers: (i) visual and somatomotor,
(ii) dorsal and ventral attention, (iv) multiple demand and limbic,
and (iv) default mode (61).
The tethering hypothesis of Buckner and Krienen (62) proposes
that this pattern arises from disproportionate expansion of the
cortical mantle during evolutionary brain enlargement, leading to
gaps between the chemical signaling gradients that pattern cortical
differentiation during development. Developmental selection in
these gaps fosters the emergence of “noncanonical” association
networks primarily interconnected with each other rather than
with more developmentally constrained peripheral sensorimotor
systems. As expected, these relatively unconstrained association
cortices are also relatively late developing (59, 63) and variable
in connectivity across individuals (64). Indeed, comparative evi-
dence indicates human-specific changes in the rate and timing
of synaptogenesis, synapse elimination, and cortical myelination,
resulting in increased plasticity into adulthood (65, 66). That
nonspecific selection for increased brain size in the human lineage
might have indirectly driven increased plasticity is suggested by
evidence of low heritability for cortical morphology (sulcal di-
mensions) vs. overall brain size in humans, a pattern that contrasts
with high heritability of both in chimpanzees (67). In any case, the
human association cortex appears particularly sensitive to envi-
ronmental and behavioral influences, providing a potent evolu-
tionary feedback mechanism between organism and environment,
which the EES refers to as reciprocal causation (56).
Such phenotypic flexibility is useful but may come at a cost
(e.g., investments in learning or temporary phenotype–environment
mismatches). Where possible, natural selection is expected to
reduce costs by “canalizing” plastic responses to recurring envi-
ronmental situations as automatic parts of normal development
(68). The tethering hypothesis suggests that such innate spe-
cializations are most likely to be found in relatively heritable
sensorimotor systems and with respect to behaviors/stimuli that
have been relatively invariant over long periods of time. Because
humans’ expanded association areas remain relatively plastic
(64) and are both late developing (59) and phylogenetically re-
cent (60), their derived cognitive features are less likely to
be directly shaped by natural selection (phylogenetically con-
structed) and more likely to result from developmental side ef-
fects (developmental construction) and modifications to the
structure of inputs they receive from more peripheral systems
(phylogenetic or developmental inflection) (69). In theory, such
modifications could arise through environmental as well as ge-
netic inheritance, including persistent changes to the physical
and social context of development brought about through niche
construction (52, 70). For example, there is widespread agree-
ment that the human brain lacks specific genetic adaptations for
literacy, and yet learning to read reliably produces functional
specialization for script perception in a particular region of the
left ventral occipitotemporal cortex known as the “visual word
form area” (3). Similar logic may apply to the enhanced mech-
anisms for complex action parsing and ToM that support ap-
prenticeship learning in a helical curriculum.
Skilled actions, such as the ballistic strikes involved in stone
knapping, often unfold too quickly to be guided by online sensory
7864 | www.pnas.org/cgi/doi/10.1073/pnas.1620738114
feedback and error correction. This limitation can be overcome
through the use of internal models that predict movements and
outcomes in advance (71), a simulation process supported by a
distributed network of frontal, parietal, and occipitotemporal re-
gions combining elements of the dorsal and ventral attention net-
works. As argued above, such models of self-action control likely
form the basis for understanding and copying the observed actions
of others through a process of matching, often referred to as
“motor resonance” (42). The assembly of complex goal-oriented
sequences from these elements is likely supported in the next tier
of cortical organization by the multiple demand system (aka
“frontoparietal control network”). This cognitive control network is
thought to support general or “fluid” intelligence through its role in
assembling structured mental programs from a series of subtasks
(72), a critical process in skill learning as reviewed above. Together,
these sensorimotor matching and control processes support the
interactive behavioral alignment that is critical to human social
learning, communication, cooperation, and bonding (73, 74).
It is debatable whether the application of increasingly sophis-
ticated motor planning and cognitive control networks to social
learning involved phylogenetic construction or is entirely explica-
ble in terms of developmental construction and inflection (42, 75),
but either scenario is consistent with the IC premise that social
learning substantially overlaps with asocial-learning mechanisms
and comes at little additional cost (15). It also fits well with the
close integration of individual and social-learning processes in
real-world skill acquisition, as envisioned by the helical curricu-
lum. Advocates of CE, however, emphasize the additional im-
portance of a specialized ToM capacity to allow truly cultural
learning (9). Does this requirement imply additional evolutionary
costs for cultural learning?
Although ToM is commonly thought of as a human specialization,
depending on phylogenetically constructed neurocognitive mecha-
nisms, Heyes and Frith (76) have recently argued that it is largely a
product of developmental inflection and cultural evolution. On this
account, low-level or “implicit” mind-reading capacities emerge di-
rectly from motor resonance properties of the action control system
discussed above. Motor resonance provides the input needed to
identify recurring relations between actions, outcomes, and internal
states, and thus to predict behavior and infer intent. This would be
largely explicable in terms of general mechanisms [e.g., statistical and
associative learning (77)] acting within developmentally constructed
systems, as favored by some IC accounts (15).
Explicit mind reading, in contrast, involves active reasoning
about mental states: in other words, the “theory” part of “theory
of mind.” Anatomically, this appears to be supported by regions
of posterior cingulate, medial frontal, and lateral temporopar-
ietal cortex associated with the so-called “default mode network”
(DMN) (78). The DMN was initially identified as a set of regions
that experience de-activation during attention-demanding tasks,
but is increasingly recognized to make a positive contribution to
abstract, internally directed tasks involving information retrieval
and integration. Examples include introspection, social cogni-
tion, autobiographical memory, future planning, narrative com-
prehension, and goal-directed working memory. This functional
profile reflects the fact that the DMN sits atop the cortical
processing hierarchy: maximally distant from peripheral senso-
rimotor systems and dominated by internal connectivity with
other association networks (61). As discussed above, its devel-
opment and function are thus expected to be highly plastic and
reliant on learning. In fact, Heyes and Frith (76) propose that
learning explicit mental theories is an inherently cultural process
requiring language-based instruction. Their view is supported by,
among other things, evidence that individual and cross-cultural
differences in caregiver use of mental state vocabulary are pre-
dictive of variation in childrens’ acquisition of ToM concepts,
such as false belief, knowledge vs. ignorance, and difference of
opinion. Insofar as mind-reading capacities are themselves seen
Stout and Hecht

as critical to language acquisition (9) and both may have been
important to Paleolithic skill reproduction (27, 55), this suggests
a deep and densely reciprocal history of biological, cultural, and
developmental interactions during the evolution of the capacities
that support cumulative culture.
An EES perspective thus charts a middle path between IC and
CE extremes. The action-parsing and ToM capacities that support
high-fidelity transmission do have a biological basis (cf. ref. 8)
and are indeed specialized and highly derived in humans (2),
but they may also be “secondarily modular” (15) products of
processes other than phylogenetic construction (9). A long his-
tory of reciprocal interaction between cultural and biological
evolution confirms the importance of considering CE processes
(2; contra ref. 8) but contradicts the view of cumulative culture
origins as a single “key event” (2) involving “one and only one
biological adaptation” (9). Disentangling the respective contri-
butions of IC and CE changes will require evidence of the se-
quence, timing, and context of evolutionary developments.
Evolution of Primate Action Systems
The gold standard for reconstructing such evolutionary develop-
ments is phylogenetic inference from comparative evidence of ex-
tant species (Fig. 2). Specialized neural machinery for visuomotor
integration is perhaps the quintessential primate adaptation to a
diurnal life in the trees (75, 79). Success in this niche was supported
by the emergence of two new brain regions. The ventral premotor
cortex (PMv) appeared anterior to primary motor cortex, and
allowed for the integration of visual input with new, higher-order
control of movement sequences. In the temporal cortex, the middle
temporal (MT) visual area (or V5) appeared as a specialized
motion-processing region. These basic adaptations are present in
all primates and were in place at or near the root of our clade.
Stout and Hecht
COLLOQUIUM
PAPER
From this early template, further adaptations emerged. Mo-
tion processing expanded from the MT to produce a dorsal
stream of “vision for action” extending into the posterior parietal
cortex (80). The internal models of movement in space processed
in this stream provide the “how” needed to execute or copy (Fig.
1) bodily actions. A second ventral stream of “vision for per-
ception” extends into the lateral and inferior temporal lobe. This
“what” pathway supports the recognition of objects, individuals,
and body parts critical for organizing goal-directed action and
emulating observed outcomes.
Whereas the presence of both dorsal and ventral visual streams
across macaques, chimpanzees, and humans reflects the ancient
origins of these systems, structural and functional differences be-
tween species provide evidence of subsequent evolutionary
changes. It is clear that, like other association cortices, temporal
regions associated with the ventral stream underwent substantial
enlargement over ape and human evolution (62) and this likely
fostered new functional capacities, ranging from semantic pro-
cessing to face recognition. However, we argue that it was a series
of key evolutionary changes to the dorsal stream that enabled in-
tegration of increasingly fine action details and complex goals
during behavior observation and execution, and thus supported the
emergence of high-fidelity social learning and cumulative culture.
One such change was the emergence of new functional regions
in the parietal cortex. Comparative fMRI studies with macaques
and humans have identified portions of human intraparietal
sulcus with novel sensitivity to 3D form-from-motion stimuli, as
well as a patch of human anterior supramarginal gyrus (aSMG)
specifically responsive to observed tool use (81). Both regions
are likely relevant to a wide array of evolutionarily relevant
object-manipulation behaviors, and both are known to be
recruited by stone tool-making activities in modern humans (82,
NEUROSCIENCE
ANTHROPOLOGY
Fig. 2. Species differences in action processing cir-
cuitry (Upper) and inferred phylogenetic origins
(Lower).
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7865
83). The aSMG in particular is believed to support a confluence
of object information from the ventral stream with dorsal stream
kinematics to manage the novel action possibilities afforded by
handheld tools (81). Although similar functional studies have not
been conducted with chimpanzees, structural evidence from
diffusion tensor imaging suggests similarity with humans.
Whereas macaques show little or no connectivity between the
aSMG and ventral stream object-processing cortex, in both hu-
mans and chimpanzees these connections are robust (84). This
tool-relevant innovation appears to predate the chimpanzee–
human split, as might be expected from the impressive tool-use
capacities of modern chimpanzees (13).
Enhanced aSMG connectivity between dorsal and ventral
steams is just one aspect of a broader pattern of changes (Fig. 2)
in action circuitry over ape and human evolution (84). In ma-
caques, this circuitry is dominated by ventral stream projections
to the ventrolateral prefrontal cortex (vlPFC) along the inferior
longitudinal fasciculus and extreme/external capsules, with rela-
tively little connectivity through the dorsal stream. Dorsal stream
projections to the vlPFC as well as the PMv through the middle
and superior longitudinal fasciculi are better developed in
chimpanzees and become quite pronounced in humans. Across
these three taxa, there is thus a trend toward the addition of
increasing dorsal stream inputs to the vlPFC in complement to
established ventral stream connectivity. We have previously
proposed that this dorsal stream enhancement may underlie the
progressive elaboration of action-parsing capacities from ma-
caques to chimpanzees and then humans (84, 85), as reflected
by these species’ increasingly complex foraging skills (13) and
capacities/propensities for bodily imitation (12).
Motor resonance mechanisms are least-developed in macaques,
which are not known to imitate manual actions. Macaque “mirror”
neurons respond to observed action goals rather than detailed
means of execution and are almost entirely unresponsive to actions
that do not involve an object (75). In contrast, chimpanzee action
observation activates nearly identical voxels to execution of the
same movements, regardless of whether they produce a physical
result on an object (85). This basic action-matching mechanism may
thus predate the chimpanzee–human split. With respect to object-
directed actions, however, chimpanzees retain a generally macaque-
like pattern of brain response dominated by the “top-down” con-
tributions of the frontal executive cortex (85, 86). Humans alone
display a more distributed pattern of occipital, temporal, parietal,
premotor, and prefrontal activation, reflecting an increased role
for bottom-up perceptual representations incorporating kinematic
and spatiotemporal details about object-directed actions (85). This
functional difference, and the structural changes that support it,
may be critical to the exceptional development of skill acquisition
and cultural learning capacities in humans.
In humans but not chimpanzees or macaques, the core action
perception circuitry includes a prominent projection to the su-
perior parietal lobule, a region associated with awareness of
one’s body in space (84). Furthermore, the third branch of the
superior longitudinal fasciculus (SLFIII), which links the inferior
parietal cortex with the PMv in monkeys, extends into more
anterior regions of the vlPFC in humans, particularly in the right
hemisphere (87). Chimpanzees again appear intermediate, with
a weak but observable extension of SLFIII into vlPFC and no
evidence of right-lateralization at the population level (87). Thus,
a robust extension of SLFIII into the right hemisphere homolog of
Broca’s area appears to be a human-specific adaptation. This re-
gion is an element of the multiple demand system discussed above,
which is thought to support the assembly of complex, multistep
action plans (88). The observed extension of human SLFIII would
thus provide an anatomical substrate for the integration of kine-
matic details into complex action goals and sequences, as required
for skill learning in a helical curriculum.
7866 | www.pnas.org/cgi/doi/10.1073/pnas.1620738114
Such learning in humans requires a degree of bodily awareness
sufficient to match variations in kinematic detail with desired
outcomes during deliberate practice. A measure of such aware-
ness that has been applied to other animals is the Mirror Self-
Recognition (MSR) test (89). Unlike enculturated humans,
mirror-naïve animals must discover de novo that the visual per-
ception of their reflection corresponds to the sensorimotor
representations of their own movements. As might be expected
from the preceding review of action perception circuitry, ma-
caques typically fail at MSR and chimpanzees are intermediate
in performance, with some passing and some failing. In fact,
chimpanzee MSR performance is predicted by individual varia-
tion in the degree of right-lateralization of SLFIII projections
into the vlPFC. In other words chimpanzees with more human-
like SLFIII connectivity show more human-like MSR behavior
(90). Because attending to one’s own movements is a critical el-
ement in the hypothesized construction of implicit mind reading
from motor resonance (76), this finding suggests further links
between dorsal stream evolution and the cognitive prerequisites
of cultural learning.
What is not yet known is the extent to which any or all of these
structural and functional differences between species are classic
“adaptations” in the sense of being canalized products of phy-
logenetic construction vs. other evolutionary processes. Even in
macaques, there is some evidence that extensive tool-training
can produce plastic alterations in dorsal stream connectivity
(91). Enlarged ape and human brains are expected to be more
developmentally plastic and subject to inflection by somatic [e.g.,
bipedal locomotion, hand morphology (92)] and sensorimotor
adaptations, and developmental niche construction (70). In fact,
research with modern humans has shown that the acquisition of
Paleolithic tool-making skills elicits plastic remodeling of dorsal
stream white matter connections, including SLFIII’s projection
into the right vlPFC, even in adults (37). Functionally, the gray
matter targeted by this projection is recruited by execution (93)
and observation (83) of relatively complex tool-making se-
quences of the kind that appeared with Late Acheulean hand-
axe technology after about 0.7 Mya (50). Such findings suggest
that further experimental studies of Paleolithic tool-making may
begin to fill in details of timing, mechanisms, and context of
evolutionary changes that occurred since the chimpanzee–human
divergence and are inaccessible to purely comparative methods.
Conclusion: An Evolving Technological Niche
The earliest stone tools (Fig. 2) predate evidence of brain
expansion by hundreds of thousands of years (32, 94) during
which their occurrence was extremely patchy, discontinuous, and
lacking in evidence of progressive change. By 2.6 Mya, early
Oldowan knapping provides some evidence for high-fidelity
cultural transmission of particular methods (35), as well as in-
creasing demands on visual, motor and attentional systems (82),
but the overall impression in this early period remains one of a
tenuous and expendable technology at the edge of contemporary
hominin capacities. It is only after about 2.0 Mya that stone tool-
making appears to become more commonplace (as indicated by
site frequency and geographic distribution), at which time it is
accompanied by evidence of brain- and body-size increase (20).
Further episodes of apparently correlated technological (50) and
brain size (20, 95) change occurred with the appearance of in-
creasingly skill-intensive Early (1.7 Mya) and Late (0.7 Mya)
Acheulean knapping. Understanding what exactly changed at
these various transitions is an important priority for future research
and will ultimately require an integration of CE approaches to
understanding technological change (19) and IC insights into the
evolutionary economics of “expensive” brains (4), with mechanistic
neuroscientific perspectives on evolving brain–behavior–culture
interactions.
Stout and Hecht

Such an approach highlights the central importance of em-
bodied skills for object manipulation and modification to the
high-fidelity social reproduction that IC and CE agree is critical
to sustained biocultural feedback. Dorsal stream action systems
constitute a critical substrate for the evolutionary-developmental
cascade that constructs the action-parsing and ToM capacities
needed for cumulative culture. These systems are themselves
sensitive to inflected input from more peripheral somatic or
sensorimotor adaptations and constructed niches that shape
early object manipulation and visual experiences (96). Object
manipulation and modification contribute to the construction of
learning niches (Fig. 1) populated by the residues of past action
(52), and provide a persistent external medium scaffolding pro-
duction of the more complex and protracted action goals and
sequences (46) that both require and reward cultural learning
through a helical curriculum.
As we have seen, stone knapping requires the multilevel in-
tegration of bodily kinematics and object affordances into goal-
oriented action sequences. This finding is supported by phylo-
genetic (87) and plastic (37) enhancements to action control
systems and demands extended investment in deliberate practice
(36). Oldowan knapping, although far from easy compared with
ape tool-use (32), remains a forgiving technology that is rela-
tively quickly acquired (36) because the limited contingency
between successive actions allows substantial latitude for error
(31). Given sufficient sensorimotor and action control capacities,
it is likely that social-learning opportunities (15) and niche-
construction processes (52) seen in socially tolerant nonhu-
mans would be sufficient for Oldowan skill acquisition (cf. ref.
29). Nevertheless intentional demonstration and linguistic in-
struction are clearly helpful (27), and such teaching did even-
tually become a vital part of human technological reproduction
(53). Archaeological evidence cannot demonstrate that a par-
ticular form of teaching was essential at a given point in pre-
history but does document transmission of quite complex and
demanding techniques by Late Acheulean times (41), some of
which modern humans find difficult to convey without explicit
verbal instruction (40, 97).
Modern human teaching relies on social affiliative bonds
that support extended interaction and motivate investment by
teachers. Such investments are a special case of the alloparental
contributions that subsidize human reproduction (4), and their
affective substrates likely evolved as part of a more general
prosocial package (98, 99) relying on strong affiliative bonds.
The interindividual behavioral alignment facilitated by reso-
nance (74) is a critical mechanism that promotes such affiliation
across many species (73), and in humans also supports pragmatic
1. Whiten A, Hinde RA, Laland KN, Stringer CB (2011) Culture evolves. Philos Trans R
Soc Lond B Biol Sci 366:938–948.
2. Boyd R, Richerson PJ, Henrich J (2011) The cultural niche: Why social learning is es-
sential for human adaptation. Proc Natl Acad Sci USA 108:10918–10925.
3. Dehaene S, Cohen L, Morais J, Kolinsky R (2015) Illiterate to literate: Behavioural and
cerebral changes induced by reading acquisition. Nat Rev Neurosci 16:234–244.
4. Isler K, Van Schaik CP (2014) How humans evolved large brains: Comparative evi-
dence. Evol Anthropol 23:65–75.
5. Odling-Smee FJ, Laland KN, Feldman MW (2003) Niche Construction: The Neglected
Process in Evolution (Princeton Univ Press, Princeton).
6. Dean LG, Vale GL, Laland KN, Flynn E, Kendal RL (2014) Human cumulative culture: A
comparative perspective. Biol Rev Camb Philos Soc 89:284–301.
7. Whiten A, Horner V, Marshall-Pescini S (2003) Cultural panthropology. Evol Anthropol
12:92–105.
8. Pinker S (2010) Colloquium paper: The cognitive niche: Coevolution of intelligence,
sociality, and language. Proc Natl Acad Sci USA 107:8993–8999.
9. Tomasello M (1999) The Cultural Origins of Human Cognition (Harvard Univ Press,
Cambridge, MA).
10. Szathmáry E, Smith JM (1995) The major evolutionary transitions. Nature 374:
227–232.
11. Boyd R, Richerson P (1996) Why culture is common but cultural evolution is rare.
Evolution of Social Behaviour Patterns in Primates and Man, eds Runciman W,
Maynard Smith J, Dunbar R (Oxford Univ Press, Oxford), pp 77–93.
Stout and Hecht
COLLOQUIUM
PAPER
collaboration, apprenticeship learning, and the development of
ToM, as discussed above. In contrast to other animals, however,
human affiliation routinely extends to include “ultrasocial” co-
operation and sharing with nonkin enabled by the use of lan-
guage to create social norms and identities, including purely
symbolic affinal and fictive kinship ties (98, 99).
A substantial body of research has linked language evolution
to exaptation of the same action control systems discussed here
(e.g., ref. 100), often with an emphasis on gesture production and
action “syntax” in the dorsal stream (55). Less emphasized has
been the critical role that such high-fidelity production (whether
manual, vocal, or artifact-mediated) plays in grounding semantic
meaning (101). Consistent physical “tokens” anchor meaning
and allow discrete recombination in a way that continuous as-
sociative networks, however complex, do not. In other words,
concrete tokenization appears critical to the transition from
broadly semantic to truly symbolic thought (57), as exemplified
by the construction of precise mathematics from approximate
numeracy using number words as tokens (102). Once learned,
this externalizing symbolic “trick” can readily be internalized as a
tool for thought, supporting such abstract concepts as social roles
that can be filled by different actual individuals. The elaboration
of action control systems supporting skilled action and behav-
ioral alignment (74) thus emerges as a foundational adaptation
enabling the modern human way of life.
This way of life might fairly be described as a human cultural
niche; however, this term is already taken to refer to a somewhat
more circumscribed (CE) theoretical proposal (2). We have thus
suggested the term “technological niche” (36) to emphasize the
critical role for object manipulation and modification indicated
by evolutionary neuroscience, as well as the necessary integration
of cultural evolutionary thinking with the economic logic of the
expensive brain to address biocultural interactions across evo-
lutionary time scales. As should be evident from the preceding
discussion, “technological” is not meant to imply a narrow focus
on artifacts apart from the broader processes that generate them.
Indeed, the now-ubiquitous term “technology” did not even
come into widespread use until well into the 20th century, when
it was adopted to describe: (i) a newly emerging and radically
transformative power going far beyond anything implied by ex-
tant conceptions of the “mechanical” or “useful” arts, and (ii) the
new configurations of materials, social and economic relations,
institutions, and regulations that come together to constitute a
unitary “thing,” like “automotive technology” (103). In this
sense, we find the term to be a particularly apt description of the
content, scope, and power of coevolutionary relationships that
conspired to produce the modern human condition.
12. Whiten A, van de Waal E (December 26, 2016) Social learning, culture and the ‘socio-
cultural brain’ of human and non-human primates. Neurosci Biobehav Rev, 10.1016/
j.neubiorev.2016.12.018.
13. Byrne RW (2016) Evolving Insight: How It Is We Can Think About Why Things Happen
(Oxford Univ Press, Oxford).
14. Burkart JM, Schubiger MN, van Schaik CP (July 28, 2016) The evolution of general
intelligence. Behav Brain Sci, 10.1017/S0140525X16000959.
15. van Schaik CP, Isler K, Burkart JM (2012) Explaining brain size variation: From social
NEUROSCIENCE
to cultural brain. Trends Cogn Sci 16:277–284.
16. Henrich J, McElreath R (2003) The evolution of cultural evolution. Evol Anthropol 12:
123–135.
17. Stout D, Human brain evolution: History or science? Rethinking Human Evolution, Vienna
Series in Theoretical Biology, ed Schwartz JH (MIT Press, Cambridge, MA), in press.
18. Cartmill M (2002) Paleoanthropology: Science or mythological charter? J Anthrop
Res 52:183–201.
19. Kolodny O, Creanza N, Feldman MW (2016) Game-changing innovations: How cul-
ture can change the parameters of its own evolution and induce abrupt cultural
ANTHROPOLOGY
shifts. PLOS Comput Biol 12:e1005302.
20. Antón SC, Potts R, Aiello LC (2014) Human evolution. Evolution of early Homo: An
integrated biological perspective. Science 345:1236828.
21. d’Errico F, Stringer CB (2011) Evolution, revolution or saltation scenario for the
emergence of modern cultures? Philos Trans R Soc Lond B Biol Sci 366:1060–1069.
22. Lewis HM, Laland KN (2012) Transmission fidelity is the key to the build-up of cu-
mulative culture. Philos Trans R Soc Lond B Biol Sci 367:2171–2180.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7867
23. Caldwell CA, Schillinger K, Evans CL, Hopper LM (2012) End state copying by humans
(Homo sapiens): Implications for a comparative perspective on cumulative culture.
J Comp Psychol 126:161–169.
24. Derex M, Godelle B, Raymond M (2013) Social learners require process information
to outperform individual learners. Evolution 67:688–697.
25. Wasielewski H (2014) Imitation is necessary for cumulative cultural evolution in an
unfamiliar, opaque task. Hum Nat 25:161–179.
26. Schillinger K, Mesoudi A, Lycett SJ (2015) The impact of imitative versus emulative
learning mechanisms on artifactual variation: Implications for the evolution of
material culture. Evol Hum Behav 36:446–455.
27. Morgan TJ, et al. (2015) Experimental evidence for the co-evolution of hominin tool-
making teaching and language. Nat Commun 6:6029.
28. Roux V, Bril B, Dietrich G (1995) Skills and learning difficulties involved in stone
knapping. World Archaeol 27:63–87.
29. Nonaka T, Bril B, Rein R (2010) How do stone knappers predict and control the
outcome of flaking? Implications for understanding early stone tool technology.
J Hum Evol 59:155–167.
30. Stout D (2002) Skill and cognition in stone tool production: An ethnographic case
study from Irian Jaya. Curr Anthropol 45:693–722.
31. Stout D, Hecht E, Khreisheh N, Bradley B, Chaminade T (2015) Cognitive demands of
lower paleolithic toolmaking. PLoS One 10:e0121804.
32. Toth N, Schick K (2009) The Oldowan: The tool making of early hominins and
chimpanzees compared. Annu Rev Anthropol 38:289–305.
33. Apel J, Knutsson K, eds (2006) Skilled Production and Social Reproduction: Aspects of
Traditional Stone-Tool Technologies. (Societas Archaeologica Upsaliensis, Uppsala,
Sweden).
34. Schillinger K, Mesoudi A, Lycett SJ (2016) Copying error, evolution, and phylogenetic
signal in artifactual traditions: An experimental approach using “model artifacts”.
J Archaeol Sci 70:23–34.
35. Stout D, Semaw S, Rogers MJ, Cauche D (2010) Technological variation in the earliest
Oldowan from Gona, Afar, Ethiopia. J Hum Evol 58:474–491.
36. Stout D, Khreisheh N (2015) Skill learning and human brain evolution: An experi-
mental approach. Camb Archaeol J 25:867–875.
37. Hecht EE, et al. (2015) Acquisition of Paleolithic toolmaking abilities involves structural
remodeling to inferior frontoparietal regions. Brain Struct Funct 220:2315–2331.
38. Magnani M, Rezek Z, Lin SC, Chan A, Dibble HL (2014) Flake variation in relation to
the application of force. J Archaeol Sci 46:37–49.
39. Faisal A, Stout D, Apel J, Bradley B (2010) The manipulative complexity of Lower
Paleolithic stone toolmaking. PLoS One 5:e13718.
40. Putt SS, Woods AD, Franciscus RG (2014) The role of verbal interaction during ex-
perimental bifacial stone tool manufacture. Lithic Technol 39:96–112.
41. Stout D, Apel J, Commander J, Roberts M (2014) Late Acheulean technology and
cognition at Boxgrove, UK. J Archaeol Sci 41:576–590.
42. Cook R, Bird G, Catmur C, Press C, Heyes C (2014) Mirror neurons: From origin to
function. Behav Brain Sci 37:177–192.
43. Laland KN, Bateson P (2001) The mechanisms of imitation. Cybern Syst 32:195–224.
44. Buccino G, et al. (2004) Neural circuits underlying imitation learning of hand actions:
an event-related fMRI study. Neuron 42:323–334.
45. Ericsson KA, Krampe RT, Tesch-Romer C (1993) The role of deliberate practice in the
acquisition of expert performance. Psychol Rev 100:363–406.
46. Stout D (2013) Neuroscience of technology. Cultural Evolution: Society, Technology,
Language, and Religion, Strungmann Forum Reports, eds Richerson PJ, Christiansen M
(MIT Press, Cambridge, MA), pp 157–173.
47. Whiten A (2015) Experimental studies illuminate the cultural transmission of per-
cussive technologies in Homo and Pan. Phil Trans R Soc B 370:20140359; erratum in
Phil Trans R Soc B 371:20150436.
48. Tennie C, Call J, Tomasello M (2009) Ratcheting up the ratchet: On the evolution of
cumulative culture. Philos Trans R Soc Lond B Biol Sci 364:2405–2415.
49. Gibson JJ (1979) The Ecological Approach to Visual Perception (Houghton Mifflin,
Boston).
50. Stout D (2011) Stone toolmaking and the evolution of human culture and cognition.
Philos Trans R Soc Lond B Biol Sci 366:1050–1059.
51. Legare CH, Nielsen M (2015) Imitation and innovation: The dual engines of cultural
learning. Trends Cogn Sci 19:688–699.
52. Fragaszy DM, et al. (2013) The fourth dimension of tool use: Temporally enduring
artefacts aid primates learning to use tools. Philos Trans R Soc B Biol Sci 368:20120410.
53. Hewlett BS, Roulette CJ (2016) Teaching in hunter-gatherer infancy. R Soc Open Sci
3:150403.
54. Burkart JM, Hrdy SB, Van Schaik CP (2009) Cooperative breeding and human cog-
nitive evolution. Evol Anthropol 18:175–186.
55. Stout D, Chaminade T (2012) Stone tools, language and the brain in human evolu-
tion. Philos Trans R Soc Lond B Biol Sci 367:75–87.
56. Laland KN, et al. (2015) The extended evolutionary synthesis: Its structure, assump-
tions and predictions. Proc R Soc Biol 282:20151019.
57. Deacon TW (1997) The Symbolic Species: The Co-Evolution of Language and the
Brain (WW Norton, New York).
58. Edelman GM (1987) Neural Darwinism (Basic Books, New York).
59. Power JD, Fair DA, Schlaggar BL, Petersen SE (2010) The development of human
functional brain networks. Neuron 67:735–748.
60. Mantini D, Corbetta M, Romani GL, Orban GA, Vanduffel W (2013) Evolutionarily
novel functional networks in the human brain? J Neurosci 33:3259–3275.
61. Margulies DS, et al. (2016) Situating the default-mode network along a principal gra-
dient of macroscale cortical organization. Proc Natl Acad Sci USA 113:12574–12579.
62. Buckner RL, Krienen FM (2013) The evolution of distributed association networks in
the human brain. Trends Cogn Sci 17:648–665.
7868 | www.pnas.org/cgi/doi/10.1073/pnas.1620738114
63. Hill J, et al. (2010) Similar patterns of cortical expansion during human development
and evolution. Proc Natl Acad Sci USA 107:13135–13140.
64. Mueller S, et al. (2013) Individual variability in functional connectivity architecture of
the human brain. Neuron 77:586–595.
65. Preuss TM (2012) Human brain evolution: From gene discovery to phenotype dis-
covery. Proc Natl Acad Sci USA 109:10709–10716.
66. Somel M, Liu X, Khaitovich P (2013) Human brain evolution: Transcripts, metabolites
and their regulators. Nat Rev Neurosci 14:112–127.
67. Gómez-Robles A, Hopkins WD, Schapiro SJ, Sherwood CC (2015) Relaxed genetic
control of cortical organization in human brains compared with chimpanzees. Proc
Natl Acad Sci USA 112:14799–14804.
68. Murren CJ, et al. (2015) Constraints on the evolution of phenotypic plasticity: Limits
and costs of phenotype and plasticity. Heredity (Edinb) 115:293–301.
69. Heyes C (2003) Four routes of cognitive evolution. Psychol Rev 110:713–727.
70. Flynn EG, Laland KN, Kendal RL, Kendal JR (2013) Target article with commentaries:
Developmental niche construction. Dev Sci 16:296–313.
71. Wolpert DM, Doya K, Kawato M (2003) A unifying computational framework for
motor control and social interaction. Philos Trans R Soc Lond B Biol Sci 358:593–602.
72. Duncan J (2010) The multiple-demand (MD) system of the primate brain: Mental
programs for intelligent behaviour. Trends Cogn Sci 14:172–179.
73. Feldman R (2017) The neurobiology of human attachments. Trends Cogn Sci 21:
80–99.
74. Hasson U, Frith CD (2016) Mirroring and beyond: Coupled dynamics as a generalized
framework for modelling social interactions. Phil Trans R Soc B 371:20150366.
75. Tramacere A, Pievani T, Ferrari PF (November 16, 2016) Mirror neurons in the tree of
life: Mosaic evolution, plasticity and exaptation of sensorimotor matching responses.
Biol Rev Camb Philos Soc, 10.1111/brv.12310.
76. Heyes CM, Frith CD (2014) The cultural evolution of mind reading. Science 344:
1243091.
77. Byrne R (1999) Imitation without intentionality. Using string parsing to copy the
organization of behaviour. Anim Cogn 2:63–72.
78. Mars RB, et al. (2012) On the relationship between the “default mode network” and
the “social brain”. Front Hum Neurosci 6:189.
79. Genovesio A, Wise SP, Passingham RE (2014) Prefrontal-parietal function: From
foraging to foresight. Trends Cogn Sci 18:72–81.
80. Milner AD, Goodale MA (1995) The Visual Brain in Action (Oxford Univ Press, Ox-
ford).
81. Orban GA, Caruana F (2014) The neural basis of human tool use. Front Psychol 5:310.
82. Stout D, Chaminade T (2007) The evolutionary neuroscience of tool making.
Neuropsychologia 45:1091–1100.
83. Stout D, Passingham R, Frith C, Apel J, Chaminade T (2011) Technology, expertise
and social cognition in human evolution. Eur J Neurosci 33:1328–1338.
84. Hecht EE, et al. (2013) Process versus product in social learning: Comparative diffu-
sion tensor imaging of neural systems for action execution-observation matching in
macaques, chimpanzees, and humans. Cereb Cortex 23:1014–1024.
85. Hecht EE, et al. (2013) Differences in neural activation for object-directed grasping in
chimpanzees and humans. J Neurosci 33:14117–14134.
86. Denys K, et al. (2004) Visual activation in prefrontal cortex is stronger in monkeys
than in humans. J Cogn Neurosci 16:1505–1516.
87. Hecht EE, Gutman DA, Bradley BA, Preuss TM, Stout D (2015) Virtual dissection and
comparative connectivity of the superior longitudinal fasciculus in chimpanzees and
humans. Neuroimage 108:124–137.
88. Duncan J, et al. (2000) A neural basis for general intelligence. Science 289:457–460.
89. Anderson JR, Gallup GG, Jr (2015) Mirror self-recognition: A review and critique of
attempts to promote and engineer self-recognition in primates. Primates 56:317–326.
90. Hecht EE, Mahovetz LM, Preuss TM, Hopkins WD (2017) A neuroanatomical pre-
dictor of mirror self-recognition in chimpanzees. Soc Cogn Affect Neurosci 12:37–48.
91. Hihara S, et al. (2006) Extension of corticocortical afferents into the anterior bank of
the intraparietal sulcus by tool-use training in adult monkeys. Neuropsychologia 44:
2636–2646.
92. Kivell TL (2015) Evidence in hand: Recent discoveries and the early evolution of
human manual manipulation. Phil Trans R Soc Lond B Biol Sci 370:20150105.
93. Stout D, Toth N, Schick K, Chaminade T (2008) Neural correlates of Early Stone Age
toolmaking: Technology, language and cognition in human evolution. Philos Trans R
Soc Lond B Biol Sci 363:1939–1949.
94. Harmand S, et al. (2015) 3.3-million-year-old stone tools from Lomekwi 3, West
Turkana, Kenya. Nature 521:310–315.
95. Rightmire GP (2004) Brain size and encephalization in early to Mid-Pleistocene
Homo. Am J Phys Anthropol 124:109–123.
96. Byrge L, Sporns O, Smith LB (2014) Developmental process emerges from extended
brain-body-behavior networks. Trends Cogn Sci 18:395–403.
97. Gärdenfors P, Högberg A (2017) The archaeology of teaching and the evolution of
Homo docens. Curr Anthrop 58:188–208.
98. Bogin B, Bragg J, Kuzawa C (2014) Humans are not cooperative breeders but practice
biocultural reproduction. Ann Hum Biol 41:368–380.
99. Hill K, Barton M, Hurtado AM (2009) The emergence of human uniqueness: Char-
acters underlying behavioral modernity. Evol Anthropol 18:187–200.
100. Arbib MA (2012) How the Brain Got Language: The Mirror System Hypothesis (Ox-
ford Univ Press, New York).
101. Clark A (2006) Language, embodiment, and the cognitive niche. Trends Cogn Sci 10:
370–374.
102. Dehaene S (1997) The Number Sense: How the Mind Creates Mathematics (Oxford
Univ Press, New York).
103. Marx L (1997) Technology: The emergence of a hazardous concept. Soc Res 64:
965–988.
Stout and Hecht
 COLLOQUIUM
PAPER
Identifying early modern human ecological niche
expansions and associated cultural dynamics
in the South African Middle Stone Age
Francesco d’Erricoa,b,1,2, William E. Banksa,c,1, Dan L. Warrend, Giovanni Sgubine, Karen van Niekerkb,f,
Christopher Henshilwoodb,f, Anne-Laure Daniaue, and María Fernanda Sánchez Goñie,g
a
CNRS, UMR 5199–De la Préhistoire à l’Actuel: Culture, Environnement et Anthropologie, Université de Bordeaux, 33615 Pessac Cedex, France; bEvolutionary
Studies Institute, University of the Witwatersrand, Witwatersrand 2050, South Africa; cBiodiversity Institute, University of Kansas, Lawrence, KS 66045-7562;
d
Biocomplexity and Biodiversity Unit, Okinawa Institute of Science and Technology, Okinawa 904-0495 Japan; eCNRS, UMR 5805–Environnements et
Paléoenvironnements Océaniques et Continentaux, Université de Bordeaux, 33615 Pessac Cedex, France; fInstitute for Archaeology, History, Culture, and
Religion, University of Bergen, 5020 Bergen, Norway; and gÉcole Pratique des Hautes Études, L’Université de Recherche Paris Sciences et Lettres, 75014
Paris, France

Edited by Marcus W. Feldman, Stanford University, Stanford, CA, and approved May 16, 2017 (received for review January 31, 2017)

The archaeological record shows that typically human cultural traits
emerged at different times, in different parts of the world, and
among different hominin taxa. This pattern suggests that their
emergence is the outcome of complex and nonlinear evolutionary
trajectories, influenced by environmental, demographic, and social
factors, that need to be understood and traced at regional scales.
The application of predictive algorithms using archaeological and
paleoenvironmental data allows one to estimate the ecological
niches occupied by past human populations and identify niche
changes through time, thus providing the possibility of investigating
relationships between cultural innovations and possible niche shifts.
By using such methods to examine two key southern Africa
archaeological cultures, the Still Bay [76–71 thousand years before
present (ka)] and the Howiesons Poort (HP; 66–59 ka), we identify a
niche shift characterized by a significant expansion in the breadth of
the HP ecological niche. This expansion is coincident with aridifica-
tion occurring across Marine Isotope Stage 4 (ca. 72–60 ka) and
especially pronounced at 60 ka. We argue that this niche shift was
made possible by the development of a flexible technological system,
reliant on composite tools and cultural transmission strategies based
more on “product copying” rather than “process copying.” These
species’ biologically dictated potential. Although some would still
argue that there is a direct link between cultural behavior and
hominin taxonomy and, as a consequence, that the typically human
secondary inheritance system only emerged with our species,
archaeological and paleogenetic research conducted over the
past 20 y challenges such a view.
First, for periods <200,000 years before the present (ka), it is
difficult to attribute a particular cognition and resulting cultural
behavior to a particular fossil species because paleogenetic evi-
dence shows that significant interbreeding occurred between
Neanderthals, Denisovans, and anatomically modern humans
(AMHs) (4–6), thus blurring the concept of fossil species that
many paleoanthropologists had in the past when interpreting
morphological differences between human remains. Each new
round of publications concerning paleogenetics shows that we are
confronted with a complex network of genetic relationships rather
than distinct and simple lines of evolutionary descent. There is no
reason to assume that such a pattern did not characterize other
phases of our lineage’s evolution.
Second, archaeological discoveries show that the cultural in-
novations generally seen as reflecting modern cognition and be-
havior did not emerge as a single package in conjunction with the

results counter the one niche/one human taxon equation. They
indicate that what makes our cultures, and probably the cultures
of other members of our lineage, unique is their flexibility and
ability to produce innovations that allow a population to shift its
ecological niche.
| |
Middle Stone Age Still Bay Howiesons Poort | archaeological record after ca. 100 ka. Such is the case with
|
ecological niche modeling paleoclimate
ANTHROPOLOGY
appearance of our species in Africa. We know that AMHs
emerged in Africa between 200 and 160 ka (7–9), but some be-
haviors considered as “modern” are present in Africa before this
speciation event. Ochre use appears at around 300 ka (10), and
laminar blade production is observed perhaps as early as 500 ka
(11). Other modern cultural traits are only observed in the African
heating of stone to facilitate knapping or retouching, pressure-
flaked bifacial projectile points, microlithic armatures, mastic-

R esearch on animal behavior has made it clear that culture
represents a second inheritance system that may have changed
the dynamics of evolution on a broad scale (1–3). Understanding
how this process has affected the evolution of our genus is a major
challenge in paleoanthropology. In what ways, and through what
phases of evolutionary history, has human culture extended beyond
culture seen in other species? Are the cultural adaptations and
associated cultural innovations that we observe in the archaeo-
logical record the direct consequence of our biological evolution,
or are they the outcome of mechanisms largely independent of it?
In our lineage, if cultural innovations were directly linked to classic
Darwinian evolutionary processes, such as isolation, random mu-
tation, selection, and speciation, one would expect a clear corre-
spondence between the emergence of a new species and a related
set of novel cultural behaviors. By shaping a new hominin species,
natural selection would provide this species with a new cognitive
setting resulting in the capacity for particular cultural innovations
or behaviors. Such a mechanism would provide the possibility for
cultural variability but would narrow its range of expression to the
ECOLOGY
facilitated hafting of stone tools, formal bone tools, abstract en-
gravings, the production of paint and pigment containers, personal
ornaments, and primary burials (12–15). Furthermore, many key
cultural innovations are present outside Africa well before AMH
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: F.d. and W.E.B. designed research; F.d., W.E.B., D.L.W., and A.-L.D.
performed research; F.d., W.E.B., D.L.W., G.S., K.v.N., and C.H. analyzed data; K.v.N. and
C.H. provided and reviewed archaeological data; A.-L.D. and M.F.S.G. interpreted paleo-
climatic data; and F.d., W.E.B., D.L.W., A.-L.D., and M.F.S.G. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1
F.d. and W.E.B. contributed equally to this work.
2
To whom correspondence should be addressed. Email: francesco.derrico@u-bordeaux.fr.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1620752114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1620752114 PNAS | July 25, 2017 | vol. 114 | no. 30 | 7869–7876

dispersal. In Europe, Neanderthals used pigment at many sites by
at least 250–200 ka. They also used complex lithic technologies,
composite tools, and complex hafting techniques by at least 180 ka
(16). At Bruniquel, France, Neanderthals broke and moved four
tons of stalagmites to build a circular structure deep within a cave
176 ka (17). At a number of sites, starting at 130 ka, they used
raptor claws and feathers, probably for symbolic activities (18, 19).
They made abstract designs on a variety of media (20, 21). Nean-
derthals in the Near East and Europe engaged very early in a va-
riety of funerary practices, including deliberate burials with simple
grave goods. The last Neanderthals in Italy and France produced
formal bone tools. They also produced a variety of personal orna-
ments consisting of animal teeth, fossils, and marine shells, some of
which were colored with ochre (22, 23). Additionally, isolated oc-
currences of innovative cultural traits are recorded at much older
sites in Europe and Asia (24), and well-established innovations
(e.g., Middle Stone Age shell beads) disappear abruptly from
the archaeological record and similar behaviors later reappear in
different forms and sometimes on different media (14, 25).
This evidence demonstrates that typically modern human
cultural traits emerged at different times, in different parts of the
world, and among different hominin taxa. Such taxa appear more
and more to be the phenotypic expression of a largely shared,
plastic cognition (26, 27), and the emergence of typically human
innovations appears to be the result of complex and nonlinear
evolutionary trajectories that need to be understood and traced
at regional scales.
It is clear that cultural innovations were triggered by several
interconnected and dynamic factors, likely biological, environ-
mental, and cultural. Because speciation does not appear to have
played a role in the emergence of key innovations, we need to
explore the potential for relationships between biology and culture
at the population level, and particularly within those past African
populations that first developed behaviors that incorporated suites
of these traits. Such an endeavor, however, is handicapped on the
biological side by a sparse Upper Pleistocene hominin fossil re-
cord, the absence of pre-Holocene paleogenetic data, and a long
history of human presence and intracontinental dispersals that
complicate interpretations of modern genetic data. Understanding
how AMHs were biologically structured in the Middle Stone Age
is also hampered by the fact that, as recently shown by genetic
analyses (6, 28) highlighting the introgression of archaic genes into
the African gene pool, they were certainly not ubiquitous across
the continent. To overcome such limitations, research has focused
on better defining the nature and chronology of the cultural en-
tities that may reflect past population structure and distributions
(29, 30), in addition to documenting the complexity of innovations
recorded in the Middle Stone Age and exploring their social and
cognitive implications (31–33). Others have attempted to identify a
correspondence between environmental or climatic variability and
the emergence of cultural innovations in the hope of identifying
causal links (34–38). These attempts, however, have no designed
means, apart from recurrence, with which to verify the hypothesis
that climate may have influenced culture, to identify the suites of
environmental parameters (i.e., the ecological niche) within which
each archaeological culture operated, or to evaluate how these
relationships varied through space and time. The emergence of
key cultural innovations in our lineage may reflect changes in the
nature of such relationships. Identifying and disentangling such
relationships is a key challenge for the involved disciplines. The
failure to do so may result in oversimplified scenarios. For ex-
ample, Ziegler et al. (38) conclude that cultural innovations during
the Middle Stone Age in southern Africa were triggered by periods
of humidity that produced higher levels of biomass and consequent
increases in human population density. This scenario, however,
only relies on the mean age of each culture and climatic conditions
associated with those means, and it does not take into consider-
ation the full age range of each recognized archaeological culture.
Furthermore, their model insinuates hiatuses in the archaeological
record following the post-Howiesons Poort (HP) that are not seen
in most southern African archaeological sequences.
7870 | www.pnas.org/cgi/doi/10.1073/pnas.1620752114
In a previous study, we stressed the need to consider the re-
lationship between past human cultures and environment as a
dynamic process that occurred at a regional level (39). We ar-
gued that to do so, one needs to develop heuristic tools that
enable the quantitative comparison and evaluation of individual
cultural trajectories, their associated behavioral changes through
time, and the mechanisms that operated behind such trends. This
approach may allow for the identification of points in time
during which human cultures substantially reorganized their
second inheritance systems, thus moving closer to the system
characteristic of historically known and present-day populations.
A regional cultural trajectory can be conceived of as a succes-
sion of cultural packages, which we term cohesive adaptive sys-
tems. A cohesive adaptive system is a cultural entity characterized
by shared and transmitted knowledge reflected by a recognizable
suite of cultural traits that a population uses to operate within both
cultural and environmental contexts (39). This concept differs from
the concept of “technocomplex” or “archaeological culture” com-
monly used in archaeology, in that exploited environmental condi-
tions (i.e., the ecocultural niche) contribute to the definition of a past
cultural adaption. When faced with successive climate changes, a
cohesive adaptive system can conserve, expand, or contract its eco-
logical niche, with “ecological niche” being defined in the Grinnel-
lian sense as the environmental and resources conditions suitable for
a species or population (40). Associated cultural traits, and the way
in which they were transmitted, may also evolve in such situations
and highlight significant changes in the way in which culture influ-
enced human populations. Research strategies have been developed
to investigate such interactions.
Predictive algorithms, originally created in the field of ecology,
are able to estimate the ecological niche occupied by a past cohesive
adaptive system (i.e., the ecocultural niche) by using the geographic
locations of archaeological sites where the cohesive adaptive system
has been recognized along with chronologically relevant paleo-
environmental data. Using these data, the predictive algorithms first
identify the environmental parameters shared among the archaeo-
logical sites and define the relationships between these parameters.
These relationships are then used to estimate a cohesive adaptive
system’s ecological niche. Another important capacity of these al-
gorithms is that they can be used to examine niches between time
periods, thereby allowing one to determine whether or not suc-
cessive populations exploited different niches. By comparing the
material cultures of two or more successive cohesive adaptive sys-
tems, and taking into account environmental frameworks within
which they operated, one can evaluate whether or not cultural in-
novations were a response to environmental fluctuations. Equally as
important, one can identify the degree of resilience of a cohesive
adaptive system to environmental change.
The goal of this study is to apply this approach to two key
Middle Stone Age archaeological cultures, the Still Bay (SB) and
the HP of southern Africa. The SB represents the first known
cultural adaptation in which technological and symbolic inno-
vations of a complexity comparable to the innovations seen in
modern hunter-gatherers appears as a coherent and recognizable
package. After a possible hiatus, we observe a different archae-
ological culture, termed the HP, characterized by dramatically
different and simplified lithic technology, as well as by markedly
different symbolic material culture. The available archaeological
and paleoenvironmental datasets of this period are of sufficient
resolution to make this period of the Middle Stone Age an ideal
laboratory for exploring how typically human behavioral pack-
ages arose and evolved in one particular region and for identi-
fying potential mechanisms at work.
Cultural and Chronological Contexts
The SB. This archaeological culture, observed at sites located in
coastal areas of southern Africa and predominantly concentrated
in southwestern regions (Fig. 1), is characterized by the production
of bifacial foliate points, often made from fine-grained, nonlocal
lithic materials (Fig. 2A). At the key site of Blombos Cave, the
majority of these points have been heat-treated before flaking with
d’Errico et al.
 COLLOQUIUM
PAPER
Fig. 1. Map of southern Africa indicating the loca-
tions of the SB (red circles) and HP (green triangles)
archaeological sites, the geographical coordinates of
which were used as occurrence inputs to estimate the
two cultures’ respective ecological niches. Sea level is
depicted at −70 m below present-day sea level.

hard and soft hammer percussion, and finished using a technique
termed pressure flaking. The latter allows for more refined shaping
of the object by giving the knapper better control over its final
form. Modern-day experiments indicate that this knapping tech-
nology requires a long period of apprenticeship. SB bifaces were
multifunctional and served as both projectiles and cutting tools.
Examinations of SB lithic assemblages (41) show that these bifaces
were often repeatedly resharpened and had long use-lives, in-
dicating that they formed a curated component of the SB lithic
toolkit. The SB is also the first archaeological culture in which
formal bone tools (i.e., artifacts made of animal osseous material
shaped with techniques, such as scraping, grinding, and incising,
specifically conceived for these materials) are observed at multiple
sites rather than as rare elements in single assemblages. Techno-
logical and functional studies show that the two different classes of
tool, projectiles and awls (Fig. 2C), were produced with different
techniques and that special attention was paid to the finishing of
the bone projectile points, suggesting that they were highly valued
Use-wear analyses indicate that they were worn for extended
periods of time (42). Other elements of SB symbolic material
culture include elaborately engraved abstract patterns on ochre
pieces (Fig. 2D), as well as more simple engravings on bone items.
Also present in assemblages are ochre pieces bearing traces indi-
cating that they were processed to produce red powder (Fig. 2E),
which likely was used for both functional and symbolic purposes.
With respect to chronology, a majority of SB sites have yielded
optically stimulated luminescence (OSL) ages that range be-
tween 76 ka and 71 ka (34, 43–45). Debate exists as to accuracy of
this range due to older OSL and thermoluminescence (TL) dates
from the Diepkloof rock shelter (45–48). Because the inexpli-
cably older set of dates from Diepkloof remains a unicum, we will
use the currently accepted chronology (45, 49, 50). Debate also
exists as to whether this culture is technologically homogeneous
or, instead, characterized by regional and temporal variability
(41). This issue, however, remains open due to a lack of chro-
nological resolution and the small number of contextually reliable

and possible status items. The SB is also the first archaeological
culture in southern Africa associated with personal ornaments.
These ornaments take the form of marine shells (Nassarius
kraussianus) that were deliberately perforated, stained with ochre,
and strung together in a variety of arrangements (33) (Fig. 2B).
ANTHROPOLOGY
archaeological assemblages.
The HP. This archaeological culture, observed in both coastal and
inland regions of southwestern and northeastern South Africa
(Fig. 1), is principally characterized by the presence of backed

ECOLOGY

Fig. 2. (Left) SB artifacts [bifacial points made of quartz and silcrete (A), perforated N. kraussianus shell beads (B), bone points and an awl (C), engraved
ochre fragments (D), and an ochre fragment shaped by grinding (E)]. (Right) HP artifacts [segment made of hornfels (F), segments made of quartz (G), flake
and segments bearing residues of mastic (H), engraved ostrich egg shells (I), ochre fragments shaped by grinding (J), and bone point and awls (K)]. Blombos
Cave (A and B), Sibudu Cave (F, G, and K), and Diepkloof Shelter (H–J) are shown. (Scale bar: 1 cm.) Images courtesy of: (A) ref. 41, (C) ref. 98, (D) ref. 12, (F) ref.
53, (G) ref. 99, (H) ref. 59, (I) ref. 61, (J) ref. 100, and (K) ref. 55. Fig. 2B courtesy of F.D. and C.H., and Fig. 2E courtesy of C.H.

d’Errico et al. PNAS | July 25, 2017 | vol. 114 | no. 30 | 7871
blades and bladelets (i.e., lithic blades steeply retouched on one
side to form crescent-shaped segments) (Fig. 2 F and G) that
were predominantly used as components in composite hunting
weapons. These tools, although not highly standardized dimen-
sionally or morphologically, were made with a lithic reduction
system that was geared toward the production of thin, straight
blades, some of which were retouched to make this culture’s fossil
directeur along with denticulated tools (29, 41, 51). Raw materials
used for the lithic technology were predominantly local or near-
local in origin, in clear contrast to what is seen for SB bifaces.
Similar to the SB, however, HP groups also sometimes heated
lithic raw materials before they were reduced to produce blades
(52) and occasionally used pressure flaking (53). Bifacial points are
absent in the HP, with the exception of a single site where speci-
mens that are smaller and of lower quality have been recovered
(54). Bone tools recovered from HP sites consist of awls, pressure
flakers, shaped splintered pieces (pièces esquillées), and small
projectile points (55) (Fig. 2K). It has been argued that HP backed
segments and bone points were used as bow-delivered arrow points
based on use-wear, fracture patterns, and morphometrics (56–58).
The interpretation that these tools were hafted is supported by the
presence of mastic remnants observed on some backed pieces (31,
59) (Fig. 2H). At present, with the exception of a perforated conus
shell found within an infant burial at Border Cave (60), personal
ornaments are lacking in HP assemblages, and undisputed sym-
bolic behavior is limited to the decoration of ostrich egg shell water
containers with a variety of abstract designs made up of linear
engravings (51, 61) (Fig. 2I). Red ochre (Fig. 2J), also sometimes
incorporated into mastic mixtures, was widely used by HP groups.
The HP has predominantly been dated with OSL and TL
techniques and appears to have lasted for a slightly longer period
than the SB. HP dates range between roughly 66 ka and 59 ka
(34, 51, 62). As with the SB, some OSL dates of the HP at
Diepkloof are significantly older (47, 48) than the corpus of
dates available from other South African sites, as well as from
other OSL dates obtained at the same site (63). Based on the fact
that the newly recalculated dates for the Diepkloof HP (63)
cluster with the HP dates from other dated contexts (50), we will
use the 66–59 ka range as the chronological interval for the HP
in this study. Shortly after ca. 59 ka, we observe the appearance
of the post-HP archaeological culture.
Paleoenvironmental Context. These two archaeological cultures oc-
curred during two very different climatic phases (Fig. 3). At the
orbital scale, the SB occurs in a phase of precession maximum
during which one observes higher seasonality and an increase in
precipitation in the Southern Hemisphere (64–67). To the contrary,
the HP is contemporaneous with a decrease in precession, with the
minimum reached toward its end (ca. 60–59 ka). This change
resulted in lower seasonality and drier conditions (SI Appendix, Fig.
S1). In addition to orbital climatic variability, SB and HP cultures
were subjected to suborbital climatic fluctuations, the so-called
Dansgaard–Oeschger (D-O) cycles expressed over Greenland by
alternating cold stadials and temperate interstadials, as well as in-
termittent and extreme cooling episodes recorded in the North
Atlantic, termed Heinrich Stadials (HSs). These millennial-scale
events are also recorded in Antarctic paleoclimatic records.
The SB occurs during a period comprising Greenland In-
terstadial (GI) 20, Greenland Stadial (GS) 20, and GI 19 (68)
(Fig. 3). This culture disappears from the archaeological record
during the initial phase of GS 19 (GS 19.2). The HP appears
toward the end of GS 19 and is present across GI 18 and GS 18
(ca. 64.4–59.4 ka, which corresponds to HS 6) (69). The suite of
diagnostic elements characteristic of this archaeological culture
is no longer present by ca. 59–58 ka, a period marked by rapid
climatic oscillations (i.e., GI 17.1, GS 17.1, GI 16.2, GS 16.2). It
is following this interval that the post-HP adaptation appears.
The impact of the D-O millennial scale climatic variability and
HSs on the Southern Hemisphere regional climates has recently
been investigated. Model experiments and climate reconstructions
suggest that GS and HS events resulted in increased sea surface
7872 | www.pnas.org/cgi/doi/10.1073/pnas.1620752114
temperatures and humidity in the South Atlantic and Southwestern
Indian Ocean (70–74). For southern Africa, Ziegler et al. (38)
examined the elemental composition of marine sediments from an
Indian Ocean core and proposed that GS and HS events are
characterized by increased erosion reflecting higher precipitation
that triggered increases in vegetation cover and biomass. Recent
research has provided direct data concerning vegetation cover and
biomass for this region. Pollen and microcharcoal records from
marine core MD96-2098, retrieved off southwestern Africa (refs.
65, 67 and this study), show repeated millennial-scale changes in
humidity during the last glacial period that also indicate, within the
uncertainties of the independent ice and marine chronologies, that
GS and HS events were associated with increases in humidity. Such
increases are inferred from peaks in microcharcoal concentration
due to grass-fueled fires and decreases in pollen from vegetation
characteristic of open environments, such as Nama Karoo and fine-
leaved savanna (Fig. 3 D and E). However, when the entire chro-
nological interval for both the SB and HP is taken into account, a
more complex climatic pattern is observed, characterized by an
alternation of wet and dry events. Despite this variability, the gen-
eral pattern revealed by all available continental proxies across the
entire range of each archaeological culture shows an overall trend
toward higher humidity during the SB and generally dryer condi-
tions during the HP. The contradictory pattern proposed by Ziegler
et al. (38) is probably due to the fact that they do not consider the
entire range of these two cultures but, instead, only look at the
humidity trends coincident with each culture’s mean age.
Materials and Methods
Paleoclimate Modeling. To estimate ecological niches exploited by the SB and
HP, we used paleoclimatic and vegetation simulations produced by Woillez
et al. (66) (SI Appendix, Paleoclimatic Simulations) for the periods of 72 ka and
60 ka. Because the two simulations are primarily constrained by orbital pa-
rameters and do not estimate suborbital variability, we used the 72 ka simu-
lation to represent climatic and environmental conditions for the SB and the
initial HP (ca. 66–63 ka) and the 60 ka simulation to represent conditions for
the terminal HP (ca. 63–59 ka). The use of the 72 ka simulation as a proxy for
climatic conditions of the initial HP is justified by the relatively high humidity
observed at the onset of HS 6, as evidenced by vegetation, fire activity, and
erosion proxies (Fig. 3 C–E, respectively). To estimate the SB and HP ecocultural
niches, we used temperature of the coldest month, maximum precipitation,
minimum precipitation, mean annual precipitation, mean annual tempera-
ture, and a measure of biomass from the relevant paleoclimatic simulations.
Ecological Niche Modeling and Hypothesis Testing. To reconstruct the potential
ecological (ecocultural) niches exploited by the SB and HP and evaluate whether
cultural changes between the two are associated with an ecological niche shift,
we constructed a georeferenced list of archaeological sites with levels that can be
securely attributed to one of these cultures (Fig. 1 and SI Appendix, Table S1).
We then used these occurrence data to conduct tests using both Bioclim (75)
and Maxent (76) predictive algorithms within the “dismo” R package (77, 78) (SI
Appendix, Ecological Niche Modeling). We use these two algorithms to explore
the differences seen when models are allowed to extrapolate freely into
combinations of environments that were unavailable during model training
(Maxent) versus models that are constrained so that they do not extrapolate
beyond the minima and maxima of the marginal environmental distributions of
the examined population (Bioclim). Due to Maxent’s ability to extrapolate, we
anticipate that similarity between different target populations will generally be
seen to be higher when environmental niches are modeled using Maxent as
opposed to Bioclim. With these two algorithms, we reconstructed both SB and
HP niches using relevant climatic outputs and simulated biomass from the 72 ka
simulation and compared these results. We also reconstructed the HP niche
using simulation outputs for 60 ka and compared these estimations with the
estimations of the SB at 72 ka. A series of Monte Carlo randomization tests was
conducted to assess the differences in the set of environments occupied by each
culture. This approach is based on widely used methods in evolutionary ecology
(the “background” or “similarity” test) (79, 80) that are used to assess whether
two populations exhibit statistically significant differences in their environ-
mental tolerances or associations (SI Appendix, Ecological Niche Modeling). We
also conducted tests using measures of niche breadth (81, 82) to determine
whether any observed differences between the two cultures’ environmental
niches represent a statistically significant expansion of the niche. Because some
of these evaluations were conducted using different climate layers for the SB
d’Errico et al.

and HP (72 ka and 60 ka, respectively), modifications that use Latin hypercube
sampling were made to the background similarity tests (SI Appendix, Ecological
Niche Modeling and Fig. S2).
Results
Niche estimations for the SB at 72 ka produced with Bioclim and
Maxent both indicate a high probability of presence primarily
restricted to the extreme southern and eastern portions of present-
day South Africa (Fig. 4 A and B). The most noticeable differences
are that the Maxent prediction includes areas in the southwestern
Cape as well as immediately coastal regions along the southeast-
ern and eastern coasts. This broader Maxent prediction is due to
this algorithm’s propensity to extrapolate into environments not
directly associated with the input occurrence data (i.e., archaeo-
logical sites). The predicted niches for the HP at 66 ka, produced
with the proxy 72 ka outputs, include those regions predicted for
the SB as well as more inland areas, including the Great Es-
carpment, the Highveld, and the Kaap Plateau, and broader areas
within the southwestern Cape and western coastal regions (Fig. 4
C and D). The niche estimations for the HP at 60 ka remain
geographically broader than the niche estimations for the SB and
still include major inland plateaus but are visibly shifted toward
d’Errico et al.
COLLOQUIUM
PAPER
Fig. 3. Climate variability during the time interval
between 90 ka and 40 ka encompassing the Middle
Stone Age cultures SB (76–71 ka, blue rectangle) and
HP (66–59 ka, red rectangle). Precession index (101)
(A), North Greenland Ice Core Project δ18O curve on
the GICC05 chronology (68) (B), Fe/Ca curve from
core CD154-17-17K collected from the Eastern Cape
margin indicating changes in river discharge (38) (C),
microcharcoal particle concentration curve from core
MD96-2098 collected off the Orange River on the
western South African margin indicating changes in
fire regime and precipitation (65) (D and this study),
Nama Karoo and fine-leaved savannah pollen per-
ANTHROPOLOGY
centage record from core MD96-2098 indicating
changes in precipitation (67) (E), and temperature
curve for Antarctica from the European Project for Ice
Coring in Antarctica ice core (102) (F). Arrows situated
between curves in C and D indicate long-term trends
in humidity during the SB and HP intervals.
the east and northeast (Fig. 4 E and F), which represent areas that
ECOLOGY
were less affected by the eastward expansion of desert areas
during Marine Isotope Stage (MIS) 4 (66).
Background similarity tests of overlap between the SB and HP
niches both modeled with Maxent using the 72 ka climatic data
produced no statistically significant result (SI Appendix, Fig. S3A
and Table S2), meaning that their respective niches are not sta-
tistically different from one another. As pointed out above, this
lack of significant difference between predictions is likely the re-
sult (Materials and Methods) of the used algorithm. To the con-
trary, these same tests using Bioclim found instead that SB and HP
niche estimations using 72 ka climate outputs were less similar
than expected by chance (I-statistic: P ∼ 0.022; SI Appendix, Fig.
S3C and Table S2). Although HP niche estimates are slightly
broader than niche estimates of the SB at 72 ka with both Maxent
and Bioclim, these differences are not statistically significant (SI
Appendix, Fig. S3 B and D and Table S2). Niche overlap between
Maxent models for the SB at 72 ka and the HP at 60 ka was
neither greater nor less than expected by chance (SI Appendix, Fig.
S3E and Table S2). However, overlap of Bioclim predictions for
the SB at 72 ka and the HP at 60 ka was significantly lower than
would be expected by chance (I-statistic: P ∼ 0.013; SI Appendix,
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7873
Fig. 4. Ecological niche predictions for the SB archaeological culture at 72 ka (A and B), the HP archaeological culture at 66 ka (C and D), and the HP ar-
chaeological culture at 60 ka (E and F) produced with Bioclim and Maxent, respectively.
Fig. S3G and Table S2), indicating that the two cultures occupied
different ecological niches. Change in niche breadth between
Maxent predictions for the SB at 72 ka and the HP at 60 ka is not
statistically different from random expectations, although the ap-
proximate P value is fairly low (P ∼ 0.11) (SI Appendix, Fig. S3F
and Table S2), suggesting that a greater sample size might es-
tablish the HP niche at 60 ka as significantly broader than the
niche SB at 72 ka. The difference in niche breadth for Bioclim
models is greater than expected by chance (P ∼ 0.027) (SI Ap-
pendix, Fig. S3H and Table S2), indicating that the HP 60 ka niche
is broader than the niche of the SB at 72 ka, and points to an
ecological niche expansion.
Discussion and Conclusions
To what extent does this study allow us to understand how human
culture extended beyond behavioral adaptations observed in other
species? Most species exhibit niche conservatism, contraction, or,
more rarely, extinction when faced with climate change (83–85).
Human populations, however, are unique in their capacity for
cumulative culture and associated complex cultural transmission
7874 | www.pnas.org/cgi/doi/10.1073/pnas.1620752114
strategies that potentially allow them to adapt to climate change
and environmental reorganization via cultural means. We observe
such a pattern between the SB and the HP of Southern Africa.
The SB was a coastal adaptation that exploited a relatively narrow
niche during mild climatic conditions across a large region. To
exploit that niche, SB populations developed a variety of complex
technologies and symbolic practices, some of which certainly
entailed costly modes of cultural transmission. A number of SB
cultural features, such as bifacial points and complex bead-
working, could only be transmitted by communication and learn-
ing strategies that emphasize imitation (high-fidelity copying) over
emulation (low-fidelity copying) (86, 87). HP populations signifi-
cantly increased the breadth of their niche compared with SB
populations. This expansion incorporated more arid and high-
altitude inland environments and demonstrates their ability to
cope successfully with the more arid climatic conditions and higher
ecological risk associated with MIS 4, particularly its latter phase.
This shift was made possible by developing a cohesive adaptive
system reliant on more flexible technologies. The variety of used
lithic raw materials, blank production techniques, and methods to
d’Errico et al.

retouch and shape those blanks to produce segments, which vary in
form and size, are indicative of a flexible toolkit, and one reliant
on composite tools in this case. With effective hafting techniques,
such a toolkit would have been easily repaired and maintained.
Due to its modular nature, the HP toolkit could be effectively used
in diverse environments. More importantly, the communicative
strategies needed to transmit the knowledge necessary to perpet-
uate this technology can be based more on “product copying”
(emulation) rather than “process copying” (imitation). In the
latter, morphological similarity is associated with the same, or very
similar, manufacturing techniques and sequences. For the former,
one would expect to see artifacts that are morphologically similar
despite being made from a variety of raw materials and tech-
niques, as is observed in the HP. Such patterns could have been
the result of a collapse of previously existing long-distance cultural
networks, leading to the formation of more local “traditions,”
again, which is exactly what we observe in HP bone and lithic
technologies (53, 55, 88). The mechanism or mechanisms that
operated behind such a process remain unclear (e.g., demographic
changes, population replacement, cultural drift). Although the SB
and HP certainly had adaptive strategies in common, it is probable
that their cultural transmission strategies differed. Considering the
niche and technological changes observed between the two cul-
tures, along with the expertise implicit in some SB technological
innovations, we propose that training to create specialists, or “selec-
tive oblique transmission” (89), was used in the SB to convey these
complex technologies effectively and that this strategy was not, or to a
greatly lesser degree, used at the HP.
Numerous studies support the hypothesis that hunter-gatherer
toolkit structure is driven, in part, by the risk of resource failure
(i.e., more diverse and complex toolkits are associated with riskier
environments) (90–92). Data do not always support this pre-
diction, however, and it has been proposed that the impact of risk
on toolkits is dependent on the scale of risk differences among the
studied populations (93). The degree of reliance on copying (94),
population size (95), and mobility (96) are other factors that may
condition toolkit structure. None of these studies, however, are
able to routinely predict what factors were implicated in shifts in
toolkit structure among early AMHs or to address the issue of how
past human niches may have changed when shifts in technology
were concomitant with major climatic changes. The approach
that we have applied here is an effective means with which to explore
relationships between climate variability and cohesive adaptive
1. Whiten A, Ayala FJ, Feldman MW, Laland KN (2017) The extension of biology
through culture. Proc Natl Acad Sci USA 114:7775–7781.
2. Whiten A (2017) Culture extends the scope of evolutionary biology in the great apes.
Proc Natl Acad Sci USA 114:7790–7797.
3. Street SE, Navarrete AF, Reader SM, Laland KN (2017) Coevolution of cultural in-
telligence, extended life history, sociality, and brain size in primates. Proc Natl Acad
Sci USA 114:7908–7914.
4. Sankararaman S, et al. (2014) The genomic landscape of Neanderthal ancestry in
present-day humans. Nature 507:354–357.
5. Sankararaman S, Mallick S, Patterson N, Reich D (2016) The Combined landscape of
Denisovan and Neanderthal ancestry in present-day humans. Curr Biol 26:1241–1247.
6. Nielsen R, et al. (2017) Tracing the peopling of the world through genomics. Nature 541:
302–310.
7. White TD, et al. (2003) Pleistocene Homo sapiens from Middle Awash, Ethiopia.
Nature 423:742–747.
8. McDougall I, Brown FH, Fleagle JG (2005) Stratigraphic placement and age of
modern humans from Kibish, Ethiopia. Nature 433:733–736.
9. Gronau I, Hubisz MJ, Gulko B, Danko CG, Siepel A (2011) Bayesian inference of ancient
human demography from individual genome sequences. Nat Genet 43:1031–1034.
10. Watts I, Chazan M, Wilkins J (2016) Early evidence for brilliant ritualized display:
Specularite use in the Northern Cape (South Africa) between ∼500 and ∼300 ka. Curr
Anthropol 57:287–310.
11. Wilkins J, Chazan M (2012) Blade production ∼500 thousand years ago at Kathu Pan
1, South Africa: Support for a multiple origins hypothesis for early Middle Pleisto-
cene blade technologies. J Archaeol Sci 39:1883–1900.
12. Henshilwood CS, d’Errico F, Watts I (2009) Engraved ochres from the Middle Stone
Age levels at Blombos Cave, South Africa. J Hum Evol 57:27–47.
13. Mourre V, Villa P, Henshilwood CS (2010) Early use of pressure flaking on lithic ar-
tifacts at Blombos Cave, South Africa. Science 330:659–662.
14. d’Errico F, Stringer CB (2011) Evolution, revolution or saltation scenario for the
emergence of modern cultures? Philos Trans R Soc Lond B Biol Sci 366:1060–1069.
d’Errico et al.
COLLOQUIUM
PAPER
systems at key moments in our evolutionary history. Its appli-
cation to other regions and periods should allow us to follow, at
regional scales, the complex interplay between cultural in-
novation, changes in modes of cultural transmission, and en-
vironmental variability. The results of the present study may be
improved in the future by producing paleoclimatic simulations
that capture millennial-scale environmental variability and by
developing and using methods (e.g., date estimations, Bayesian
age modeling) that would allow one to attribute archaeological
site levels more precisely to millennial-scale climatic phases. Al-
though the former is technically possible, using such models will
not be productive as long as the latter remains beyond our grasp,
at least at present. By capturing the main climatic trends charac-
teristic of the end of MIS 5 and MIS 4, our paleoclimatic simu-
lations appear appropriate for examining culture–environment
relationships when one considers the degree of chronological
uncertainty associated with the two targeted cultures.
Our results demonstrate that in some early AMH regional
cultural trajectories, niche expansion was not always associated
with cultural complexification (an opposite case is discussed in
ref. 97). In this study’s case, complex cultural behaviors and
inferred transmission strategies were replaced during a period of
pronounced aridification with more flexible adaptations that
were used to exploit a broader ecological niche. Increased cul-
tural complexity and elaborated social learning strategies ap-
parently were not always necessary for a culture to expand its
ecological niche. Our findings support the view that the path
followed by past human populations to produce adaptations and
cultural traits, which most researchers would qualify as typically
human, is not the outcome of classic Darwinian evolutionary
processes in which the appearance of a new niche is often as-
sociated with a new species. Rather, the innovations character-
istic of the HP represent cultural exaptation: innovations that use
existing skills, techniques, and ideas in new ways. The consoli-
dation of these innovations depends on a population’s ability to
develop, when necessary, new modes of cultural transmission
that allow such innovations to be maintained through time.
ACKNOWLEDGMENTS. This research was conducted with the financial
support of the Agence Nationale de la Recherche ANR-10-LABX-52 and the
ANTHROPOLOGY
European Research Council’s Advanced Grant TRACSYMBOLS 249587 awarded
under the Seventh Framework Programme.
15. Henshilwood CS, et al. (2011) A 100,000-year-old ochre-processing workshop at
Blombos Cave, South Africa. Science 334:219–222.
16. Mazza PPA, et al. (2006) A new Palaeolithic discovery: Tar-hafted stone tools in a
European Mid-Pleistocene bone-bearing bed. J Archaeol Sci 33:1310–1318.
17. Jaubert J, et al. (2016) Early Neanderthal constructions deep in Bruniquel Cave in
southwestern France. Nature 534:111–114.
ECOLOGY
18. Romandini M, et al. (2014) Convergent evidence of eagle talons used by late
Neanderthals in Europe: A further assessment on symbolism. PLoS One 9:e101278.
19. Radovčic D, Sršen AO, Radovčic J, Frayer DW (2015) Evidence for Neandertal jewelry:
Modified white-tailed eagle claws at Krapina. PLoS One 10:e0119802.
20. Soressi M, D’Errico F (2007) Pigments, gravures, parures: Les comportements sym-
boliques controversés des Néandertaliens. Les Néandertaliens. Biologie et Cultures,
Documents préhistoriques 23, eds Vandermeersch B, Maureille B (Éditions du CTHS,
Paris), pp 297–309. French.
21. Rodríguez-Vidal J, et al. (2014) A rock engraving made by Neanderthals in Gibraltar.
Proc Natl Acad Sci USA 111:13301–13306.
22. Zilhão J, et al. (2010) Symbolic use of marine shells and mineral pigments by Iberian
Neandertals. Proc Natl Acad Sci USA 107:1023–1028.
23. Caron F, d’Errico F, Del Moral P, Santos F, Zilhão J (2011) The reality of Neandertal
symbolic behavior at the Grotte du Renne, Arcy-sur-Cure, France. PLoS One 6:e21545.
24. Joordens JCA, et al. (2015) Homo erectus at Trinil on Java used shells for tool pro-
duction and engraving. Nature 518:228–231.
25. d’Errico F, et al. (2009) Out of Africa: Modern human origins special feature: Addi-
tional evidence on the use of personal ornaments in the Middle Paleolithic of North
Africa. Proc Natl Acad Sci USA 106:16051–16056.
26. Dehaene S, Cohen L (2007) Cultural recycling of cortical maps. Neuron 56:384–398.
27. Stout D, Hecht EE (2017) Evolutionary neuroscience of cumulative culture. Proc Natl
Acad Sci USA 114:7861–7868.
28. Racimo F, Sankararaman S, Nielsen R, Huerta-Sánchez E (2015) Evidence for archaic
adaptive introgression in humans. Nat Rev Genet 16:359–371.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7875
29. Henshilwood CS (2012) Late Pleistocene techno-traditions in Southern Africa: A re-
view of the Still Bay and Howiesons Poort, c. 75–59 ka. J World Prehist 25:205–237.
30. Scerri EML, Groucutt HS, Jennings RP, Petraglia MD (2014) Unexpected technological
heterogeneity in northern Arabia indicates complex Late Pleistocene demography at
the gateway to Asia. J Hum Evol 75:125–142.
31. Wadley L, Hodgskiss T, Grant M (2009) From the cover: Implications for complex
cognition from the hafting of tools with compound adhesives in the Middle Stone
Age, South Africa. Proc Natl Acad Sci USA 106:9590–9594.
32. Wadley L, et al. (2011) Middle Stone Age bedding construction and settlement
patterns at Sibudu, South Africa. Science 334:1388–1391.
33. Vanhaeren M, d’Errico F, van Niekerk KL, Henshilwood CS, Erasmus RM (2013)
Thinking strings: Additional evidence for personal ornament use in the Middle Stone
Age at Blombos Cave, South Africa. J Hum Evol 64:500–517.
34. Jacobs Z, et al. (2008) Ages for the Middle Stone Age of southern Africa: Implications
for human behavior and dispersal. Science 322:733–735.
35. Osborne AH, et al. (2008) A humid corridor across the Sahara for the migration of early
modern humans out of Africa 120,000 years ago. Proc Natl Acad Sci USA 105:16444–16447.
36. Armitage SJ, et al. (2011) The southern route “out of Africa”: Evidence for an early
expansion of modern humans into Arabia. Science 331:453–456.
37. Compton JS (2011) Pleistocene sea-level fluctuations and human evolution on the
southern coastal plain of South Africa. Quat Sci Rev 30:506–527.
38. Ziegler M, et al. (2013) Development of Middle Stone Age innovation linked to rapid
climate change. Nat Commun 4:1905.
39. d’Errico F, Banks WE (2013) Identifying mechanisms behind Middle Paleolithic and
Middle Stone Age cultural trajectories. Curr Anthropol 54:S371–S387.
40. Peterson AT, et al. (2011) Ecological Niches and Geographic Distributions (Princeton
Univ Press, Princeton).
41. Soriano S, et al. (2015) The Still Bay and Howiesons Poort at Sibudu and Blombos:
Understanding Middle Stone Age technologies. PLoS One 10:e0131127.
42. d’Errico F, Henshilwood C, Vanhaeren M, van Niekerk K (2005) Nassarius kraussianus
shell beads from Blombos Cave: Evidence for symbolic behaviour in the Middle Stone
Age. J Hum Evol 48:3–24.
43. Jacobs Z, Duller GAT, Wintle AG (2003) Optical dating of dune sand from Blombos
Cave, South Africa: II–single grain data. J Hum Evol 44:613–625.
44. Jacobs Z, Duller GAT, Wintle AG, Henshilwood CS (2006) Extending the chronology
of deposits at Blombos Cave, South Africa, back to 140 ka using optical dating of
single and multiple grains of quartz. J Hum Evol 51:255–273.
45. Jacobs Z, Hayes EH, Roberts RG, Galbraith RF, Henshilwood CS (2013) An improved
OSL chronology for the Still Bay layers at Blombos Cave, South Africa: Further tests
of single-grain dating procedures and a re-evaluation of the timing of the Still Bay
industry across southern Africa. J Archaeol Sci 40:579–594.
46. Tribolo C, et al. (2009) Thermoluminescence dating of a Stillbay–Howiesons Poort se-
quence at Diepkloof Rock Shelter (Western Cape, South Africa). J Archaeol Sci 36:730–739.
47. Guérin G, Murray AS, Jain M, Thomsen KJ, Mercier N (2013) How confident are we in the
chronology of the transition between Howieson’s Poort and Still Bay? J Hum Evol 64:314–317.
48. Tribolo C, et al. (2013) OSL and TL dating of the Middle Stone Age sequence at
Diepkloof Rock Shelter (South Africa): A clarification. J Archaeol Sci 40:3401–3411.
49. Archer W, Pop CM, Gunz P, McPherron SP (2016) What is Still Bay? Human bio-
geography and bifacial point variability. J Hum Evol 97:58–72.
50. Jacobs Z, Roberts RG (2017) Single-grain OSL chronologies for the Still Bay and
Howieson’s Poort industries and the transition between them: Further analyses and
statistical modelling. J Hum Evol 107:1–13.
51. Henshilwood CS, et al. (2014) Klipdrift Shelter, southern Cape, South Africa: Pre-
liminary report on the Howiesons Poort layers. J Archaeol Sci 45:284–303.
52. Delagnes A, et al. (2016) Early evidence for the extensive heat treatment of silcrete in the
Howiesons Poort at Klipdrift Shelter (Layer PBD, 65 ka), South Africa. PLoS One 11:e0163874.
53. de la Peña P (2015) Refining our understanding of Howiesons Poort lithic technology:
The evidence from Grey Rocky Layer in Sibudu Cave (KwaZulu-Natal, South Africa).
PLoS One 10:e0143451.
54. de la Peña P, Wadley L, Lombard M (2013) Quartz bifacial points in the Howiesons
Poort of Sibudu. S Afr Archaeol Bull 68:119–136.
55. d’Errico F, Backwell LR, Wadley L (2012) Identifying regional variability in Middle
Stone Age bone technology: The case of Sibudu Cave. J Archaeol Sci 39:2479–2495.
56. Backwell L, d’Errico F, Wadley L (2008) Middle Stone Age bone tools from the
Howiesons Poort layers, Sibudu Cave, South Africa. J Archaeol Sci 35:1566–1580.
57. Lombard M, Phillipson L (2010) Indications of bow and stone-tipped arrow use
64,000 years ago in KwaZulu-Natal, South Africa. Antiquity 84:635–648.
58. Bradfield J, Lombard M (2011) A macrofracture study of bone points used in experimental
hunting with reference to the South African middle stone age. S Afr Archaeol Bull 66:67.
59. Charrié-Duhaut A, et al. (2013) First molecular identification of a hafting adhesive in
the Late Howiesons Poort at Diepkloof Rock Shelter (Western Cape, South Africa).
J Archaeol Sci 40:3506–3518.
60. d’Errico F, Backwell L (2016) Earliest evidence of personal ornaments associated with
burial: The Conus shells from Border Cave. J Hum Evol 93:91–108.
61. Texier P-J, et al. (2010) From the cover: A Howiesons Poort tradition of engraving
ostrich eggshell containers dated to 60,000 years ago at Diepkloof Rock Shelter,
South Africa. Proc Natl Acad Sci USA 107:6180–6185.
62. Wadley L, Mohapi M (2008) A segment is not a monolith: Evidence from the Ho-
wiesons Poort of Sibudu, South Africa. J Archaeol Sci 35:2594–2605.
63. Jacobs Z, Roberts RG (2015) An improved single grain OSL chronology for the sedimentary
deposits from Diepkloof Rockshelter, Western Cape, South Africa. J Archaeol Sci 63:175–192.
64. Partridge TC, Demenocal PB, Lorentz SA, Paiker MJ, Vogel JC (1997) Orbital forcing of
climate over South Africa: A 200,000-year rainfall record from the Pretoria saltpan. Quat
Sci Rev 16:1125–1133.
7876 | www.pnas.org/cgi/doi/10.1073/pnas.1620752114
65. Daniau A-L, et al. (2013) Orbital-scale climate forcing of grassland burning in
southern Africa. Proc Natl Acad Sci USA 110:5069–5073.
66. Woillez M-N, et al. (2014) Impact of precession on the climate, vegetation and fire
activity in southern Africa during MIS4. Climate of the Past 10:1165–1182.
67. Urrego DH, Sánchez Goñi MF, Daniau A-L, Lechevrel S, Hanquiez V (2015) Increased
aridity in southwestern Africa during the warmest periods of the last interglacial.
Climate of the Past 11:1417–1431.
68. Rasmussen SO, et al. (2014) A stratigraphic framework for abrupt climatic changes
during the Last Glacial period based on three synchronized Greenland ice-core records:
Refining and extending the INTIMATE event stratigraphy. Quat Sci Rev 106:14–28.
69. Sánchez Goñi MF, Bard E, Landais A, Rossignol L, d’Errico F (2013) Air-sea temperature de-
coupling in western Europe during the last interglacial-glacial transition. Nat Geosci 6:837–841.
70. Broccoli AJ, Dahl KA, Stouffer RJ (2006) Response of the ITCZ to Northern Hemi-
sphere cooling. Geophys Res Lett 33:L01702.
71. Stouffer RJ, et al. (2006) Investigating the causes of the response of the thermo-
haline circulation to past and future climate changes. J Clim 19:1365–1387.
72. Barker S, et al. (2009) Interhemispheric Atlantic seesaw response during the last
deglaciation. Nature 457:1097–1102.
73. Kanner LC, Burns SJ, Cheng H, Edwards RL (2012) High-latitude forcing of the South
American summer monsoon during the Last Glacial. Science 335:570–573.
74. Marino G, et al. (2013) Agulhas salt-leakage oscillations during abrupt climate
changes of the Late Pleistocene. Paleoceanography 28:599–606.
75. Nix HA (1986) A biogeographic analysis of Australian elapid snakes. Atlas of Elapid
Snakes of Australia, Australian Flora and Fauna Series, ed Longmore R (Australian
Government Publishing Service, Canberra), pp 4–15.
76. Phillips SJ, Anderson RP, Schapire RE (2006) Maximum entropy modeling of species
geographic distributions. Ecol Modell 190:231–259.
77. Hijmans RJ, Phillips SJ, Leathwick J, Elith J (2017) dismo: Species Distribution Modeling.
Available at https://cran.r-project.org/web/packages/dismo/index.html. Accessed January
11, 2017.
78. R Core Team (2013) R: A Language and Environment for Statistical Computing (R
Foundation for Statistical Computing, Vienna).
79. Warren DL, Glor RE, Turelli M (2008) Environmental niche equivalency versus con-
servatism: Quantitative approaches to niche evolution. Evolution 62:2868–2883.
80. Warren DL, Glor RE, Turelli M (2010) ENMTools: A toolbox for comparative studies of
environmental niche models. Ecography 33:607–611.
81. Levins R (1968) Evolution in Changing Environments (Princeton Univ Press, Princeton).
82. Mandle L, et al. (2010) Conclusions about niche expansion in introduced Impatiens
walleriana populations depend on method of analysis. PLoS One 5:e15297.
83. Parmesan C, Yohe G (2003) A globally coherent fingerprint of climate change im-
pacts across natural systems. Nature 421:37–42.
84. Wiens JJ, Graham CH (2005) Niche conservatism: Integrating evolution, ecology, and
conservation biology. Annu Rev Ecol Evol Syst 36:519–539.
85. Peterson AT (2011) Ecological niche conservatism: A time-structured review of evi-
dence. J Biogeogr 38:817–827.
86. Tennie C, Call J, Tomasello M (2009) Ratcheting up the ratchet: On the evolution of
cumulative culture. Philos Trans R Soc Lond B Biol Sci 364:2405–2415.
87. Whiten A, McGuigan N, Marshall-Pescini S, Hopper LM (2009) Emulation, imitation,
over-imitation and the scope of culture for child and chimpanzee. Philos Trans R Soc
Lond B Biol Sci 364:2417–2428.
88. Porraz G, et al. (2013) Technological successions in the Middle Stone Age sequence
of Diepkloof Rock Shelter, Western Cape, South Africa. J Archaeol Sci 40:3376–3400.
89. d’Errico F, Banks WE (2015) The archaeology of teaching: A conceptual framework.
Camb Archaeol J 25:859–866.
90. Torrence R (2001) Hunter-gatherer technology: Macro- and microscale patterns. Hunter-
Gatherers: An Interdisciplinary Perspective, Biosocial Society Symposium Series, eds Panter-
Brick C, Layton R, Rowley-Conwy P (Cambridge Univ Press, Cambridge, UK), pp 73–98.
91. Collard M, Kemery M, Banks S (2005) Causes of toolkit variation among hunter-gatherers:
A test of four competing hypotheses. Can J Archaeol J Can D’Archéologie 29:1–19.
92. Read D (2008) An interaction model for resource implement complexity based on
risk and number of annual moves. Am Antiq 73:599–625.
93. Collard M, Buchanan B, Morin J, Costopoulos A (2011) What drives the evolution of
hunter-gatherer subsistence technology? A reanalysis of the risk hypothesis with
data from the Pacific Northwest. Philos Trans R Soc Lond B Biol Sci 366:1129–1138.
94. Rendell L, et al. (2011) How copying affects the amount, evenness and persistence of
cultural knowledge: Insights from the social learning strategies tournament. Philos
Trans R Soc Lond B Biol Sci 366:1118–1128.
95. Kline MA, Boyd R (2010) Population size predicts technological complexity in Oce-
ania. Proc R Soc Lond B Biol Sci 277:2559–2564.
96. Collard M, Buchanan B, O’Brien MJ, Scholnick J (2013) Risk, mobility or population
size? Drivers of technological richness among contact-period western North Amer-
ican hunter-gatherers. Philos Trans R Soc Lond B Biol Sci 368:20120412.
97. Banks WE, d’Errico F, Zilhão J (2013) Human-climate interaction during the Early
Upper Paleolithic: Testing the hypothesis of an adaptive shift between the Proto-
Aurignacian and the Early Aurignacian. J Hum Evol 64:39–55.
98. d’Errico F, Henshilwood CS (2007) Additional evidence for bone technology in the
southern African Middle Stone Age. J Hum Evol 52:142–163.
99. de la Peña P, Wadley L (2014) Quartz knapping strategies in the Howiesons Poort at
Sibudu (KwaZulu-Natal, South Africa). PLoS One 9:e101534.
100. Dayet L, Texier P-J, Daniel F, Porraz G (2013) Ochre resources from the Middle Stone Age
sequence of Diepkloof Rock Shelter, Western Cape, South Africa. J Archaeol Sci 40:3492–3505.
101. Laskar J, et al. (2004) A long-term numerical solution for the insolation quantities of
the Earth. Astron Astrophys 428:261–285.
102. Jouzel J, et al. (2007) Orbital and millennial Antarctic climate variability over the
past 800,000 years. Science 317:793–796.
d’Errico et al.

Cumulative cultural learning: Development
and diversity
Cristine H. Legarea,1
a
Department of Psychology, The University of Texas at Austin, Austin, TX 78712
Edited by Andrew Whiten, University of St. Andrews, St. Andrews, United Kingdom, and accepted by Editorial Board Member Andrew G. Clark June 16, 2017
(received for review January 14, 2017)
The complexity and variability of human culture is unmatched by any
other species. Humans live in culturally constructed niches filled with
artifacts, skills, beliefs, and practices that have been inherited,
accumulated, and modified over generations. A causal account of
the complexity of human culture must explain its distinguishing
characteristics: It is cumulative and highly variable within and across
populations. I propose that the psychological adaptations supporting
cumulative cultural transmission are universal but are sufficiently
flexible to support the acquisition of highly variable behavioral
repertoires. This paper describes variation in the transmission prac-
tices (teaching) and acquisition strategies (imitation) that support
cumulative cultural learning in childhood. Examining flexibility and
variation in caregiver socialization and children’s learning extends our
understanding of evolution in living systems by providing insight into
the psychological foundations of cumulative cultural transmission—
the cornerstone of human cultural diversity.
|
cumulative culture cultural evolution | cross-cultural comparison |
|
teaching imitation
H uman and nonhuman animals engage in behaviors that are
culturally created and subsequently transmitted (1–3).
Long-term studies of nonhuman animal species in their natural
habitats have demonstrated that many species respond to, and
learn from, social information (4–9). Nonhuman animals also
transmit group-specific behavior through what could be consid-
ered rudimentary forms of cultural transmission (10–14). How-
ever, cultural transmission in humans differs markedly from that
in nonhuman animals in both its extent and structural complexity
(15, 16). The psychological foundations of cultural complexity
are multifaceted. Explanations require drawing together de-
velopmental, cross-cultural, and comparative research that ex-
tends biology through culture.
Culture is defined as “group-typical behaviors shared by members
of a community that rely on socially learned and transmitted in-
formation” (17). Humans are “ultra” cultural (18); they live in
culturally constructed niches filled with artifacts, skills, beliefs, and
practices that have been inherited, accumulated, and modified over
generations (19–22). What explains the technological and social
complexity of human culture? A causal account must explain the
distinguishing characteristics of human culture: It is cumulative,
transmitted horizontally within groups and vertically across
generations, and varies within and between populations. Cu-
mulative culture is a process by which innovations are pro-
gressively incorporated into a population’s stock of skills and
knowledge, generating more complex repertoires (23–26).
Cumulative culture requires psychological adaptations that
ensure the high-fidelity transmission of knowledge, skills, and
practices (27–29). However, innovation is also necessary to
ensure cultural and individual adaptation to novel and changing
challenges (30–33).
Cumulative cultural transmission accelerates innovation be-
cause each generation can build upon the technologies passed
down by previous generations (34). Much of human technology is
too complex and sophisticated to be recreated within individual
lifetimes (35). The growth of cultural complexity is not exponential
www.pnas.org/cgi/doi/10.1073/pnas.1620743114
COLLOQUIUM
PAPER
or linear but instead is a process of punctuated accumulation; it
involves the conservation of some features, incremental in-
novation, and occasionally dramatic qualitative shifts (36).
The diversity of skills, practices, beliefs, and values among pop-
ulations is another distinguishing feature of human culture. Cultural
groups are heterogeneous populations of individuals that differ
along complex ecological, social, and structural variables. Socially
acquired and transmitted behaviors vary more distinctly among
human populations than in any other species (37). Cultural vari-
ability is one of our species’ most distinctive features, and a causal
account of human culture must explain its diversity. The psycho-
logical adaptations supporting cumulative cultural transmission are
hypothesized to be universal features of human psychology, but they
must be sufficiently flexible to support the acquisition of highly
variable skill sets and behavioral repertoires (38).
What psychological adaptations explain the species-specific ca-
pacity to accumulate and build upon the cultural innovations of
previous generations? To what extent do cultural transmission
practices (teaching) and cultural acquisition strategies (imitation)
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
vary across populations? How do caregivers use teaching to
transmit information, skills, and practices to children? How do
children use imitation to acquire the knowledge and skills of their
groups? The objective of this paper is to answer these questions
using data on teaching and imitation from developmental and
cross-cultural research.
One potential explanation for cross-species variation in cultural
complexity is social learning (2). Social learning is defined as
“learning that is influenced by observation of or interaction with
another animal (typically a conspecific) or its products” (39,
p. 207). Young children are well equipped with a complex repertoire
of social learning capacities (1, 40). Cumulative culture trans-
mission requires a particular kind of social learning that allows the
accumulation of successful modifications over time through a
process of cultural ratcheting (41–43). For example, humans im-
proved upon the Oldowan single-face stone tool, used more or less
intact for a million years, by creating bifacial Acheulean handaxes
with dramatically improved functionality—an example of cultural
continuity followed by punctuated innovation (44).
Cumulative cultural learning is psychologically prepared by a
set of adaptations that facilitate the transmission and acquisition
of information within and across generations (29, 45–47).
Teaching, high-fidelity imitation, and language are three linked
abilities that work in concert to support cultural transmission in
humans (48). Teaching and imitation reflect the distinction be-
tween instructed and imitative learning (43). Language allows
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: C.H.L. wrote the paper.
The author declares no conflict of interest.
This article is a PNAS Direct Submission. A.W. is a guest editor invited by the Editorial
Board.
1
Email: legare@austin.utexas.edu.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7877–7883
information transfer between individuals, supporting both teach-
ing and imitation. These cognitive abilities are supported by a
psychological system that has evolved to understand the minds of
others and to navigate complex social group behavior (49–51).
Well-documented cognitive biases reinforce cultural transmission,
including preferences for similar others (homophily) (52) and
proclivities for conformity (53), consensus (54–56), prestige (57,
58), and social norms (59–62).
If teaching and imitation provide the foundation for cumulative
culture, they should be early developing and universal (63). They
also must afford the capacity to respond flexibly to diverse onto-
genetic contexts and cultural ecologies (38, 64–66). Understanding
cultural continuity and variation in teaching and imitation pro-
vides insight into the process by which cumulative culture allows
humans to adapt to highly diverse environments. Teaching and
imitation conserve cultural knowledge, thus increasing the po-
tential for innovation or modification at the group level, which
further increases cultural complexity (67). Greater cultural com-
plexity increases the repertoire of socially transmitted beliefs and
practices, which increases the prevalence and necessity of teaching
(46). High-fidelity transmission may be more central to main-
taining cumulative culture than to innovation (68). Teaching and
imitation conserve and transmit both group-specific behavioral
repertoires and cultural innovations.
To understand variability in teaching and imitation better,
research must be conducted on childrearing environments and
practices across diverse populations. The great majority of psy-
chological research has been conducted in populations that are
unrepresentative of human culture globally and historically—
those from Western, educated, industrialized, rich, and demo-
cratic (WEIRD) backgrounds (38, 69). A growing literature within
developmental psychology and the anthropology of childhood aims
to correct the bias in studying WEIRD populations within the dis-
cipline (18, 37, 38, 70–78).
Using evidence from cross-cultural, developmental research, I
will first describe variation in cultural transmission practices and
then cultural acquisition strategies. My objective is to provide
insight into the origins of variation in cumulative cultural
learning and the processes by which knowledge is acquired and
transmitted during development. A comprehensive account of
teaching and imitation requires systematic study of variations in
childrearing practices and beliefs.
Variation in Cultural Transmission Practices
The universal goals of childrearing include promoting the survival,
health, and cultural competency of children (79). Children, in
collaboration with their caregivers and peers, interact in ways that
ensure the transmission of cultural practices and beliefs across
generations (37, 80). Teaching, defined as “a behavior in which
one animal intends that another learn some skill or acquire some
bit of information or knowledge that it did not have previously”
(48, p. 374), promotes the efficient transfer of information and is a
recurrent feature of human cultural transmission.
Human caregivers are unique among animals in their moti-
vation to transmit information to children through teaching (81).
Human adults expect children to learn and provide assistance
when needed (82). The ability to share psychological states and
intent with others during (intersubjectivity and metacognition),
to engage in mutually recognized, shared focus (joint attention),
and to engage in collaborative and coordinated interactions
contributes to efficient scaffolding and teaching (83–87).
Substantial quantitative and qualitative variation exists in
teaching both within (88) and among (89) populations. For ex-
ample, in different populations, caregivers respond differently to
infants’ emotional displays (90), speak to and structure their
infants’ social interactions and expectations in distinct ways (74,
91), and display variability in the modalities (e.g., physical, visual,
vocal) used to transmit information to infants (92, 93).
7878 | www.pnas.org/cgi/doi/10.1073/pnas.1620743114
Teaching is not a monolithic process. It consists of a repertoire
of cultural transmission strategies that vary based on the kind of
information or skill being transmitted, the effort required, and
beliefs about how children learn. Kline (77) developed a taxon-
omy of teaching based on data from a teaching ethogram for
cross-cultural human research (TEACH) to describe the varia-
tion and function of different pedagogical styles. For example,
when teaching by social tolerance, a teacher grants the learner
access for close observation. When teaching by opportunity
provisioning, teachers provide access to activities that are too
difficult or dangerous for the learner to explore independently,
without modification. When teaching by evaluative feedback, the
teacher provides positive or negative reinforcement of the learner’s
behavior through positive or negative verbal or gestural feedback,
positive or negative consequences, teasing, warning of danger, or
commands to stop, to say or to do. When teaching by social or local
enhancement, the teacher directs the learner’s attention toward
the task at hand. Direction of this kind can include calling atten-
tion to an object or person and commands to watch. When en-
gaging in direct active teaching, the teacher makes relevant
aspects of the task accessible or observable. Direct active teaching
involves defining the boundaries of what is to be learned and can
include direct communication, abstract communication, and
demonstration.
Opportunity provisioning and direct active teaching are rela-
tively effortful compared with the other types of teaching, because
the behaviors required are less compatible with teachers’ ongoing
behaviors, requiring an interruption of the teacher’s behavior at
some cost. Some of the indicators in the TEACH ethogram are
established in the literature as behavioral markers of teaching,
including ostensive cues and behaviors that facilitate shared at-
tention (48). Pedagogical style varies predictably based on the costs
and benefits of the mode of transmission, the cultural domain and
complexity, learner, and teacher identity (94, 95).
Direct active teaching has much in common with didactic
pedagogy (i.e., adults structuring and guiding children’s learning).
When engaging in direct active teaching, caregivers in WEIRD
populations scaffold and manage children’s learning environment,
often engaging in extensive face-to-face interaction, eye contact,
and instruction (92). According to this pedagogical model, the
caregiver is a teacher—a model of child socialization based on
Western formal educational practices. The extensive reliance on
direct active teaching may be a relatively recent historical phe-
nomenon in caregiver–child interactions and reflects cultural be-
liefs about children’s learning, the structure of formal educational
institutions, and the extensive body of abstract knowledge and
skills children are expected to master (e.g., literacy and numerical
computation). Direct active teaching is such a normative feature
of cultural transmission in WEIRD populations that some are
calling for limiting its use because of its potentially detrimental
effects on self-directed discovery and exploration (96).
Populations vary widely in the amount of direct active teaching
in which caregivers engage. For example in Fiji, direct active
teaching is relatively rare compared with less time-intensive and
costly forms of teaching, such as teaching by social tolerance or
allowing children to observe behaviors they may need to learn
(77, 95, 97). Cross-cultural research in the United States and
Vanuatu has demonstrated that, in contrast to caregivers from
Vanuatu, caregivers from the United States rely heavily on ver-
bal communication. They scaffold through language by asking
children questions, encourage planning, and provide high levels
of verbal praise and encouragement. Caregivers in the United
States also use extensive verbal instruction and repetition to in-
troduce new objects or unfamiliar tasks and to establish common
ground in information sharing (98). They engage in high levels of
visual contact with children, consistent with previous research
on joint attention in WEIRD populations. In contrast, caregivers
Legare

from Vanuatu use substantially more nonverbal forms of com-
munication, such as gesture and physical touch (92).
Ethnographic accounts of caregiver–child interaction empha-
size variation in parental ethnotheories (i.e., cultural beliefs)
about children’s capacity to self-educate through observational
learning (99, 100). Children routinely engage in third-party ob-
servation of adult activity and often learn by close observation
without being directly addressed or involved (101–104). Caregiver
expectations that will children learn through attentive observation
before participating impacts the caregiver’s pedagogical style (105).
If caregivers expect children to learn through observation instead
of through interactive conversation, they may be less likely to en-
gage in verbal scaffolding or to direct active teaching (77). Formal
education also impacts childrearing practices and values (76, 93,
106). These practices include how parents interact with their
infants (107), direct children’s attention (108), and use verbal
instruction (109).
There is substantial variation in how caregivers structure
children’s learning opportunities (102, 110–112). The kinds of
tasks caregivers and children engage in together vary (113), as
does the amount of time children spend with nonparental care-
givers and peers (79, 114). Cultural groups also vary in the degree
to which children are segregated from or are participants in adult
economic and social activity (100, 115). For example, children
living in communities that rely on labor-intensive subsistence ag-
riculture are expected to assist adults in subsistence-based labor
(e.g., cooking, planting and harvesting crops, and helping with the
childcare of younger siblings) at a young age (116).
Variation in cultural transmission practices also reflects the
kinds of skills and behaviors children must acquire. For example,
learning a complex or abstract skill often requires direct active
instruction to acquire that skill efficiently. Caregivers play a critical
role in transmitting the beliefs, skills, and practices of particular
populations. Cultural transmission alone does not explain high-
fidelity cultural acquisition. Young children are adept at acquir-
ing the beliefs and practices of the groups they are born into, an
extraordinary learning achievement that requires substantial flex-
ibility (38). Next I review evidence for high-fidelity imitation, de-
scribe evidence for variation between populations, and discuss the
implications for acquiring knowledge.
Variation in Cultural Acquisition Practices
Our species-typical proclivity for high-fidelity imitation is critical for
cumulative cultural transmission (42, 117, 118). High-fidelity imi-
tation plays a central role in both horizontal and vertical trans-
mission of group-specific cultural practices. Young children possess
cognitive and communication systems that support the transmission
of complex technical skills and social conventions (46, 65, 75).
Children learn the skills and practices of their communities by
imitating others. The ability and motivation to engage in high-
fidelity copying allows children to acquire an extraordinary va-
riety of skills and information they otherwise would not be able
to acquire through direct exploration or experimentation alone
(29). For acquired behavior to count as cultural, it must dis-
seminate in a social group and remain stable across generations
(119, 120). The conservation of knowledge and skills across
generations supports individual and group-level innovation
(121). The propensity for overimitation, or copying actions that
are causally irrelevant to achieving an instrumental end goal
(122, 123), develops early. Children often copy when uncertain
about the underlying causal structure of a behavior. This pro-
clivity is useful, given that a vast amount of behavior that chil-
dren acquire is opaque from the perspective of physical causality
(124, 125). High-fidelity imitation is an adaptive human strategy
facilitating more rapid social learning of instrumental skills than
would be possible if copying required a full causal representation
of an event (126).
Legare
COLLOQUIUM
PAPER
Cumulative culture requires the high-fidelity transmission of
two qualitatively different behaviors: instrumental knowledge
and skills (e.g., how to keep warm during winter) and social
conventional knowledge and skills (e.g., how to perform a cer-
emonial dance) (127). Acquiring the behavior of other group
members may be the function of an individual-level adaptation
for imitation in our species. Thus, the transmission of cumulative
culture across generations can be seen, in part, as a product of
our propensity for imitative flexibility (128).
The unique demands of acquiring instrumental skills and social
conventions provide insight into when children imitate and when
they innovate. The objective of imitating instrumental behavior is
reproducing the end goal by discerning which actions are causally
relevant to producing the desired outcome (127). Attending to the
causal relationship between the actions and the end goal allows for
innovation and variability in the reproduction of the behavior and,
as a result, lower-fidelity imitation. In contrast, the objective of
imitating conventional behavior is reproducing all the steps in the
process (129), which requires attending to the way in which the
behavior ought to be executed. In contrast to imitating in-
strumental behaviors, imitating conventional behaviors requires
consistently high-fidelity imitation. Children may encode causally
irrelevant actions not because they think that they are causally
efficacious in some way, or even to demonstrate shared intentions,
but rather to conform to social conventions (130). Although
learning an instrumental skill often allows for variability and in-
novation in methods of execution, learning social conventions
requires close conformity to the way other group members per-
form the actions.
Imitation has social functions, such as encoding normative
behavior (131). The adaptive benefits of group membership have
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
favored individuals who engage in affiliative behaviors, such as
high-fidelity imitation (132, 133). Children imitate social con-
ventions as a means of affiliation with group members (127).
High-fidelity imitation also may function as a reinclusion be-
havior in reaction to the threat of social exclusion from an in-
group in childhood in ways that parallel the increase in motor
mimicry following social exclusion by in-group members ob-
served in adults (134). Children ostracized by in-group members
display higher levels of anxiety and engage in higher imitative
fidelity of a group convention than children ostracized by out-
group members (135). They imitate instrumental tasks with
higher fidelity when primed with ostracism (136, 137). When
status or inclusion within a group is threatened, children may be
particularly motivated to enhance their standing in a group
through affiliative behavior such as high-fidelity imitation.
Imitation is used to acquire instrumental skills as well as to
engage in social conventions such as rituals. However, it is often
difficult to determine whether a behavior is instrumental or con-
ventional based on observation of the behavior alone. For exam-
ple, lighting a candle could have an instrumental goal (lighting a
dark room) or a conventional goal (worshiping a deity). How do
children determine whether a behavior is instrumental or con-
ventional? Young children are highly sensitive to contextual var-
iation in social information (138). Children use a number of social
and contextual cues when making inferences about the goal of
behavior. Cues to conventionality increase imitative fidelity. One
is causal opacity (i.e., lack of a physical causal mechanism). A
second is consensus (i.e., multiple actors performing the same
actions). A third is synchrony (i.e., multiple actors performing the
same actions at the same time) (56). Children are also highly
sensitive to verbal cues to conventionality and to the presence of a
social norm (65, 127, 139). Even infants are sensitive to language
cues to conventionality (140).
There is both continuity and variation in imitative flexibility across
populations (141–143). For example, children in industrialized,
Western populations (e.g., the United States) and subsistence-based,
non-Western populations (e.g., Vanuatu) imitate conventional
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7879
tasks with higher fidelity than instrumental tasks—an example of
continuity. Children in Vanuatu, however, engage in higher imitative
fidelity of instrumental tasks than in the United States, a potential
consequence of greater socialization for conformity in some pop-
ulations than in others (139, 144).
Cues to conventionality also increase expectations for confor-
mity and attention to behavioral variation. Children’s accuracy in
detecting differences between the performances of two actors is
greater when an action is interpreted as a social convention, po-
tentially because of expectations for conformity to conventional
behavior (127). The social conventionality of an action may trigger
affiliative behavior through conformity, motivating greater atten-
tion to detail and alertness to deviations from procedure. Children
also transmit conventional behavior with higher fidelity than in-
strumental tasks when teaching a peer (65).
Imitative flexibility improves over the course of childhood. For
example, there are age-related improvements in object memory-
based imitation between 2 and 5 y of age (145, 146). Children’s
understanding of the social and contextual cues that distinguish
instrumental from conventional behavior increases with age
(127, 139). They may become more sensitive to these cues as a
result of learning about social conventionality (50, 147). Un-
derstanding the development of imitative flexibility requires ex-
amining the extent to which caregivers scaffold this ability.
Caregivers in the United States adjust their interactions with
children according to the goal of the behavior. For example, they
encourage higher-fidelity imitation of social conventions than of
instrumental tasks and, conversely, encourage more creativity
and innovation for instrumental tasks than for social conven-
tions. They also engage in more encouragement, demonstration,
and monitoring when teaching their children conventional tasks
than when teaching instrumental behavior (148).
Adults across a wide range of global populations view high-
fidelity imitation as an efficient method of learning. For exam-
ple, parents in the United States and Vanuatu encourage chil-
dren to conform to the behavior of others (144). However, there
is variation in beliefs about the relationship between conformity
and competency. For example, when evaluating US children, US
adults are more likely to endorse low- than high-conformity
children as intelligent, often citing creativity as a justification
for their judgments. In contrast, Vanuatu adults are more likely
to endorse high- than low-conformity children as intelligent and
are more likely to endorse high-conformity children as well be-
haved than are US adults (144). The perceived relations between
intelligence, conformity, and creativity vary across populations.
Perceived intelligence is critical to social esteem and status
within a group. The variation in beliefs about indicators of in-
telligence based on conformity and creativity has implications for
the kinds of behaviors that are transmitted as well as for the
kinds of behaviors associated with prestige within groups.
As novice learners, children must acquire the practices and
beliefs of the group they are born into. To understand cultural
continuity and variability in cultural acquisition strategies better,
research must be conducted with children across diverse pop-
ulations. There is substantial evidence that high-fidelity imitation
is universal in human children. However, imitation, like teaching,
is not a monolithic capacity. Efficient cultural learning requires
flexible imitation of instrumental skills and social conventions.
The skills and conventions children must acquire vary enor-
mously among populations, as do expectations for conformity
versus innovation. Cross-cultural data demonstrate that variation
of this kind impacts the extent to which children engage in high-
fidelity imitation versus innovation.
Recent decades have produced a large literature on cumula-
tive cultural transmission. Children, in collaboration with their
caregivers and peers, interact in ways that ensure the transmission
of cultural practices and beliefs across generations. However, we
currently lack a complete causal explanatory account of cultural
7880 | www.pnas.org/cgi/doi/10.1073/pnas.1620743114
variation. Next I describe future directions for research designed to
explain variation between populations.
Explaining Cultural Variation
Despite evidence that cultural transmission practices and cul-
tural acquisition strategies vary across populations, to date we
cannot predict and explain the sources of this variation. Do
caregivers in populations with lower levels of Western-style ed-
ucation engage in less direct active teaching and more observa-
tional learning? If so, is this difference explained by participation
in Western-style education or other factors such as social orga-
nization or degree of market participation? Do caregivers in
hierarchical populations engage in more active teaching than
caregivers in egalitarian populations? Do peers, older siblings, or
cousins use different teaching styles than caregivers, and does
age-heterogeneity of the peer group impact learning? More re-
search is needed to collect the kind of demographic and mixed-
methodological data required to answer such questions.
Cultural groups vary along multiple continua. These include level
of integration into the global economic marketplace, social orga-
nization, urbanicity, kinship networks, peer-group age heterogene-
ity, and formal versus informal education. Systematic comparisons
among multiple groups will provide much stronger support for
causal claims that a particular variable of interest is responsible
for variation in dependent measures (149). For example, study-
ing Melanesian populations in Yasawa Islands, Fiji, and Tanna,
Vanuatu, would allow a comparison of populations that are similar
in terms of subsistence agricultural practices and limited exposure
to Western-style education but are different in terms of social or-
ganization (hierarchical social organization in Fiji versus egalitarian
chiefdoms in Vanuatu). Conducting research with multiple pop-
ulations that are similar along some variables but different along
others will (i) reveal the impact of sources of variation on outcomes;
(ii) prevent inadvertently describing idiosyncratic features of par-
ticular cultural contexts; and (iii) provide opportunities to reveal
social and psychological processes not possible if data were collected
only from a narrow range of populations.
The dearth of systematic research outside Western pop-
ulations presents a major impediment to theoretical progress in
the psychological sciences in general (47) and to the develop-
mental sciences in particular (18). Despite growing recognition
that most of what we know about child development is based on
a very narrow sample of children, cross-cultural developmental
studies are still rare, often unsystematic, and typically rely on
convenience sampling (29, 150). A new path forward in devel-
opmental science is needed to understand better the ontogeny of
a species that inhabits diverse cultural ecologies and faces complex
adaptive problems.
Building a comprehensive understanding of cultural trans-
mission practices and acquisition strategies requires studying
cultural contexts that differ in theoretically relevant ways. There
is a pressing need for systematic, cross-cultural, and mixed-
methodological research on this topic. The lack of infrastructure
for conducting research across multiple field sites has previously
posed a major impediment to understanding cultural variation.
Collaborative networks of international fieldsites are needed to
generate data from diverse populations—an undertaking that
requires the expertise and cooperation of multiple international
and interdisciplinary partners. Another obstacle to conducting
research of this kind is gaining approval to work in diverse
populations. Connections need to be established with diverse
communities and relationships developed based on trust and
respect, an issue that is even more critical when working with
children. In each community being studied, a network of local
research assistants and translators must be established and
maintained, and special care must be taken to ensure that the
methodologies and stimuli used in research are culturally salient
and appropriate.
Legare

Summary
The unparalleled intellectual success of humans is widely at-
tributed to our ability for cumulative cultural transmission, a
process by which we take the discoveries, behaviors, and inven-
tions of others and build upon them further to create increasingly
complex reserves of socially heritable knowledge and technology
(121). Evidence for culture in nonhuman species continues to
grow, but there are few candidate examples of cumulative culture
outside humans’ distinctively complex achievements. Human
culture is uniquely variable in nature, as exemplified by the ex-
traordinary diversity across technological skills and social prac-
tices within and among populations. Human psychological
flexibility provides the foundation for cultural diversity and is a
prerequisite for cumulative culture. It allows humans to build
upon established behaviors by relinquishing old solutions and
flexibly switching to more productive or efficient ones (128).
Cultural transmission practices and acquisition strategies
support cumulative culture: our species-specific capacity to ac-
cumulate and build upon the cultural innovations of previous
generations. A comprehensive account of teaching and imitation
requires systematic study of cultural variation and continuity in
childrearing practices (151). Variable socialization strategies
support different culturally specific childrearing goals. Teaching
practices reflect the values, educational institutions, and skill sets
of diverse cultural and ecological contexts. Children around the
globe use imitation flexibly to acquire the specific practices,
beliefs, and values of their groups. Future cross-cultural research
on teaching and imitation will enrich our understanding of cog-
nitive and social development and will substantially increase our
knowledge about the developmental origins of a psychological
hallmark of our species—cumulative cultural transmission.
The psychological foundations of cultural complexity are
multifaceted. Explanations that extend biology through cul-
ture require drawing together developmental, cross-cultural, and
1. Herrmann E, Call J, Hernàndez-Lloreda MV, Hare B, Tomasello M (2007) Humans
have evolved specialized skills of social cognition: The cultural intelligence hypoth-
esis. Science 317:1360–1366.
2. van Schaik CP, Burkart JM (2011) Social learning and evolution: The cultural in-
telligence hypothesis. Philos Trans R Soc Lond B Biol Sci 366:1008–1016.
3. Whiten A, van Schaik CP (2007) The evolution of animal ‘cultures’ and social in-
telligence. Philos Trans R Soc Lond B Biol Sci 362:603–620.
4. Aplin LM, et al. (2015) Experimentally induced innovations lead to persistent culture
via conformity in wild birds. Nature 518:538–541.
5. Fragaszy D, Visalberghi E (2004) Socially biased learning in monkeys. Learn Behav 32:
24–35.
6. Leadbeater E (2015) What evolves in the evolution of social learning? J Zool 295:
4–11.
7. Perry S, et al. (2003) Social conventions in wild white‐faced capuchin monkeys: Evi-
dence for traditions in a neotropical primate. Curr Anthropol 44:241–268.
8. Plotnik JM, Lair R, Suphachoksahakun W, de Waal FBM (2011) Elephants know when
they need a helping trunk in a cooperative task. Proc Natl Acad Sci USA 108:
5116–5121.
9. Rendell L, et al. (2010) Why copy others? Insights from the social learning strategies
tournament. Science 328:208–213.
10. Cantor M, et al. (2015) Multilevel animal societies can emerge from cultural trans-
mission. Nat Commun 6:8091.
11. Fragaszy DM, Perry S (2003) The Biology of Traditions: Models and Evidence (Cam-
bridge Univ Press, New York).
12. Garland EC, et al. (2013) Humpback whale song on the Southern Ocean feeding
grounds: Implications for cultural transmission. PLoS One 8:e79422.
13. Laland KN, Galef BG (2009) The Question of Animal Culture (Harvard Univ Press,
Cambridge, MA).
14. van Leeuwen EJC, Cronin KA, Haun DBM (2014) A group-specific arbitrary tradition
in chimpanzees (Pan troglodytes). Anim Cogn 17:1421–1425.
15. Dean LG, Vale GL, Laland KN, Flynn E, Kendal RL (2014) Human cumulative culture: A
comparative perspective. Biol Rev Camb Philos Soc 89:284–301.
16. Johnson-Pynn J, Fragaszy DM, Cummins-Sebree S (2003) Common territories in
comparative and developmental psychology: Quest for shared means and meaning
in behavioral investigations. Int J Comp Psychol 16:1–27.
17. Laland KN, Hoppitt W (2003) Do animals have culture? Evol Anthropol 12:150–159.
18. Nielsen M, Haun D (2016) Why developmental psychology is incomplete without
comparative and cross-cultural perspectives. Proc Biol Sci 371:20150071.
19. Odling-Smee FJ, Laland KN, Feldman MW (2003) Niche Construction: The Neglected
Process in Evolution (Princeton Univ Press, Princeton).
Legare
COLLOQUIUM
PAPER
comparative research (18, 38). The vast majority of studies
attempting to elucidate the evolutionary origins and ontogenetic
processes of cultural learning focus on children raised in WEIRD
societies (69, 151). Children living in technologically complex
cultural environments provide excellent opportunities to study the
early-developing capacity to adopt, capitalize upon, and build in-
creasingly sophisticated and opaque technologies. These pop-
ulations, however, do not reflect the childrearing environments
that Homo sapiens and their close ancestors experienced through
much of their cultural evolutionary history or the diverse ways in
which children are raised across the world today. Cross-cultural
research highlights not only the varied effects of the ontogenetic
environment on behavior and cognition but also strengthens
claims about universal and phylogenetically endowed mechanisms.
Children experience enculturation from infancy through their
interaction with caregivers, artifacts, and cultural institutions.
For this reason, it is also necessary to look to other species to
understand better how evolution has shaped the human mind.
Chimpanzees, arguably the second most cultural extant species
(81), are an ideal comparative sample for studying the mecha-
nisms and processes which may be unique to human culture or
inherited from our shared ancestors. Studying a wide age range
of chimpanzees, raised in different environments, will also in-
crease our understanding of the evolutionary origins of when and
why developmental processes shaped the hominin mind.
Humans engage in a wider variety of socially acquired and
transmitted behaviors that vary more distinctly across commu-
nities than any other animal species. Data from cross-cultural,
comparative, and developmental research are needed to increase
our understanding of the evolution of cumulative cultural
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
transmission. Examining the flexibility and variation in cultural
transmission practices and acquisition strategies provides insight
into the psychological foundations of cumulative cultural trans-
mission—the cornerstone of human cultural diversity.
20. Scott-Phillips TC, Laland KN, Shuker DM, Dickins TE, West SA (2014) The niche con-
struction perspective: A critical appraisal. Evolution 68:1231–1243.
21. Whiten A, Hinde RA, Laland KN, Stringer CB (2011) Culture evolves. Philos Trans R
Soc Lond B Biol Sci 366:938–948.
22. Boyd R, Richerson PJ, Henrich J (2011) The cultural niche: Why social learning is es-
sential for human adaptation. Proc Natl Acad Sci USA 108:10918–10925.
23. Pagel MD (2012) Wired for Culture: Origins of the Human Social Mind (W.W. Norton,
New York).
24. Pradhan GR, Tennie C, van Schaik CP (2012) Social organization and the evolution of
cumulative technology in apes and hominins. J Hum Evol 63:180–190.
25. Kurzban R, Barrett HC (2012) Behavior. Origins of cumulative culture. Science 335:
1056–1057.
26. Whiten A, Erdal D (2012) The human socio-cognitive niche and its evolutionary or-
igins. Philos Trans R Soc Lond B Biol Sci 367:2119–2129.
27. Creanza N, Kolodny O, Feldman MW (2017) Cultural evolutionary theory: How cul-
ture evolves and why it matters. Proc Natl Acad Sci USA 114:7782–7789.
28. Chudek M, Henrich J (2011) Culture-gene coevolution, norm-psychology and the
emergence of human prosociality. Trends Cogn Sci 15:218–226.
29. Legare CH, Nielsen M (2015) Imitation and innovation: The dual engines of cultural
learning. Trends Cogn Sci 19:688–699.
30. Arbilly M, Laland KN (2017) The magnitude of innovation and its evolution in social
animals. Proc Biol Sci 284:20162385.
31. Carr K, Kendal RL, Flynn EG (2016) Eureka!: What is innovation, how does it develop,
and who does it? Child Dev 87:1505–1519.
32. Henrich J, McElreath R (2007) Dual inheritance theory: The evolution of human
cultural capacities and cultural evolution. Oxford Handbook of Evolutionary
Psychology, eds Dunbar R, Barrett L (Oxford Univ Press, Oxford,UK), pp 555–570.
33. Lotem A, Halpern JY, Edelman S, Kolodny O (2017) The evolution of cognitive
mechanisms in response to cultural innovations. Proc Natl Acad Sci USA 114:
7915–7922.
34. Smaldino PE, Richerson PJ (2013) Human cumulative cultural evolution as a form of
distributed computation. Handbook of Human Computation, ed Michelucci P
(Springer, New York), pp 979–992.
35. Muthukrishna M, Henrich J (2016) Innovation in the collective brain. Proc Biol Sci
371:20150192.
36. Kolodny O, Creanza N, Feldman MW (2015) Evolution in leaps: The punctuated
accumulation and loss of cultural innovations. Proc Natl Acad Sci USA 112:
E6762–E6769.
37. Konner M (2010) The Evolution of Childhood: Relationships, Emotion, Mind (Harvard
Univ Press, Cambridge, MA).
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7881
38. Legare CH, Harris PL (2016) The ontogeny of cultural learning. Child Dev 87:633–642.
39. Heyes CM (1994) Social learning in animals: Categories and mechanisms. Biol Rev
Camb Philos Soc 69:207–231.
40. Nagell K, Olguin RS, Tomasello M (1993) Processes of social learning in the tool use
of chimpanzees (Pan troglodytes) and human children (Homo sapiens). J Comp
Psychol 107:174–186.
41. Mesoudi A (2017) Pursuing Darwin’s curious parallel: Prospects for a science of cul-
tural evolution. Proc Natl Acad Sci USA 114:7853–7860.
42. Tennie C, Call J, Tomasello M (2009) Ratcheting up the ratchet: On the evolution of
cumulative culture. Philos Trans R Soc Lond B Biol Sci 364:2405–2415.
43. Tomasello M (2016) The ontogeny of cultural learning. Current Opinion in
Psychology 8:1–4.
44. Stout D, Hecht EE (2017) Evolutionary neuroscience of cumulative culture. Proc Natl
Acad Sci USA 114:7861–7868.
45. Dean LG, Kendal RL, Schapiro SJ, Thierry B, Laland KN (2012) Identification of the
social and cognitive processes underlying human cumulative culture. Science 335:
1114–1118.
46. Hewlett BS, Roulette CJ (2016) Teaching in hunter-gatherer infancy. R Soc Open Sci
3:150403.
47. Mesoudi A, Chang L, Murray K, Lu HJ (2015) Higher frequency of social learning in
China than in the West shows cultural variation in the dynamics of cultural evolu-
tion. Proc Biol Sci 282:201442209.
48. Tomasello M, Kruger AC, Ratner HH (1993) Cultural learning. Behav Brain Sci 16:
495–552.
49. Bjorklund DF, Ellis BJ (2014) Children, childhood, and development in evolutionary
perspective. Dev Rev 34:225–264.
50. Diesendruck G, Markson L (2011) Children’s assumption of the conventionality of
culture. Child Dev Perspect 5:189–195.
51. Hrdy SB (2009) Mothers and Others: The Evolutionary Origins of Mutual Understanding
(Harvard Univ Press, Cambridge, MA).
52. Haun DBM, Over H (2014) Like me: A homophily-based account of human culture.
Epistemological Dimensions of Evolutionary Psychology, ed Breyer T (Springer, New
York), pp 117–130.
53. Muthukrishna M, Morgan TJH, Henrich J (2016) The when and who of social learning
and conformist transmission. Evol Hum Behav 37:10–20.
54. Claidière N, Whiten A (2012) Integrating the study of conformity and culture in
humans and nonhuman animals. Psychol Bull 138:126–145.
55. Corriveau KH, Fusaro M, Harris PL (2009) Going with the flow: Preschoolers prefer
nondissenters as informants. Psychol Sci 20:372–377.
56. Herrmann PA, Legare CH, Harris PL, Whitehouse H (2013) Stick to the script: The
effect of witnessing multiple actors on children’s imitation. Cognition 129:536–543.
57. Chudek M, Heller S, Birch S, Henrich J (2012) Prestige-biased cultural learning: By-
stander’s differential attention to potential models influences children’s learning.
Evol Hum Behav 33:46–56.
58. Henrich J (2009) The evolution of costly displays, cooperation, and religion: Credi-
bility enhancing displays and their implications for cultural evolution. Evol Hum
Behav 30:244–260.
59. Kenward B (2012) Over-imitating preschoolers believe unnecessary actions are nor-
mative and enforce their performance by a third party. J Exp Child Psychol 112:
195–207.
60. Haun DB, van Leeuwen EJ, Edelson MG (2013) Majority influence in children and
other animals. Dev Cogn Neurosci 3:61–71.
61. Rakoczy H, Schmidt MFH (2013) The early ontogeny of social norms. Child Dev
Perspect 7:17–21.
62. Schmidt MFH, Butler LP, Heinz J, Tomasello M (2016) Young children see a single
action and infer a social norm: Promiscuous normativity in 3-year-olds. Psychol Sci 27:
1360–1370.
63. Gergely G, Csibra G (2006) Sylvia’s recipe: The role of imitation and pedagogy in the
transmission of human culture. Roots of Human Sociality: Culture, Cognition and
Human Interaction, eds Enfield NJ, Levenson SC (Berg Publishers, Oxford, UK), pp
229–255.
64. Barrett HC, et al. (2016) Small-scale societies exhibit fundamental variation in the
role of intentions in moral judgment. Proc Natl Acad Sci USA 113:4688–4693.
65. Clegg JM, Legare CH (2016a) Instrumental and conventional interpretations of be-
havior are associated with distinct outcomes in early childhood. Child Dev 87:
527–542.
66. House BR, et al. (2013) Ontogeny of prosocial behavior across diverse societies. Proc
Natl Acad Sci USA 110:14586–14591.
67. Enquist M, Strimling P, Eriksson K, Laland K, Sjostrand J (2010) One cultural parent
makes no culture. Anim Behav 79:1353–1362.
68. Lewis HM, Laland KN (2012) Transmission fidelity is the key to the build-up of cu-
mulative culture. Philos Trans R Soc Lond B Biol Sci 367:2171–2180.
69. Henrich J, Heine SJ, Norenzayan A (2010) The weirdest people in the world? Behav
Brain Sci 33:61–83, discussion 83–135.
70. Arnett JJ (2008) The neglected 95%: Why American psychology needs to become less
American. Am Psychol 63:602–614.
71. Barrett HC, et al. (2013) Early false-belief understanding in traditional non-Western
societies. Proc Biol Sci 280:20122654.
72. Berl REW, Hewlett BS (2015) Cultural variation in the use of overimitation by the Aka
and Ngandu of the Congo Basin. PLoS One 10:e0120180.
73. Blake PR, et al. (2015) The ontogeny of fairness in seven societies. Nature 528:
258–261.
74. Broesch T, Bryant GA (February 27, 2017) Fathers’ infant-directed speech in a small-
scale society. Child Dev, 10.1111/cdev.12768.
7882 | www.pnas.org/cgi/doi/10.1073/pnas.1620743114
75. Callaghan T, et al. (2011) Early social cognition in three cultural contexts. Monogr
Soc Res Child Dev, 10.1111/j.1540-5834.2011.00603.x.
76. Gaskins S, Paradise R (2010) Learning Through Observation. The Anthropology of
Learning and Childhood, eds Lancy DF, Bock J, Gaskins S (Alta Mira, Lanham, MD), pp
85–117.
77. Kline MA (2015) How to learn about teaching: An evolutionary framework for the
study of teaching behavior in humans and other animals. Behav Brain Sci 38:e31.
78. Shneidman L, Gaskins S, Woodward A (2016) Child-directed teaching and social
learning at 18 months of age: Evidence from Yucatec Mayan and US infants. Dev Sci
19:372–381.
79. LeVine JL (2007) Ethnographic studies of childhood: A historical overview. Am
Anthropol 109:247–260.
80. Otto H, Keller H (2014) Different Faces of Attachment: Cultural Variations on a
Universal Human Need (Cambridge Univ Press, Cambridge, UK).
81. Whiten A (2017) Culture extends the scope of evolutionary biology in the great apes.
Proc Natl Acad Sci USA 114:7790–7797.
82. Kruger AC, Tomasello M (1996) Cultural learning and learning culture. The
Handbook of Human Development and Education, eds Olson D, Torrance N
(Blackwell, Oxford, UK), pp 369–387.
83. Csibra G, Gergely G (2009) Natural pedagogy. Trends Cogn Sci 13:148–153.
84. Marsh KL, Richardson MJ, Schmidt RC (2009) Social connection through joint action
and interpersonal coordination. Top Cogn Sci 1:320–339.
85. Nielsen M (2012) Imitation, pretend play, and childhood: Essential elements in the
evolution of human culture? J Comp Psychol 126:170–181.
86. Shneidman L, Woodward AL (2016) Are child-directed interactions the cradle of
social learning? Psychol Bull 142:1–17.
87. Tomasello M, Carpenter M, Call J, Behne T, Moll H (2005) Understanding and sharing
intentions: The origins of cultural cognition. Behav Brain Sci 28:675–691, discussion
691–735.
88. Mesoudi A, Chang L, Dall SRX, Thornton A (2016) The evolution of individual and
cultural variation in social learning. Trends Ecol Evol 31:215–225.
89. Hewlett BS, Fouts HN, Boyette AH, Hewlett BL (2011) Social learning among Congo
Basin hunter-gatherers. Philos Trans R Soc Lond B Biol Sci 366:1168–1178.
90. Broesch T, Rochat P, Olah K, Broesch J, Henrich J (2016) Similarities and differences in
maternal responsiveness in three societies: Evidence from Fiji, Kenya, and the United
States. Child Dev 87:700–711.
91. Kärtner J, Crafa D, Chaudhary N, Keller H (2016) Reactions to receiving a gift—
Maternal scaffolding and cultural learning in Berlin and Delhi. Child Dev 87:712–722.
92. Little EE, Carver LJ, Legare CH (2016) Cultural variation in triadic infant-caregiver
object exploration. Child Dev 87:1130–1145.
93. Keller H, et al. (2006) Cultural models, socialization goals, and parenting ethno-
theories: A multicultural analysis. J Cross Cult Psychol 37:155–172.
94. Demps K, Zorondo-Rodríguez F, García C, Reyes-García V (2012) Social learning
across the life cycle: Cultural knowledge acquisition for honey collection among the
Jenu Kuruba, India. Evol Hum Behav 33:460–470.
95. Kline MA, Boyd R, Henrich J (2013) Teaching and the life history of cultural trans-
mission in Fijian villages. Hum Nat 24:351–374.
96. Bonawitz E, et al. (2011) The double-edged sword of pedagogy: Instruction limits
spontaneous exploration and discovery. Cognition 120:322–330.
97. Kline MA (2016) TEACH: An ethogram-based method to observe and record teaching
behavior. Field Methods, 10.1177/1525822X16669282.
98. Clark EV, Bernicot J (2008) Repetition as ratification: How parents and children place
information in common ground. J Child Lang 35:349–371.
99. Harkness S, Super C (2002) Culture and Parenting, ed Bornstein M, Handbook of
parenting (Erlbaum, Mahwah), pp 253–280.
100. Keller H (2007) Cultures of infancy (Lawrence Erlbaum Associates, Mahwah, NJ).
101. Chavajay P, Rogoff B (2002) Schooling and traditional collaborative social organi-
zation of problem solving by Mayan mothers and children. Dev Psychol 38:55–66.
102. Mejía-Arauz R, Rogoff B, Dexter A, Najafi B (2007) Cultural variation in children’s
social organization. Child Dev 78:1001–1014.
103. Peck JG, Gregory RJ (2005) A brief overview of the Old New Hebrides.
Anthropologist 7:269–282.
104. Tobin J, Hsueh Y, Karasawa M (2009) Preschool in Three Cultures Revisited: China,
Japan, and the United States (Univ of Chicago Press, Chicago).
105. Rogoff B, Paradise R, Arauz RM, Correa-Chávez M, Angelillo C (2003) Firsthand
learning through intent participation. Annu Rev Psychol 54:175–203.
106. Correa-Chávez M, Rogoff B (2009) Children’s attention to interactions directed to
others: Guatemalan Mayan and European American patterns. Dev Psychol 45:
630–641.
107. Kärtner J, Keller H, Yovsi RD (2010) Mother-infant interaction during the first
3 months: The emergence of culture-specific contingency patterns. Child Dev 81:
540–554.
108. Silva KG, Correa-Chávez M, Rogoff B (2010) Mexican-heritage children’s attention
and learning from interactions directed to others. Child Dev 81:898–912.
109. LeVine RA, LeVine S, Schnell-Anzola B, Rowe M, Dexter E (2012) Literacy and
Mothering: How Women’s Schooling Changes the Lives of the World’s Children
(Oxford Univ Press, New York).
110. Keller H Kärtner (2013) Development—The culture-specific solution of universal
developmental tasks. Advances in Culture and Psychology, eds Gelfand ML, Chiu C-Y,
Hong YY (Oxford Univ Press, New York), pp 63–116.
111. Ochs E, Schieffelin BB (2001) Language acquisition and socialization: Three de-
velopmental stories and their implications. Linguistic Anthropology: A Reader, ed
Duranti A (Cambridge Univ Press, Cambridge, UK), pp 296–328.
112. Whiting BB, Whiting JW (1975) Children of Six Cultures: A Psycho-Cultural Analysis
(Harvard Univ Press, Cambridge, MA).
Legare

113. Cole M (1990) Cognitive development and formal schooling: The evidence from
cross-cultural research. Vygotsky and Education, ed Moll LC (Cambridge Univ Press,
Cambridge, UK), pp 89–110.
114. Gaskins S (2006) Cultural perspectives on infant-caregiver interaction. The Roots of
Human Sociality: Culture, Cognition, and Human Interaction, eds Enfield NJ,
Levinson SC (Berg, Oxford), pp 279–298.
115. Gauvain M, Munroe RL (2009) Contributions of societal modernity to cognitive de-
velopment: A comparison of four cultures. Child Dev 80:1628–1642.
116. Lancy DF (2015) The Anthropology of Childhood: Cherubs, Chattel, Changelings
(Cambridge Univ Press, Cambridge, UK).
117. Heyes C (2009) Evolution, development and intentional control of imitation. Philos
Trans R Soc Lond B Biol Sci 364:2293–2298.
118. Schillinger K, Mesoudi A, Lycett SJ (2015) The impact of imitative versus emulative
learning mechanisms on artifactual variation: Implications for the evolution of
material culture. Evol Hum Behav 36:446–455.
119. Claidière N, Sperber D (2010) Imitation explains the propagation, not the stability of
animal culture. Proc Biol Sci 277:651–659.
120. Mathew S, Perreault C (2015) Behavioural variation in 172 small-scale societies in-
dicates that social learning is the main mode of human adaptation. Proc Biol Sci
282:20150061.
121. Henrich J (2015) The Secret of Our Success: How Culture Is Driving Human
Evolution, Domesticating Our Species, and Making Us Smarter (Princeton Univ
Press, Princeton).
122. Horner V, Whiten A (2005) Causal knowledge and imitation/emulation switching in
chimpanzees (Pan troglodytes) and children (Homo sapiens). Anim Cogn 8:164–181.
123. Lyons DE, Damrosch DH, Lin JK, Macris DM, Keil FC (2011) The scope and limits of
overimitation in the transmission of artefact culture. Philos Trans R Soc Lond B Biol
Sci 366:1158–1167.
124. Legare CH, Souza AL (2012) Evaluating ritual efficacy: Evidence from the supernat-
ural. Cognition 124:1–15.
125. Legare CH, Watson-Jones RE (2015) The evolution snd ontogeny of ritual. The
Handbook of Evolutionary Psychology, ed Buss DM (Wiley & Sons, Hoboken, NJ),
pp 829–847.
126. Toelch U, Bruce M, Newson L, Richerson P, Reader S (2014) Individual consistency and
flexibility in human social information use. Proc Biol Sci 281:20132864.
127. Legare CH, Wen NJ, Herrmann PA, Whitehouse H (2015) Imitative flexibility and the
development of cultural learning. Cognition 142:351–361.
128. Davis SJ, Vale GL, Schapiro SJ, Lambeth SP, Whiten A (2016) Foundations of cumu-
lative culture in apes: Improved foraging efficiency through relinquishing and
combining witnessed behaviours in chimpanzees (Pan troglodytes). Sci Rep 6:35953.
129. Call J, Carpenter M, Tomasello M (2005) Copying results and copying actions in the
process of social learning: Chimpanzees (Pan troglodytes) and human children
(Homo sapiens). Anim Cogn 8:151–163.
130. Keupp S, Behne T, Rakoczy H (2013) Why do children overimitate? Normativity is
crucial. J Exp Child Psychol 116:392–406.
131. Over H, Carpenter M (2012) Putting the social into social learning: Explaining both
selectivity and fidelity in children’s copying behavior. J Comp Psychol 126:182–192.
Legare
COLLOQUIUM
PAPER
132. Watson-Jones RE, Legare CH (2016) The social functions of group rituals. Curr Dir
Psychol Sci 25:42–46.
133. Wen N, Herrmann P, Legare C (2016) Ritual increases children’s affiliation with in-
group members. Evol Hum Behav 37:54–60.
134. Chartrand TL, Lakin JL (2013) The antecedents and consequences of human behav-
ioral mimicry. Annu Rev Psychol 64:285–308.
135. Watson-Jones RE, Whitehouse H, Legare CH (2016) In-group ostracism increases
high-fidelity imitation in early childhood. Psychol Sci 27:34–42.
136. Over H, Carpenter M (2009) Priming third-party ostracism increases affiliative imi-
tation in children. Dev Sci 12:F1–F8.
137. Watson-Jones RE, Legare CH, Whitehouse H, Clegg JM (2014) Task-specific effects of
ostracism on imitative fidelity in early childhood. Evol Hum Behav 35:204–210.
138. Buchsbaum D, Gopnik A, Griffiths TL, Shafto P (2011) Children’s imitation of causal
action sequences is influenced by statistical and pedagogical evidence. Cognition
120:331–340.
139. Clegg JM, Legare CH (2016b) A cross-cultural comparison of children’s imitative
flexibility. Dev Psychol 52:1435–1444.
140. Scott JC, Henderson AME (2013) Language matters: Thirteen-month-olds un-
derstand that the language a speaker uses constrains conventionality. Dev Psychol
49:2102–2111.
141. Nielsen M, Tomaselli K (2010) Overimitation in Kalahari Bushman children and the
origins of human cultural cognition. Psychol Sci 21:729–736.
142. Nielsen M, Mushin I, Tomaselli K, Whiten A (2014) Where culture takes hold:
“Overimitation” and its flexible deployment in Western, Aboriginal, and Bushmen
children. Child Dev 85:2169–2184.
143. Nielsen M, Mushin I, Tomaselli K, Whiten A (2016) Imitation, collaboration and their
interaction among Western and Indigenous Australian preschool children. Child Dev
87:795–806.
144. Clegg JM, Wen NJ, Legare CH (2017) Is non-conformity WEIRD? Cultural variation in
adults’ beliefs about children’s competency and conformity. J Exp Psychol Gen 146:
428–441.
145. Rakoczy H, Brosche N, Warneken F, Tomasello M (2009) Young children’s un-
derstanding of the context-relativity of normative rules in conventional games. Br J
Dev Psychol 27:445–456.
146. Subiaul F, Schilder B (2014) Working memory constraints on imitation and emula-
tion. J Exp Child Psychol 128:190–200.
147. Köymen B, et al. (2014) Children’s norm enforcement in their interactions with
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
peers. Child Dev 85:1108–1122.
148. Clegg JM, Legare CH (2017) Parents scaffold flexible imitation during early child-
hood. J Exp Child Psychol 153:1–14.
149. Hewlett BS, Lamb ME, Shannon D, Leyendecker B, Schölmerich A (1998) Culture and
early infancy among central African foragers and farmers. Dev Psychol 34:653–661.
150. Jensen LA (2012) Bridging universal and cultural perspectives: A vision for de-
velopmental psychology in a global world. Child Dev Perspect 6:98–104.
151. Nielsen M, Haun D, Kärtner J, Legare CH (2017) The persistent sampling bias in
developmental psychology: A call to action. J Exp Child Psychol 162:31–38.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7883
Young children communicate their ignorance and
ask questions
Paul L. Harrisa,1, Deborah T. Bartza, and Meredith L. Rowea
a
Graduate School of Education, Harvard University, Cambridge, MA 02138
Edited by Andrew Whiten, University of St. Andrews, St. Andrews, United Kingdom, and accepted by Editorial Board Member Andrew G. Clark April 8, 2017
(received for review January 12, 2017)
Children acquire information, especially about the culture in which
they are being raised, by listening to other people. Recent evidence
has shown that young children are selective learners who preferen-
tially accept information, especially from informants who are likely
to be representative of the surrounding culture. However, the extent
to which children understand this process of information trans-
mission and actively exploit it to fill gaps in their knowledge has not
been systematically investigated. We review evidence that toddlers
exhibit various expressive behaviors when faced with knowledge
gaps. They look toward an available adult, convey ignorance via
nonverbal gestures (flips/shrugs), and increasingly produce verbal
acknowledgments of ignorance (“I don’t know”). They also produce
comments and questions about what their interlocutors might
know and adopt an interrogative stance toward them. Thus, in
the second and third years, children actively seek information
from interlocutors via nonverbal gestures or verbal questions
and display a heightened tendency to encode and retain such
sought-after information.
ignorance | questions | children | communication
A n influential body of research in developmental psychology
has shown that infants are cognitively attuned to stable
properties of the world: They possess core knowledge (1), a set of
concepts enabling them to make sense of events and transfor-
mations in the physical, biological, and psychological domains.
Moreover, building on that core knowledge, young children
gradually construct a variety of deeper conceptual and causal in-
sights within each of those domains (2–4). Alongside this portrayal
of the child as a young scientist who steadily builds up a co-
herent and objective conception of the natural world, recent
developmental research has paid increasing attention to the
ways in which infants and young children can also be viewed
as anthropologists. They are cultural learners, receptive to
information from other people, including caregivers, adult
members of their group, and peers, especially regarding the
distinctive languages, beliefs, and practices of the culture that
they live in (5–8).
Much of this recent research has emphasized that despite their
receptivity to the information provided by other people, young
children are selective about their informants. More specifically,
they appraise potential informants along a variety of dimensions,
including their familiarity (9, 10), their prior accuracy (9, 11, 12),
apparent group membership (13, 14), and degree of consensus
(15–17). Such selectivity is likely to facilitate children’s acquisition
of those beliefs and norms that are representative of their culture.
In highlighting children’s appraisal of, and receptivity to, poten-
tial informants, research on early cultural learning has tended to
ignore children’s self-appraisals and their concomitant information
seeking. However, unlike other species, cultural learning by human
children is often based on the testimony and teaching of others (i.e.,
on nonverbal or verbal messages deliberately aimed at informing
naive or ignorant learners). Granted that distinctive mode of cul-
tural learning, it is plausible that even from an early age, children
communicate gaps in their knowledge and ask for pertinent in-
formation. In this view, young children are not just selective recipients
7884–7891 | PNAS | July 25, 2017 | vol. 114 | no. 30
of the information that is made available by the surrounding com-
munity; they also remedy their own ignorance by adopting an in-
terrogative stance toward potential informants.
Below, we review recent findings on children’s appraisal of
informant consensus, highlighting the research lacuna just
mentioned. We then turn to research focusing on young child-
ren’s appraisal of themselves, especially their states of ignorance,
as well as the emergence of the interrogative stance.
Sensitivity to Consensus and Uncertainty
Research on children’s appraisal of informant consensus has drawn
on approaches to cultural learning grounded in evolutionary theo-
rizing. Adopting this approach, Morgan et al. (17) asked how far
children would be swayed by varying degrees of consensus among
their informants when making numerical judgments. Children
ranging from 3 to 7 y of age were asked to say which of two displays,
each containing 10–30 dots, was numerically greater. Consistent
with prior findings (18), children were better at choosing the nu-
merically larger display the greater the difference in size between
the two displays, as indexed by the dot ratio (i.e., the ratio of the
difference between the displays relative to the size of the smaller
display); the gradient of this improvement in accuracy for easier
compared with more difficult trials was steep for older children but
shallow for younger children.
Having made a decision about any given pair of displays,
children were offered feedback by 10 informants who each either
agreed or disagreed with their decision, with the number of in-
formants who agreed versus disagreed varying (from 0 to 10)
from one trial to the next. Thus, children might be confronted
with unanimous agreement with their initial decision, unanimous
disagreement, or any split between those two extremes. After this
social feedback, children were invited to make a second decision
about the two displays, thereby completing that particular trial.
All age groups were prone to stick with their initial decision, and,
surprisingly, they did so even on difficult trials. Moreover, Fig. 1
shows that the tendency to stick with an initial decision became
stronger with age, irrespective of the ease or difficulty of the
trial. Nevertheless, children’s overall tendency to stick to their
initial decision was tempered by their sensitivity to social feed-
back, and the pattern of that sensitivity changed considerably
with age. Fig. 2 indicates that 7-y-olds displayed a so-called
“conformist” bias: They were disproportionately sensitive to
the majority opinion among informants. In contrast, 6-y-olds
displayed a linear or proportionate response: The greater the
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: P.L.H., D.T.B., and M.L.R. designed research; D.T.B. performed re-
search; P.L.H. and D.T.B. analyzed data; and P.L.H. and M.L.R. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. A.W. is a guest editor invited by the Editorial
Board.
1
To whom correspondence should be addressed. Email: paul_harris@gse.harvard.edu.
www.pnas.org/cgi/doi/10.1073/pnas.1620745114

Fig. 1. Probability that children stick with their initial decision for the case
of five versus five informants such that whether or not children stick is based
on the initial information they gathered via observation of the displays
(asocial information) and their sticking tendency. Children tend to stick with
their initial decision across all trial ratios (i.e., irrespective of trial difficulty),
although the tendency to stick is slightly lower on the more difficult (low dot
ratio) trials. The tendency to stick sharply increases with age; the oldest
children (7-y-olds) display a >80% chance of sticking (color-coding of the ages of
3, 4, 5, 6, and 7 y is provided in Fig. 2). Reprinted with permission from ref. 17.
number of informants choosing a given option, the greater was
the likelihood that 6-y-olds responded similarly. Finally, younger
children were swayed by unanimity among informants but
showed little sensitivity to social feedback that fell short of
unanimity. For example, their final decisions were roughly the
same whether two of the 10 informants judged like them and
eight did not, or the reverse. In sum, although the exact nature of
children’s reactions to disagreement among informants changed
sharply with age, children were sensitive to social feedback at all
ages. Moreover, older children displayed the type of conformist
bias (i.e., a disproportionate sensitivity to nontotal majorities)
that evolutionary theory has identified as a highly effective
strategy for widespread cultural dissemination (19). Hence,
children’s tendency to stick with their initial judgment cannot be
attributed to any overall insensitivity to social feedback.
An alternative explanation of children’s tendency to stick, and
to stick even on difficult trials, is that they lack an ability to
monitor their own knowledge states. They treat what effectively
amounts to a random judgment on difficult trials and a well-
founded judgment on easy trials as more or less equivalent. In
this view, young children ultimately have little ability to differen-
tiate cognitive states that, in principle, ought to be quite distinct:
notably, states of ignorance, in which only a guess can be made,
and states of knowledge, in which a judgment can be made with a
high probability of its being correct. If this hypothesis is correct, it
implies that children are poor at weighing social feedback against
their own asocial information. Having little awareness of the ep-
istemic standing of their own asocial information, they do not
appropriately calibrate their deference to social information.
However, even if young children are insensitive to the cer-
tainty versus uncertainty of their numerical judgments, it is un-
likely that they are insensitive to the standing of their cognitive
states across all domains of knowledge. Indeed, as elaborated
below, recent evidence suggests that even 2-y-olds have some
Harris et al.
COLLOQUIUM
PAPER
metacognitive awareness, especially in the context of ongoing
dialogue with an interlocutor. In the next section, we argue that
very young children display five interlinked abilities: (i) an un-
derstanding of the nature of communication, especially its power
to convey information from an informant to a recipient; (ii) an
ability to signal their ignorance to an interlocutor; (iii) an ability
to talk cogently about knowledge and ignorance; (iv) an ability to
communicate their desire for information via gestures and
questions; and (v) an ability to monitor the extent to which the
information requested of an informant does or does not remedy
their ignorance.
An Early Understanding of Communication
In the course of the second year, when the ability of human infants
to communicate with words remains limited, they nonetheless
display a basic understanding of the way that communication
works. They understand that requests and assertions can be
communicated from one person to another so that the recipient is
likely to end up with information that he or she can act upon, and
may indeed favor relative to prior information based on first-hand
observation (20). Thus, infants show some understanding of the
way that communication can guide the actions and update the
knowledge base of a recipient. Moreover, they display an un-
derstanding of the impact of communication, not simply when
they seek information from a potential informant or when they
supply information to a recipient but also when, as a third party,
they witness or eavesdrop on an exchange between two other
people. In such contexts, infants appear to encode the message
supplied by the informant and to work out its likely impact on the
actions and knowledge of the recipient.
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
The following body of experimental work illustrates these
basic points. Krehm et al. (21) had infants aged 9 and 11 mo
watch while an informant expressed her preference for one of
two objects by reaching for and manipulating her preferred
Fig. 2. Probability that children stick with a given decision (e.g., the right-
hand side display of dots) for a trial of intermediate difficulty. The 7-y-olds
show a conformist bias by responding disproportionately to majorities that
fall short of unanimity. The 6-y-olds display a proportionate response to the
number of informants endorsing their decision. Younger children, especially
3- and 4-y-olds, are only affected by informant feedback when there is com-
plete unanimity; they are prone to ignore informant feedback when there is
disagreement among informants. Reprinted with permission from ref. 17.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7885
object. A recipient then appeared who expressed no specific
preference for either object insofar as she handled both. In the
subsequent test phase, the informant reappeared and pointed to
her preferred object as the recipient watched. Infants expressed
more surprise (by looking longer) when the recipient handed the
informant her nonpreferred object as opposed to the one that
she had pointed at. By implication, infants expected the recipient
to understand which object the informant wanted, given her
pointing gesture, and to respond accordingly. A control condi-
tion consolidated this interpretation. If the informant gestured
with an open fist rather than a point, or if the recipient closed
her eyes rather than watched the informant’s gesture, the se-
lective pattern of looking disappeared. This selective pattern was
displayed by both 9- and 11-mo-olds, and indeed irrespective of
whether infants had started to point themselves.
Martin et al. (22) obtained similar findings when the informant
signaled that she wanted a preferred object, not via a pointing
gesture but by saying “koba.” This lexical item was unfamiliar to
the 12-mo-olds being tested. Nevertheless, they tended to construe
it as a request by the informant for her preferred object, again as
indexed by the pattern of looking that they displayed when the
recipient did or did not comply in terms of the particular object
that she handed to the informant. Infants expressed more surprise
(looked longer) if the recipient handed the informant the non-
preferred object. Control conditions indicated that infants’ con-
strual of the informant’s signal as a request was restricted to
speech-like utterances. If the informant coughed rather than
spoke, or produced a vocalization (“Oooh!”) rather than a lexical
item, the pattern of selective looking disappeared.
Thus, at the very beginning of the second year, infants are not
inattentive or uncomprehending bystanders with respect to ongoing
patterns of communication. They grasp that particular signals can
be interpreted as requests for a particular object. Moreover, their
construal of dialogic communication is such that they expect the
recipient to interpret the requests appropriately and to act ac-
cordingly. This construal of dyadic communication is appropriately
confined to certain types of signals, notably a pointing gesture or the
production of a lexical item (including one that is novel) rather than
a hand movement, a cough, or a vocalization.
Song et al. (23) asked if older infants, aged 18 mo, would un-
derstand not just a request for an object, as conveyed by a pointing
gesture or lexical item, but an assertion, and notably an assertion
that could, in principle, update the recipient’s knowledge base.
Infants watched as an adult repeatedly placed a ball in a box,
withdrew it, replaced it, and eventually left the room. A second
adult who had witnessed the actions of the first adult then moved
the ball to a cup and covered it with a lid. The first adult returned
to retrieve the ball, but before her making any attempt to retrieve
it, she was provided with information about its new location by the
second adult: “The ball is in the cup.” Alternatively, the second
adult made an uninformative remark that did not indicate the
ball’s new location: “I like the cup.” In the informative condition,
infants expressed surprise (looked longer) when the returning
adult searched in the now empty box rather than in the cup where
she had just been told that the ball was located. In the un-
informative condition, by contrast, infants were more surprised if
the returning adult appeared to know that the object was in the
cup, as indexed by her searching there rather than in the box
where she had left it. Moreover, in line with the findings for re-
quests discussed above, a pointing response by the second adult
was also construed by 18-mo-olds as an informative assertion that
was likely to guide the search behavior of the returning adult.
Eighteen-month-olds also display some facility in decoding the
information conveyed by head gestures as well as hand gestures.
As in the studies described above, Fusaro and Harris (24)
arranged for infants to witness a minidialogue between two adults
and then probed their construal of that dialogue. One adult sought
information about the location of a hidden object by pointing to
7886 | www.pnas.org/cgi/doi/10.1073/pnas.1620745114
each of two boxes in succession and asking: “Is it here?” A second
adult answered with a nod to one query and a shake of the head to
the other. When infants were then prompted to find the object,
they typically selected the correct box. Effectively, infants were
able not only to note the difference between the two head gestures
of the second adult but to tie each gesture to a query about a
particular location indicated by the first adult.
Taken together, these findings imply that infants aged 12–
18 mo possess a relatively abstract comprehension of the nature
of communication. They realize that certain signals, such as
pointing gestures, lexicalized speech, and head gestures, can
provide information about what an informant wants or knows.
They expect the recipient of those communicative signals to
construe and respond to them appropriately, by compliance if a
request has been made and by acting in relation to new in-
formation about the state of the world if it has been supplied. By
implication, human infants have a basic understanding of the way
that communication conveys information between two interloc-
utors. Moreover, they do so at an age when their own production
of spoken language remains limited. Accordingly, when infants
proceed to exploit the rich communicative power of language,
they are likely to situate that power within a broad understanding
of the way that human communication operates, particularly
their realization that communication can function to provide an
interlocutor with information.
Granted that young infants understand how communication
operates, we may ask whether they build on that understanding
by actively eliciting information rather than remaining passive
observers or recipients of information that is on offer. To opti-
mize the elicitation of information, it would be helpful for infants
to possess four interrelated abilities: the ability to signal their
own ignorance, to talk about knowledge and ignorance, to pro-
duce interrogative acts of communication, and to gauge the ad-
equacy of the replies received. In the following sections, evidence
will be reviewed showing that human children display each of
these abilities in the course of the second and third years, es-
pecially in the context of an ongoing dialogue with an adult.
Signaling Ignorance
Nonhuman animals appear to possess at least some metacognitive
capacity. They are capable of monitoring their own uncertainty in
that they deliberately withhold a response when faced with a dif-
ficult discrimination between different choices (25). There is also
evidence that chimpanzees and young children (aged 27–32 mo)
appropriately seek out additional visual evidence in the context of
uncertainty about the location of a hidden object. For example, if
they have had the opportunity to observe in which of several tubes
a desirable object has been hidden, both species search promptly
in that particular tube. However, if they have not seen the hiding
and do not know in which particular tube the object was hidden,
they are likely to bend their head or body to look inside the
available tubes before searching in the one where the hidden
object can be seen; alternatively, they opt for a smaller reward in a
known location (26, 27). By implication, both chimpanzees and
children realize when they do not know, or have not seen, an
object’s location and act accordingly. They proceed to gather more
location information before searching accurately on the basis of
that newly gathered information, or they opt for a less desirable
object in a known location.
In these cases, neither the chimpanzee nor the child commu-
nicates ignorance to another individual. Rather, in the context of
ignorance, they engage in visual exploration or opt out of
searching. However, recent evidence indicates that human in-
fants are capable of signaling their ignorance. Goupil et al. (28)
trained 20-mo-old infants to ask their caregiver for guidance if
they were uncertain of a hidden object’s location. More specifi-
cally, infants watched as a toy was hidden in one of two opaque
containers. On so-called “possible” trials, the infant observed the
Harris et al.

hiding of the object. By contrast, on so-called “impossible” trials,
the hiding was carried out behind a curtain so that infants could
not tell which container the object was hidden inside. In either
case, the two containers were subsequently occluded for a delay
ranging from 3 to 12 s and then uncovered once more. Infants
were taught to indicate which container they had remembered the
object to be in by pointing to its location. The container indicated
was then moved forward so that the infant could either recover the
toy (if correct) or find the container to be empty (if incorrect).
Note that depending on the nature of the hiding, and on the
length of time that the two containers were subsequently occluded,
trials varied in terms of the likelihood that infants could know and
remember where the object was hidden. Thus, on possible trials,
especially when the delay was short, remembering the object’s
location was relatively easy. However, on possible trials when the
delay was longer, remembering was more difficult. Finally, on
impossible trials, remembering was precluded because the initial
hiding of the object had not been witnessed.
Half the infants were taught in a prior training session to ask
their caregiver for help when needed. In this training session,
their pointing responses on impossible trials were ignored. In-
stead, caregivers waited until infants turned to look at them in
the eyes and then provided help by pushing the correct container
forward and saying: “Here it is, look.” Thus, infants were ef-
fectively taught that, when uncertain of the object’s location, they
should turn to look at their caregiver, who would then help them
to identify the correct container.
Several results showed that infants in the trained group pro-
duced this help-seeking signal in an appropriate fashion (i.e., when
they were unsure of the object’s location). First, compared with
infants in the untrained group, infants in the trained group proved
to more accurate on those occasions when they did point. This
greater accuracy was because, when they were uncertain, instead
of pointing with a considerable risk of error, they were likely to
seek help by looking toward their caregiver. In addition, the
trained infants who asked for help were more likely to do so when
the experimental setup created uncertainty. Thus, they were more
likely to ask for help on impossible compared with possible trials,
and, within the set of possible trials, they were more likely to ask
for help if the containers had been occluded for a longer delay.
Taken together, these findings provide strong evidence that in-
fants aged 20 mo are able to monitor their ignorance or un-
certainty and can learn to signal that uncertainty by gazing at a
potential informant, notably a caregiver, in such circumstances.
Despite the impressive and systematic nature of such un-
certainty monitoring and help seeking, the findings also point to
the critical role of training. More specifically, control infants who
received no initial training did sometimes look at their caregiver.
However, such responses were no more frequent with greater
delay lengths and were no more frequent for impossible com-
pared with possible trials. Thus, even if these gaze responses
were aimed at prompting help from the caregiver, there was no
evidence that they signaled uncertainty because their production
was not positively correlated with the experimental conditions
producing uncertainty. By implication, although 20-mo-olds do
spontaneously look toward a caregiver, and indeed may do so
with the expectation that helpful information will be supplied,
training might be needed if such looks are to be produced in a
strategic fashion to signal uncertainty. More generally, this study
provides persuasive evidence that infants have some awareness
of their own uncertainty or ignorance, echoing findings with
nonhuman primates, but it provides no evidence that they are
prone to signal ignorance or uncertainty in a spontaneous fash-
ion even if it shows that they can be trained to do so.
When do young children begin to signal their ignorance
spontaneously? Limited observational evidence suggests that in
the course of the second year, human toddlers will sometimes
spontaneously express their ignorance via a distinctive nonverbal
Harris et al.
COLLOQUIUM
PAPER
flip (or shrug) gesture. Acredolo and Goodwyn (29) report a case
study of a child whose communicative gestures were studied
from the age of 12–17 mo. Starting at the age of 15 mo, the child
produced a gesture that appeared to signal ignorance. She
shrugged her shoulders and flipped the palms of her hands up-
ward and out to the side. However, because this case study was a
study of a single child, it is unclear whether production of this
gesture is widely used to signal ignorance or is produced in only a
small minority of families and by a small minority of children. It
is also unclear whether the child observed by these researchers
was especially precocious in her communication skills or repre-
sentative of the communication patterns displayed by typically
developing toddlers.
To examine these issues, Bartz (30) analyzed data from 64 chil-
dren included in the Language Development Project (31), a longi-
tudinal study of early language development in which the families of
the children constituted a representative sample of the US pop-
ulation in terms of education and socioeconomic status. The project
researchers recorded children’s everyday interactions with care-
givers in their homes every 4 mo from the age of 14 mo onward for
a 90-min period. The recordings from eight successive visits (at 14,
18, 22, 26, 30, 34, 38, and 42 mo of age) were analyzed with the goal
of identifying the age of emergence and prevalence of flip gestures.
Fig. 3 shows the cumulative number of children who had pro-
duced at least one of various types of flip gesture across this 18-mo
period. Fig. 3 also shows the cumulative number of children whom
coders judged to be expressing ignorance via their flip gesture. Fi-
nally, Fig. 3 shows the cumulative number of children who produced
the explicit verbal utterance, “I don’t know.” Inspection of Fig. 3
shows that, consistent with the earlier case study of a single child,
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
the flip gesture expressing ignorance emerged among some children
in the course of the second year. At 22 mo, one-fifth of the sample
had been observed producing a flip to signal ignorance, and by
42 mo, almost half had done so. Verbal statements of ignorance
emerged somewhat later but rose sharply in frequency across the
same period, eventually becoming more widespread.
These findings build on the findings of Goupil et al. (28) by
showing that deliberate teaching and reinforcement are not re-
quired for the production of gestures signaling ignorance. Many
children produce such a signal in the course of everyday in-
teraction outside the laboratory. The results also raise the pos-
sibility that such signals are produced not just in the context of
goal-directed behavior, such as in the search for a hidden object,
but in the context of an ongoing dialogue in which an adult poses
a question that the child is unable to answer. By implication, it
would be wrong to assume that signals of ignorance arise only in
problem-solving contexts where children face a practical di-
lemma or obstacle and turn to an adult for help in resolving it.
70
60
Cumulative number of children
Children who ever produced
50 flips
40 Children who ever produced I
DON'T KNOW flips
30 Children who ever said "I don't
know"
20
10
0
14 18 22 26 30 34 38 42
Fig. 3. Cumulative number of children who ever produced a flip, produced
an I DON’T KNOW flip, or said, “I don’t know” at each age point (14–42 mo).
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7887
The data suggest that expressions of ignorance also occur in the
context of conversation.
In an experimental study, this conclusion was examined more
systematically (30). Children (aged 16–37 mo) were asked a series of
questions by an adult, only some of which they could easily answer
based on their existing knowledge. More specifically, they were
shown a mix of pictures and asked the name for each of the entities
depicted. Some pictures depicted familiar, easy-to-name entities
(e.g., book, bird), whereas others depicted unfamiliar, hard-to-name
entities (e.g., unusual hardware item). The pattern of responding
was different for the unfamiliar entities compared with the familiar
entities. Not only did children make more naming errors and pro-
duce more filled speech pauses (e.g., “umm”), but they were also
more likely to look toward an adult (either the experimenter or
their mother) to ask for information (e.g., “What’s that?”) or to say
“I don’t know.” This differential pattern of responding was apparent
among younger infants (16–27 mo) but was more systematic among
older infants (28–37 mo), especially with respect to filled speech
pauses and requests for information.
Taken together, these studies show that toddlers communicate
their uncertainty in various ways. They communicate by looking
toward an adult and by producing a filled speech pause, a flip
gesture, an explicit affirmation of ignorance, or a question to an
interlocutor. Admittedly, when they look at an adult or produce
a filled pause or a flip, such responses might reflect behavioral
uncertainty rather than metacognitive awareness of ignorance.
However, such a parsimonious interpretation seems less appro-
priate when toddlers begin to affirm their ignorance verbally.
Note also that there was only a modest developmental lag be-
tween the production of flips and the emergence of verbal af-
firmations of ignorance. In the next section, such affirmations
are scrutinized in more detail.
Talking About Knowledge and Ignorance
The scope of children’s metacognitive awareness can be illumi-
nated by analyzing their production of the cognitive verb “know”
in the context of everyday conversations with caregivers (32).
Arguably, young children are aware only of gaps in their
knowledge. They might have little or no awareness of when their
information retrieval processes operate smoothly. For example,
they might register occasions when they cannot readily respond
to questions about an object’s name or location but ignore or fail
to register occasions when they can successfully answer. In this
view, children would be likely to deny that they have knowledge
(“I don’t know. . .”) but unlikely to affirm that they do have
knowledge (“I know. . .”). A further question concerns children’s
insight into the cognitive states of other people. Do they talk
about the ignorance or knowledge of other people in the same
way as they talk about their own, or is there any asymmetry
between talk about the self and talk about others?
To answer these questions, the spontaneous utterances of three
children were analyzed. Two children were English-speaking
(Adam, a middle-class, African-American child and Sarah, a
white, working-class child), whose early language had been recorded
by Brown (33) and his colleagues at regular intervals. The utter-
ances of each child could be retrieved via the child language data
exchange system, CHILDES (34). All utterances produced by the
two children that included the mental verb know from the age of
27 mo (the age at which recordings had begun) to the age of 36 mo
were analyzed. The third child, Qianqian (芊芊), was a Mandarin-
speaking child whose utterances had been recorded and transcribed
from the age of 16 to 39 mo by her mother, a psycholinguist.
Qianqian’s production of the verb zhi1dao4 was analyzed. Similar
to the phrase “know that” in English, zhi1dao4 is an epistemic verb
that is used in the context of factual knowledge. [Note that, in
contrast to English, Mandarin uses a different verb (i.e., hui4) for
the phrase “know how” (as in “know how to dance”) (35)].
7888 | www.pnas.org/cgi/doi/10.1073/pnas.1620745114
Did the three children simply use the word know by echoing its
production in an immediately prior utterance by their interlocutor
or did they introduce the word know into the conversation in an
autonomous fashion? The same pattern emerged for all three
children: The large majority of children’s references to the word
know were autonomous, rather than echoes of their interlocutor’s
prior utterance. Next, utterances were analyzed to determine
whether children referred only to their own cognitive states or also
made references to the cognitive states of their interlocutor or to
the cognitive states of a third party. The majority of references
were indeed to children’s own cognitive states. Nevertheless,
children also referred quite often to the cognitive states of their
interlocutor. By contrast, references to a third party, someone not
participating in the conversation, were rare.
Granted that children talked about their own cognitive states as
well as the cognitive states of their interlocutor, an analysis was
conducted to assess whether the pragmatic function of the utter-
ances was similar or different for these two persons. More spe-
cifically, the proportions of affirmations (“I know. . .” or “You
know. . .”), denials (“I don’t know. . .” or “You don’t know. . .”),
and questions (“Do I know. . .?” or “Do you know. . .?”) that in-
volved a reference to the self compared with the interlocutor were
compared. These proportions varied across the three pragmatic
functions. In the case of affirmations, children produced them
with respect to both the self and their interlocutor. Denials and
questions, by contrast, exhibited a strongly asymmetrical pattern.
Children often denied their own knowledge (“I don’t know. . .”)
but very rarely denied the knowledge of their interlocutor (“You
don’t know. . .”). Conversely, children often asked questions about
their interlocutor’s knowledge (“Do you know. . .?”) but never
asked questions about their own knowledge (“Do I know. . .?”).
This asymmetry in the pattern of production for denials compared
with questions was marked, but it was based on only three chil-
dren. To establish its existence firmly, the utterances of a further
eight English-speaking children drawn from the CHILDES data-
base were also analyzed (36). An identical pattern emerged for all
eight children: Denials were almost invariably produced with re-
spect to the self rather than the interlocutor, whereas questions
were invariably produced with respect to the interlocutor rather
than the self.
Returning to the two questions guiding this study, the data
show that 2-y-olds do not simply talk about their ignorance. They
also affirm that they possess particular items of knowledge. In
addition, the pattern of talk about the self is different from the
pattern of talk about the interlocutor: Denials of knowledge are
frequent for the self (“I don’t know”) but not for the in-
terlocutor, and questions about knowledge are frequent for the
interlocutor (“Do you know?”) but not for the self.
The exact explanation for this asymmetry warrants further
investigation (36), but its existence points to the following pos-
sibility for early communication between young children and
their interlocutors. On the one hand, children monitor their own
cognitive states: They are aware of knowing some items of in-
formation and affirm possessing that knowledge, and they are
also aware of lacking other items of information and deny having
that knowledge. Their monitoring of other people’s knowledge is
more circumspect. They sometimes affirm, but almost never
deny, that an interlocutor knows something. Rather, they ask an
interlocutor about what he or she knows. Given children’s
awareness of what they do not know (as indexed by their explicit
denials) combined with their receptivity to the possibility that an
interlocutor might know (as indexed by their questions), and
given also their understanding of the way that communication
can pass knowledge between an informant and a recipient, it is
feasible for them to turn to other people for information when
they do not know something. In particular, it would make sense
for them to ask information-seeking questions. In the next sec-
tion, we review the onset of such questions.
Harris et al.

The Onset of Information-Seeking Questions
A long tradition of developmental research has investigated the
emergence of joint attention in infancy: the capacity to turn the
head and eyes toward a target that is pointed out by a caregiver and
the reciprocal capacity to call a caregiver’s attention to objects via
pointing. The emergence of pointing follows a stable developmental
timetable. At around 8 mo, whole-handed pointing starts to
emerge, and at around 11 mo, index finger pointing starts to emerge
across markedly different cultural settings (37). Despite this stable
timetable, more hand gestures, including pointing gestures, were
observed between caregivers and their infants in China than in
Holland, and more were observed in Holland than in the Yucatan
(38). In all three settings, however, pointing was a dyadic or re-
ciprocal mode of engagement. Thus, when a caregiver pointed, the
infant often reciprocated with a point to the same target (within
10 s), and vice versa.
The standard functional interpretation of infant pointing has
been that it serves either to request an out-of-reach object or to
establish joint attention to an object of interest (39). More re-
cently, however, it has been proposed that pointing can serve an
interrogative function (40). Thus, an infant point can imply not
just “I want this” or “Look at this” but also “What is this?” with
the expectation that the interlocutor will respond with pertinent
information. Begus and Southgate (41) report evidence supporting
this emphasis on the interrogative function of pointing. Sixteen-
month-olds proved to be more likely to point to an object if an
available informant appeared to be knowledgeable. A female in-
formant sat facing the infant, and a series of novel objects was
presented behind her but in view of the infant. Infants often
pointed out these novel objects to her. They appeared to be sig-
naling that they wanted her to provide information about the
objects because they pointed out novel objects less often if she had
proven to be an unreliable informant. Thus, they were less likely to
point to the novel objects if she had previously named familiar
objects incorrectly and now appeared unsure of the names of the
novel objects. A follow-up study suggested that the experimenter’s
prior naming errors were especially important in reducing infants’
interrogative points. If the experimenter simply called attention to
the objects (e.g., “Wow, look at this!”) and then appeared unsure
how to name the novel objects, infants still pointed them out.
When infants receive information in the wake of an interrogative
point, how well do they process that information, and do they pro-
cess it more effectively than unsolicited information? To examine
these questions, Begus et al. (42) presented 16-mo-old infants with
two objects at once. When infants pointed to one of the two objects,
the experimenter modeled an action either on the indicated object
or on the alternative object. After a 10-min delay, the demonstration
object was presented again and infants were given an opportunity to
imitate the action they had seen demonstrated. Infants reproduced
the actions demonstrated on the objects they had pointed at signif-
icantly more than those actions demonstrated on the nonchosen
objects. Moreover, this difference emerged even though infants had
been equally attentive visually during the demonstrations on both
types of objects. A follow-up experiment showed that this difference
in copying was due to learning being facilitated when infants’
pointing was responded to rather than hindered when their pointing
was ignored. Thus, even when infants’ pointing was not ignored and
a single object was presented, copying was still inferior to when two
objects were presented and the experimenter consistently offered a
demonstration on the one that infants pointed to. By implication,
infants’ pointing at a given object had been aimed at eliciting in-
formation about it and that information was better encoded than
information they had not aimed to elicit.
Toddlers’ early word learning provides more evidence for the
information-seeking role of pointing. Lucca and Wilbourn (43)
presented 18-mo-olds with pairs of unfamiliar objects, and when
the infants targeted one of them via selective pointing, reaching,
Harris et al.
COLLOQUIUM
PAPER
or looking, the experimenter provided a novel name for the
targeted object. Infants subsequently showed greater learning of
names for the objects they had targeted via pointing compared
with reaching or looking. By implication, infants were especially
receptive to learning a novel name if it was supplied in the wake
of their interrogative point toward it.
This early emerging disposition toward interrogative commu-
nication is not just a subtle behavior whose diagnosis requires the
tools of the laboratory. Chouinard (44) asked parents of infants
aged 12–17 mo and 18–23 mo to keep a diary in which they noted
instances in which they judged their toddler to be asking a
question. Despite the limited verbal ability of the infants, espe-
cially in the younger group, parents had little difficulty in iden-
tifying instances of questions. For example, one younger child
was beside her mother when she was unpacking groceries. The
child noticed a kiwi fruit, a fruit that was novel to her; picked it
up; and, as she showed it to her mother with a puzzled expres-
sion, produced a vocalization “Uh?” The mother interpreted her
communication as a question about the name or identity of the
fruit, roughly: “What’s this?” Further analysis showed that such
“questions” rose in frequency in the course of the second year.
Overall, the available evidence indicates that toddlers use
several means in the course of the second year to prompt an
interlocutor to supply them with information. They use pointing,
showing, and vocalization, either separately or in combination, in
advance of any capacity to formulate a question in words fully.
Monitoring an Informant’s Replies
We have argued that infants understand how communication can
provide information and ask questions when they do not know
something. Indeed, from the age of ∼18 mo onward, children ask an
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
increasing proportion of questions aimed at gathering information
as opposed to questions that make other types of requests (e.g., for
permission, for clarification) (44). When asking such information-
seeking questions, do they monitor the replies that they receive? In
particular, do they differentiate between a satisfactory answer, one
that dispels their ignorance, and an unsatisfactory answer that does
not? To examine this issue, Chouinard (44) looked at what children
said in reply to an informative answer versus an uninformative an-
swer. When adults failed to supply the information they sought,
children were likely to persist with their questions.
Extending this analysis, Frazier et al. (45) focused on the “why” or
“how” questions (i.e., the explanation-seeking questions) of six
English-speaking children whose language had been recorded reg-
ularly from 2–5 y of age. Children reacted differently depending on
whether they received a satisfactory explanation or not. Following a
satisfactory explanation, they were likely to acknowledge their
agreement or to ask a new, follow-up question on the same topic. By
contrast, when they were not given a satisfactory explanation, they
were likely to persist with their initial question or to offer an ex-
planation of their own. A follow-up study confirmed that explanatory
information is also better remembered. Thus, when preschoolers
received an explanation for a puzzling illustration, they were more
likely to remember that information than a nonexplanation. Indeed,
children often misremembered nonexplanations, converting them
into explanations via appropriate elaboration (46).
Conclusions and Implications
In the course of the second year, children begin to communicate
their doubt or ignorance in various ways: through nonverbal
gestures, explicit statements of their ignorance (“I don’t know”),
as well as information-seeking questions. Nonhuman primates
also indicate their uncertainty: They act differently depending on
whether they are sure or unsure of what to do next. In particular,
they suppress responding in a situation where a mistaken re-
sponse would impose costs. However, despite important paral-
lels, notably the implication that all primates are able to monitor
their own level of certainty or doubt, the two bodies of evidence
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7889
also diverge. Toddlers readily express doubt or ignorance in the
context of communication; their flips, “I don’t know” utterances, and
information-seeking questions are directed at an interlocutor. Con-
ceivably, some of these signals could be produced when children are
alone and faced with uncertainty. For example, having searched
unsuccessfully, a toddler might produce a flip gesture, signaling un-
certainty about the object’s location and/or where to search next.
However, pending more evidence of children’s expressive gestures
when they are alone, it is plausible that the majority of such meta-
cognitive signals are produced in a communicative context, not as an
adjunct to ongoing solo behavior. More generally, these signals ap-
pear to serve two interwoven functions. First, they provide an answer
to an interlocutor’s question: an admission of ignorance that
amounts to a well-formed turn in an ongoing conversation. Second,
when formulated as questions, they convey not just ignorance but
also a request that the interlocutor offer help by supplying missing
information. That children aim to elicit missing information with
their questions is underlined by their differential reactions to in-
formative vs. uninformative replies.
Young children’s facility in communicating their ignorance and in
asking questions appears to build on their foundational insight into
the way that testimony works (20). As described earlier, infants aged
12–18 mo understand that someone who lacks information (e.g.,
regarding the location of an object) can be provided with that in-
formation via the gestures or vocalization of an informant. The
present review indicates that toddlers and young children go beyond
that basic insight. They produce avowals of ignorance and adopt an
interrogative stance. Moreover, the interrogative stance appears to
involve not simply the seeking of information from others via
pointing and/or questions but an accompanying state of in-
formational receptivity (i.e., a motivational readiness to encode and
retain the information thereby elicited). Thus, information about
object names or object functions is better retained if it is received in
1. Spelke ES, Kinzler KD (2007) Core knowledge. Dev Sci 10:89–96.
2. Carey S (2009) The Origin of Concepts (Oxford Univ Press, New York).
3. Gopnik A, Wellman HM (2012) Reconstructing constructivism: Causal models, Bayes-
ian learning mechanisms, and the theory theory. Psychol Bull 138:1085–1108.
4. Wellman HM (2014) Making Minds (Oxford Univ Press, New York).
5. Harris PL, Koenig MA (2006) Trust in testimony: How children learn about science and
religion. Child Dev 77:505–524.
6. Harris PL (2012) Trusting What You’re Told: How Children Learn from Others (Harvard
Univ Press, Cambridge, MA).
7. Legare CH, Harris PL (2016) The ontogeny of cultural learning. Child Dev 87:633–642.
8. Harris PL, Corriveau KH (2011) Young children’s selective trust in informants. Philos
Trans R Soc Lond B Biol Sci 366:1179–1187.
9. Corriveau K, Harris PL (2009) Choosing your informant: Weighing familiarity and re-
cent accuracy. Dev Sci 12:426–437.
10. Lucas AJ, et al. (December 29, 2016) The development of selective copying: Children’s
learning from an expert versus their mother. Child Dev, 10.1111/cdev.12711.
11. Koenig MA, Clément F, Harris PL (2004) Trust in testimony: Children’s use of true and
false statements. Psychol Sci 15:694–698.
12. Koenig MA, Harris PL (2005) Preschoolers mistrust ignorant and inaccurate speakers.
Child Dev 76:1261–1277.
13. Kinzler KD, Corriveau KH, Harris PL (2011) Children’s selective trust in native-accented
speakers. Dev Sci 14:106–111.
14. Chen EE, Corriveau KH, Harris PL (2013) Children trust a consensus composed of
outgroup members–but do not retain that trust. Child Dev 84:269–282.
15. Corriveau KH, Harris PL (2010) Preschoolers (sometimes) defer to the majority in
making simple perceptual judgments. Dev Psychol 46:437–445.
16. Corriveau KH, Fusaro M, Harris PL (2009) Going with the flow: Preschoolers prefer
nondissenters as informants. Psychol Sci 20:372–377.
17. Morgan TJH, Laland KN, Harris PL (2015) The development of adaptive conformity in
young children: Effects of uncertainty and consensus. Dev Sci 18:511–524.
18. Halberda J, Feigenson L (2008) Developmental change in the acuity of the “Number
Sense”: The Approximate Number System in 3-, 4-, 5-, and 6-year-olds and adults. Dev
Psychol 44:1457–1465.
19. Boyd R, Richerson PJ (1985) Culture and the Evolutionary Process (Univ of Chicago
Press, Chicago).
20. Harris PL, Lane JD (2014) Infants understand how testimony works. Topoi (Dordr) 33:
443–458.
21. Krehm M, Onishi KH, Vouloumanos A (2014) Infants under 12 months understand
that pointing is communicative. J Cogn Dev 15:527–538.
22. Martin A, Onishi KH, Vouloumanos A (2012) Understanding the abstract role of
speech in communication at 12 months. Cognition 123:50–60.
7890 | www.pnas.org/cgi/doi/10.1073/pnas.1620745114
response to an interrogative gesture rather than supplied in an
unsolicited fashion (42, 43), and explanatory information is better
retained than nonexplanatory information (46).
Research on the child’s theory of mind has highlighted the
important role of language and conversation in promoting child-
ren’s insight into the way that the mind works (47). The primary
focus of that research has been on children’s developing insight
into false beliefs. The present findings point to a more basic in-
sight that conversation is likely to promote. In several of the
studies reported above, children signaled their ignorance in the
context of an ongoing conversation with an adult. A speculative
but plausible implication is that involvement in conversation can
serve as a constant tutorial for children with respect to the range
and depth of their ignorance. To the extent that children are
prone to engage in conversation with better informed interlocu-
tors, they are likely to discover that their existing knowledge is
limited and fragmentary, albeit open to expansion if they pose
appropriate questions. Granted that children vary in the quantity
of speech that they are exposed to by their caregivers (48) in the
extent to which that speech is directive rather than discursive (49),
is tightly focused on the immediate situation or includes an ex-
ploration of situations and events displaced from the here and now
(50), and includes satisfactory answers to children’s causal ques-
tions (51), we can anticipate that children will grow up with
markedly different assessments of the scope of human knowledge,
the magnitude of their own comparative ignorance, and the po-
tential role of question asking in mitigating that ignorance.
ACKNOWLEDGMENTS. Collection and coding of the data displayed in Fig. 3
were supported by the Eunice Kennedy Shriver National Institute of Child
Health and Human Development of the NIH under Award P01HD040605
(Principal Investigator, Susan Goldin-Meadow). The content is solely the re-
sponsibility of the authors and does not necessarily represent the official
views of the NIH.
23. Song H-J, Onishi KH, Baillargeon R, Fisher C (2008) Can an agent’s false belief be
corrected by an appropriate communication? Psychological reasoning in 18-month-
old infants. Cognition 109:295–315.
24. Fusaro M, Harris PL (2013) Dax gets the nod: Toddlers detect and use social cues to
evaluate testimony. Dev Psychol 49:514–522.
25. Smith JD (2009) The study of animal metacognition. Trends Cogn Sci 13:389–396.
26. Call J, Carpenter M (2001) Do apes and children know what they have seen? Anim
Cogn 3:207–220.
27. Neldner K, Collier-Baker E, Nielsen M (2015) Chimpanzees (Pan troglodytes) and
human children (Homo sapiens) know when they are ignorant about the location of
food. Anim Cogn 18:683–699.
28. Goupil L, Romand-Monnier M, Kouider S (2016) Infants ask for help when they know
they don’t know. Proc Natl Acad Sci USA 113:3492–3496.
29. Acredolo LP, Goodwyn SW (1985) Symbolic gesturing in language development: A
case study. Hum Dev 28:40–49.
30. Bartz DT (2017) Young children’s meta-ignorance. EdD thesis (Harvard University,
Cambridge, MA).
31. Goldin-Meadow S, et al. (2014) New evidence about language and cognitive develop-
ment based on a longitudinal study: Hypotheses for intervention. Am Psychol 69:
588–599.
32. Harris PL, Yang B, Cui Y (2017) “I don’t know”: Children’s early talk about knowledge.
Mind Lang 32:283–307.
33. Brown R (1973) A First Language (Allen & Unwin, London).
34. MacWhinney B, Snow C (1985) The child language data exchange system. J Child Lang
12:271–295.
35. Tardif T, Wellman HM (2000) Acquisition of mental state language in Mandarin- and
Cantonese-speaking children. Dev Psychol 36:25–43.
36. Harris PL, Ronfard S, Bartz D (2017) Young children’s developing conception of
knowledge and ignorance: Work in progress. Eur J Dev Psychol 14:221–232.
37. Liszkowski U, Brown P, Callaghan T, Takada A, de Vos C (2012) A prelinguistic gestural
universal of human communication. Cogn Sci 36:698–713.
38. Salomo D, Liszkowski U (2013) Sociocultural settings influence the emergence of
prelinguistic deictic gestures. Child Dev 84:1296–1307.
39. Bates E, Camaione L, Volterra V (1975) The acquisition of performatives prior to
speech. Merrill Palmer Q Behav Dev 21:205–226.
40. Southgate V, van Maanen C, Csibra G (2007) Infant pointing: Communication to co-
operate or communication to learn? Child Dev 78:735–740.
41. Begus K, Southgate V (2012) Infant pointing serves an interrogative function. Dev Sci
15:611–617.
42. Begus K, Gliga T, Southgate V (2014) Infants learn what they want to learn: Re-
sponding to infant pointing leads to superior learning. PLoS One 9:e108817.
Harris et al.

43. Lucca K, Wilbourn MP (December 29, 2016) Communicating to learn: Infants’ pointing
gestures result in optimal learning. Child Dev, 10.1111/cdev.12707.
44. Chouinard MM (2007) Children’s questions: A mechanism for cognitive development.
Monogr Soc Res Child Dev 72:1–112; discussion 113–126.
45. Frazier BN, Gelman SA, Wellman HM (2009) Preschoolers’ search for explanatory in-
formation within adult-child conversation. Child Dev 80:1592–1611.
46. Frazier BN, Gelman SA, Wellman HM (2016) Young children prefer and remember
satisfying explanations. J Cogn Dev 17:718–736.
47. Harris PL, de Rosnay M, Pons F (2005) Language and children’s understanding of
mental states. Curr Dir Psychol Sci 14:69–73.
Harris et al.
COLLOQUIUM
PAPER
48. Huttenlocher J, Vasilyeva M, Waterfall HR, Vevea JL, Hedges LV (2007) The varieties of
speech to young children. Dev Psychol 43:1062–1083.
49. Hart B, Risley T (1992) American parenting of language-learning children: Persisting
differences in family-child interactions observed in natural home environments. Dev
Psychol 28:1096–1105.
50. Rowe ML (2012) A longitudinal investigation of the role of quantity and quality of
child-directed speech in vocabulary development. Child Dev 83:1762–1774.
51. Kurkul KE, Corriveau KH (January 27, 2017) Question, explanation, follow-up: A
mechanism for learning from others? Child Dev, 10.1111/cdev.12726.
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7891
Changes in cognitive flexibility and hypothesis search
across human life history from childhood to
adolescence to adulthood
Alison Gopnika,1, Shaun O’Gradya, Christopher G. Lucasb, Thomas L. Griffithsa, Adrienne Wentea, Sophie Bridgersc,
Rosie Aboodyd, Hoki Funga, and Ronald E. Dahle
a
Department of Psychology, University of California, Berkeley, CA 94720; bSchool of Informatics, University of Edinburgh, Edinburgh EH1 2QL, United
Kingdom; cDepartment of Psychology, Stanford University, Stanford, CA 94305; dDepartment of Psychology, Yale University, New Haven, CT 06520;
and eSchool of Public Health, University of California, Berkeley, CA 94720
Edited by Andrew Whiten, University of St. Andrews, St. Andrews, United Kingdom, and accepted by Editorial Board Member Andrew G. Clark April 8, 2017
(received for review January 18, 2017)
How was the evolution of our unique biological life history related
to distinctive human developments in cognition and culture? We
suggest that the extended human childhood and adolescence
allows a balance between exploration and exploitation, between
wider and narrower hypothesis search, and between innovation
and imitation in cultural learning. In particular, different develop-
mental periods may be associated with different learning strate-
gies. This relation between biology and culture was probably
coevolutionary and bidirectional: life-history changes allowed
changes in learning, which in turn both allowed and rewarded
extended life histories. In two studies, we test how easily people
learn an unusual physical or social causal relation from a pattern of
evidence. We track the development of this ability from early
childhood through adolescence and adulthood. In the physical
domain, preschoolers, counterintuitively, perform better than
school-aged children, who in turn perform better than adolescents
and adults. As they grow older learners are less flexible: they are
less likely to adopt an initially unfamiliar hypothesis that is
consistent with new evidence. Instead, learners prefer a familiar
hypothesis that is less consistent with the evidence. In the social
domain, both preschoolers and adolescents are actually the most
flexible learners, adopting an unusual hypothesis more easily than
either 6-y-olds or adults. There may be important developmental
transitions in flexibility at the entry into middle childhood and in
adolescence, which differ across domains.
causal reasoning | social cognition | cognitive development | adolescence |
life history
O ne of the most distinctive biological features of human
beings is our unusual life history. Compared with our closest
primate relatives, we have a dramatically extended childhood,
including an exceptionally long middle childhood and adoles-
cence. Moreover, humans have shorter interbirth intervals than
our closest primate relatives, producing an even greater number
of less-capable children (1). There is evidence for other human
adaptations that helped cope with this flood of needy young. In
contrast to our closest primate relatives, human children enjoy
the benefits of care from three sources in addition to biological
mothers: pair-bonded fathers (2), alloparents (3), and post-
menopausal women, in particular, grandmothers (4).
It may seem evolutionarily paradoxical that humans would have
developed a life history that includes such expensive and vulner-
able young for such a long period. However, across many different
species, including birds and both placental and marsupial mam-
mals, there is a very general (although not perfect) correlation
between relative brain size, intelligence and a reliance on learning,
and an extended period of immaturity (5–7). This correlation
suggests a relation between our distinctive human life history and
our equally distinctive large brains and reliance on learning, par-
ticularly cultural learning. Such a relation between biology and
7892–7899 | PNAS | July 25, 2017 | vol. 114 | no. 30
culture would have been coevolutionary and bidirectional: life-
history changes allowed changes in cultural learning, which in
turn both allowed and rewarded extended life histories. In this
way, culture could have extended biology.
A number of researchers have suggested that our life history is
related to our learning abilities (8–10). But what might this re-
lation be like in more detail? It is possible that the extended
human childhood and adolescence is simply a waiting period in
which a large brain can grow or cultural learning can take place
(11). However, both developmental psychology and neuroscience
suggest that there may be more substantive differences in
learning and plasticity in different developmental periods, dif-
ferences that could contribute to human intelligence and culture.
We argue that there may be a developmental trade-off be-
tween cognitive abilities that allow organisms to learn the
structure of a new physical or social environment, abilities that
are characteristic of children, and the more adult abilities that
allow skilled action on a familiar environment. Empirical evi-
dence suggests that children may sometimes be better, and
particularly more flexible, learners than adults. Ideas from the
literature on developmental neuroscience, machine learning, and
cultural learning may help to characterize and explain these
developmental differences more precisely.
We go on to test these ideas by examining cognitive flexibility
across the developmental periods of preschool, middle-childhood,
adolescence, and adulthood, in both the physical and social domain.
When Younger Learners Do Better. Younger learners usually have
more difficulty with cognitive tasks than older children and
adults. Young children have characteristic deficits in executive
function, working memory, attentional focus, and control (12,
13). These are precisely the same abilities required for per-
forming complex skilled actions swiftly and effectively in adult-
hood. Indeed, human children are so dependent on others partly
because of their deficits in these areas.
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: A.G., S.O., C.G.L., T.L.G., A.W., and R.E.D. designed research; A.G.,
S.O., A.W., S.B., R.A., and H.F. performed research; A.G., S.O., and C.G.L. analyzed data;
and A.G. and S.O. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. A.W. is a guest editor invited by the Editorial
Board.
1
To whom correspondence should be addressed. Email: gopnik@berkeley.edu.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1700811114/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1700811114

However, at the same time that their executive abilities are so
limited, human children learn a tremendous amount about the
world easily and rapidly. They quickly and spontaneously learn
about the causal structure of their physical and social environ-
ments, constructing intuitive theories of the physical, biological,
and psychological world (e.g., ref. 14).
There is also empirical evidence that younger learners sometimes,
counterintuitively, actually outperform older ones on learning tasks,
showing more flexibility. Younger mice learn to reverse a learned
rule more easily than postpubertal mice (15). Older monkeys show
neural plasticity when they learn an auditory or tactile pattern, but
only when the pattern is relevant to their goals; juveniles extract the
patterns and demonstrate plasticity independently of goals (16).
Among humans, younger learners are more able to learn new
linguistic distinctions than older learners (17, 18) and they are
better at imagining new uses for a tool (19). Younger children
also remember information that is outside the focus of goal-
directed attention better than adults and older children (20, 21).
We have recently found that preschool children also outperform
older children and adults on abstract social (22) and physical (23)
causal learning problems (24). In particular, younger learners are
more likely to infer an initially unlikely causal hypothesis from a
pattern of evidence. These kinds of causal learning are especially
relevant for human evolution. Theories of the evolution of cogni-
tion stress the adaptive value of human abilities to learn both the
psychological and social causal relationships that are involved in
“theory of mind” and “Machiavellian intelligence,” and the physi-
cal causal relationships that underpin tool use (25, 26).
These findings suggest empirically that children might be es-
pecially flexible learners. But why would this be?
Neuroscience: Trade-Offs Between Executive Function and Plasticity.
Neuroscientists have investigated the origins of both the in-
creased executive control and decreased plasticity that come with
age. One set of developments involves synaptic changes. In the
early period of development, many more new synaptic connec-
tions are made than in adulthood. With age some of these neural
connections are strengthened but others are pruned, trans-
forming a more flexible, sensitive, and plastic brain into a more
effective and controlled one (27, 28).
Increasing executive control is also related to the development
of prefrontal areas of the brain and their increasing connection
to other brain areas. However, neuroscientists have also argued
that strong frontal control has costs for exploration and learning
(29). Interference with prefrontal control areas through trans-
cranial direct current stimulation leads to a wider range of re-
sponses on a “divergent thinking” task (30), and during learning
there is a characteristic release of frontal control (31).
The adolescent brain undergoes particular changes. There is
significant maturational development in prefrontal areas and in
areas thought to be involved in self-perception and social cog-
nition (32), which may indicate increased plasticity. However,
there is also evidence for enhanced consolidation and pruning in
adolescence (33), which might suggest a period of less flexibility.
Computation: Trade-Offs Between Exploitation and Exploration, and
Narrow and Broad Search. The trade-off between executive func-
tion and plasticity in the neuroscience literature parallels an-
other trade-off that appears in machine learning. Reinforcement
learning algorithms make an important distinction between pe-
riods of exploration, in which the system gathers information
about potential actions and outcomes, and exploitation, in which
information gathering is replaced by taking the actions most
likely to maximize reward (34). Human life histories can be
interpreted as a unique solution to the explore/exploit tension,
with low executive control and high plasticity early in life maxi-
mizing exploration, and increased executive function and lower
plasticity maximizing reward as we switch to exploitation.
Gopnik et al.
COLLOQUIUM
PAPER
The different strategies that learners might engage in—and
their consequences for those learners—can also be characterized
more precisely by considering cognitive development from the
perspective of a probabilistic model approach to cognition. This
approach, inspired by statistical methods that are widely used in
artificial intelligence and machine learning, has become in-
creasingly influential in cognitive science (e.g., refs. 35–40).
This approach applies particularly naturally to learning the causal
structure of the environment. Probabilistic models of cognition use
sophisticated causal models to specify the probability of observing a
particular statistical pattern of evidence if a causal hypothesis is true
(41). This makes it possible to use Bayesian inference to determine
the probability that the hypothesis is true given that evidence (42).
Rather than simply generating a yes or no decision about whether a
particular hypothesis is true, Bayesian inference evaluates multiple
hypotheses and assigns probabilities to those hypotheses (14, 35–
40). Many studies have presented children with evidence patterns
and alternative hypotheses that might explain those patterns, and
found that children characteristically choose hypotheses that
Bayesian inference suggests should be more probable (14).
However, Bayesian inference comes at a cost: the significant
computational cost of evaluating hypotheses. It is impossible for
any system, human or computer, to consider and compare all of
the possible hypotheses relevant to a realistic learning problem.
Computer scientists and statisticians often use “sampling” to
help solve this problem—stochastically selecting some hypothe-
ses rather than others—and there is evidence that people, in-
cluding young children, do something similar (43–45).
The sampling process, however, presents learners with a di-
lemma. A learner can conduct a narrow search, only revising
PSYCHOLOGICAL AND
current hypotheses when the evidence is particularly strong and
COGNITIVE SCIENCES
making small adjustments to accommodate new evidence. This
strategy is most likely to quickly yield a “good enough” solution
that will support immediate effective action. But it also means
that the learner may miss a better alternative that is farther from
the current hypothesis, such as a hypothesis about an unusual
causal relation.
Alternatively, a learner can conduct a more exploratory search,
moving to new hypotheses with only a small amount of evidence,
and trying out hypotheses that are less like the current hypotheses.
This strategy is less efficient if the learner’s starting hypothesis is
reasonably good, and may mean that the learner wastes time con-
sidering unlikely possibilities. But it may also make the learner more
likely to adopt genuinely new solutions.
There is a related contrast in the algorithms that are used in
computer science. Drawing on an analogy to statistical physics,
computer scientists have explored the consequences of using
narrower “low-temperature” versus broader “high-temperature”
searches. Continuing the analogy, “simulated annealing” (46) is
one of the best ways of resolving the tension between these two
strategies. Learners who begin with a broader higher-temperature
search and gradually move to a narrower low-temperature search
are most likely to find the optimal solution, just as in metallurgy
heating a metal and then cooling it leads to the most robust
structure. Moreover, as in physical cases of annealing, there may be
multiple rounds of this process. We have argued for a similar de-
velopmental pattern with early broad exploratory sampling followed
by a later narrower search (23, 24). Our hypothesis is that childhood
and adolescence may be evolution’s way of performing simulated
annealing, and hence resolving the explore/exploit trade-off.
Cultural Learning: Trade-Offs Between Imitation and Innovation. The
causal learning problems where children do better can also be
recast as cultural learning problems, and understood in relation
to the cultural learning literature. Consider a learner who ob-
serves someone else performing a complicated series of actions
with artifacts that produce an effect. The learner might approach
this information in several ways. First, the learner might simply
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7893
reproduce the actions in detail. Alternatively, the learner might
apply existing causal knowledge to the situation, and bring about
the effect more directly. These two forms of learning have been
the focus of the extensive “overimitation” literature, starting with the
classic Horner and Whiten study (47).
Human preschoolers are sensitive to information about phys-
ical events and actor’s intentions in deciding how faithfully to
imitate, and there are also developmental and cultural differ-
ences in how imitation takes place (48–52). Learners of all ages
may use their existing causal and cultural knowledge to interpret
the actions of another person and to decide whether and how
faithfully to imitate those actions.
However, they might also use another person’s demonstration
to discover a new or unexpected causal relationship. For exam-
ple, consider a Pleistocene learner who sees an expert produce a
flake from one side of a rock by hitting it on the other side (53),
or a modern learner who watches an expert swipe to find a photo
on a phone. The learner might simply imitate the demonstrator
exactly. Alternatively, she might use her existing causal knowl-
edge to bring about the result (hitting the rock at the place where
she wants it to flake or using a keyboard command).
However, a learner might also use this information to infer an
unexpected abstract causal principle (distant force or touch ac-
tivation). She could then use this principle to design innovative
actions beyond the demonstration, shaping other tools or trying
other swipes for other commands. This kind of learning would
both enable learners to adopt innovations in an intelligent way
and to create innovations themselves.
This approach also applies to social and psychological causal
learning. Imagine that a learner hears a complex narrative de-
scribing a series of human actions, again a classic cultural, as well
as causal, learning scenario. The learner might simply encode the
actions as they are described, recording what the actors did. She
might interpret those actions in terms of an existing psycholog-
ical schema. Alternatively, she might use the information in the
narrative to infer new psychological or social relations.
As in the physical case, this last option might lead to both the
adoption and creation of social and psychological innovations.
Consider a learner who hears a story in which Sam and John live
together and share a bedroom. She might interpret this story in
terms of her existing cultural schemas (perhaps Sam and John
are close friends with a small apartment). She might also, how-
ever, use the story to make a broader inference about the pos-
sibility of same-sex marriage.
These alternative forms of cultural learning exemplify the
explore/exploit tension. The first two strategies, namely, exact
imitation or reliance on causal knowledge, are likely to lead to
quick and mostly effective actions. Entertaining the unlikely new
causal relation is both more cognitively demanding and more
risky. In the long run, however, it may confer an advantage in
dealing with changing and variable environments.
Human learners of all ages may use all these strategies to some
extent. However, our hypothesis is that learners at different
developmental stages may be more or less likely to use different
strategies. In particular, more protected and more behaviorally
variable younger learners may be more likely to adopt new hy-
potheses than older learners. In fact, the causal learning tasks in
our earlier research, in which younger learners do better than
older ones, involve precisely these kinds of scenarios. Learners
infer a new causal relation from a demonstration or narrative.
This developmental difference may also help resolve the ten-
sion between imitation and innovation in cultural learning (48).
Human children are adept at imitation. However, the flexibility
of childhood cognition may also help allow innovations to be
adopted and to spread. Young children are rarely the source of
complex technical innovations; actually designing and producing
an effective tool, for example, is a challenging task that requires
both innovation and executive skill (48, 54). However, innovations
7894 | www.pnas.org/cgi/doi/10.1073/pnas.1700811114
that are effortful and rare when they first appear within a gener-
ation can become effortlessly and widely adopted by the next
generation. In fact, among nonhuman animals, cultural innovations
are often first produced, adopted, and spread by juveniles (55–58).
Continuous Knowledge Acquisition vs. Discontinuous Developmental
Transition. There are two complementary mechanisms that might
lead to a developmental shift from broader exploration to nar-
rower exploitation. One is simply the accumulation of knowledge
itself. As we learn more and grow more confident in our beliefs,
we are less likely to change those beliefs. From a Bayesian per-
spective, development proceeds from a relatively “flat” prior,
where different hypotheses have more similar probabilities, to a
more “peaked” distribution, where some hypotheses are much
more likely than others, as a learner accumulates knowledge. In
Bayesian models a flatter prior would automatically lead to
broader search.
Another complementary possibility, building on the literature
discussed above, is that maturation and general experience lead
to different degrees of plasticity and flexibility and different
search strategies, independent of accumulated knowledge. There
might be nonlinear changes at points of developmental transi-
tion, such as the transition from early to middle childhood at
around 6 y or in adolescence, rather than a simple continuous
change with accumulated knowledge.
In particular, although adolescents have more accumulated
experience than younger children, there is evidence, as noted
above, that adolescence may also be a period of enhanced
plasticity and learning (59, 60), especially for social domains (32,
61), in part through the privileging of social information pro-
cessing and the salience of social rewards in decision making (62,
63). Cultural innovations, such as new socially significant forms
of language, dress, or music often first appear in adolescents.
Adolescence might be an extra round of annealing in the social
sphere. However, there is also evidence that adolescence may be
a period of pruning and consolidation.
In fact, two contrasting developmental patterns characterize ad-
olescence (64, 65). On some measures, such as cognitive control,
and self-regulation, there is a relatively linear trajectory from
childhood through adolescence to adulthood. On others, such as
sensation-seeking and risk-taking, both forms of exploration, there
is a marked increase associated with the onset of puberty, and an
inverted U pattern peaking in adolescence and then declining.
There is extensive research on risk-taking and decision-making in
adolescence but, to our knowledge, no research on causal learning.
Current Studies. We approach these questions by extending two
earlier causal learning experiments. Where the original experiments
contrasted preschoolers with either 6-y-old children or adults, we
report results covering the entire developmental span from pre-
school to adulthood, with special focus on the transition to middle
childhood and adolescence, periods not explored previously. This
approach allows us to explore learning across human life history,
and to ask whether there are distinctive developmental transitions.
Both experiments have the same logic. We contrast two hy-
potheses about how objects or people work: one that is initially
more likely, at least for adults, and one that is more unusual. In
Exp. 1 we contrast the hypothesis that individual objects activate
a machine with the hypothesis that particular combinations of
objects do. In Exp. 2 we contrast the hypothesis that someone
took a risk because of their personal traits with the hypothesis
that they took the risk because of the situation they encountered.
In one condition, participants receive covariation evidence that
supports the likely hypothesis. In a second, otherwise identical con-
dition, they receive covariation evidence that supports the unlikely
hypothesis. In a third baseline condition, participants do not receive
evidence either way. We record whether participants of different ages
adopt the likely or unlikely hypothesis in each condition.
Gopnik et al.

The different conditions allow us to control for alternative
factors that might influence performance on these tasks. In the
first two conditions, supporting the likely hypothesis or the un-
likely one, the participants see similar agents perform similar
actions on similar objects; all that differs is the covariation be-
tween causes and effects. Moreover, both conditions require that
the learner attend to and use the particular pattern of data
presented in the demonstration. Whether they adopt the likely or
unlikely hypothesis, the learner still has to attend to the specific
details of the evidence to answer correctly. Differences in per-
formance, then, should reflect differences in causal learning
rather than more general information-processing, linguistic, or
motivational factors.
Exp. 1: Reasoning About the Causes of Physical Events
In an earlier study, Lucas et al. (23) found that, across three
different experiments, with different participants and designs,
preschool children learned an unusual abstract physical causal
relationship but adults had difficulty.
In the second experiment of that study, preschool children and
adults were presented with a machine that lights up when you place
certain patterns of blocks on top, and were told that “blicketness”
makes the machine go. First, in a training trial, participants saw
unambiguous covariation evidence suggesting that the machine
operated according to a general logical rule. In one condition, the
machine operated on a disjunctive “or” rule: each block either ac-
tivated the machine or did not. Accounts of adult causal reasoning
suggest that this disjunctive rule is the default assumption for adults
(e.g., ref. 66). In the other condition, the machine operated on a
more unusual conjunctive “and” rule: two blocks had to be placed
on the machine at the same time to make it activate. Four-year-old
children and adults in both conditions then saw an ambiguous test
trial with new blocks that was consistent with either general prin-
ciple. In a baseline condition, participants only saw the ambiguous
trial without the training trials. In each condition, participants were
then asked whether each block was or was not a “blicket” and were
asked to activate the machine.
Children learned the appropriate general rule in each condi-
tion and applied it to the ambiguous case. Adults applied the
default disjunctive rule in the ambiguous case even when the
earlier evidence weighed against it.
In Exp. 1 we used exactly the same methods across the entire
developmental range, including 6- to 7-y-olds, 9- to 11-y-olds, and
12- to 14-y-olds. Fig. 1 provides a visual display of the pattern of
evidence used for training and test trials.
We extended the contrast between preschoolers and adults to
include school-aged children and adolescents. This approach
A B C AB AC BC
A B C AB AC BC
Test Trials
D D D E DF DEF DF
Fig. 1. Schematic of the procedure for Exp. 1. The yellow rectangle repre-
sents the machine’s activation. “Disjunctive” training provides evidence of
the more common, disjunctive hypothesis. “Conjunctive” training provides
support for the less common conjunctive hypothesis. “Test” trials presented
ambiguous evidence about the “D” object.
Gopnik et al.
COLLOQUIUM
PAPER
Fig. 2. Proportion of participants labeling test objects as “blickets” with SEs.
allowed us to examine the transitions from early to middle
childhood, from middle childhood to adolescence, and from
adolescence to adulthood. Would there be differences between
preschoolers and school-age children? Would adolescents be less
flexible and more like adults? Or might they be more flexible
than school-aged children and adults with the inverted U pat-
tern? Finally, would there be a continuous change as children
accumulated more knowledge or more discontinuous changes at
developmental transitions?
Results.
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
Blicket judgments. We combined new data collected from younger
school-aged children (6- to 7-y-olds), older preadolescent chil-
dren (9- to 11-y-olds), and young adolescents (12-to 14-y-olds)
with the data from 4-y-olds and adults tested with the identical
method in Lucas et al. (23).
If the observers believe the machine operates on an unusual
conjunctive rule, requiring multiple blickets to operate, they
should say that F, D, and possibly E are blickets and use multiple
objects to make the machine go. If observers believe that the
machine works on the “disjunctive” rule, in contrast, they should
say that F is a blicket but that D and E are not and put single
objects on the machine. [The evidence that E is a blicket is less
strong than the evidence for D, so participants should be less likely
to say that E is a blicket than D (23).] (See SI Appendix, Table S5
for analysis of E judgments, consistent with these predictions.)
Fisher’s exact tests revealed no significant differences between
conditions or ages for the unambiguous F object; as predicted all
of the age groups in all of the conditions said that F was a blicket
(means ranged from 0.7 to 0.96).
Fig. 2 presents the proportion of participants in each age
group labeling the critical D test object as a blicket by condition.
Because the dependent measure is a binary response, we used
comparisons of generalized linear models to identify the statis-
tical model with the best fit to the data. Results of model com-
parisons can be found in SI Appendix, Table S4.
A model predicting the binary D judgment from condition and
age group with no interactions was best fit to the data. Post hoc
tests using Tukey’s honest significant differences (HSD) for D
object judgments revealed a significant difference between the
conjunctive (M = 0.52, SE = 0.02) and the disjunctive (M = 0.13,
SE = 0.01; t = −0.391, P < 0.001) and baseline (M = 0.15, SE =
0.01; t = −0.374, P < 0.001) conditions, and there was no sig-
nificant difference between the disjunctive and baseline condi-
tions (t = −0.017, P = 0.923).
In addition to the model comparisons, we conducted planned
comparisons for the theoretically crucial developmental con-
trasts in the critical conjunctive condition, using Fisher’s exact
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7895
tests. These were the transition to school-age (4- vs. 6-y-olds) and
to adolescence (12- to 14-y-olds vs. 6- to 7- and 9- to 11-y-olds,
and vs. adults). Four-year-olds (M = 0.92, SE = 0.01) were sig-
nificantly more likely to label D a blicket than 6- to 7-y-olds (M =
0.56, SE = 0.02; P < 0.01). Six- to 7-y-olds and 9- to 11-y-olds
(M = 0.6, SE = 0.02; P = 1) did not differ but both 6- to 7-y-olds
and 9- to 11-y-olds labeled D as a blicket significantly more than
12- to 14-y-olds (M = 0.28, SE = 0.02; P < 0.05, in both cases).
However, adolescents (12- to 14-y-olds) judgments did not differ sig-
nificantly from the judgments of adults (M = 0.25, SE = 0.02; P = 1.0).
Thus, within the new data collected in this study, we saw some
evidence for both middle childhood and adolescent transitions.
Intervention choices. We also analyzed participants’ choices when
they were asked to activate the machine. Fig. 3 displays the pro-
portion of participants choosing multiple items, indicating that they
thought more than one object was necessary to activate the ma-
chine. There was more variability in this open-ended response than
in the yes/no blicket judgments. However, the general pattern was
similar. In particular, adolescents and adults were more likely to
choose single objects to make the machine go, suggesting that they
had genuinely concluded that the machine worked disjunctively,
and did not simply use the word “blicket” differently than younger
participants.
Again, we used a generalized linear model (see SI Ap-
pendix, Table S6 for details). The model with the best fit to
the data predicted the single vs. multiple object use from con-
dition, age group, and the interaction between condition and
age group.
As with the blicket judgment measure, we made planned
comparisons for the conjunctive condition using Fisher’s exact
tests, focusing on the school-aged and adolescent transitions.
These tests showed that 4-y-olds (M = 0.84, SE = 0.01) were
more likely to use multiple objects to activate the machine than
6- to 7-y-olds (M = 0.53, SE = 0.02; P < 0.05), again suggesting a
middle childhood transition. With this measure, 6- to 7-y-olds
and 9- to 11-y-olds (M = 0.63, SE = 0.02) did not differ signifi-
cantly from 12- to 14-y-olds and adolescents did not differ sig-
nificantly from adults.
Discussion. These results suggest that, in this task, as learners
grow older and have more experience they become less sensitive
to the evidence and more reliant on their prior beliefs. They
increasingly prefer disjunctive explanations to conjunctive ones,
even when the evidence weighs in the opposite direction.
The results from both the blicket judgments and interventions
suggest a developmental transition at the entry to middle child-
hood and the blicket judgment results also suggest a transition at
Fig. 3. Proportion of participants choosing either single or multiple items
for intervention choice with SEs.
7896 | www.pnas.org/cgi/doi/10.1073/pnas.1700811114
adolescence, rather than just a continuous change with in-
creasing knowledge. School-aged children are similar to each
other and less flexible than preschoolers, adolescents and adults
are similar, and both are less flexible than preschoolers and
school-aged children (Fig. 2).
Exp. 2: Reasoning About the Causes of Actions
In the second experiment we turned from physical causality to
social and psychological causality. Classic findings in social psy-
chology show that Western adults attribute actions to the stable
internal personal traits of an actor despite countervailing evi-
dence, the “fundamental attribution error” (67). These findings
suggest that adults rely on existing causal hypotheses rather than
modifying those hypotheses in the face of evidence.
In one study, for example, an experimenter instructed half the
participants in a group to write and read aloud an essay sup-
porting Castro and the other half to write and read an essay
opposing him. Despite the obvious evidence that the essays were
the result of the situation, participants reported that people in
the first group were more left wing than those in the second (68).
Among adults, this trait bias tends to become stronger with age
(69) and it appears to be stronger in some cultures than others:
American and European middle-class participants show a stronger
trait bias than Hong Kong, Mainland Chinese, Japanese, and
Korean participants (70).
How does this bias develop in childhood? Seiver et al. (22)
presented preschool children with a scenario in which two dolls
either played or refused to play on two potentially risky toys. The
covariation evidence supported either a person or situation attri-
bution. Then they asked the children to explain why the actors
played or refused to play on the toys. Four-year-olds accurately
made person or situation attributions depending on the evidence.
Six-year-olds, however, showed a trait bias. They made more
person attributions than 4-y-olds, even when the covariation in-
formation supported a situation attribution. In Exp. 2 we extend
this previous work to study the developmental changes in learning
over childhood and adolescence.
We included an adult sample to ensure that adults would in-
deed show a trait bias in this task. Adding 9- to 11-y-old and 12-
to 14-y-old samples let us test whether the previously discovered
transition from 4- to 6-y-olds was part of a continuous de-
velopmental decline, or reflected a particular transition into
middle childhood.
We could also examine adolescence. Like adults, adolescents
have extensive experience of their particular culture and the trait
assumptions that go with it. There might be a developmental
progression toward the adult pattern, as in Exp. 1. However,
adolescents are also especially sensitive to social information and
strongly motivated to explain peer behavior (59). They might be
more sensitive to social evidence, and more likely to override a
trait bias than adults. We might then expect something more like
the inverted U of risk-taking and sensation-seeking.
Results. We combined data from 9- to 11-y-olds, 12- to 14-y-olds,
and adults with the data from preschoolers and 6-y-old children
presented in Seiver et al. (22). We recorded how often partici-
pants explained the dolls’ actions in terms of situations—the
initially unlikely hypothesis—and used this to assign a “situation”
score from 0 to 2. Fig. 4 shows performance across age in the
person condition, where the evidence supports a trait attribution,
the situation condition, where the evidence supports a situation
attribution, and a baseline condition, which didn’t support either
explanation. Linear regression analyses were used to predict the
attribution score from age group and condition.
Model comparisons showed that the model with the best fit to
the data predicted situation attribution score from age group
and condition, as well as interactions between the two variables
[F(14, 525) = 15.43, P < 0.001, adjusted R2 = 0.273] (details of
Gopnik et al.

the model comparisons can be found in SI Appendix, Table S12).
Fig. 4 plots the average situation attribution score for each age
group by condition.
As in Exp. 1, we also performed planned comparisons for the
crucial age transitions in the situation condition, using t tests.
The critical situation condition revealed whether participants
would adopt the unlikely situation hypothesis given evidence, or
would instead attribute actions to traits as they did in the person
and baseline conditions.
In the Seiver et al. (22) data the 6-y-olds, but not the 4-y-olds,
showed a trait bias in the situation condition, suggesting a tran-
sition at school age. In this experiment, we also tested the ado-
lescent transition by comparing the 12- to 14-y-olds to 6- and
9-y-olds and to adults. The adolescents showed an interesting
pattern, unlike the pattern in Exp. 1, which appeared to be re-
sponsible for the interaction effect in the model. Adolescent re-
sponses in the situation condition differed both from adults and
younger children in an inverted U pattern. In the situation condition
12- to 14-y-olds (t = −4.1048, P < 0.001) made more situation
attributions than adults, and 12- to 14-y-olds also made signif-
icantly more situation attributions than 6-y-olds (t = −2.34, P = 0.02),
although they were not significantly different from 9- to 11-y-olds.
We also performed additional analyses using Tukey’s HSD
test. Participants in both the person (M = 0.2, SE = 0.02; t =
−0.531, P < 0.001) and baseline (M = 0.49, SE = 0.03; t = −0.531,
P < 0.001) conditions provided significantly fewer situation attri-
butions than those in the situation (M = 1.02, SE = 0.03) condition.
There was not a significant difference between the baseline and
person conditions, suggesting a trait bias.
Given the interaction, we also used Tukey’s HSD tests to ex-
amine age differences separately for each condition. There were
no significant age differences in attribution scores in the person
condition; all age groups produced trait explanations when these
explanations were congruent with the data, and rarely made
situation attributions.
The baseline condition allowed us to assess participants’ judg-
ments when no evidence was available (their “prior” in Bayesian
terms). Post hoc Tukey tests revealed that 4-y-olds (M = 0.93,
SE = 0.08) provided significantly more situation attributions than
both 12- to 14-y-olds (M = 0.24, SE = 0.06; t = −0.694, P = 0.001)
and adults (M = 0.38, SE = 0.05; t = −0.55, P = 0.004). Although
both 6-y-olds (M = 0.43, SE = 0.1; t = −0.49, P = 0.09) and 9- to
11-y-olds (M = 0.55, SE = 0.11; t = −0.386, P = 0.49) provided
fewer situation attributions than 4-y-olds, these differences did not
reach statistical significance. This finding suggested that a trait
bias developed around 6 y and was maintained with age.
Discussion. In the person condition, participants of all ages mostly
made trait attribution explanations, in accordance with the evi-
dence. In the baseline condition, with no evidence, there was a
decrease in situation explanations with age. Accumulating ex-
perience may have led to a trait bias.
In the situation condition, in which the learners had to infer
the unusual hypothesis, there was an interesting developmental
reversal, with an inverse U pattern. Twelve- to 14-y-olds were
less likely to make trait attributions than either 6-y-olds or
adults. In other words, although the adolescents had developed
a strong bias to begin with, they overcame that bias when they
received contradictory evidence. The adolescents showed the
largest gap between the baseline condition and the situation
condition.
These findings support the idea that adolescents may be par-
ticularly interested in discovering new social possibilities. This
finding is consistent with the fact that, compared with adults,
adolescents show greater activation in brain regions associated
with self-perception and social cognition (71, 72), and that ad-
olescents are often at the forefront of social change.
Gopnik et al.
COLLOQUIUM
PAPER
Fig. 4. Average attribution scores by age group and condition with SEs. YO,
year old.
Finally, these results suggest that changes in flexibility are not
solely because of the accumulation of knowledge. The adoles-
cents should have accumulated more knowledge than the
younger children and this was reflected in their trait bias in the
baseline condition. However, the adolescents were also the most
flexible social thinkers; they were most able to overcome prior
biases in the face of new evidence.
General Discussion and Conclusions
These results support the suggestion that the extended human
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
period of immaturity allows a period of flexible hypothesis search
in cultural learning. In both studies, we also found some evidence
for developmental transitions, particularly from early to middle
childhood and at adolescence.
The crucial conditions involved cases where the evidence
and the existing hypotheses were in conflict, the conjunctive
condition in Exp. 1 and the situation condition in Exp. 2. In
both studies 4-y-olds and 6- to 7-y-olds were significantly dif-
ferent in these conditions. In both studies, however, we did
not see significant differences between 6- to 7-y-olds and 9- to
11-y-olds.
Similarly, we found evidence for a transition in adolescence in
both studies in these conditions, but this transition went in op-
posite directions. In the physical case, in the conjunctive condi-
tion adolescents were similar to adults but less flexible than
either 6-y-olds or 9- to 11-y-olds. Like adults, the adolescents
seemed reluctant to revise physical knowledge they had already
acquired. In the social case, however, in the situation condition
adolescents were more flexible than either 6-y-olds or adults.
This finding is consistent with the idea that adolescents are more
tuned to the social domain than the physical one, and are willing
to entertain new social possibilities.
These findings also raise the question of the interaction be-
tween biological and environmental factors in the unfolding of
life history. The findings in the baseline conditions suggest that
children are gradually accumulating more knowledge and that
this may play a role in the decline of cognitive flexibility.
However, the discontinuous pattern in the conjunctive and
situation conditions suggests that other factors also play a role.
Biological changes like puberty may play a role in the adolescent
transitions. There may also be more complex interactions be-
tween the changing life experiences that come with different
developmental stages and hypothesis search and flexibility. Ad-
olescence is not only a time of biological change; it is also a time
of new social motivation and experience. Similarly, there is a
complex interaction between biological changes at around 6 y
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7897
and experiences such as school in our culture, or more informal
apprenticeships in cultures without formal schooling.
It is also plausible that a playful protected environment may
lead to more flexible, exploratory and childlike learning, even in
adulthood, and that even in childhood, stressful or resource-poor
environments may lead to less flexibility and a more adult-like
emphasis on exploitation (see, e.g., refs. 73 and 74).
These issues are all worthy of exploration, as are extensions of
these studies to new domains. The physical causal learning re-
sults in Exp. 1 have been replicated in low socioeconomic status
preschoolers in Peru and the United States,* but more extensive
cross-cultural comparisons, including the social tasks and
extending to forager and small-scale agricultural cultures, would
also be important. The current findings do, however, suggest a
relation between biology and culture, in particular between the
distinctive childhood and adolescence of our life history and our
equally distinctive ability to learn about and create new social
and physical environments.
Methods
Data from the new participants in this study can be found on the Open
Science Framework (https://osf.io) under the profile for Shaun O’Grady.
Exp. 1.
Participants. Children aged 6- to 7-y-old, (n = 90), 9- to 11-y-old (n = 90), and 12-
to 14-y-old (n = 86) participated. We combined these new data with that
reported for preschoolers and adults in Exp. 2 of Lucas et al. (23) to com-
pare performance from preschool to adulthood. For all participants in both
experiments reported here, parents provided written informed consent
and the child participants provided either written assent (9- to 14-y-olds) or
verbal assent (4- to 7-y-olds) in accordance with protocols approved by the
University of California, Berkeley Committee for the Protection of
Human Subjects.
Procedure. Participants from each age group were randomly assigned to one
of three conditions: two training conditions (conjunctive and disjunctive
conditions) and a third condition with no training, termed the baseline
condition. In each condition the participants were shown nine different
blocks (A, B, C, A2, B2, C2, D, E, and F). Participants were presented with a
machine and were informed that “blicketness” makes the machine light up
and play music.
In both of the training conditions, the experimenter placed individual blocks or
combinations of blocks on the machine in the same order (Fig. 1). In the con-
junctive condition the machine only activated when the experimenter placed
both A and C on the machine at the same time, providing evidence that supports
a conjunctive rule about the machine’s operation. In the disjunctive condition
the machine activated any time either A or C were placed on the machine,
suggesting that only one of the two blocks was needed. After the two training
trials participants saw one test trial with three new items: D, E, and F. The test
trials provided ambiguous information that could support either the conjunctive
or disjunctive rule (i.e., D and F are both blickets or just F is a blicket). In the
baseline condition, participants were not given any prior training about the rule
*Wente AO, et al., Cognitive Development Society Meeting, October 9–11, 2015, Colum-
bus, OH.
1. Hill K, Kaplan H (1999) Life history traits in humans: Theory and empirical studies.
Annu Rev Anthropol 28:397–430.
2. Chapais B (2009) Primeval Kinship: How Pair-Bonding Gave Birth to Human Society
(Harvard Univ Press, Cambridge, MA).
3. Hrdy SB (2011) Mothers and Others (Harvard Univ Press, Cambridge, MA).
4. Hawkes K, O’Connell JF, Jones NG, Alvarez H, Charnov EL (1998) Grandmothering, men-
opause, and the evolution of human life histories. Proc Natl Acad Sci USA 95:1336–1339.
5. Bennett PM, Harvey PH (1985) Brain size, development and metabolism in birds and
mammals. J Zool 207:491–509.
6. Weisbecker V, Goswami A (2010) Brain size, life history, and metabolism at the
marsupial/placental dichotomy. Proc Natl Acad Sci USA 107:16216–16221.
7. Street SE, Navarrete AF, Reader SM, Laland KN (2017) Coevolution of cultural in-
telligence, extended life history, sociality, and brain size in primates. Proc Natl Acad
Sci USA 114:7908–7914.
8. Bruner JS (1972) Nature and uses of immaturity. Am Psychol 27:687–708.
9. Konner M (2010) The Evolution of Childhood: Relationships, Emotion, Mind (Harvard
Univ Press, Cambridge, MA).
7898 | www.pnas.org/cgi/doi/10.1073/pnas.1700811114
for operating the machine, but instead were presented with two ambiguous test
trials. We recorded results from the second test trial but there were no signifi-
cant differences between them.
The three conditions only differed in the covariation between the blocks
and the machine. In all three conditions, at the end of both training and test
trials, the experimenter pointed to each item individually and asked the
participant if that item was a blicket or not a blicket. Finally, the experi-
menter then gestured to the set of three objects and asked the participant,
“Which of these [gesturing to the three test objects] would you use to turn
on the machine?”
Exp. 2.
Participants. The same 9- to 11-y-olds (n = 90) and 12- to 14-y-olds (n = 86) in Exp.
1 also participated in this experiment. Order of administration of the tasks was
counterbalanced to avoid interference; there were no order effects. An addi-
tional 240 adult participants were recruited for an online version of this exper-
iment via Amazon’s Mechanical Turk. We combined these data with the original
data from Seiver et al. (22) for 4- and 6-y-olds.
Procedure and coding. Participants were randomly assigned to one of three
conditions in which two dolls interacted with two toys. Subjects assigned to the
situation condition saw two dolls play on one toy four times and then saw those
same dolls avoid playing on a second toy four times. This pattern of covariation
should suggest that something about the situation caused the pattern of ac-
tions (i.e., “her friend played on the bicycle” or “the trampoline is danger-
ous”). Those assigned to the person condition saw one doll play on both toys
four times, whereas the other doll avoided playing on both toys four times.
This evidence should suggest that the actions resulted from an inherent
trait of the doll, and produce trait-based explanations, such as “she’s the
type of doll that gets scared/is brave” or “she knows how to ride a bike.”
Finally, in a baseline condition, participants saw one doll play on one toy
four times, whereas the other doll avoided the other toy four times. Partici-
pants in this condition could not rely on covariation information to make
attributions because they had not seen how each doll acted on the other toy.
After they watched the dolls interact with the toys, each participant was asked
why each doll either played or did not play on the second toy.
Explanations referring to an enduring characteristic of the doll were coded
as “person” attributions and were given a score of 0 (e.g., “Because she
might be more brave than the other one”). When an explanation referenced
an aspect of the toy or situation, the response was coded as a “situation”
attribution and given a score of 1 (e.g., “The trampoline doesn’t have any
edges”). Some explanations referred to both personal traits and situational
factors and were coded as “interactions” and given a score of 0.5. See SI
Appendix, Table S9 for a list of example responses by category. Reliability
coding was conducted on 16% of the responses by a second coder who was
blind to condition, and interrater reliability was high (Cohen’s κ = 0.967, P <
0.001). Coded explanation responses for each participant were summed to
provide a “situation” attribution score for each participant.
Analyses. All analyses in both experiments were performed using the R
statistical programming language (75). Preliminary analyses revealed no
effect of block shape, doll name, toy, or the order in which the dolls played.
Linear regression models found no effect of gender of the participants or
the experimenter in either experiment (see SI Appendix, Tables S3 and S11).
ACKNOWLEDGMENTS. This work was funded by the National Science
Foundation Graduate Research Fellowship under Grant DGE 1106400 (to
S.O.); National Science Foundation Grant BCS-331620 (to A.G. and T.L.G.);
and grants from the Bezos Foundation and McDonnell Foundation (to A.G.).
10. Bjorklund DF, Green BL (1992) The adaptive nature of cognitive immaturity. Am
Psychol 47:46–54.
11. Bogin BA, Smith BH (1996) Evolution of the human life cycle. Am J Hum Biol 8:703–716.
12. Munakata Y, Casey BJ, Diamond A (2004) Developmental cognitive neuroscience:
Progress and potential. Trends Cogn Sci 8:122–128.
13. Carlson SM (2005) Developmentally sensitive measures of executive function in pre-
school children. Dev Neuropsychol 28:595–616.
14. Gopnik A, Wellman HM (2012) Reconstructing constructivism: Causal models, Bayes-
ian learning mechanisms, and the theory theory. Psychol Bull 138:1085–1108.
15. Johnson C, Wilbrecht L (2011) Juvenile mice show greater flexibility in multiple choice
reversal learning than adults. Dev Cogn Neurosci 1:540–551.
16. Buonomano DV, Merzenich MM (1998) Cortical plasticity: From synapses to maps.
Annu Rev Neurosci 21:149–186.
17. Werker JF, Hensch TK (2015) Critical periods in speech perception: New directions.
Annu Rev Psych 66:173–196.
18. Kuhl PK (2004) Early language acquisition: Cracking the speech code. Nat Rev
Neurosci 5:831–843.
Gopnik et al.

19. German TP, Defeyter MA (2000) Immunity to functional fixedness in young children.
Psychon Bull Rev 7:707–712.
20. Plebanek DJ, Sloutsky VM (April 1, 2017) Costs of selective attention: When children
notice what adults miss. Psychol Sci, 10.1177/0956797617693005.
21. Sloutsky VM, Fisher AV (2004) When development and learning decrease
memory. Evidence against category-based induction in children. Psychol Sci 15:
553–558.
22. Seiver E, Gopnik A, Goodman ND (2013) Did she jump because she was the big sister
or because the trampoline was safe? Causal inference and the development of social
attribution. Child Dev 84:443–454.
23. Lucas CG, Bridgers S, Griffiths TL, Gopnik A (2014) When children are better (or at
least more open-minded) learners than adults: Developmental differences in learning
the forms of causal relationships. Cognition 131:284–299.
24. Gopnik A, Griffiths TL, Lucas CG (2015) When younger learners can be better (or at
least more open-minded) than older ones. Curr Dir Psychol Sci 24:87–92.
25. Byrne RW (1995) The Thinking Ape: Evolutionary Origins of Intelligence (Oxford Univ
Press, Oxford).
26. Byrne R, Whiten A (1989) Machiavellian Intelligence: Social Expertise and the
Evolution of Intellect in Monkeys, Apes, and Humans (Oxford Univ Press, Oxford).
27. Huttenlocher PR (1990) Morphometric study of human cerebral cortex development.
Neuropsychologia 28:517–527.
28. Huttenlocher PR (2009) Neural Plasticity (Harvard Univ Press, Cambridge, MA).
29. Thompson-Schill SL, Ramscar M, Chrysikou EG (2009) Cognition without control:
When a little frontal lobe goes a long way. Curr Dir Psychol Sci 18:259–263.
30. Chrysikou EG, et al. (2013) Noninvasive transcranial direct current stimulation over
the left prefrontal cortex facilitates cognitive flexibility in tool use. Cogn Neurosci 4:
81–89.
31. Bassett DS, et al. (2011) Dynamic reconfiguration of human brain networks during
learning. Proc Natl Acad Sci USA 108:7641–7646.
32. Blakemore SJ, Choudhury S (2006) Development of the adolescent brain: Implications
for executive function and social cognition. J Child Psychol Psychiatry 47:296–312.
33. Lebel C, Beaulieu C (2011) Longitudinal development of human brain wiring con-
tinues from childhood into adulthood. J Neurosci 31:10937–10947.
34. Kaelbling LP, Littman ML, Moore AW (1996) Reinforcement learning: A survey. J Art
Int Res 4:237–285.
35. Gopnik A, et al. (2004) A theory of causal learning in children: Causal maps and Bayes
nets. Psychol Rev 111:3–32.
36. Perfors A, Tenenbaum JB, Griffiths TL, Xu F (2011) A tutorial introduction to Bayesian
models of cognitive development. Cognition 120:302–321.
37. Tenenbaum JB, Kemp C, Griffiths TL, Goodman ND (2011) How to grow a mind:
Statistics, structure, and abstraction. Science 331:1279–1285.
38. Gopnik A (2012) Scientific thinking in young children: Theoretical advances, empirical
research, and policy implications. Science 337:1623–1627.
39. Xu F, Kushnir T (2013) Infants are rational constructivist learners. Curr Dir Psychol Sci
22:28–32.
40. Kushnir T, Xu F, eds (2012) Rational Constructivism in Cognitive Development (Aca-
demic, Cambridge, MA), Vol 43.
41. Pearl J (2009) Causality (Cambridge Univ Press, Cambridge, UK), 2nd Ed.
42. Griffiths TL, Chater N, Kemp C, Perfors A, Tenenbaum JB (2010) Probabilistic models
of cognition: Exploring representations and inductive biases. Trends Cogn Sci 14:
357–364.
43. Bonawitz E, Denison S, Griffiths TL, Gopnik A (2014) Probabilistic models, learning
algorithms, and response variability: Sampling in cognitive development. Trends
Cogn Sci 18:497–500.
44. Denison S, Bonawitz E, Gopnik A, Griffiths TL (2013) Rational variability in children’s
causal inferences: The sampling hypothesis. Cognition 126:285–300.
45. Bonawitz E, Denison S, Gopnik A, Griffiths TL (2014) Win-stay, lose-sample: A simple
sequential algorithm for approximating Bayesian inference. Cognit Psychol 74:35–65.
46. Kirkpatrick S, Gelatt CD, Jr, Vecchi MP (1983) Optimization by simulated annealing.
Science 220:671–680.
47. Horner V, Whiten A (2005) Causal knowledge and imitation/emulation switching in
chimpanzees (Pan troglodytes) and children (Homo sapiens). Anim Cogn 8:164–181.
Gopnik et al.
COLLOQUIUM
PAPER
48. Legare CH, Nielsen M (2015) Imitation and innovation: The dual engines of cultural
learning. Trends Cogn Sci 19:688–699.
49. Williamson RA, Meltzoff AN (2011) Own and others’ prior experiences influence
children’s imitation of causal acts. Cogn Dev 26:260–268.
50. Buchsbaum D, Gopnik A, Griffiths TL, Shafto P (2011) Children’s imitation of causal
action sequences is influenced by statistical and pedagogical evidence. Cognition 120:
331–340.
51. Berl RE, Hewlett BS (2015) Cultural variation in the use of overimitation by the Aka
and Ngandu of the Congo Basin. PLoS One 10:e0120180.
52. Legare CH (2017) Cumulative cultural learning: Development and diversity. Proc Natl
Acad Sci USA 114:7877–7883.
53. Morgan TJH, et al. (2015) Experimental evidence for the co-evolution of hominin tool-
making teaching and language. Nat Commun 6:6029.
54. Beck SR, Apperly IA, Chappell J, Guthrie C, Cutting N (2011) Making tools isn’t child’s
play. Cognition 119:301–306.
55. Aplin LM, et al. (2015) Experimentally induced innovations lead to persistent culture
via conformity in wild birds. Nature 518:538–541.
56. Kawamura S (1959) The process of sub-culture propagation among Japanese ma-
caques. Primates 2:43–60.
57. Aplin LM, Sheldon BC, McElreath R (2017) Conformity does not perpetuate sub-
optimal traditions in a wild population of songbirds. Proc Natl Acad Sci USA 114:
7830–7837.
58. Perry SE, Barrett BJ, Godoy I (2017) Older, sociable capuchins (Cebus capucinus) invent
more social behaviors, but younger monkeys innovate more in other contexts. Proc
Natl Acad Sci USA 114:7806–7813.
59. Crone EA, Dahl RE (2012) Understanding adolescence as a period of social-affective
engagement and goal flexibility. Nat Rev Neurosci 13:636–650.
60. Piekarski DJ, et al. (2017) Does puberty mark a transition in sensitive periods for
plasticity in the associative neocortex? Brain Res 1654:123–144.
61. Blakemore SJ, Mills KL (2014) Is adolescence a sensitive period for sociocultural pro-
cessing? Annu Rev Psychol 65:187–207.
62. Cardoos SL, et al. (2017) Social status strategy in early adolescent girls: Testosterone
and value-based decision making. Pscyhoneuroendocrinolgy 81:14–21.
63. Braams BR, Peters S, Peper JS, Güroğlu B, Crone EA (2014) Gambling for self, friends,
and antagonists: Differential contributions of affective and social brain regions on
adolescent reward processing. Neuroimage 100:281–289.
64. Braams BR, van Duijvenvoorde ACK, Peper JS, Crone EA (2015) Longitudinal changes
in adolescent risk-taking: A comprehensive study of neural responses to rewards,
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
pubertal development, and risk-taking behavior. J Neurosci 35:7226–7238.
65. Steinberg L, et al. (2017) Around the world, adolescence is a time of heightened
sensation seeking and immature self-regulation. Dev Sci, 10.1111/desc.12532.
66. Cheng PW (1997) From covariation to causation: A causal power theory. Psychol Rev
104:367–405.
67. Kelley HH (1967) Attribution theory in social psychology. Nebraska Symposium on
Motivation 15:192–238.
68. Jones EE, Harris VA (1967) The attribution of attitudes. J Exp Soc Psychol 3:1–24.
69. Horhota M, Blanchard-Fields F (2006) Do beliefs and attributional complexity influ-
ence age differences in the correspondence bias? Soc Cogn 24:310–337.
70. Morris MW, Nisbett RE, Peng K (1995) Causal attribution across domains and cultures.
Causal Cognition: A Multidisciplinary Debate, eds Sperber D, Premack D, Premack AJ
(Clarendon Press, Oxford), pp 577–612.
71. Pfeifer JH, et al. (2009) Neural correlates of direct and reflected self-appraisals in
adolescents and adults: When social perspective-taking informs self-perception. Child
Dev 80:1016–1038.
72. Pfeifer JH, Lieberman MD, Dapretto M (2007) “I know you are but what am I?!”:
Neural bases of self- and social knowledge retrieval in children and adults. J Cogn
Neurosci 19:1323–1337.
73. Gee DG, et al. (2013) Early developmental emergence of human amygdala-prefrontal
connectivity after maternal deprivation. Proc Natl Acad Sci USA 110:15638–15643.
74. Nettle D, Frankenhuis WE, Rickard IJ (2013) The evolution of predictive adaptive re-
sponses in human life history. Proc Biol Sci, 10.1098/rspb.2013.1343.
75. R Core Team (2012) R: A language and environment for statistical computing. (R
Foundation for Statistical Computing, Vienna).
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7899
How language shapes the cultural inheritance
of categories
Susan A. Gelmana,1 and Steven O. Robertsa
a
Department of Psychology, University of Michigan, Ann Arbor, MI 48109
Edited by Andrew Whiten, University of St. Andrews, St. Andrews, United Kingdom, and accepted by Editorial Board Member Andrew G. Clark April 12, 2017
(received for review January 12, 2017)
It is widely recognized that language plays a key role in the
transmission of human culture, but relatively little is known about
the mechanisms by which language simultaneously encourages
both cultural stability and cultural innovation. This paper examines
this issue by focusing on the use of language to transmit categories,
focusing on two universal devices: labels (e.g., shark, woman) and
generics (e.g., “sharks attack swimmers”; “women are nurturing”).
We propose that labels and generics each assume two key princi-
ples: norms and essentialism. The normative assumption permits
transmission of category information with great fidelity, whereas
essentialism invites innovation by means of an open-ended, place-
holder structure. Additionally, we sketch out how labels and ge-
nerics aid in conceptual alignment and the progressive “looping”
between categories and cultural practices. In this way, human lan-
guage is a technology that enhances and expands the categoriza-
tion capacities that we share with other animals.
language | categories | essentialism | norms | children
I t is broadly agreed that language is a distinctive human ca-
pacity and a powerful engine of cultural transmission. As such,
language is important to the theme of this special issue (1). No
matter how sophisticated the cultural transmission systems of
nonhuman species (and they are astonishingly sophisticated; see
papers in this issue) (2), they proceed without language. We can
thus ask what language distinctively contributes to cultural
transmission in humans and (more speculatively, but impor-
tantly) what language may distinctively contribute to cultural
evolution in humans. Recent evidence from language learning in
children provides new insights into these questions.
In this paper, we focus specifically on a key universal element
of language, category labels (e.g., dogs, gold, women, Muslims),
and their central role in the transmission and evolution of cat-
egory representations. The argument, in brief, is that category
labels work in an almost paradoxical way to ensure stability in the
transmission process, but simultaneously to permit and even
foster conceptual change. On the one hand, words are conven-
tional and prescriptive, and provide a stable representation that
is easily shared with great fidelity, but on the other hand, words
have an open-ended “placeholder” structure that invites in-
novation. We suggest that this dual capacity contributes to what
is distinctive in human cultural evolution.
Propositions vs. Presuppositions
Maynard Smith and Szathmáry argued that language is one of
the major transitions in the evolution of complexity, specifically
in the intergenerational transmission of information: “We accept
[the origin of human speech] as being the decisive step in the
origin of specifically human society” (3). Kirby et al. (4) similarly
note that “Language is unique in being a system that supports
unlimited heredity of cultural information, allowing our species
to develop a unique kind of open-ended adaptability.” And Pagel
(5) likewise refers to “language’s role in the transmission of the
information that makes our societies possible.”
The most obvious way that language transmits information is
via explicit declarative propositions (e.g., “You can crack open a
7900–7907 | PNAS | July 25, 2017 | vol. 114 | no. 30
nut using a rock”; “I’ll give you my money if you put down that
gun”; “Don’t trust Joe—he lies constantly”), which can share
ideas, negotiate trades, deceive enemies, impress potential mates,
affect reputations, and so forth. The expressive capacity of human
language is virtually unlimited because of its hierarchical, combi-
natorial structure (6). In contrast to the communication systems of
other organisms (even those as impressive as whales, bees, or
vervet monkeys), human language is generative: it permits in-
finitely many messages to be constructed out of a limited number
of elements. This remarkably flexible system has obvious survival
value, as it is used in the “cognitive arms race” of competitive
feedback loops implicated in cooperative interactions that involve
and must deal with cheating and cheating-detection (7).
However, much of what human language conveys is not ex-
plicitly articulated via propositional content, but rather is implied
via presupposition, implicature, and other forms of inference (8).
Four examples follow.
(i) Language marks social identity through variation. There are
roughly 6,000 human languages around the globe, mutually un-
intelligible, and (with rare exception) fully learnable only in
childhood. These aspects materially affect with whom one can
communicate and coordinate, and from whom one can learn. Even
among those who speak the same language, accent and dialect
reveal a person’s cultural origins, and so serve as honest signals to
identity, with consequences for whom others choose to interact
with and which models others trust to imitate and learn from (9).
(ii) Language directs a person’s attention in the moment by
means of structural features of the grammar. Different linguistic
communities focus on different aspects of experience, and in so
doing indicate what is important (10). For example, Japanese has
an honorific system that requires a speaker to decide level of
politeness; Quechua has an evidential system for expressing how
a speaker comes to know something: directly seeing vs. hearsay.
There is debate regarding the role of these differences on non-
linguistic cognition (11, 12). But at a minimum, these structural
features affect a person’s thinking in the moment of speaking
(13), including what information gets encoded and transmitted
within a social interaction.
(iii) Language transmits information through a rich system
of pragmatic implications (14, 15). Communication involves in-
ferring the speaker’s intentions, a complex process that builds on
theory-of-mind capacities (16). Pragmatic inferences not only
allow a listener to infer a speaker’s meaning, but also to learn
about properties of the world (17).
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: S.A.G. and S.O.R. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. A.W. is a guest editor invited by the Editorial
Board.
1
To whom correspondence should be addressed. Email: gelman@umich.edu.
www.pnas.org/cgi/doi/10.1073/pnas.1621073114

(iv) Language provides cognitive tools that aid with recall,
transmission, and manipulation of concepts that otherwise can be
difficult to hold in mind. For example, number words help
speakers of a language, such as English or Turkish, remember and
communicate exact cardinalities of sets; speakers of a language
without number words (such as Pirahã) perform poorly on nu-
merical reasoning tasks that tap into these processes (18). In this
way, language provides tools much like any other informational
technology, such as Arabic numerals, written language, the abacus,
or even computers (19).
Our focus is on this last sort of presupposition: language as a
cognitive tool. We focus on how human languages represent cate-
gories, by means of two universal devices: labels and generics. We
argue that these devices convey two important presuppositions: that
categories are normative and that categories have essences. We
further suggest that these presuppositions are both constraining
(leading to stability) and generative (leading to innovation) in the
process of cultural transmission.
Categories as Cultural Inheritance
Categories are mental representations in which perceptibly distinct
entities are treated as alike (e.g., the category “apple” permits one
to identify a variety of different apples as edible). Every animal
species uses categories to organize their representations of experi-
ence, identify newly encountered instances, and make predictive
inferences, from pigeons identifying food to voles identifying kin.
Categories are also a foundational component of socially trans-
mitted behaviors, such as tool use (categories are needed to identify
potential tools), vocalizations to warn of danger (categories are
needed to identify predators), or rituals to maintain group cohesion
(social categories are needed to decide whom to copy) (20, 21).
For humans, categories themselves are a key part of our cultural
inheritance, which is to say, they exhibit learned, socially trans-
mitted variation that cannot be explained by genetic or environ-
mental factors (2). We are not born with a fixed set of categories
(no one is born knowing of screwdrivers, or that whales are mam-
mals, or that girls wear pink). Nor do we simply pick up on dis-
continuities in the biological world; rather, human categories have a
cultural overlay. We see this in categories of natural kinds, social
kinds, and artifacts, all of which display tremendous linguistic, cul-
tural, and regional variation. Classifications of the natural world
vary in which animals, plants, or substances are classified as edible,
which are classified as medicinal, and which are classified as clean/
unclean (22). Classifications of the social world vary in how gender,
race, and social hierarchies are organized (23, 24). Classifications of
artifacts vary in the very entities there are to be classified, with
distinct types of tools, clothing, furniture, and so forth, as well as
category boundaries (25). And there is marked linguistic variation
in the classification of dimensions of experience, including color,
number, time, space, emotions, even senses (26, 27).
Although categories can be acquired asocially by individuals via
direct observations and interactions with the world, human languages
provide a socially transmitted system for efficiently communicating
information about which categories there are, what belongs in those
categories, and which attributes those categories possess. Universally,
languages use two devices for the intergenerational transmission of
categories: labels (names for categories, such as “shark” or “woman”)
and generics (generalizations about named categories, such as “sharks
attack swimmers” or “women are nurturing”).
Labels express concepts that have some cultural significance;
whereas there are indefinitely many concepts one can generate
(e.g., “items weighing more than 500 grams,” including vultures
and the Oxford English Dictionary but not a small grapefruit),
only a subset of these ideas are lexicalized, and of these, only a
subset are maintained in a language over time (28). Words are
distinctive to humans in their number (typically about 50,000 in
an adult speaker, many of which are names for things), con-
ceptual precision (e.g., chase vs. flee), and need to be learned
Gelman and Roberts
COLLOQUIUM
PAPER
(29). Children devote considerable time and effort to amassing
words, learning roughly 14,000 words by age 6, which averages to
learning nearly one new word every waking hour from 18 mo
through 6 y of age (30). Although category membership can be
inferred without language (e.g., recognizing an animal as a snake
based on its shape and movement), labels have informational
capacity beyond direct observation. Even for young children, they
can convey surprising category membership of an individual item
(e.g., that a legless lizard is not a snake) (31) or introduce wholly
new categorical distinctions (e.g., to distinguish animals based on
subtle variations in antennae rather than overall body shape) (32).
Generics are generalizations that refer to a category directly (e.g.,
“birds fly”) (33). In contrast to specific utterances (e.g., “Did you
see that bird?”), generics convey information that extends beyond
the current context and indeed in principle cannot be demonstrated
directly. Importantly, generics are input to children’s developing
knowledge systems, as they are frequent in child-directed speech
and acquired by about 2.5 y of age (34, 35). As soon as children
master the syntactic prerequisites for expressing generics in their
language (e.g., in English: plurals, articles, and tense), they produce
generics (“Does lions crawl?”; “I don’t like babies that cry”), com-
prehend generics as kind-referring and distinct from specific refer-
ence, and recall whether information was provided using generic
or specific language (35–37). Generics have been attested in all
documented languages, including pidgins and Pirahã (38–42).
Children’s early capacity to learn generics is particularly striking,
given that generic referents are abstract (one cannot point to a
kind, only to instances of a kind) and their semantics cannot be
reduced to a particular quantity (unlike “some,” “most,” or “all”)
(43–45). For example, although “Lions have manes” is acceptable
despite applying only to male lions, “Lions are male” is semanti-
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
cally unacceptable, and preschool children understand this (46).
Two Presuppositions: Norms and Essences
On a strict reading, labels communicate the category to which
something belongs, and generics communicate some fact, opinion,
or belief about a category. This is the explicit informational value of
these expressions. However, labels and generics in actual in-
terpersonal use imply more than these literal meanings, and indeed
we would argue that appropriate use of these expressions requires
understanding these implications. Next, we review two conceptual
presuppositions embedded in the use of labels and generics: that
categories are normative (i.e., conventional and prescriptive) and
that categories have essences.
Categories as Normative. A social norm is a shared, socially con-
structed, context-specific rule that indicates what is (or is not)
socially appropriate (21, 47). Labels are fundamentally normative
in that they are conventional (i.e., shared with other members of
the speech community, a principle required for their successful
use) (48, 49). A person who did not appreciate the normative
value of labels might arbitrarily substitute vocalizations of their
own invention for words that they hear from others (e.g., you call
that a hammer, but I’ll call it a blicket), and the whole commu-
nicative enterprise would never get off the ground. By the time of
their first word, children appreciate the conventionality principle,
expecting novel labels used by one speaker to be understood by
others within the speech community (50). Not all behaviors are
treated the way that labels are treated; for example, infants as-
sume that preferences are individually varying rather than shared
or conventional (51). Infants also appreciate that language oper-
ates via a division of linguistic labor, whereby more knowledgeable
members of the community can be trusted to provide accurate
labeling (52). From an early age, children are sensitive to social
variation in labelers, for example preferentially accepting labels
from adults over children and experts over novices (53), an ex-
pectation that fosters conformity. A powerful consequence of this
principle is that even a simple relabeling can shift children’s label
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7901
use, with only minimal explanation, as in the following examples of
parents looking through a picture book with their children (34):
i) Child: That’s kangaroo. (Pointing to an aardvark.)
Mother: Well, that looks like a kangaroo, but it’s called
an aardvark.
Child: Aardvark.
ii) Child: That’s a snake. (Referring to an eel.)
Mother: It looks like a snake, doesn’t it? It’s called an eel. It’s
like a snake, only it lives in the water. And there’s another one.
In experimental settings as well, children accept relabelings
from others, even when they compete with perceptual evidence
that children directly experience (54).
Generic information is likewise assumed to be conventional and
shared with others rather than idiosyncratic, private, or subjective.
Even in prelinguistic communication, an action that is displayed to
others is more often assumed to be generic than information that is
done for the actor himself or herself (55). With regard to language,
young children treat generic statements as conveying information
that is widely known (56–58). Generics are particularly frequent in
pedagogical (information transmitting) contexts, such as book
reading, and when taking on a pedagogical role, such as talking to a
more ignorant interlocutor or pretending to be a teacher (59, 60).
Although generics can express idiosyncratic or subjective per-
spectives (e.g., “Vegemite is delicious!”), expressing this gener-
ically implies a general truth, even to preschool children (61).
Category-referring language is normative in a stronger, pre-
scriptive sense as well. That is, labels and generics imply that a
feature linked to a category not only is but also should be (62,
63). This is particularly so for generic language, which expresses
norms that may even compete with statistical observations: “Boys
don’t cry” is deemed true—despite being demonstrably false—
because it expresses a norm (64–66). Similarly, generics such as
“Scientists care about the truth” express abstract values rather
than descriptively accurate features (67). Parents likewise pro-
duce generics that express prescriptive norms that conflict with
the reality in the moment (e.g., “Remember, we don’t stand up
on chairs”; “Oh, no, you don’t pull on books”; see http://childes.
talkbank.org/access/Eng-NA/Brown.html).
Generic language leads to normative judgments, even when the
categories are novel and the content is innocuous. In a series of ex-
periments, children 4–13 y of age learned of two novel groups
that contrasted with one another in some harmless behavior,
such as the music they listen to or the food they eat (e.g.,
Hibbles listen to one kind of music, and Glerks listen to another
kind of music). Children reported that it was “not OK” for an
individual to fail to conform to the group behavior (e.g., for a
Hibble to listen to music that is more typical of Glerks) (68). In
other words, children interpreted an unfamiliar descriptive
regularity as if it were prescriptive (see also refs. 69–71 for
additional evidence that descriptive and prescriptive norms are
conflated in children’s and adults’ concepts). Language plays an
important role in licensing this normative response: when the
vignettes depicted individuals (not groups) that received cate-
gory labels in either specific statements (e.g., “This Hibble
listens to this kind of music”) or generic statements (e.g.,
“Hibbles listen to this kind of music”), children made normative
judgments; when the vignettes depicted individuals without cat-
egory labels or generics (e.g., “This listens to this kind of mu-
sic”), children did not make normative judgments (72). Thus,
category labels and generic statements license a prescriptive
reading of novel, innocuous behaviors: they imply that mem-
bers of the labeled group should behave a certain way. The
establishment of norms is itself a mechanism that fosters the
stability of group behaviors (70).
7902 | www.pnas.org/cgi/doi/10.1073/pnas.1621073114
Essentialism.
[Essence is] the very being of anything, whereby it is what it is. And thus
the real internal, but generally . . . unknown constitution of things,
whereon their discoverable qualities depend, may be called their essence.
Locke (73)
A striking aspect of human categories—and the words that ex-
press them—is that they often defy appearances: stick-bugs look like
sticks, pyrite looks like gold. It is not surprising that scientific cate-
gories extend beyond the obvious, given that the natural world
provides evolved mechanisms that lead appearances to mislead, in-
cluding homologies, camouflage, mimicry, and sexual dimorphism.
What is notable is that nonscientists, including children, share the
expectation that categories have hidden structure and that words in
ordinary language (e.g., bug, gold) capture this structure (74). This
expectation contrasts with classic theories of cognitive development,
which propose that young children are “perceptually bound” think-
ers, and that concepts shift from similarity-based to conceptually
based over development (75, 76).
We refer to this assumption as “psychological essentialism”: an
intuitive belief that categories of the natural world share not just
observable features, but also a deeper, nonobvious reality: they
“carve up nature at its joints” (74, 77, 78). Thus, tigers share more
than a certain size, gait, striped fur, and ferocity, but also internal
parts, temperament, instincts, as well as an innate, unchanging
tiger “essence.” This essence might be blood, DNA, or even an
unspecified, unknown placeholder, an expectation that there is an
essence without knowing what it might be. For example, young
children report that an animal’s behavior is caused by its own
insides or energy before they can have detailed expectations about
the particular form that such causal force might take (79–81).
Evidence for psychological essentialism comes from research with
adults as well as young children (74, 82). Even in infancy, children
expect members of a category to share internal, nonobvious, or
causal similarities, even in the face of superficial dissimilarities (31,
83–85). Boundaries between categories are treated as discontinuous
and objectively correct, and category membership itself is viewed as
immutable (24, 86–89). Category members are thought to have in-
nate potential that resists environmental influences (90–92). Internal
bodily organs are thought to have the power to modify the recipient’s
behavior (93, 94). That essentialist beliefs have been documented in
young children and across a variety of cultural contexts suggests that
essentialism is a fundamental component of human cognition (23,
95–99). Although which categories are essentialized varies cross-
culturally, especially for categories of people (such as race, gender,
or ethnicity) (100), essentialism of both natural kinds and social
kinds has been broadly and consistently documented (101–103).
Essentialized social categories have important implications for evo-
lutionarily significant behaviors in humans, including patterns of
affiliation, mating, reproduction, and conflict. For example, essen-
tialized social categories are often conceptualized as less human and
more threatening than nonessentialized social categories (104, 105),
and both children and adults are reluctant to share resources with
members of essentialized out-groups (101, 106).
Essentialist expectations are linked to category labels. Hearing
that a pterodactyl is a “dinosaur,” that a swaddled baby is a
“boy,” or that a child received the heart of a “monkey” leads to
the inference that the pterodactyl does not live in a nest, that the
baby will grow up to like football regardless of its upbringing, and
that the donated heart will confer a slight but inevitable uptick in
one’s tendency to eat bananas. Essentialist expectations attach
also to wholly novel labels applied to wholly novel categories
(107–109). This is not to say that labels automatically trigger
essentialist reasoning; they do not (74, 110). However, when a
label is applied to a category that has some coherent conceptual
basis (e.g., shared features) then essentialist beliefs follow (32).
Gelman and Roberts

Labels may play a particularly important role for social cate-
gories, given how culturally variable they are (86, 111).
Generics also facilitate the social transmission of essentialism.
They have two semantic features that support essentialism: they
express properties that are timeless and nonaccidental (e.g.,
“birds have hollow bones”), and they minimize within-category
variability (e.g., “birds lay eggs,” even though only adult females
do so). Preschool children appreciate both these points (46, 58).
Moreover, hearing novel generics about novel categories leads to
more within-category inferences (36, 112), assumption of core
features (57), and essentialist inferences about that category,
above and beyond labeling per se (108, 109).
Elsewhere we have argued that essentialism is a flawed on-
tology that oversimplifies by underestimating within-category
variability, overestimating between-category differences, and
assigning too much causal significance to imputed essences (74).
Viewing biological categories as immutable, and viewing varia-
tion as only superficial, contributes to persistent misconceptions
about evolutionary processes, genetics, and other aspects of
science (113–116). Attributing existing patterns of social ineq-
uities to hidden, inherent, and inalterable causes in individuals is
an oversimplification that ignores structural and historical fac-
tors (117, 118) and contributes to a variety of social ills, including
stereotyping, prejudice, and discrimination (102, 119–125). For
example, essentialist beliefs about gender promote disrespect
and lowered expectations toward girls and women in schools and
academia (104, 126), and essentialist beliefs about “Blackness”
predict the perception of Black people as less than human, which
subsequently predicts greater violence toward Black children and
increased rates of applying the death penalty toward Black men
in the United States courts system (127, 128).
So then, why do we essentialize? In the words of Medin (78),
“psychological essentialism is bad metaphysics . . . [but] may
prove to be good epistemology.” In other words, essentialism is
factually wrong but heuristically useful. Essentialism promotes
learning and conceptual change by providing a placeholder
structure that promotes the search for underlying causes and
modifications over time. The evidence reviewed above demon-
strates the placeholder notion of essentialized concepts in three
interrelated respects: (i) children expect items with the same
label to share nonobvious similarities that they have not yet
learned; (ii) children are guided by labeling and generics even
when in competition with children’s own direct experiences (e.g.,
“a whale isn’t a fish”; “boys don’t cry”); and (iii) labeling and
generics operate according to a “division of linguistic labor,”
whereby children defer to more expert others to inform them
about the classifications and generalizations of experience (129–
131). Importantly, these placeholder expectations, in turn, permit
and promote conceptual innovation, because children’s classifica-
tions build upon the expertise of others, and because children are
motivated to search for underlying causal similarities that members
of a category share. That even young children view categories in
essentialist ways suggests that categories are not just structures for
organizing what is already known, but placeholders for further
knowledge that is expected to accrue. The meaning of a word is not
a list of known features or learned facts. Rather, a word serves as an
invitation to form a category (132) and to extend and modify it with
growing knowledge and expertise. “Dog” is not a tag for a fixed set
of observed features, but rather a pointer to “things of that nature,”
where the “nature” will be filled in via learning and input from
others. Here we endorse Putnam’s (133) famous assertion that
“‘meanings’ just ain’t in the head!” Words refer to placeholder
concepts that do not have fixed content and thus can be modified.
Language “is not a mirror of our inner states but a complement to
them. It serves as a tool whose role is to extend cognition in ways
that on-board devices cannot” (19).
Gelman and Roberts
COLLOQUIUM
PAPER
Conformity and Innovation
Biological evolution requires inheritance and mutations. Simi-
larly, cultural evolution requires both conformity and innovation
(134–137), and we suggest that the linguistic presuppositions
discussed above—norms and essentialism—contribute to both
these processes. Because labels and generics are fundamentally
normative (conventional and prescriptive), they provide stable
representations that are easily shared with great fidelity. Because
labels are understood to be conventional and shared among
members of one’s language group, a child who hears a word in
context rapidly maps an initial meaning on the basis of a single
exposure (30, 138), although a full understanding emerges more
slowly and gradually (30, 139). Because generic information is
viewed as not only descriptively accurate but also as prescriptively
correct, children may judge that failure to conform to generically
stated category regularities (such as acting at variance with one’s
in-group) is wrong or even risks punishment (140). At the same
time, the open-ended, placeholder structure of essentialism, also
implied by labels and generics, invites category change. This is
perhaps most evident in the ease with which children accept
counterintuitive labels offered by knowledgeable others (e.g.,
accepting that a legless lizard is not a snake) as well as the in-
ductive potential of labels and generics, wherein new information
is rapidly learned and generalized to new instances. These pre-
suppositions suggest a transmission process that fosters change, at
the same time that it resists change in the transmission process.
In this section we suggest two additional mechanisms by which
the linguistic representation of categories may promote cultural
transmission and cultural evolution: they transform variable in-
put into categorical representations, and (in the case of human
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
kinds) they involve looping effects between categories and the
people being categorized.
From Variable Input to Categorical Representations. A uniquely human
aspect of language is that it takes variable, idiosyncratic experiences
and transforms them into discrete, symbolic, shared representations
(28). The world is a complex, continuously dynamic array of sensory
inputs, and no two people experience identical environmental cues.
The experience of categories is thus doubly variable: in the range of
instances that an individual encounters and in the experiences of
individuals across the language community. Language reduces and
regularizes this remarkable variety. Consider the use of a simple
word, “bird,” which extends from hummingbirds to dodos, from
downy chicks to vicious birds of prey. We converge on a shared
label, regardless of our varied experiences: which birds we have seen
or heard, which ones we have owned or eaten, whether our expe-
rience comes from real-world encounters, plush toys, or Big Bird.
This gap between the variability of experience and the com-
monality of labels presents a puzzle: “If biological and real world
constraints are not enough then how is it nevertheless possible
for a group to arrive at a sufficiently shared set of conceptual
distinctions to make language possible?” (141). In other words,
the transmission of language requires conceptual alignment or
compatible mental representations that are abstracted away from
varying experiences and knowledge bases (142, 143).
We suggest that the manner in which labels and generics ab-
stract away from experience aids in conceptual alignment. Cat-
egory labels abstract away from the particulars that make
individuals unique (a small poodle and a large Great Dane are
both “dogs”), and generics abstract away from any particular
context (“birds fly,” even when the only birds in sight are pen-
guins) (144, 145). Speakers don’t require shared experiences to
have a shared system of communication. A 12-mo-old infant and
a biologist can communicate with the word “dog,” despite radi-
cally different understandings.
Generics are particularly well-suited for expressing abstract,
shared representations because, as noted earlier, they systematically
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7903
underplay variations in experience by glossing over exceptions. Ge-
nerics are not disconfirmed by counterexamples (the existence of a
nonflying bird does not disconfirm the generic claim that birds fly)
(45, 46), which means that generic messages can trump a listener’s
personal experiences. People produce generics about features they
consider conceptually important (e.g., dangerous or distinctive), even
when they know them to be variably present in a category, but those
who hear such generics (whether adults or young children) tend to
assume that the feature is almost universally present among category
members (43, 146). This results in systematic distortions in the
transmission process, from variability to category-wide consistency.
The Looping Effect of Human Kinds. Hacking (147) speaks of a
“looping effect” in social categories: specifically, that classifica-
tions of people have cognitive consequences for those that are
classified, which feedback into these same classifications:
To create new ways of classifying people is also to change how we can
think of ourselves, to change our sense of self-worth, even how we re-
member our own past. This in turn generates a looping effect, because
people of the kind behave differently and so are different. This is to say
the kind changes, and so there is new causal knowledge to be gained and
perhaps, old causal knowledge to be jettisoned. . . . that new knowledge in
turn becomes part of what is to be known about members of the kind,
who change again. . . . Kinds are modified, revised classifications are
formed, and the classified change again, loop upon loop” (147).
We have sketched out linguistic mechanisms that may con-
tribute to this looping effect. Labels and generics stake out cat-
egories, which then are altered through human action to reify
such categories. In contrast to Hacking, however, we see this
looping effect not only for categories to which one belongs, but
also for categories of others. History is replete with modifications
that differentiate groups. Thus, for example, male/female dif-
ferences are exaggerated by differences in clothing, hairstyles,
gait, bodily deformations (e.g., foot-binding), and styles of
speech. Modifications may be imposed (e.g., Jews in World War
II Germany being required to wear stars) or chosen (e.g., fash-
ions worn by self-identified hipsters). Social groups may be
physically separated, either by explicit policy (e.g., segregationist
policies toward Blacks in the southern United States; Japanese
internment camps in the United States during World War II) or
by other practices and constraints (e.g., low-income families re-
stricted to neighborhoods with unclean water and air). Concepts
of human kinds may lead to a cyclical pattern in which cultural
practices lead groups to appear more distinct from one another,
which confirms the categorizations, leading to more differenti-
ating practices, and so forth. Viewing social kinds as having deep
differences has cycling effects on behaviors that contribute to the
reality of that social kind.
Norms and Essentialism in Nonhuman Species
Are nonhuman animals also capable of learning categories with
prescriptive implications and a nonobvious basis? This question is
timely, given recent discoveries of remarkably sophisticated cate-
gorization and social transmission abilities in nonhuman animals
(see other papers in this issue). For example, consider an in-
genious experiment demonstrating that chimpanzees conform to
cultural (descriptive) norms of tool use (148). The researchers first
taught a high-ranking chimpanzee one of two manners of tool use
to obtain food out of a puzzle box (e.g., using either a poking or a
lifting motion). When let loose within the group, other members
picked up the demonstrated solution strategy, even adhering to
the method common in the group after having successfully used
the alternative method. Certainly language was not required.
Nonhuman primates are also capable of categorizing based on
nonperceptible features. For example, baboons engage in so-
phisticated categorizations of conspecifics, with dominance hi-
erarchies that simultaneously rank by individual rank and family
7904 | www.pnas.org/cgi/doi/10.1073/pnas.1621073114
group using matrilineal kinship, friendships, and causal theories
(149). These categories have a nonobvious basis (e.g., infants
“inherit” the rank of the mother) and are learned (e.g., members
need to learn which individuals fall into which group). Seyfarth
and Cheney propose: “. . . when it comes to recognizing matri-
lineal kin groups, baboons are ‘essentialists’ . . . They act as if the
members of kin groups ‘have essences or underlying natures that
make them the things that they are’” (149).
Monkeys and great apes can also track category membership
across radical featural transformations, and privilege kind (essential
features) over superficial appearance (surface features). For ex-
ample, one study presented rhesus macaques with food items in
which the inner identity was transformed (e.g., an apple was dis-
guised as a coconut) (150). After a piece of this transformed fruit
was placed in a container, if the animal reached in and found a
piece that matched the appearance rather than the inner kind, they
continued to search for another piece, indicating that they had been
expecting the sample to match the inner kind and not the appear-
ance. The researchers interpret the findings as “evidence that ma-
caques share this one primitive aspect of psychological essentialism”
(150). Similarly, in another study, bonobos, orangutans, and chim-
panzees viewed a transformation process in which one piece of food
was disguised to look like another (e.g., a carrot slice was disguised
as a banana slice) (151). When given a choice between a true piece
of banana versus a disguised piece of carrot that only looked like a
banana, animals preferred the true banana. The authors interpret
this as “a kind of psychological essentialism, perhaps the phyloge-
netically and ontogenetically most basic one” (151). Again, language
was not required to consider an appearance–reality conflict and to
privilege the inner identity.
These impressive capacities demonstrate that humans share
with at least other primates the ability to categorize based on
subtle, nonperceptible cues, and the ability to conform to nor-
mative regularities (although conformity is substantially greater
in humans) (152). Indeed, norms and essentialism may precede
language in human development, as preverbal infants infer
general ways of interacting with objects from pedagogical dem-
onstrations, evaluate others based on their social interactions,
categorize based on nonobvious features, and distinguish indi-
viduals from kinds (55, 80, 153–155).
Nonetheless, we suggest four key respects in which human
language may be unique in fostering the social transmission and
evolution of categories.
Efficiency in Transmitting Category Information. First and most
obviously, labels and generic language ensure speed, fidelity, and
ease of transmitting category information, by means of an overt
and stable representational format. This would be difficult
(perhaps impossible) to achieve by means of actions alone. (Note
that language is not necessarily more efficient for transmitting all
sorts of information. For example, showing the location of an
object is likely more efficiently done by pointing; teaching
weaving is likely more efficiently done by demonstration.) Con-
sider the case of conveying that an item is not what it appears to
be. The studies with nonhuman primates required a lengthy and
rather elaborate shared context (the transformation process
itself), carried out by an expert with special tools and procedural
know-how. Someone who was not present during this demon-
stration would not have access to the relevant information.
Contrast this with the human language case, which efficiently
corrects a misconception with a single sentence (“This looks like
a banana, but really it’s a carrot”). Anyone who hears the new
label—even a nonexpert or young child—could then share it with
others, ensuring a transmission chain. Consider, too, the case of
conveying the scope of a feature: if eating a mushroom makes
you sick, is it because of that particular mushroom (e.g., maybe it
rotten or was sprayed by pesticides) or mushrooms of that type
more generally? Again, this is efficiently conveyed via generic
Gelman and Roberts

language (“Death cap mushrooms are poisonous”), but difficult
(perhaps impossible) to convey nonlinguistically. Notably, ge-
neric information is conveyed equally well whether it expresses
preference or avoidance, whereas nonlinguistic social learning
mechanisms may be asymmetric in this regard (e.g., the Norway
rat can learn which foods to try by sniffing the breath of con-
specifics, but cannot learn which foods to avoid by sniffing the
breath of a sick conspecific) (156, 157).
Conceptual Innovation and Change. Language can evoke concep-
tual change, not just providing new information (e.g., that dan-
delions are edible) but also abandoning an old classificatory
framework (e.g., learning that plants are alive). Human classifi-
cation systems undergo reorganizations throughout history, and
naming patterns have shifted to accommodate these changes
(158). It is less clear that nonhumans engage in conceptual
change. Consider the sad case of seabirds that consume plastic
they find in the ocean, resulting in poisoning and malnutrition.
They do so because the chemical odor of the plastic is similar to
that of dimethyl sulfide, a compound found in marine algae
(159). The birds are effectively tricked into eating plastic because
it smells like food. Thus, a categorization capacity that was useful
for locating food went awry when the environment changed. It is
not clear how one could convince seabirds to abandon this
classification system, even when it’s a matter of survival.
Scope of Application. In nonhuman animals, the examples we have
seen of sophisticated social transmission, adherence to group
norms, and nonobvious categorizations fall within a narrow set of
domains, primarily involving food and within-group social rela-
tions (e.g., mating, dominance). In contrast, human norms and
essentialism extend beyond content with obvious survival value
to include any aspect of experience. Essentialism applies to
natural substances, living kinds, human social groups, personal
characteristics, diseases, and in some respects even artifacts (31,
82, 113, 160–165). Similarly, normative expectations extend to a
vast array of behaviors, including which clothing to wear, which
music to listen to, or which games to play (68).
From Models to Morals. Although nonhuman animals are capable
of conforming to high-ranking group members (copying modeled
behaviors) and “punishing” others by retaliating when they are
wronged, we are unaware of evidence that they display moral
condemnation or punishment of nonconformity in others. For
example, in one study with chimpanzees, an actor could punish a
thief by depriving them of food reward (via trapdoor) (166). The
actor only retaliated when their own food was stolen, not when
another chimpanzee’s food was stolen. This is in sharp contrast
to the findings with young children, who exhibit strong moral
evaluations of others (47, 167, 168). One might say that social
1. Lotem A, Halpern JY, Edelman S, Kolodny O (2017) The evolution of cognitive mechanisms
in response to cultural innovations. Proc Natl Acad Sci USA 114:7915–7922.
2. Whiten A (2017) A second inheritance system: The extension of biology through
culture. Interface Focus, in press.
3. Maynard Smith JM, Szathmáry E (1997) The Major Transitions in Evolution (Oxford
Univ Press, Oxford).
4. Kirby S, Cornish H, Smith K (2008) Cumulative cultural evolution in the laboratory:
An experimental approach to the origins of structure in human language. Proc Natl
Acad Sci USA 105:10681–10686.
5. Pagel M (2017) Darwinian perspectives on the evolution of human languages.
Psychon Bull Rev 24:151–157.
6. Chomsky N (1975) Aspects of the Theory of Syntax (MIT Press, Cambridge, MA).
7. Pinker S, Bloom P (1990) Natural language and natural selection. Behav Brain Sci 13:
707–727.
8. Mannheim B (2015) The social imaginary, unspoken in verbal art. The Routledge
Handbook of Linguistic Anthropology, ed Bonvillain N (Routledge, New York), pp 44–61.
9. Kinzler KD, Corriveau KH, Harris PL (2011) Children’s selective trust in native-
accented speakers. Dev Sci 14:106–111.
10. Hill JH, Mannheim B (1992) Language and world view. Annu Rev Anthropol 21:
381–406.
Gelman and Roberts
COLLOQUIUM
PAPER
transmission processes in nonhuman animals provide models of
what behaviors are possible (i.e., models), whereas social trans-
mission processes in human animals provide models of what
behaviors are appropriate (i.e., morals).
Conclusions
The evolution of culture involves not only behavioral practices and
material artifacts, but also the representation of these practices
and artifacts in the human mind, including categories. Cultural
evolution (as opposed to mere change) entails an increase in di-
versity and complexity; it cannot just be the recycling of behaviors
(169, 170). We suggest at least three ways that categories can be
said to evolve. First, as technologies evolve, so do the categories
they belong to and the labels that express them. For example, in
English we have words for hammers, trucks, and violas (all
invented technologies), and in many languages, old words are
refitted to accommodate new inventions (e.g., “fire vehicle” means
train in Mandarin). Young children have no difficulty acquiring
these words; they do not lag behind the acquisition of categories
that were in our distant evolutionary history. Second, as theories
change and evolve, so do our labeled categories: a whale is no
longer a fish; Pluto is no longer a planet; “female hysteria” is no
longer a disease. And third, human kinds arguably become in-
creasingly diversified and complex by means of “looping effects.”
Here, however, it is important to note that the evolution of cat-
egories can be negative as well as positive. Cumulative cultural
change can be a good thing: tools get more sophisticated, social
organizations get more complex, means of food production get
more varied. But there is negative ratcheting as well, in the form of
bigotry, polarization, and the perpetuation of social hierarchies.
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
Levinson notes the special role of language in the process of
enculturation of cognition: “. . . language appears to play a cru-
cial role [in how culture gets into the head]: it is learnt far earlier
than most aspects of culture, is the most highly practiced set of
cultural skills, and is a representation system that is at once
public and private, cultural and mental” (171).
In the case of learning categories, we suggest that cumulative
cultural evolution is enhanced by labels and generics, which provide
a simple yet powerful means of passing along the wisdom (and
prejudices) of prior generations. In this way, language enhances and
expands (nonlinguistic) capacities to categorize that we share with
other animals. A full understanding of this process will require
studying how it intersects with a variety of other important cognitive
capacities that are present early in human development, including
theory of mind, alertness to testimony, attention to ritual, and a
drive for causal understandings (134, 172–174).
ACKNOWLEDGMENTS. We thank Bruce Mannheim and two anonymous
reviewers for very helpful comments on an earlier draft.
11. Gentner D, Goldin-Meadow S, eds (2003) Language in Mind: Advances in the Study
of Language and Thought (MIT Press, Cambridge, MA).
12. Gleitman L, Papafragou A (2013) Relations between language and thought.
Handbook of Cognitive Psychology, ed Reisberg D (Oxford Univ Press, New York).
13. Slobin DI (1991) Learning to think for speaking: Native language, cognition, and
rhetorical style. Pragmatics 1:7–25.
14. Grice HP (1975) Logic and conversation. Syntax and Semantics 3: Speech Acts, eds
Cole P, Morgan J (Academic, New York), pp 41–58.
15. Levinson SC (2000) Presumptive Meanings: The Theory of Generalized Conversational
Implicature (MIT Press, Cambridge, MA).
16. Sperber D, Wilson D (2002) Pragmatics, modularity and mind‐reading. Mind Lang 17:3–23.
17. Horowitz AC, Frank MC (2016) Children’s pragmatic inferences as a route for
learning about the world. Child Dev 87:807–819.
18. Frank MC, Everett DL, Fedorenko E, Gibson E (2008) Number as a cognitive tech-
nology: Evidence from Pirahã language and cognition. Cognition 108:819–824.
19. Clark A, Chalmers D (1998) The extended mind. Analysis 58:7–19.
20. Watson-Jones RE, Legare CH (2016) The social functions of group rituals. Curr Dir
Psychol Sci 25:42–46.
21. Legare CH (2017) Cumulative cultural learning: Development and diversity. Proc Natl
Acad Sci USA 114:7877–7883.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7905
22. Atran S, Medin DL (2008) The Native Mind and the Cultural Construction of Nature
(MIT Press, Cambridge, MA).
23. Astuti R, Solomon GE, Carey S (2004) Constraints on conceptual development: A case
study of the acquisition of folkbiological and folksociological knowledge in Mada-
gascar. Monogr Soc Res Child Dev 69:1–135, vii–viii, discussion 136–161.
24. Diesendruck G, Goldfein-Elbaz R, Rhodes M, Gelman S, Neumark N (2013) Cross-
cultural differences in children’s beliefs about the objectivity of social categories.
Child Dev 84:1906–1917.
25. Malt BC, Majid A (2013) How thought is mapped into words. Wiley Interdiscip Rev
Cogn Sci 4:583–597.
26. Barrett LF, Mesquita B, Gendron M (2011) Context in emotion perception. Curr Dir
Psychol Sci 20:286–290.
27. Regier T, Kay P (2009) Language, thought, and color: Whorf was half right. Trends
Cogn Sci 13:439–446.
28. Pagel M (2012) Wired for Culture: Origins of the Human Social Mind (Norton, New York).
29. Pinker S, Jackendoff R (2005) The faculty of language: What’s special about it?
Cognition 95:201–236.
30. Carey S (1978) The child as word learner. Linguistic Theory and Psychological Reality,
eds Bresnan J, Miller G, Halle M (MIT Press, Cambridge, MA), pp 264–293.
31. Gelman SA, Markman EM (1986) Categories and induction in young children.
Cognition 23:183–209.
32. Gelman SA, Davidson NS (2013) Conceptual influences on category-based induction.
Cognit Psychol 66:327–353.
33. Carlson GN, Pelletier FJ, eds (1995) The Generic Book (Univ Chicago Press, Chicago).
34. Gelman SA, Coley JD, Rosengren KS, Hartman E, Pappas A (1998) Beyond labeling:
The role of maternal input in the acquisition of richly structured categories. Monogr
Soc Res Child Dev 63:I–V, 1–148, discussion 149–157.
35. Gelman SA, Goetz PJ, Sarnecka BW, Flukes J (2008) Generic language in parent-child
conversations. Lang Learn Dev 4:1–31.
36. Graham SA, Gelman SA, Clarke J (2016) Generics license 30-month-olds’ inferences
about the atypical properties of novel kinds. Dev Psychol 52:1353–1362.
37. Gelman SA, Raman L (2007) This cat has nine lives? Children’s memory for genericity
in language. Dev Psychol 43:1256–1268.
38. Maurer P, Meeuwis M; APiCS Consortium (2013) Generic noun phrases in subject
function. The Atlas of Pidgin and Creole Language Structures, eds Michaelis SM,
Maurer P, Haspelmath M, Huber M (Oxford Univ Press, New York), pp 114–117.
39. Everett DL (2009) Pirahã culture and grammar: A response to some criticisms. Lang
85:405–442.
40. Gelman SA, Sánchez Tapia I, Leslie SJ (2016) Memory for generic and quantified
sentences in Spanish-speaking children and adults. J Child Lang 43:1231–1244.
41. Mannheim B, Gelman SA, Escalante C, Huayhua M, Puma R (2010) A developmental
analysis of generic nouns in Southern Peruvian Quechua. Lang Learn Dev 7:1–23.
42. Tardif T, Gelman SA, Fu X, Zhu L (2012) Acquisition of generic noun phrases in
Chinese: Learning about lions without an “-s”. J Child Lang 39:130–161.
43. Cimpian A, Brandone AC, Gelman SA (2010) Generic statements require little evi-
dence for acceptance but have powerful implications. Cogn Sci 34:1452–1482.
44. Cimpian A, Gelman SA, Brandone AC (2010) Theory-based considerations influence
the interpretation of generic sentences. Lang Cogn Process 25:261–276.
45. Leslie SJ (2008) Generics: Cognition and acquisition. Philos Rev 117:1–47.
46. Brandone AC, Cimpian A, Leslie SJ, Gelman SA (2012) Do lions have manes? For
children, generics are about kinds rather than quantities. Child Dev 83:423–433.
47. Rakoczy H, Schmidt MF (2013) The early ontogeny of social norms. Child Dev
Perspect 7:17–21.
48. Clark EV (1992) Conventionality and contrast: Pragmatic principles with lexical
consequences. Frames, Fields, and Contrasts: New Essays in Semantic and Lexical
Organization, eds Kittay EF, Lehrer A (Erlbaum, Hillsdale, NJ), pp 171–188.
49. Saussure FD (1915) Cours de Linguistique Générale (Payot, Paris).
50. Sabbagh MA, Henderson AM (2007) How an appreciation of conventionality shapes
early word learning. New Dir Child Adolesc Dev (115):25–37.
51. Henderson AM, Woodward AL (2012) Nine-month-old infants generalize object la-
bels, but not object preferences across individuals. Dev Sci 15:641–652.
52. Jaswal VK, Markman EM (2007) Looks aren’t everything: 24-month-olds’ willingness
to accept unexpected labels. J Cogn Dev 8:93–111.
53. Koenig MA, Harris PL (2005) Preschoolers mistrust ignorant and inaccurate speakers.
Child Dev 76:1261–1277.
54. Lane JD, Harris PL, Gelman SA, Wellman HM (2014) More than meets the eye: Young
children’s trust in claims that defy their perceptions. Dev Psychol 50:865–871.
55. Csibra G, Gergely G (2009) Natural pedagogy. Trends Cogn Sci 13:148–153.
56. Cimpian A, Scott RM (2012) Children expect generic knowledge to be widely shared.
Cognition 123:419–433.
57. Cimpian A, Markman EM (2009) Information learned from generic language be-
comes central to children’s biological concepts: Evidence from their open-ended
explanations. Cognition 113:14–25.
58. Hollander MA, Gelman SA, Raman L (2009) Generic language and judgements about
category membership: Can generics highlight properties as central? Lang Cogn
Process 24:481–505.
59. Gelman SA, Tardif T (1998) A cross-linguistic comparison of generic noun phrases in
English and Mandarin. Cognition 66:215–248.
60. Gelman SA, Ware EA, Manczak EM, Graham SA (2013) Children’s sensitivity to the knowl-
edge expressed in pedagogical and nonpedagogical contexts. Dev Psychol 49:491–504.
61. Holubar TF, Markman EM (2013) Preschoolers’ understanding of preferences is
modulated by linguistic framing. Cooperative Minds: Social Interaction and Group
Dynamics, Proceedings of the 35th Annual Meeting of the Cognitive Science Society,
eds Knauff M, Sebanz N, Pauen M, Wachsmuth I (Cognitive Science Society, Austin,
TX), pp 603–608.
7906 | www.pnas.org/cgi/doi/10.1073/pnas.1621073114
62. Orvell A, Kross E, Gelman SA (2017) How “you” makes meaning. Science 355:
1299–1302.
63. Orvell A, Kross E, Gelman SA, That’s how ‘you’ do it: Generic you expresses norms in
early childhood. J Exp Child Psych, in press.
64. Leslie SJ (2015) “Hillary Clinton is the only man in the Obama administration”: Dual
character concepts, generics, and gender. Analytic Philos 56:111–141.
65. Prasada S, Dillingham EM (2009) Representation of principled connections: A win-
dow onto the formal aspect of common sense conception. Cogn Sci 33:401–448.
66. Wodak D, Leslie SJ, Rhodes M (2015) What a loaded generalization: Generics and
social cognition. Philos Compass 10:625–635.
67. Knobe J, Prasada S, Newman GE (2013) Dual character concepts and the normative
dimension of conceptual representation. Cognition 127:242–257.
68. Roberts SO, Gelman SA, Ho AK (2017) So it is, so it shall be: Group regularities license
children’s prescriptive judgments. Cogn Sci 41:576–600.
69. Bear A, Knobe J (November 11, 2016) Normality: Part descriptive, part prescriptive.
Cognition, 10.1016/j.cognition.2016.10.024.
70. Eriksson K, Strimling P, Coultas JC (2015) Bidirectional associations between de-
scriptive and injunctive norms. Organ Behav Hum Decis Process 129:59–69.
71. Tworek CM, Cimpian A (2016) Why do people tend to infer “ought” from “is”? The
role of biases in explanation. Psychol Sci 27:1109–1122.
72. Roberts SO, Ho AK, Gelman SA (2017) Group presence, category labels, and generic
statements influence children to treat descriptive group regularities as prescriptive.
J Exp Child Psychol 158:19–31.
73. Locke J (1959) An Essay Concerning Human Understanding (Dover, New York), Vol 2. Reprint.
74. Gelman SA (2003) The Essential Child: Origins of Essentialism in Everyday Thought
(Oxford Univ Press, New York).
75. Fisher AV, Sloutsky VM (2005) When induction meets memory: Evidence for gradual
transition from similarity-based to category-based induction. Child Dev 76:583–597.
76. Inhelder B, Piaget J (1964) The Early Growth of Logic in the Child (Norton, New York).
77. Keil FC, Richardson DC (1999) Species, stuff, and patterns of causation. Species: New
Interdisciplinary Essays, ed Wilson RA (MIT Press, Cambridge, MA).
78. Medin DL (1989) Concepts and conceptual structure. Am Psychol 44:1469–1481.
79. Gottfried GM, Gelman SA (2005) Developing domain-specific causal-explanatory
frameworks: The role of insides and immanence. Cogn Dev 20:137–158.
80. Setoh P, Wu D, Baillargeon R, Gelman R (2013) Young infants have biological ex-
pectations about animals. Proc Natl Acad Sci USA 110:15937–15942.
81. Simons DJ, Keil FC (1995) An abstract to concrete shift in the development of bi-
ological thought: The insides story. Cognition 56:129–163.
82. Haslam N, Rothschild L, Ernst D (2000) Essentialist beliefs about social categories. Br J
Soc Psychol 39:113–127.
83. Booth AE (2014) Conceptually coherent categories support label-based inductive
generalization in preschoolers. J Exp Child Psychol 123:1–14.
84. Sobel DM, Yoachim CM, Gopnik A, Meltzoff AN, Blumenthal EJ (2007) The blicket
within: Preschoolers’ inferences about insides and causes. J Cogn Dev 8:159–182.
85. Walker CM, Lombrozo T, Legare CH, Gopnik A (2014) Explaining prompts children to
privilege inductively rich properties. Cognition 133:343–357.
86. Rhodes M, Gelman SA (2009) A developmental examination of the conceptual
structure of animal, artifact, and human social categories across two cultural con-
texts. Cognit Psychol 59:244–274.
87. Roberts SO, Gelman SA (2015) Do children see in Black and White? Children’s and
adults’ categorizations of multiracial individuals. Child Dev 86:1830–1847.
88. Gelman SA, Wellman HM (1991) Insides and essences: Early understandings of the
non-obvious. Cognition 38:213–244.
89. Keil FC (1989) Concepts, Kinds, and Cognitive Development (MIT Press, Cambridge, MA).
90. Meyer M, Gelman SA (2016) Gender essentialism in children and parents: Implica-
tions for the development of gender stereotyping and gender-typed preferences.
Sex Roles 75:409–421.
91. Taylor MG, Rhodes M, Gelman SA (2009) Boys will be boys; cows will be cows:
Children’s essentialist reasoning about gender categories and animal species. Child
Dev 80:461–481.
92. Ware EA, Gelman SA (2014) You get what you need: An examination of purpose-
based inheritance reasoning in undergraduates, preschoolers, and biological ex-
perts. Cogn Sci 38:197–243.
93. Meyer M, Gelman SA, Roberts SO, Leslie SJ (November 17, 2016) My heart made me
do it: Children’s essentialist beliefs about heart transplants. Cogn Sci, 10.1111/
cogs.12431.
94. Meyer M, Leslie SJ, Gelman SA, Stilwell SM (2013) Essentialist beliefs about bodily
transplants in the United States and India. Cogn Sci 37:668–710.
95. Atran S, et al. (2001) Folkbiology doesn’t come from folkpsychology: Evidence from
Yukatek Maya in cross-cultural perspective. J Cogn Cult 1:3–42.
96. Moya C, Boyd R, Henrich J (2015) Reasoning about cultural and genetic transmission:
Developmental and cross‐cultural evidence From Peru, Fiji, and the United States on
how people make inferences about trait transmission. Top Cogn Sci 7:595–610.
97. del Río MF, Strasser K (2011) Chilean children’s essentialist reasoning about poverty.
Br J Dev Psychol 29:722–743.
98. Sousa P, Atran S, Medin D (2002) Essentialism and folkbiology: Evidence from Brazil.
J Cogn Cult 2:195–223.
99. Waxman S, Medin D, Ross N (2007) Folkbiological reasoning from a cross-cultural
developmental perspective: Early essentialist notions are shaped by cultural beliefs.
Dev Psychol 43:294–308.
100. Haslam N, Holland E, Karasawa M (2013) Essentialism and entitativity across cultures. Culture
and Group Processes, eds Yuki M, Brewer M (Oxford Univ Press, New York), pp 17–37.
101. Rhodes M, Leslie SJ, Saunders K, Dunham Y, Cimpian A (February 22, 2017) How does
social essentialism affect the development of inter-group relations? Dev Sci, 10.1111/
desc.12509.
Gelman and Roberts

102. Diesendruck G, Menahem R (2015) Essentialism promotes children’s inter-ethnic bias.
Front Psychol 6:1180.
103. Pauker K, Xu Y, Williams A, Biddle AM (2016) Race essentialism and social contextual
differences in children’s racial stereotyping. Child Dev 87:1409–1422.
104. Haslam N, Whelan J (2008) Human natures: Psychological essentialism in thinking
about differences between people. Soc Personal Psychol Compass 2/3:1297–1312.
105. Leyens JP, et al. (2001) Psychological essentialism and the differential attribution of
uniquely human emotions to ingroups and outgroups. Eur J Soc Psychol 31:395–411.
106. Bastian B, Haslam N (2008) Immigration from the perspective of hosts and immigrants:
Roles of psychological essentialism and social identity. Asian J Soc Psychol 11:127–140.
107. Gelman SA, Heyman GD (1999) Carrot-eaters and creature-believers: The effects of lexi-
calization on children’s inferences about social categories. Psychol Sci 10:489–493.
108. Gelman SA, Ware EA, Kleinberg F (2010) Effects of generic language on category
content and structure. Cognit Psychol 61:273–301.
109. Rhodes M, Leslie SJ, Tworek CM (2012) Cultural transmission of social essentialism.
Proc Natl Acad Sci USA 109:13526–13531.
110. Gelman SA, Waxman SR (2007) Looking beyond looks: Comments on Sloutsky, Kloos,
and Fisher (2007). Psychol Sci 18:554–555, discussion 556–557.
111. Diesendruck G (2003) Categories for names or names for categories? The interplay be-
tween domain-specific conceptual structure and language. Lang Cogn Process 18:759–787.
112. Gelman SA, Star JR, Flukes J (2002) Children’s use of generics in inductive inferences.
J Cogn Dev 3:179–199.
113. Gelman SA, Rhodes M (2012) Two-thousand years of stasis. How psychological es-
sentialism impedes evolutionary understanding. Evolution Challenges: Integrating
Research and Practice in Teaching and Learning about Evolution, eds Rosengren KS,
Brem S, Evans EM, Sinatra G (Oxford Univ Press, Cambridge, UK), pp 3–21.
114. Shtulman A, Schulz L (2008) The relation between essentialist beliefs and evolu-
tionary reasoning. Cogn Sci 32:1049–1062.
115. Dar-Nimrod I, Heine SJ (2011) Genetic essentialism: On the deceptive determinism of
DNA. Psychol Bull 137:800–818.
116. Leslie SJ (2013) Essence and natural kinds: When science meets preschooler intuition.
Oxford Studies in Epistemology 4:108–165.
117. Bonilla-Silva E (1997) Rethinking racism: Toward a structural interpretation. Am
Sociol Rev 62:465–480.
118. Lee J, Bean FD (2007) Reinventing the color line: Immigration and America’s new
racial/ethnic divide. Soc Forces 86:561–586.
119. Cimpian A, Salomon E (2014) The inherence heuristic: An intuitive means of making
sense of the world, and a potential precursor to psychological essentialism. Behav
Brain Sci 37:461–480.
120. Williams MJ, Eberhardt JL (2008) Biological conceptions of race and the motivation
to cross racial boundaries. J Pers Soc Psychol 94:1033–1047.
121. Bastian B, Haslam N (2006) Psychological essentialism and stereotype endorsement.
J Exp Soc Psychol 42:228–235.
122. Chao MM, Hong YY, Chiu CY (2013) Essentializing race: Its implications on racial
categorization. J Pers Soc Psychol 104:619–634.
123. Gaither SE, et al. (2014) Essentialist thinking predicts decrements in children’s
memory for racially ambiguous faces. Dev Psychol 50:482–488.
124. Ho AK, Roberts SO, Gelman SA (2015) Essentialism and racial bias jointly contribute
to the categorization of multiracial individuals. Psychol Sci 26:1639–1645.
125. Kraus MW, Keltner D (2013) Social class rank, essentialism, and punitive judgment.
J Pers Soc Psychol 105:247–261.
126. Leslie SJ, Cimpian A, Meyer M, Freeland E (2015) Expectations of brilliance underlie
gender distributions across academic disciplines. Science 347:262–265.
127. Goff PA, Jackson MC, Di Leone BAL, Culotta CM, DiTomasso NA (2014) The essence of
innocence: Consequences of dehumanizing Black children. J Pers Soc Psychol 106:526–545.
128. Goff PA, Eberhardt JL, Williams MJ, Jackson MC (2008) Not yet human: Implicit
knowledge, historical dehumanization, and contemporary consequences. J Pers Soc
Psychol 94:292–306.
129. Gelman SA (2009) Learning from others: Children’s construction of concepts. Annu
Rev Psychol 60:115–140.
130. Lutz DJ, Keil FC (2002) Early understanding of the division of cognitive labor. Child
Dev 73:1073–1084.
131. Markman EM, Jaswal VK (2003) Commentary on Part II: Abilities and assumptions
underlying conceptual development. Early Category and Concept Development:
Making Sense of the Blooming, Buzzing Confusion, eds Rakison D, Oakes L (Ox-
ford Univ Press, New York), pp 384–402.
132. Waxman SR, Markow DB (1995) Words as invitations to form categories: Evidence
from 12- to 13-month-old infants. Cognit Psychol 29:257–302.
133. Putnam H (1975) The meaning of ‘meaning’ Mind, Language, and Reality (Cam-
bridge Univ Press, Cambridge, UK), pp 215–271.
134. Legare CH, Nielsen M (2015) Imitation and innovation: The dual engines of cultural
learning. Trends Cogn Sci 19:688–699.
135. Tomasello M (2009) The Cultural Origins of Human Cognition (Harvard Univ Press,
Cambridge, MA).
136. Creanza N, Kolodny O, Feldman MW (2017) Cultural evolutionary theory: How cul-
ture evolves and why it matters. Proc Natl Acad Sci USA 114:7782–7789.
Gelman and Roberts
COLLOQUIUM
PAPER
137. d’Errico F, et al. (2017) Identifying early modern human ecological niche expansions
and associated cultural dynamics in the South African Middle Stone Age. Proc Natl
Acad Sci USA 114:7869–7876.
138. Markson L, Bloom P (1997) Evidence against a dedicated system for word learning in
children. Nature 385:813–815.
139. Horst JS, Samuelson LK (2008) Fast mapping but poor retention by 24-month-old
infants. Infancy 13:128–157.
140. Rhodes M (2012) Naïve theories of social groups. Child Dev 83:1900–1916.
141. Steels L (2011) Modeling the cultural evolution of language. Phys Life Rev 8:339–356.
142. Pickering MJ, Garrod S (2006) Alignment as the basis for successful communication.
Res Lang Comput 4:203–228.
143. Stolk A, Verhagen L, Toni I (2016) Conceptual alignment: How brains achieve mutual
understanding. Trends Cogn Sci 20:180–191.
144. Edmiston P, Lupyan G (2015) What makes words special? Words as unmotivated
cues. Cognition 143:93–100.
145. Gelman SA, Raman L (2003) Preschool children use linguistic form class and prag-
matic cues to interpret generics. Child Dev 74:308–325.
146. Brandone AC, Gelman SA, Hedglen J (2015) Children’s developing intuitions about
the truth conditions and implications of novel generics versus quantified statements.
Cogn Sci 39:711–738.
147. Hacking I (1995) The looping effects of human kinds. Causal Cognition: A
Multidisciplinary Debate, eds Sperber D, Premack D, Premack AJ (Clarendon Press,
Oxford, UK), pp 351–394.
148. Whiten A, Horner V, de Waal FB (2005) Conformity to cultural norms of tool use in
chimpanzees. Nature 437:737–740.
149. Seyfarth RM, Cheney DL (2009) The evolution of social categories. Neurobiology of.
Umwelt, eds Bethoz A, Christen A (Springer, Berlin), pp 69–87.
150. Phillips W, Shankar M, Santos LR (2010) Essentialism in the absence of language?
Evidence from rhesus monkeys (Macaca mulatta). Dev Sci 13:F1–F7.
151. Cacchione T, Hrubesch C, Call J, Rakoczy H (2016) Are apes essentialists? Scope and
limits of psychological essentialism in great apes. Anim Cogn 19:921–937.
152. Haun DB, Rekers Y, Tomasello M (2014) Children conform to the behavior of peers;
other great apes stick with what they know. Psychol Sci 25:2160–2167.
153. Hamlin JK, Wynn K, Bloom P (2007) Social evaluation by preverbal infants. Nature
450:557–559.
154. Newman GE, Herrmann P, Wynn K, Keil FC (2008) Biases towards internal features in
infants’ reasoning about objects. Cognition 107:420–432.
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
155. Dewar K, Xu F (2007) Do 9-month-old infants expect distinct words to refer to kinds?
Dev Psychol 43:1227–1238.
156. Galef BG, McQuoid LM, Whiskin EE (1990) Further evidence that Norway rats do not
socially transmit learned aversions to toxic baits. Anim Learn Behav 18:199–205.
157. Galef BG, Laland KN (2005) Social learning in animals: Empirical studies and theo-
retical models. Bioscience 55:489–499.
158. Love AC (2015) Conceptual Change in Biology (Springer, New York).
159. Savoca MS, Wohlfeil ME, Ebeler SE, Nevitt GA (2016) Marine plastic debris emits a
keystone infochemical for olfactory foraging seabirds. Sci Adv 2:e1600395.
160. Regnier D (2015) Clean people, unclean people: The essentialisation of ‘slaves’
among the southern Betsileo of Madagascar. Soc Anthropol 23:152–168.
161. Gelman SA, Heyman GD, Legare CH (2007) Developmental changes in the coherence
of essentialist beliefs about psychological characteristics. Child Dev 78:757–774.
162. Cooper JA, Marsh JK (2015) The influence of expertise on essence beliefs for mental
and medical disorder categories. Cognition 144:67–75.
163. Gelman SA (2013) Artifacts and essentialism. Rev Phil Psychol 4:449–463.
164. Nemeroff C, Rozin P (1994) The contagion concept in adult thinking in the United
States: Transmission of germs and of interpersonal influence. Ethos 22:158–186.
165. Newman GE (2016) An essentialist account of authenticity. J Cogn Cult 16:294–321.
166. Riedl K, Jensen K, Call J, Tomasello M (2012) No third-party punishment in chim-
panzees. Proc Natl Acad Sci USA 109:14824–14829.
167. Riedl K, Jensen K, Call J, Tomasello M (2015) Restorative justice in children. Curr Biol
25:1731–1735.
168. Göckeritz S, Schmidt MH, Tomasello M (2014) Young children’s creation and trans-
mission of social norms. Cogn Dev 30:81–95.
169. Lewis HM, Laland KN (2012) Transmission fidelity is the key to the build-up of cu-
mulative culture. Philos Trans R Soc Lond B Biol Sci 367:2171–2180.
170. Heyes C (2016) Who knows? Metacognitive social learning strategies. Trends Cogn
Sci 20:204–213.
171. Levinson SC (2005) Comment on: Cultural constraints on grammar and cognition in
Piraha by Daniel L. Everett. Curr Anthropol 46:637–638.
172. Wellman HM (2014) Making Minds: How Theory of Mind Develops (Oxford Univ
Press, New York).
173. Harris PL, Lane JD (2014) Infants understand how testimony works. Topoi 33:
443–458.
174. Walker CM, Gopnik A (2014) Toddlers infer higher-order relational principles in
causal learning. Psychol Sci 25:161–169.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7907
Coevolution of cultural intelligence, extended life
history, sociality, and brain size in primates
Sally E. Streeta,b,1, Ana F. Navarretea, Simon M. Readerc, and Kevin N. Lalanda,1
a
Centre for Social Learning and Cognitive Evolution, School of Biology, University of St. Andrews, St. Andrews KY16 9AJ, United Kingdom; bDepartment of
Anthropology, Durham University, Durham DH1 3LE, United Kingdom; and cDepartment of Biology, McGill University, Montreal, QC H3A 1B1, Canada
Edited by Marcus W. Feldman, Stanford University, Stanford, CA, and approved May 30, 2017 (received for review January 15, 2017)
Explanations for primate brain expansion and the evolution of human
cognition and culture remain contentious despite extensive research.
While multiple comparative analyses have investigated variation in
brain size across primate species, very few have addressed why
primates vary in how much they use social learning. Here, we evaluate
the hypothesis that the enhanced reliance on socially transmitted
behavior observed in some primates has coevolved with enlarged
brains, complex sociality, and extended lifespans. Using recently
developed phylogenetic comparative methods we show that, across
primate species, a measure of social learning proclivity increases with
absolute and relative brain volume, longevity (specifically reproductive
lifespan), and social group size, correcting for research effort. We also
confirm relationships of absolute and relative brain volume with
longevity (both juvenile period and reproductive lifespan) and social
group size, although longevity is generally the stronger predictor.
Relationships between social learning, brain volume, and longevity
remain when controlling for maternal investment and are therefore not
simply explained as a by-product of the generally slower life history
expected for larger brained species. Our findings suggest that both
brain expansion and high reliance on culturally transmitted behavior
coevolved with sociality and extended lifespan in primates. This
coevolution is consistent with the hypothesis that the evolution of
large brains, sociality, and long lifespans has promoted reliance on
culture, with reliance on culture in turn driving further increases in brain
volume, cognitive abilities, and lifespans in some primate lineages.
|
cultural evolution social learning | brain evolution | primates |
phylogenetic comparative analysis
B rain expansion is unquestionably a distinctive feature of
primate, and especially human, evolution. Primate brain
expansion is evident regardless of whether the brain is measured
in absolute terms, in relation to body size, or as the size of the
neocortex relative to the rest of the brain (1), and irrespective of
whether it is better characterized by variation in a single size
dimension (2) or mosaic evolution of component parts (3). The
striking variation in brain size in nonhuman primates, across
three orders of magnitude (4), has long demanded an evolu-
tionary explanation (5). Although the cognitive implications of
cross-species variation in whole brain size remain contentious
and require further investigation (5–7), evolutionary increases in
overall brain size in primates reflect neuroanatomical changes
that are plausibly linked to increases in general cognitive abilities.
For instance, larger primate brains have more neurons in absolute
terms (8–11), with coordinated expansion particularly in the neo-
cortex and cerebellum (12), potentially supporting a greater diversity
of cognitive functions (7, 10). In support of this idea, overall brain
size increases with broad measures of cognitive ability in primates,
including performance in laboratory tests of learning and cognition
across primate genera (13) and performance in experimental mea-
sures of behavioral inhibition across primate species (14).
At ∼1,500 g (15), human brains are at least three times heavier
than those of any other primate species (1). However, humans
are also extreme in their long lifespan, social complexity, cog-
nition, and cultural capabilities (16, 17), raising questions about
whether large brains, long lives, complex cognition, and advanced
7908–7914 | PNAS | July 25, 2017 | vol. 114 | no. 30
cultural capabilities evolved independently or coevolved through di-
rectly reinforcing processes. Enlarged brains, enhanced cognition, and
highly developed social learning abilities co-occur not only in primate
species but also in some cetaceans and birds (18–22), raising the pos-
sibility of a key role for social learning and culture in brain evolution
and intelligence in multiple, independent animal lineages (23–27).
Across primates, support for multiple, nonexclusive hypotheses
for enlarged brain (particularly neocortex) size has been identified
in comparative studies, emphasizing the roles of social complexity
(e.g., group size) (28, 29), ecological intelligence (e.g., dietary
complexity) (30, 31), technical intelligence (e.g., tool use and
technical innovation) (21, 25, 32), and behavioral complexity (e.g.,
innovativeness, social learning, and tactical deception) (21, 25, 33).
Further, several comparative studies have found that larger
brained primates have slower life histories, including longer juvenile
periods and overall lifespans (e.g., ref. 29). Although mutually
reinforcing evolutionary processes have been proposed to account
for this association (16), recent comparative analyses suggest that
lifespan increases with brain size in mammals instead due to de-
velopmental costs: i.e., it requires a longer period of maternal in-
vestment to support offspring with greater natal and postnatal brain
growth, requiring a slower life history strategy of which longer life-
span is a by-product (34). Primates, however, are potentially distinct
from most mammalian taxa in their unusually large, neuron-dense
brains (8–11) and in the extensive occurrence of socially transmitted
behavior exhibited in some lineages (e.g., refs. 35–37). Whether the
association between extended life history and enlarged brain size is
best explained by a cognitive or developmental mechanism in pri-
mates specifically remains to be explored. Further, despite many
previous comparative analyses of brain size and relevant predictors
in primates, comparative analyses have not yet directly explored the
evolutionary relationships between brain expansion, cultural com-
plexity, sociality, and longevity in analyses that include all of these
variables, with control for relevant potentially confounding variables.
Here, in a comparative analysis of primate species, we directly
test the widely held view that encephalization, sociality, longev-
ity, and reliance on culture have coevolved (16, 23–27, 32, 38).
We use a quantitative behavioral measure of reliance on culture:
specifically, the number of unique reports (i.e., richness) of social
learning per species from a sample of relevant published litera-
ture (21, 39) (henceforth referred to simply as “social learning”)
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: S.E.S. and K.N.L. designed research with contributions from A.F.N.
and S.M.R.; S.E.S. performed research; S.E.S. analyzed data; and S.E.S., A.F.N., S.M.R., and
K.N.L. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1
To whom correspondence may be addressed. Email: knl1@st-andrews.ac.uk or
sallystreet13@gmail.com.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1620734114/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1620734114

(see SI Appendix for further details on this measure). We use
Bayesian phylogenetic mixed models to investigate a cluster of
related hypotheses concerned with the evolutionary relationships
between social learning, brain volume, group size, and lifespan.
Below, we specify and test four predictions, each of which is
independent; however, if all are supported, it would imply sup-
port for a cluster of related, and mutually consistent, ideas
concerning the factors underlying evolutionary expansions of
brains, cognition, and culture.
Prediction 1: Social Learning Increases with Absolute and
Relative Brain Volume
This expectation follows from the hypotheses that (i) high levels
of knowledge and skill are required for primates to exploit high-
quality, difficult-to-access dietary resources, with these skills
primarily acquired through social learning (16, 23–27, 32, 38, 40)
and (ii) the energy so acquired is critical to developing and
running a large brain (16, 27). Previous comparative analyses
have identified positive associations of social learning with the
absolute and relative sizes of brain components, primarily the
neocortex (21, 25). Here, we extend these analyses to overall
brain size measured as endocranial volume (ECV), allowing for
much larger (at least threefold) sample sizes, far more repre-
sentative of the diversity in brain size across primate species (4).
Prediction 2: Social Learning Increases with Longevity
This expectation follows from the hypotheses that (i) extended life
history, particularly a longer lifespan and period of juvenile de-
pendence, facilitates the acquisition, exploitation, and social trans-
mission of life skills (16, 23, 40) and (ii) cultural knowledge promotes
survival and long lives (25–27) by acting as a “cognitive buffer,”
enhancing survival in challenging environmental conditions through
behavioral responses (41, 42). Complex skills frequently take time to
learn; therefore, longer lifespans potentially provide more time for
relevant experience to accrue, more time for adults to benefit from
knowledge acquired earlier in life, and more time for parents to pass
on relevant skills to offspring (16, 23, 26, 27, 40). If an extended
juvenile period in particular is critical for the acquisition of adaptive
socially transmitted behavior (16), we expect that juvenile period has
a strong association with social learning richness. However, costly
investment in learning socially transmitted skills may pay off in later
life only across a long reproductive lifespan (16); therefore, we may
expect the association between social learning and longevity to be
driven more strongly by increases in reproductive lifespan. If there is
a specific relationship of social learning with longevity, not con-
founded by relationships of either with absolute or relative brain size,
we should still find this association even when controlling for brain
volume and body mass. Furthermore, if reliance on socially trans-
mitted behavior is related to longevity via a cognitive buffer mech-
anism rather than as a by-product of a relationship between social
learning, brain volume, and slower life history traits due to de-
velopmental constraints, this relationship should remain when con-
trolling for the potentially confounding effect of maternal investment
(measured as the sum of gestation and lactation periods) (34).
Prediction 3: Social Learning Increases with Group Size
This expectation follows from several theoretical and empirical
analyses showing that large social groups support greater amounts
of adaptive cultural knowledge (e.g., refs. 43–46) and broader
hypotheses that stable social grouping supports the evolution of
reliance on social learning (e.g., ref. 20). If the relationship
of social learning to group size is not confounded by associations
of either trait with absolute brain volume, relative brain volume, or
longevity, this prediction should hold when controlling for brain
volume, body mass, and longevity measures.
Street et al.
COLLOQUIUM
PAPER
Prediction 4: Absolute and Relative Brain Volume Increases
with Longevity
Across mammals, a relationship between adult brain mass and
longevity is not supported when controlling for maternal in-
vestment, suggesting that developmental constraints associated
with investing in large-brained offspring underpin this associa-
tion (34). However, if associations of longevity with absolute and
relative brain volume remain when maternal investment is in-
cluded in analyses of primate species, the relationship between
brain volume and lifespan is not confounded with maternal in-
vestment, thus potentially indicative of a cognitive, rather than
solely developmental, mechanism underpinning this relationship
in primates specifically, even if not across mammals more gen-
erally. Additionally, if longevity is related to brain volume in-
dependently of any potentially confounding effect of social group
size, these associations should remain intact when group size is
included in statistical models.
Results
Prediction 1: Social Learning and Brain Volume. As predicted, social
learning richness increases with both absolute brain volume
(<1% β coefficients in the posterior distribution crossing zero,
n = 150) (SI Appendix, Table S1i) and relative brain volume (3%
β crossing zero, n = 150) (SI Appendix, Table S1ii).
EVOLUTION
Prediction 2: Social Learning and Longevity. As predicted, social
learning richness increases with longevity (<1% β crossing zero,
n = 117) (SI Appendix, Table S2 A, i and Fig. 1A). We found no
evidence that social learning increases with juvenile period length,
however (58% β crossing zero, n = 101) (SI Appendix, Table S2 B,
i and Fig. 1A). Rather, social learning increases with reproductive
lifespan specifically (0% β crossing zero, n = 92) (SI Appendix,
Table S2 C, i). Relationships between social learning and lon-
gevity, and between social learning and reproductive lifespan, re-
main intact when maternal investment (summed gestation and
lactation time) is included as an additional predictor (2%, <1% β
crossing zero, n = 87, n = 82, respectively) (SI Appendix, Table S2
A, ii and C, ii) whereas maternal investment itself does not predict
social learning in these models (≥35% β crossing zero) (SI Appendix,
Table S2 A, ii and C, ii). Relationships between social learning and
longevity or reproductive lifespan are also not confounded by
those between social learning and absolute or relative brain vol-
ume, as they remain when either brain volume or both brain
volume and body mass are included as additional predictors (<1%
β crossing zero, n = 111, n = 89) (SI Appendix, Table S2 A, iii and
iv and C, iii and iv). However, brain volume itself does not predict
social learning when included alongside longevity measures (>22%
β crossing zero) (SI Appendix, Table S2 A, iii and iv and C, iii and iv).
Prediction 3: Social Learning and Group Size. As predicted, we found
a positive association between group size and social learning (<1%
β crossing zero, n = 167) (SI Appendix, Table S3i and Fig. 1A).
This association is independent of the relationship between social
learning and longevity or reproductive lifespan, as it remains when
either of these life history traits is included (4% β crossing zero,
5% β crossing zero, n = 111, n = 89) (SI Appendix, Table S3 ii, A
and B). The relationship between group size and social learning is
also not confounded by the association of either trait with absolute
or relative brain volume, because it remains when either brain
volume or both brain volume and body mass are included as ad-
ditional predictors (<4% β crossing zero, n = 140) (SI Appendix,
Table S3 iii and iv). Both absolute and relative brain volume have,
however, a weaker effect on social learning when group size is
included as an additional predictor (2%, 7% β crossing zero)
(SI Appendix, Table S3 iii and iv) compared with models
without group size (<1%, 3% β crossing zero) (SI Appendix,
Table S1 i and ii).
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7909
 A B C

Fig. 1. Posterior distributions of β coefficients for the effects of longevity, juvenile period, and group size on (A) social learning richness, (B) absolute brain
volume, and (C) relative brain volume (i.e., brain volume accounting for body mass). Here, we present effects from the simplest models, including only either
longevity, juvenile period, or group size as independent variables, together with research effort and body mass for the social learning model, and body mass
for the relative brain model. However, these results are not strongly affected by the inclusion of additional potentially confounding variables (Methods,
Results, and SI Appendix). Percentages indicate the percentage of posterior estimates that cross zero in the opposite to the predicted direction for each effect.
Distributions shifted substantially away from zero indicate evidence for effects of predictor variables in the corresponding direction whereas those centered
close to zero indicate little or no evidence for effects of predictor variables.

Prediction 4: Predictors of Brain Volume. We confirmed the expec- Discussion

ted positive association of absolute brain volume with social
group size (3% β crossing zero, n = 151) (SI Appendix, Table S4i
and Fig. 1B). Absolute brain volume also increases with lon-
gevity, juvenile period length, and reproductive lifespan (<1% β
crossing zero, n = 112, n = 98, n = 90) (SI Appendix, Table S4 ii,
A, B, and C and Fig. 1B). Relationships between absolute brain
volume and longevity, juvenile period, and reproductive lifespan
remain intact when maternal investment is included in the
model, which itself also increases with brain volume (all <1% β
crossing zero, n = 84, n = 86, n = 79) (SI Appendix, Table S4 iii,
A, B, and C). Relationships between longevity, juvenile period,
and reproductive lifespan with absolute brain volume are in-
dependent of the association of brain volume and group size,
remaining intact when group size is included as an additional
predictor (all <1% β crossing zero, n = 106, n = 95, n = 87) (SI
Appendix, Table S4 iv, A, B, and C) whereas group size is a rel-
atively weak predictor when included with longevity or re-
productive lifespan (>6% β crossing zero) (SI Appendix, Table S4
iv, A and C).
Similarly, relative brain volume increases with social group size
(4% β crossing zero, n = 151) (SI Appendix, Table S5i and Fig.
1C). Relative brain volume also increases with longevity, juvenile
period, and reproductive lifespan (<1% β crossing zero, n = 112,
n = 98, n = 90) (SI Appendix, Table S5 ii, A, B, and C and Fig. 1C).
Again, associations between relative brain volume and all three
life history measures remain intact when controlling for maternal
investment, which itself also increases with relative brain volume
(all <1% β crossing zero, n = 84, n = 86, n = 79) (SI Appendix,
Table S5 iii, A, B, and C). The relationship between relative brain
volume and life history length is not confounded by social group
size because all three measures remain intact when group size is
added to the model (<1% β crossing zero, n = 106, n = 95, n = 87)
(SI Appendix, Table S5 iv, A, B, and C). When included with
longevity or reproductive lifespan, however, group size is not
strongly supported as a predictor of relative brain volume (>12%
β crossing zero) (SI Appendix, Table S5 iv, A and C).
Parameters from all statistical models are reported in full in SI
Appendix, Tables S1–S5. All results reported in the main text
refer to models including great apes, but none of our main re-
sults are qualitatively affected by removing these species (n = 4)
from analyses (SI Appendix, Tables S1–S5). Variation in social
learning, longevity, group size, and brain volume data across
primate genera is illustrated in Fig. 2.
We investigated the widely held view that cultural intelligence,
extended life history, sociality, and brain size have coevolved in
nonhuman primates (16, 23–27). Using Bayesian phylogenetic
generalized linear mixed models, we found a positive relation-
ship between reliance on culture (as measured by reported
richness of social learning, corrected for research effort) and
measures of both absolute and relative brain volume. Earlier
studies had established positive relationships between primate
social learning and both absolute and ratio measures of the size of
the “executive brain” (combined neocortex and striatum volume)
(25) and that social learning, as a component of a composite
measure of general cognitive ability, increases with absolute and
ratio measures of neocortex size and with executive brain ratio
(21). Here, we found that these associations generalize further to
overall brain size measured as endocranial volume, across a sub-
stantially larger (>3×) sample of primate species. Although its
occurrence in insects demonstrates that large brains, in absolute
terms, are not a prerequisite for social learning (47), evolutionary
expansions in brain size may support more efficient, high-fidelity,
or more diverse forms of social transmission (25), due to increases
in, for instance, cross-modal integration of perceptual and motor
information and the general computational power and flexibility
required to implement sophisticated learning strategies (40, 48).
Evolutionary expansion of the primate brain is driven substantially
by visual specialization (5, 49, 50) and coordinated expansion of
the neocortex and cerebellum, with likely corresponding increases
in fine visuo-motor control, which may underpin the ability to
replicate complex behavioral sequences inherent to high-fidelity
social learning (5, 12). In turn, more effective social learning po-
tentially allows individuals to garner high-quality dietary resources
that can be invested in brain growth (16). Therefore, while the
cognitive mechanisms underpinning social learning largely remain
to be established (51), it remains highly plausible that evolutionary
increases in overall brain size are associated with elevated social
learning capabilities. However, our finding that brain volume does
not predict social learning when accounting for longevity, together
with strong links between social learning and longevity, and longevity
and brain size, suggests that the association of social learning and
brain volume may be indirect, mediated via increased longevity.
Results support a similar picture for relationships between social
learning, group size, and brain volume. While we cannot rule out
the possibility that the reduced sample size and likely correspond-
ing reduction in power with the inclusion of additional variables

7910 | www.pnas.org/cgi/doi/10.1073/pnas.1620734114 Street et al.

 COLLOQUIUM
PAPER
Galagoides
Otolemur
Galago
A
Euoticus
Nycticebus
Loris
Perodicticus
Arctocebus
Daubentonia
Varecia
Lemur
Hapalemur
Eulemur
Propithecus
Lepilemur
Cheirogaleus
Mirza
Microcebus
Tarsius
Pithecia
Chiropotes
B
Cacajao
Saimiri
Cebus
Saguinus
Leontopithecus
Callithrix
Callimico
Aotus
Lagothrix
Ateles
Alouatta
Pongo
Pan
Gorilla
Symphalangus
Hylobates
Presbytis
Trachypithecus
C
Semnopithecus
Pygathrix
Nasalis
Colobus
Macaca
Papio
Social learning Theropithecus
Lophocebus
Brain volume Mandrillus
Group size Cercocebus
Longevity Erythrocebus
Cercopithecus
Allenopithecus

Fig. 2. Summary of raw data on social learning, absolute brain volume, group size, and longevity for 52 primate genera, using the consensus phylogeny
from 10ktrees (65). For illustration purposes only, all data are summarized as genus-level means, standardized with minimum 0 and maximum 1. Also for

EVOLUTION
illustration purposes only, social learning is displayed as a proportion of research effort whereas, in statistical analyses, social learning is controlled for
research effort by including research effort as an independent variable. Images show (A) bearded capuchin (Cebus libidinosus), (B) chimpanzees (Pan
troglodytes), and (C ) guinea baboons (Papio papio), illustrating lineages that represent convergent coevolution of high social learning abilities, large
brain volumes, complex social relationships, and long lifespans. (A) Courtesy of Flickr/Bart van Dorp, (B) courtesy of Flickr/USAID in Africa, and (C) courtesy of Flickr/
William Warby.

accounts for the loss of a direct relationship between social
learning and brain size, these results are consistent with a previous
exploratory phylogenetic path analysis showing that social learning
and brain volume are related indirectly via links with dietary,
social, or life history traits (32).
The positive relationship between social learning and longevity
we identify supports the idea that longer lifespans provide spe-
cies reliant on culture more time to learn novel skills, more time
to “cash in” on those skills once learned, and more time to pass
them on to their offspring (16, 23, 40). Additionally, longer
lifespans may confer greater opportunity for behavioral innova-
tions, providing the raw material for social transmission, because
longer lifespans are positively associated with greater propensity
to innovate in birds (52) and in primates (ref. 32, albeit in-
directly). Culturally acquired knowledge is typically adaptive and
may often promote growth and survival of both learners and
their dependent young, and thereby extend lifespans (25–27) via
a cognitive buffer effect whereby social learning allows individ-
uals to adapt behaviorally to challenging environments (41, 42).
These benefits may be sufficient to compensate for negative fit-
ness consequences associated with reliance on social learning,
such as increased risk of social transmission of parasites (39).
Although hypotheses for the coevolution of lifespan and culture
propose that increases in both juvenile period and overall lifespan
are related to reliance on culturally transmitted knowledge (e.g.,
ref. 16), here, we found that the association between social learn-
ing and longevity is driven by an increased reproductive lifespan,
rather than an extended period of juvenile dependence. It
remains possible that a link between extended juvenile periods
and social learning capabilities will be identified in future studies
using novel social learning measures, such as those based on
experimental tests. Nonetheless, our current findings suggest
that an extended reproductive lifespan, during which enhanced
fitness benefits of earlier costly investment in learning skills
for survival can be reaped, primarily drives the association
between social learning and lifespan that we identify here.
Our finding that the relationship between longevity and social
learning remains when measures of maternal investment are
included in analyses supports these functional arguments and
argues against an interpretation solely in terms of developmental
constraints, in primates at least. Therefore, in primates, the
combination of social learning with large brains may provide a
cognitive buffer against environmental unpredictability, im-
proving survival and permitting long lives. Primates may con-
trast with most mammalian lineages in this regard due to the
unusually extensive reliance on culturally transmitted behav-
ior seen in certain lineages (e.g., refs. 35–37), perhaps nec-
essary for social learning to buffer individuals sufficiently
against environmental risks.
Our finding of a positive relationship between social learning
and group size supports the expectation that large, stable social
groups support greater amounts of adaptive cultural knowledge
and facilitate a greater reliance on social learning (20). Although
this hypothesis is well-established in theoretical models (e.g., refs.
43, 44) and has found recent empirical support in human historical
(45) and experimental (46) studies, previous comparative phylo-
genetic analyses have failed to find this relationship across primate
species (21, 25). The fact that we found a positive association here
most likely reflects the greater power of our analyses compared
with earlier studies, due to the availability of a larger group size
database (53) and phylogenetic comparative methods that adjust
phylogenetic signal according to the traits included in the model
(SI Appendix, Methods), contrasting with the older independent
contrasts method, which effectively assumes a maximum level of
phylogenetic signal and can therefore be overly conservative (54).
The relationship between social learning and group size remains
when longevity, brain volume, and body mass are included and
therefore seems not to be simply a by-product of the relationship
between group size and absolute or relative brain volume, or
confounded by life history traits.
Both large social groups and extended longevity (including
increases in juvenile period and reproductive and total lifespan)
are associated with enlarged brain volume, whether measured in
absolute terms or relative to body mass. Group size has proven a

Street et al. PNAS | July 25, 2017 | vol. 114 | no. 30 | 7911
robust predictor of measures of brain size, particularly relative
neocortex size (29, 55, 56), and it remains an important predictor
of both absolute and relative whole brain volume, as well as
social learning, in our analyses. Thus, our findings support pre-
vious studies claiming an important role for social intelligence in
primate brain evolution (e.g. 29, 55–57). However, when in-
cluded together with longevity, longevity is independently related
to brain volume whereas group size becomes a fairly weak pre-
dictor. This result may be significant because the association of
brain volume and longevity is usually not regarded as directly
causally relevant in brain evolution (e.g., ref. 29). Further, a
recently published comparative analysis suggests that dietary
factors, rather than sociality, are the primary drivers of increased
relative brain size in primates (31). It remains to be seen whether
these findings generalize to measures of neocortex volume, ar-
guably more relevant to social intelligence (29, 55–57). None-
theless, together, these results reinforce an emerging consensus
that sociality is not the sole driver of primate brain evolution but
rather is embedded in a nexus of evolutionary conditions that
favor brain expansion, including dietary, ecological, life history,
and behavioral factors (12, 16, 21, 25, 29, 32).
Across mammals more broadly, the relationship between adult
brain mass and longevity is accounted for by patterns of maternal
investment and is generally interpreted as a manifestation of de-
velopmental costs of producing larger brained offspring, rather than
necessarily due to any cognitive or behavioral mechanism (34).
Here, however, we found that the associations of longevity with
absolute and relative brain volume remain when controlling
for maternal investment. Therefore, in primates, compared with
mammals in general (34), variation in adult brain size across species
cannot be fully accounted for by patterns of maternal investment,
and the relationship between brain size and lifespan is potentially
indicative of a cognitive buffering (41, 42), rather than solely de-
velopmental, mechanism through which cultural intelligence facil-
itates survival. This contrast can perhaps be explained by divergent
scaling relationships between brain volume and neuron number
(potentially a more relevant correlate of cognitive capacity) (7, 10,
12) in primates compared with other mammalian lineages. Unlike
nonprimate mammalian lineages, such as rodents, in which neuron
size increases and neuron density decreases with increased brain
volume, in primates, the number of neurons increases approxi-
mately isometrically with brain volume (8–11). Therefore, in pri-
mates, larger brains may confer stronger benefits in terms of
increased cognitive function and behavioral flexibility com-
pared with other mammalian lineages. Overall, together with
the strong relationship between social learning and longevity,
these findings are consistent with the hypotheses that cultural
knowledge facilitates survival and that extended longevity fa-
cilitates the acquisition, exploitation, and social transmission
of life skills (16, 23, 25–27, 40).
Our finding that longevity is a strong, and potentially causally
significant, predictor of both brain volume and social learning
richness is evocative of the argument that intelligence and life-
history length have coevolved in humans because our intellectual
abilities allowed us to exploit high-quality, but difficult-to-access,
food resources, with the nutrients gleaned “paying” for brain
growth, and with increased longevity favored because it allowed
more time to cash in on complex, and difficult to master, for-
aging skills, with fitness benefits that pay off later in life (16).
High levels of knowledge, skill, coordination, and strength are
required to exploit the high-quality dietary resources consumed
by humans and other apes. Consistent with this idea, the most
common use of social learning in primates seems to be in ac-
quiring foraging skills, as ∼50% of reports of social learning in
a prior compilation occurred within the context of foraging
(25, 58). Complex tool use and extractive foraging abilities re-
quire time to acquire, but, in larger brained animals, an ex-
tended learning phase, during which productivity is low, can be
7912 | www.pnas.org/cgi/doi/10.1073/pnas.1620734114
compensated for by higher productivity during the adult period,
provided there is an intergenerational flow of both food and
knowledge from old to young (59). Our results are therefore
broadly consistent with a cultural intelligence explanation (23–
27) manifested in particular primate lineages showing high re-
liance on social learning, in which selection for efficient social
learning has allowed energy gains in diet, which in turn fueled
brain growth, and generated selection for extended longevity.
Previous comparative phylogenetic analyses have found social
learning to covary positively with rates of behavioral innovation
and tool use in primates (21, 25). Additionally, the best supported
graphs in exploratory phylogenetic path analyses link technical
innovation directly to brain size and social learning and non-
technical innovation indirectly to brain volume via diet and life-
history measures (32). Together with the current study, this body of
findings is consistent with the hypothesis that cultural intelligence,
as manifested by a cluster of behavioral traits, including social
learning, innovation, and tool use, may have been a significant
driver of primate brain evolution. However, we highlight two notes
of caution in particular. First, the majority of primates exhibit
comparatively little social learning (Fig. 2) (at least, as reflected in
our database), which implies that any selection for cultural in-
telligence has operated primarily in a small number of large-
brained primate lineages. Second, our social learning measure is
largely based on observational reports, not controlled experimental
tests, whereas social learning is challenging to identify from ob-
servation alone (21, 25). However, this approach provides a more
naturalistic comparative measure of social learning in comparison
with those based on experimental tests, representing a far broader
range of primate behavioral diversity, necessary for large-scale
comparative investigations (21, 25, 32, 39, 60). Results based on
patterns of observational accounts of social learning across species
should be valuable in informing and directing future, larger scale
comparative experimental investigations of variation in social
learning abilities across species (21, 39, 61).
One comprehensive way to interpret these findings is to recog-
nize multiple waves of selection for enlarged brains and enhanced
cognition in primates. In addition to selection for the cognitive skills
required for complex social lives (29) and dietary niches (31)
characteristic of some primate taxa, our results imply a likely later
bout of selection for cultural intelligence among a restricted number
of large-brained primate lineages. The latter notably include the
great apes, but also other independent lineages such as capuchins
and baboons (Fig. 2), as our results are not contingent on the in-
clusion of great apes. Plausibly, complex sociality and foraging may
have led to the evolution of large-brained primate lineages, some of
which passed a critical threshold in reliance on socially learned
behaviors, leading to mutually reinforcing selection for increased
brain size, cognitive abilities, and reliance on social learning and
innovation, mediated by conferred increases in longevity and diet
quality. The twin challenges of complex socioecological niches and
reliance on culture may therefore best account for the evolution of
large brains, advanced cognition, and extended lifespans in pri-
mates. However, our analyses do not allow the direction of causality
to be inferred, and other interpretations, for instance, in which large
brains evolved for other reasons, subsequently allowing for gains in
social and cultural complexity, are equally supported by the findings
presented here.
Our results do, however, strongly suggest a strong coevolutionary
relationship among cultural intelligence, brain size, sociality, and
life-history length in primates. Although we have focused here on
nonhuman primates, broader comparative trends support the idea
that enlarged brain size, general cognitive abilities, and reliance on
culture may have coevolved in other long-lived, highly social line-
ages, including some birds (e.g., corvids and parrots) and toothed
whales (18–20, 22). These associations may be mutually reinforcing
(24), with positive feedback loops reaching their zenith in humans,
Street et al.
 COLLOQUIUM
PAPER
who are extreme in their encephalization, intelligence, culture, and controlled for body mass in models in which life history traits predicted so-
lifespan (23, 62). cial learning as the outcome variable, due to the well-established association
of larger adult body size with slower life histories (e.g., ref. 67). For models
Methods including longevity, we reran analyses including maternal investment as an
additional predictor to account for its potentially confounding effect on
Data Compilation. All data used in analyses were obtained from existing
brain volume and longevity (34). Namely, if associations of brain volume
published datasets, referenced in full below, with additional details in SI
and/or social learning with longevity are confounded with maternal in-
Appendix, Methods.
vestment, we expect to find that, when included together with longevity,
Endocranial volume (ECV, in cubic centimeters) and body mass (in grams)
only maternal investment is a strong predictor of brain volume and/or social
data were obtained from ref. 4. Because ECV reflects the interior volume of
learning (as in ref. 34). Models including longevity as a predictor were also
the cranial cavity, including not only the volume of the brain but also the
rerun using either juvenile period length (age of sexual maturity) or re-
volume of protective structures of the brain, such as the meninges (4), and
productive lifespan (longevity minus juvenile period), to investigate whether
does not allow for separate estimates of the volumes of individual brain compo-
any identified relationships with longevity were driven by increases in juvenile
nents, it is a relatively crude brain measure (6). Nonetheless, ECV is strongly and
period length, reproductive lifespan, or both. To investigate whether group
near isometrically related to brain mass in primates (4), which is itself related ap-
size and longevity predicted brain volume and social learning independently
proximately isometrically to neuron number (8–11). Moreover, brain volume esti-
of each other, we ran additional models in which both group size and lon-
mates from ECV (hereafter “brain volume”) are available for around three times
gevity were included as predictors. We reran all analyses without great apes,
more primate species (n = 184 species) (SI Appendix, Methods) than for volumes
a potentially influential group due to their high social learning richness and
of individual brain structures (neocortex, cerebellum, etc.; typically ∼60 species)
large brains (Fig. 2), and due to potential researcher biases toward identi-
(e.g., ref. 63), allowing for analyses far more representative of the range of in-
fying social learning in apes compared with monkeys (SI Appendix, Methods).
terspecific variation in primate brain size (4). Further, because size estimates from
We found that none of our key findings were affected, demonstrating that
brain tissue can be influenced by variation in environmental effects, such as the
our results are robust to removal of potential outliers and to possible biases
age and life experience of the individual, along with variation in preservation
associated with this group (SI Appendix, Tables S1–S5).
techniques (6), ECV may be a more consistent measure of species-typical brain
To analyze data, we used Bayesian phylogenetic generalized linear mixed
size than those derived from direct measurements of volume or mass (4).
models, which allow for control for phylogenetic nonindependence and for
Data on social learning richness and a measure of research effort were
modeling non-Gaussian response variables, using the R package MCMCglmm
obtained from ref. 21 via the DataDryad digital repository (64) (see SI Appendix,
(68). Where brain volume was the response variable, Gaussian models were

EVOLUTION
Methods for full details on the social learning measure, illustrative examples,
used with all variables log-10 transformed, diffuse normal priors for the fixed
and discussion of its reliability). Briefly, social learning richness is the number of
effects with a mean of 0 and a large variance (1010), and inverse-Wishart priors
reports of unique social learning behaviors per primate species, primarily from
for the phylogenetic and residual variance (with V = 1, ν = 0.002). Where social
a literature sample of >4,000 articles from primate behavior journals (from
learning was the response variable, Gaussian models were not appropriate
1925 to 2000) (21). Instances of social learning were identified using keywords
due to the highly skewed distribution of this variable; we therefore used
(e.g., “social learning,” “cultural transmission,” and “traditional”) to minimize
Poisson models, with all predictor variables log-10 transformed and with
subjectivity in the collation of reports from the literature (21, 25). Although
nontransformed response variables. Poisson models used the same priors for
identifying social learning from literature reports of nonhuman primate be-
the fixed effects and residual variance as for the Gaussian models, with a
havior is inherently challenging, this approach allows for a quantitative be-
parameter-expanded prior (V = 1, ν = 1, αμ = 0, and αV = 252) for the phy-
havioral measure of social learning across a large sample of diverse primate
logenetic random effect (68, 69). Although a large proportion of the species
species, supporting far larger scale comparative analyses than would be pos-
included in analyses had zero records of social learning, these species are still
sible using data from controlled experiments alone (21, 25, 32, 39, 60). Ex-
informative due to the inclusion of research effort in all models (SI Appendix,
perimental approaches to measuring social learning across species are associated
Methods). Further, preliminary analyses established that Poisson models
with their own particular challenges, especially in comparability and ecological
without a zero-inflation term were appropriate for our data (SI Appendix,
validity of behavioral tests, and limited statistical power due to smaller sample
Methods).
sizes (21, 25, 61). We account for broad-scale species differences in research ef-
Markov chain Monte Carlo (MCMC) analyses were run with a sufficient
fort, here estimated using the number of papers published in the Zoological
number of iterations and thinning to return effective sample sizes of >1,000
Record (between 1993 and 2001, total 7,288 articles) (21) (see SI Appendix,
for all parameters (SI Appendix, Methods). Chain convergence and adequate
Methods for further information).
performance were confirmed by visual inspection of trace plots and checking
Data on social group size and life history traits (gestation length, weaning
effective sample sizes. From each model, we report the mean h2 (a measure of
age, age of sexual maturity, and maximum longevity) were obtained from the
phylogenetic signal equivalent to Pagel’s λ) (70) and mean β coefficient esti-
PanTheria dataset (53). As a measure of maternal investment, we summed
mate from posterior distributions. To assess the strength of evidence for fixed
gestation length and weaning age (following ref. 34). Reproductive lifespan
effects, we use the percentage of posterior β coefficient estimates crossing zero
was calculated as age of sexual maturity subtracted from maximum lon-
in the direction opposite to predictions (as in refs. 39, 71, and 72, for example).
gevity. Comparative datasets were matched to a dated consensus phylogeny
Posterior distributions shifted substantially away from zero in a positive or
for 301 primate species (10kTrees version 3, using GenBank taxonomy) (65).
negative direction indicate support for positive or negative associations, re-
Taxonomic mismatches were resolved using the 10kTrees Translation table
spectively, between fixed effects and outcome variables. Conversely, posterior
and the International Union for Conservation of Nature (IUCN) Red List
distributions centered on zero or overlapping substantially with zero indicate a
website (66).
lack of evidence for any relationship between the fixed effects and outcome
variables. Here, all associations are predicted to be positive in direction. As a
Statistical Analyses. To test predictions, we ran a series of statistical models in
measure of model fit, we used a pseudo R2, estimated as the squared Pearson’s
which the outcome variables were always either brain volume or social
correlation between fitted values and observed data (73). No analysis reported
learning, fitting independent variables that correspond to specific predicted
a variance inflation factor (VIF) above 5, demonstrating that multicollinearity
associations, along with appropriate potentially confounding variables. Ac-
was not a concern in our analyses (SI Appendix, Methods).
counting for the effects of multiple variables is essential in comparative
studies of brain evolution, due to multiple potential correlates (29). We
ACKNOWLEDGMENTS. We thank Chris Venditti for advice regarding the
analyzed brain volume both in absolute terms, and relative to body mass, by
implementation of phylogenetic Poisson models. Research was supported in
variably including body mass as an additional predictor variable. Where part by European Research Council Advanced Grant “Evoculture” 232823 (to
social learning was the outcome variable, research effort was always in- K.N.L.), John Templeton Foundation Grant 23807 (to K.N.L. and S.M.R.), and
cluded as a predictor to account for its effect on the number of records Natural Sciences and Engineering Research Council of Canada Grants
of social learning in the primate behavioral literature (21, 25). We also 418342-2012 and 429385-2012 (to S.M.R.).

1. Striedter GF (2005) Principles of Brain Evolution (Sinauer, Sunderland, MA).
2. Finlay BL, Darlington RB (1995) Linked regularities in the development and evolution
of mammalian brains. Science 268:1578–1584.
3. Barton RA, Harvey PH (2000) Mosaic evolution of brain structure in mammals. Nature
405:1055–1058.
4. Isler K, et al. (2008) Endocranial volumes of primate species: Scaling analyses using a
comprehensive and reliable data set. J Hum Evol 55:967–978.
5. Barton RA (2006) Primate brain evolution: Integrating comparative, neurophysio-
logical, and ethological data. Evol Anthropol 15:224–236.
6. Healy SD, Rowe C (2007) A critique of comparative studies of brain size. Proc Biol Sci
274:453–464.
7. Chittka L, Niven J (2009) Are bigger brains better? Curr Biol 19:R995–R1008.
8. Herculano-Houzel S, Collins CE, Wong P, Kaas JH (2007) Cellular scaling rules for
primate brains. Proc Natl Acad Sci USA 104:3562–3567.

Street et al. PNAS | July 25, 2017 | vol. 114 | no. 30 | 7913
 9. Herculano-Houzel S (2009) The human brain in numbers: A linearly scaled-up primate
brain. Front Hum Neurosci 3:31.
10. Herculano-Houzel S (2011) Brains matter, bodies maybe not: The case for examining
neuron numbers irrespective of body size. Ann N Y Acad Sci 1225:191–199.
11. Herculano-Houzel S, Manger PR, Kaas JH (2014) Brain scaling in mammalian evolution
as a consequence of concerted and mosaic changes in numbers of neurons and av-
erage neuronal cell size. Front Neuroanat 8:77.
12. Barton RA (2012) Embodied cognitive evolution and the cerebellum. Philos Trans R
Soc Lond B Biol Sci 367:2097–2107.
13. Deaner RO, Isler K, Burkart J, van Schaik C (2007) Overall brain size, and not en-
cephalization quotient, best predicts cognitive ability across non-human primates.
Brain Behav Evol 70:115–124.
14. MacLean EL, et al. (2014) The evolution of self-control. Proc Natl Acad Sci USA 111:
E2140–E2148.
15. Later W, et al. (2010) Is the 1975 Reference Man still a suitable reference? Eur J Clin
Nutr 64:1035–1042.
16. Kaplan H, Hill K, Lancaster J, Hurtado AM (2000) A theory of human life history
evolution: Diet, intelligence, and longevity. Evol Anthropol 9:156–185.
17. Boyd R, Silk JB (2012) How Humans Evolved (Norton, New York).
18. Emery NJ, Clayton NS (2004) The mentality of crows: Convergent evolution of in-
telligence in corvids and Apes. Science 306:1903–1907.
19. Emery NJ (2006) Cognitive ornithology: The evolution of avian intelligence. Philos
Trans R Soc B Biol Sci 361:23–43.
20. Rendell L, Whitehead H (2001) Culture in whales and dolphins. Behav Brain Sci 24:
309–324.
21. Reader SM, Hager Y, Laland KN (2011) The evolution of primate general and cultural
intelligence. Philos Trans R Soc B Biol Sci 366:1017–1027.
22. Hunt GR, Gray RD (2003) Diversification and cumulative evolution in New Caledonian
crow tool manufacture. Proc Biol Sci 270:867–874.
23. Boyd R, Richerson PJ (1985) Culture and the Evolutionary Process (Univ of Chicago
Press, Chicago).
24. Wilson AC (1985) The molecular basis of evolution. Sci Am 253:164–173.
25. Reader SM, Laland KN (2002) Social intelligence, innovation, and enhanced brain size
in primates. Proc Natl Acad Sci USA 99:4436–4441.
26. Whiten A, van Schaik CP (2007) The evolution of animal “cultures” and social in-
telligence. Philos Trans R Soc B Biol Sci 362:603–620.
27. van Schaik CP, Burkart JM (2011) Social learning and evolution: The cultural in-
telligence hypothesis. Philos Trans R Soc Lond B Biol Sci 366:1008–1016.
28. Whiten A, Byrne RW (1997) Machiavellian Intelligence II: Extentions and Evaluations
(Cambridge Univ Press, Cambridge, UK).
29. Dunbar RIM, Shultz S (2007) Understanding primate brain evolution. Philos Trans R
Soc Lond B Biol Sci 362:649–658.
30. Clutton-Brock TH, Harvey PH (1980) Primates, brains and ecology. J Zool 190:309–323.
31. DeCasien AR, Williams SA, Higham JP (2017) Primate brain size is predicted by diet but
not sociality. Nat Ecol Evol 1:0112.
32. Navarrete AF, Reader SM, Street SE, Whalen A, Laland KN (2016) The coevolution of
innovation and technical intelligence in primates. Philos Trans R Soc Lond B Biol Sci
371:20150186.
33. Byrne RW, Corp N (2004) Neocortex size predicts deception rate in primates. Proc Biol
Sci 271:1693–1699.
34. Barton RA, Capellini I (2011) Maternal investment, life histories, and the costs of brain
growth in mammals. Proc Natl Acad Sci USA 108:6169–6174.
35. Whiten A, et al. (1999) Cultures in chimpanzees. Nature 399:682–685.
36. van Schaik CP, et al. (2003) Orangutan cultures and the evolution of material culture.
Science 299:102–105.
37. Perry S, et al. (2003) Social conventions in wild white-faced capuchin monkeys. Curr
Anthropol 44:241–269.
38. Henrich J (2015) The Secret of Our Success: How Culture Is Driving Human Evolution,
Domesticating Our Species, and Making Us Smarter (Princeton Univ Press, Princeton).
39. McCabe CM, Reader SM, Nunn CL (2015) Infectious disease, behavioural flexibility and
the evolution of culture in primates. Proc Biol Sci 282:20140862.
40. Laland KN (2017) Darwin’s Unfinished Symphony: How Culture Made the Human
Mind (Princeton Univ Press, Princeton).
41. Sol D (2009) Revisiting the cognitive buffer hypothesis for the evolution of large
brains. Biol Lett 5:130–133.
42. González-Lagos C, Sol D, Reader SM (2010) Large-brained mammals live longer. J Evol
Biol 23:1064–1074.
7914 | www.pnas.org/cgi/doi/10.1073/pnas.1620734114
43. Henrich J (2004) Demography and cultural evolution: How adaptive cultural processes
can produce maladaptive losses—the Tasmanian case. Am Antiq 69:197–214.
44. Powell A, Shennan S, Thomas MG (2009) Late Pleistocene demography and the ap-
pearance of modern human behavior. Science 324:1298–1301.
45. Kline MA, Boyd R (2010) Population size predicts technological complexity in Oceania.
Proc Biol Sci 277:2559–64.
46. Derex M, Beugin M-P, Godelle B, Raymond M (2013) Experimental evidence for the
influence of group size on cultural complexity. Nature 503:389–391.
47. Leadbeater E, Chittka L (2007) Social learning in insects: From miniature brains to
consensus building. Curr Biol 17:R703–R713.
48. Street SE, Laland KN (2017) Social learning, intelligence, and brain evolution. The
Wiley Handbook of Evolutionary Neuroscience, ed Shepherd SV (John Wiley, Chichester,
UK), pp 495–513.
49. Barton RA (1998) Visual specialization and brain evolution in primates. Proc Biol Sci
265:1933–1937.
50. Barton RA (2004) Binocularity and brain evolution in primates. Proc Natl Acad Sci USA
101:10113–10115.
51. Heyes C (2012) What’s social about social learning? J Comp Psychol 126:193–202.
52. Sol D, Sayol F, Ducatez S, Lefebvre L (2016) The life-history basis of behavioural in-
novations. Philos Trans R Soc Lond B Biol Sci 371:20150187.
53. Jones KE, et al. (2009) PanTHERIA: A species-level database of life history, ecology,
and geography of extant and recently extinct mammals. Ecology 90:2648.
54. Carvalho P, Diniz-Filho JAF, Bini LM (2006) Factors influencing changes in trait cor-
relations across species after using phylogenetic independent contrasts. Evol Ecol 20:
591–602.
55. Dunbar RIM (1995) Neocortex size and group size in primates: A test of the hy-
pothesis. J Hum Evol 28:287–296.
56. Dunbar RIM (1998) The social brain hypothesis. Evol Anthropol 6:178–190.
57. Dunbar RIM (1992) Neocortex size as a constraint on group size in primates. J Hum
Evol 22:469–493.
58. Reader SM (2000) Social learning and innovation: Individual differences, diffu-
sion dynamics and evolutionary issues. PhD dissertation (University of Cambridge,
Cambridge, UK).
59. Kaplan HS, Robson AJ (2002) The emergence of humans: The coevolution of in-
telligence and longevity with intergenerational transfers. Proc Natl Acad Sci USA 99:
10221–10226.
60. Lefebvre L, Reader SM, Sol D (2004) Brains, innovations and evolution in birds and
primates. Brain Behav Evol 63:233–246.
61. Bates LA, Byrne RW (2007) Creative or created: Using anecdotes to investigate animal
cognition. Methods 42:12–21.
62. Pagel M (2012) Evolution: Adapted to culture. Nature 482:297–299.
63. Reader SM, MacDonald K (2003) Environmental variability and primate behavioural
flexibility. Animal Innovation, eds Reader SM, Laland KN (Oxford Univ Press, Oxford),
pp 83–116.
64. Reader SM, Hager Y, Laland KN (2011) Data from: The evolution of primate general
and cultural intelligence. Dryad Digital Repository. Available at dx.doi.org/10.5061/
dryad.t0q94. Accessed November 23, 2016.
65. Arnold C, Matthews LJ, Nunn CL (2010) The 10kTrees Website: A new online resource
for primate phylogeny. Evol Anthropol 19:114–118.
66. IUCN (2016) The IUCN Red List of Threatened Species. Version 2016-2. Available at
www.iucnredlist.org. Accessed November 24, 2016.
67. West HER, Capellini I (2016) Male care and life history traits in mammals. Nat
Commun 7:11854.
68. Hadfield JD (2010) MCMC methods for multi-response generalized linear mixed
models: The MCMCglmm R package. J Stat Softw 33:1–22.
69. Hadfield J (2016) MCMCglmm Course Notes. Available at: ftp://cran.r-project.org/pub/
R/web/packages/MCMCglmm/vignettes/CourseNotes.pdf. Accessed January 5, 2017.
70. Hadfield JD, Nakagawa S (2010) General quantitative genetic methods for compar-
ative biology: Phylogenies, taxonomies and multi-trait models for continuous and
categorical characters. J Evol Biol 23:494–508.
71. Capellini I, Baker J, Allen WL, Street SE, Venditti C (2015) The role of life history traits
in mammalian invasion success. Ecol Lett 18:1099–1107.
72. Allen WL, Street SE, Capellini I (2017) Fast life history traits promote invasion success
in amphibians and reptiles. Ecol Lett 20:222–230.
73. Zheng B, Agresti A (2000) Summarizing the predictive power of a generalized linear
model. Stat Med 19:1771–1781.
Street et al.

The evolution of cognitive mechanisms in response to
cultural innovations
Arnon Lotema, Joseph Y. Halpernb, Shimon Edelmanc, and Oren Kolodnyd,1
a
Department of Zoology, Tel Aviv University, Tel Aviv 6997801, Israel; bDepartment of Computer Science, Cornell University, Ithaca, NY 14850; cDepartment
of Psychology, Cornell University, Ithaca, NY 14850; and dDepartment of Biology, Stanford University, Stanford, CA 94305
Edited by Kevin N. Laland, University of St. Andrews, St. Andrews, United Kingdom, and accepted by Editorial Board Member Andrew G. Clark April 29, 2017
(received for review January 20, 2017)
When humans and other animals make cultural innovations, they
also change their environment, thereby imposing new selective
pressures that can modify their biological traits. For example,
there is evidence that dairy farming by humans favored alleles for
adult lactose tolerance. Similarly, the invention of cooking possibly
affected the evolution of jaw and tooth morphology. However,
when it comes to cognitive traits and learning mechanisms, it is
much more difficult to determine whether and how their evolution
was affected by culture or by their use in cultural transmission. Here
we argue that, excluding very recent cultural innovations, the
assumption that culture shaped the evolution of cognition is both
more parsimonious and more productive than assuming the
opposite. In considering how culture shapes cognition, we suggest
that a process-level model of cognitive evolution is necessary and
offer such a model. The model employs relatively simple coevolv-
ing mechanisms of learning and data acquisition that jointly
construct a complex network of a type previously shown to be
capable of supporting a range of cognitive abilities. The evolution
of cognition, and thus the effect of culture on cognitive evolution,
is captured through small modifications of these coevolving learn-
ing and data-acquisition mechanisms, whose coordinated action is
critical for building an effective network. We use the model to show
how these mechanisms are likely to evolve in response to cultural
phenomena, such as language and tool-making, which are asso-
ciated with major changes in data patterns and with new compu-
tational and statistical challenges.
| |
tool-making language evolution niche construction | phenomena such as language and tool-making (each related to
|
cognitive evolution social learning
A n open question in the study of culture and cognitive evo-
lution is whether (and to what extent) cognitive mechanisms,
especially those viewed as advanced or sophisticated, evolved in
response to social-learning challenges or are merely the product of
domain-general mechanisms (1–3). According to one view—still
widely held in cognitive science and evolutionary psychology—cog-
nitive adaptations take the form of specialized brain modules (or
neuronal mechanisms) that evolved for specific, often social pur-
poses, such as “imitation” (4, 5), “mind reading” (6, 7), “cheating
detection” (8), or most famously, language acquisition (9, 10). These
ideas have been criticized on theoretical and empirical grounds (11,
12), and the debate around them demonstrates our limited un-
derstanding of the evolution of cognition, its relationship to the
evolution of social behavior and, in some organisms, culture.
The question of whether culture and social behavior shape the
evolution of the brain is, in our view, best considered using the
evolutionary framework of niche construction (13–16): that is,
culture and social behavior change the ecological niche to which
cognitive traits must adapt in the same manner that nest-building
by birds changes the ecological niche in which their nestlings evolve.
For example, animals’ ability to learn from each other may have
initially been a by-product of domain-general associative learn-
ing mechanisms that did not evolve for social learning (1).
However, as soon as these mechanisms enabled social learning
and were recruited by it for regular use, social learning and its
outcomes also became part of their ecological niche. From that
www.pnas.org/cgi/doi/10.1073/pnas.1620742114
COLLOQUIUM
PAPER
moment on, learning mechanisms that need not have initially been
specifically social were also selected according to their ability to
support social learning. If that is the case, one can certainly claim
that these mechanisms were adapted or shaped to serve their new
social function (although using the term “evolved for” may still be
premature without knowing the degree of genetic modification
and specialization).
Similarly, when social learning enables the accumulation or
spread of shared group behaviors—these days recognized as the
formation of “culture” (17, 18)—this culture becomes the new
ecological niche for all of the learning mechanisms that con-
tribute to it, and therefore has the potential to shape their evolution.
Thus, in theory, given sufficient evolutionary time, cultural phe-
EVOLUTION
nomena that are adaptive for the individual, and whose acquisition is
supported by advanced learning or cognitive skills, such as the ability
to imitate or to learn language, are expected to select for improve-
ments in these cognitive skills (see also ref. 19). In practice, however,
clear evidence showing the effect of culture on cognition is lacking,
and alternative accounts for the evolution of advanced cognition
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
and culture through domain-general learning principles cannot
be ruled out (1, 2, 20). As a result, whether and how culture
really shapes the evolution of cognition is still under debate.
In what follows, we first clarify some of the theoretical issues in
this debate, using two recent controversies in the fields of language
evolution and social learning. We then offer a process-level ap-
proach to cognitive evolution that may be useful in predicting what
aspects of learning and cognition are likely to coevolve with cul-
ture. Finally, we use the model to demonstrate how cultural
one of the two controversies discussed earlier) are likely to shape
cognition, given their association with changes in data distribution
and with new computational and statistical challenges.
Can Culture Evolve Without Shaping Cognition? On
Parsimony, Likelihood, and Scientific Productivity
Evolution takes time, so it is clear that very recent cultural in-
novations, such as cars, computers, cellular phones, or the In-
ternet, could not have yet generated detectable effects (or
perhaps any effect at all) on the evolution of cognition. But what
about relatively ancient cultural phenomena, such as song-
learning in birds or tool-making and language acquisition in
humans? Although there is evidence for the effect of human
culture on biological traits and gene frequencies (21), evidence
for specific effects of human culture on learning and cognitive
mechanisms is mostly circumstantial. This evidence includes
This paper results from the Arthur M. Sackler Colloquium of the National Academy of
Sciences, “The Extension of Biology Through Culture,” held November 16–17, 2016, at the
Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering
in Irvine, CA. The complete program and video recordings of most presentations are available
on the NAS website at www.nasonline.org/Extension_of_Biology_Through_Culture.
Author contributions: A.L., J.Y.H., S.E., and O.K. designed research, performed research,
and wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. K.N.L. is a guest editor invited by the Editorial
Board.
1
To whom correspondence should be addressed. Email: okolodny@stanford.edu.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7915–7922
signs of selection on genes implicated in brain growth, learning,
and cognition (22–24) that may be attributed to human culture
(21), and differences in gene expression in the brain between
human and nonhuman primates (25, 26) that may be interpreted
similarly. There is also recent evidence relating structural changes
in the human brain to Paleolithic tool-making abilities (27, 28),
but additional work is still needed to clarify the direction of
causation between culture and cognition (we will return to dis-
cuss these findings toward the end of the paper). Finally, when
animal culture is considered, a recent study (29) suggests an
effect of culture on learning in songbirds: a larger repertoire size
is found in species that developed open-ended learning ability.
The lack of clear empirical evidence for the effect of culture on
cognition is not surprising, given that cognitive mechanisms and
their genetic underpinning are still poorly understood, making it
difficult to track their evolution (as opposed to that of clearly de-
fined biological traits). As a result, much of the debate is focused on
theoretical arguments of plausibility and likelihood, which may be
interpreted differently by psychologists and evolutionary biologists.
We use two examples of such controversies to illustrate this prob-
lem and to suggest a methodologically productive resolution.
Problem 1: The Evolution of Social-Learning Mechanisms. Social
learning is broadly defined as learning that is influenced by ob-
servation or interaction with other individuals or with the
products of their behavior (30). This definition leaves open the
question of whether social-learning mechanisms have evolved
specifically to serve their social function or whether they are
domain-general associative learning mechanisms that are also
used to learn socially. In a series of thought-provoking papers,
Heyes and her colleagues (1, 2, 31–33) have demonstrated that
most mechanisms of social learning and imitation that are nor-
mally viewed as specialized adaptations for social life (4, 5, 34)
can also be explained by domain-general associative learning
principles. In this light, and in the absence of convincing evi-
dence to the contrary, they also suggested that there is no need
to posit that these domain-general mechanisms were shaped by
their social or cultural function (35). In other words, it would
appear more parsimonious to assume that these mechanisms did
not evolve beyond their initial domain-general form. However,
this appeal to parsimony is somewhat misleading in evolutionary
contexts and time scales, where changes are actually to be
expected (36, 37). In fact, for most evolutionary biologists, it
would appear highly unlikely that some learning mechanisms
would be used for many generations to serve social functions, yet
remain unaffected by this new social niche. It would almost be
like expecting the surface of the moon to remain unmarked with
craters after millions of years of exposure to space debris.
To explain why, imagine a population of finches that expands
its range. The beaks of these finches have a certain morphology
that evolved to handle seed types present in the original habitat;
the expanded habitat includes novel types of seeds. In this set-
ting, the assumption that after many generations there would be
no evolutionary change in beak morphology is not at all parsi-
monious. For a biological trait to remain unchanged over time,
an active process of stabilizing selection is necessary, otherwise it
will change through directional selection or genetic drift. That is,
the probability that the new types of seeds will not affect the
previous stabilizing selection regime is extremely small.
Returning to the evolution of learning, and following the same
reasoning, it is difficult to imagine how adding a new function for a
basic learning mechanism would not affect its evolution. As in the
beak example, the change may be subtle, and merely quantitative.
This is in part because a bird has only one beak that must serve
many different functions (from foraging for different types of food
to feather-preening and nest-building); it cannot be specialized
into different kinds of beaks. However, even in this case of a single
multipurpose adaptation, every new function must affect evolu-
tion; we would expect a more significant change if specialization is
possible. Similarly, even if all cognitive functions are supported by
the same domain-general learning mechanisms, it is unlikely that
7916 | www.pnas.org/cgi/doi/10.1073/pnas.1620742114
social learning and culture could evolve without somehow affect-
ing these mechanisms. In reality, of course, unlike in the beak
example, learning mechanisms are not necessarily constrained to
be uniform across all domains; there is plenty of evidence for
adaptive specialization in associative learning mechanisms (e.g.,
refs. 38–40).
The key point in this evolutionary argument is that, even in the
absence of supportive evidence, it is more parsimonious to as-
sume that learning mechanisms were shaped by their social
function and, if the relevant evidence is lacking, that we have
simply failed to find it, than to assume the opposite. The as-
sumption of no change requires us to posit a lack of genetic
variance in learning mechanisms, which runs counter to sub-
stantial evidence (39, 41–45), or else to explain how the selection
regime was miraculously unaffected by the new social niche. The
assumption of evolutionary change may also be viewed as sci-
entifically more productive, as it encourages further research
(46). This implies that a useful working hypothesis should be that
“culture did shape cognition” and that we have to find out how.
Demonstrating that social learning and the so-called “mirror
neurons” phenomenon can be explained by associative learning
principles (1, 2) is important and consistent with our view. But
given the evolutionary argument above, the demonstration need
not imply that these associative learning mechanisms did not
evolve beyond their basic state (see also ref. 47). Indeed, the ar-
gument suggests that associative learning mechanisms are the
building blocks of cognitive evolution, and are finely tuned to
serve new social, cultural, and other advanced functions (48). This
view may help us make the shift from postulating “black box”
adaptations that evolve “for” particular social purposes to having
credible process-level mechanistic models of cognitive evolution.
Problem 2: The Evolution of Language and Memory Constraints. The
same evolutionary argument discussed above is also relevant to
the question of whether or not human language shaped the
evolution of cognition (16, 49–52). It implies that it is highly
unlikely that the use of language for at least several thousand
generations [or even more (53–55)] failed to affect some aspects
of learning and cognition. The real question is not whether or
not it did, but rather what aspects were affected and how. For
example, Chater et al. have pointed out correctly that many as-
pects of human language change too fast for genetic evolution to
respond (56). The authors used computer simulations to show
that even in the presence of genetic variation, cultural conven-
tions of language are like “moving targets” for natural selection,
making the evolution of genetic adaptations to specific languages
highly implausible. However, although this analysis makes a
convincing argument against strong nativism (the claim that
there is a significant innate component to human language), it
also implies that genes for language can evolve if they serve
general skills for language learning that are stable over time (49,
56, 57). In other words, the need to learn a language may still
select for many general cognitive abilities, such as better mem-
ory, computational abilities, or greater attention to verbal input.
Indeed, it does not seem possible that language could evolve
without affecting such abilities.
Interestingly, in a more recent paper, Christiansen and Chater
(58) have extended their view that language cannot shape the
evolution of cognition by proposing that the limited sensory
memory span—the window of time during which a linguistic ut-
terance is retained in its entirety—creates a bottleneck that
strongly constrains language and cannot evolve to become wider.
The authors suggest that language has evolved to cope with this
memory limitation, but that the evolution of this memory limita-
tion was not affected by language. As we noted elsewhere (59),
this assumption runs counter to the evolutionary argument above
and to substantial evidence for genetic variance in memory pa-
rameters (60–63). In a reply to this criticism, Chater and Chris-
tiansen (64) explained that the memory bottleneck cannot be
viewed as a genetically variable trait that can respond freely to
selection because it “emerges from the computational architecture
Lotem et al.

of the brain.” This answer, however, merely kicks the can down the
road, by moving the problem from the domain of memory to that of
the computational architecture of the brain. The same arguments
hold: although constrained by many factors, the computational
architecture of the brain is shaped by the sum of selective pressures
arising from the need to accommodate the multiple functions of the
brain that influence the organism’s fitness. Even if the memory
bottleneck emerges as a product of this architecture, it can still
evolve as long as this architecture evolves. If the challenge of
processing and using language, which affects individuals’ fitness and
has been in place for thousands of generations, has played a role in
shaping this architecture and its emergent properties, then, as one
of these emergent properties, memory should have been affected.
Specifically, as we proposed in the past (59) and explain further
below, the selective pressure exerted by language learning may have
acted to limit the working-memory buffer, as this may be useful for
coping with the computational challenges involved in data seg-
mentation and network construction.
Our two examples—social learning and memory bottleneck—
suggest that, excluding very recent cultural innovations, it is
unlikely that culture could have evolved without shaping learning
and cognition. This forces us to think more specifically about
how culture shapes cognition, which requires, as we claim next,
adopting a process-level approach to cognitive evolution: that is,
a mechanistic model that explains a behavior or an ability as the
outcome of a process. Such a model may also help provide a useful
structure for reexamining the two problems outlined above.
Why Do We Need a Process-Level Approach to Theorize
About Culture and Cognition?
Whereas it is relatively easy to see how natural selection acts on
clearly defined morphological traits, such as limbs, bones, or
coloration, with cognitive traits that are not well understood, it is
difficult to tell what is actually evolving. Cognition is not a
physical trait, but an emergent property of processes that are
carried out by multiple mechanisms, most of which involve
learning. Thus, to consider how culture shapes the evolution of
cognition, we must explain how such mechanisms work and how
they can be modified by natural selection. The importance of
using mechanistic models in the study of behavioral evolution is
increasingly recognized (65–68), but most attempts to integrate
evolutionary theory and cognition are still based on modeling the
evolution of learning rules that are far too simple to capture
complex cognition (69–74). To understand how culture shapes
the evolution of cognitive mechanisms, such as those serving
imitation, theory of mind, or language acquisition, it is necessary
to have models that explain how such mechanisms work and how
they could evolve.
Clearly, given the immense complexity of the brain, any at-
tempt to propose a general process-level model of advanced
cognition would be ambitious. However, we believe that it is
possible and necessary to start by constructing models that cap-
ture some of the key working principles of advanced cognitive
mechanisms in a manner that suffices to explain their evolution.
An analogy that may clarify our approach is the apparent chal-
lenge in explaining the evolution of the eye. The vertebrate eye is
highly complex; it is initially hard to see how it could have
evolved. However, with a minimal understanding of how the eye
works, the “magic” is removed (75). The basic eye model is a
layer of photosensitive cells; the visual acuity it provides can
gradually improve as it buckles into a ball-like shape, looking
(and working) more and more like a pinhole camera. This sketch
ignores many details, and is far from explaining everything about
eyes and vision, but is sufficient to resolve the puzzle.
This is the kind of modeling approach that we seek for explaining
cognitive evolution. Specifically, we do not seek a fully detailed
neuronal-level model of brain and cognition. Instead, we want a
minimal set of principles that suffice to explain how simple oper-
ational units, capable of only the most basic forms of learning, can
jointly and gradually create the much more sophisticated mecha-
nisms of advanced cognition. [Powerful algorithms using deep
Lotem et al.
COLLOQUIUM
PAPER
learning (76, 77), with less emphasis on biological and behavioral
realism (78), are being developed and applied to challenging
tasks, including: perceptual parsing, associative learning, and the
learning of conceptual contingencies (79, 80).]
Once we have such a minimal model, we can consider how
small variations in the basic operational units and their param-
eters can build better, or different, cognitive mechanisms and
how this evolutionary process can be shaped by culture. Over the
past few years, we have developed such a model and explored its
ability to explain a range of phenomena. In the following sec-
tions, we briefly describe this model and use it to consider how
culture may shape the evolution of cognition and, in particular,
how such a model may help resolve the abovementioned con-
troversies regarding social-learning mechanisms and memory
constraints on language.
A Process-Level Model of Cognitive Evolution
The model presented here has already been described in several
of our previous papers (81–87). Some of the main aspects of the
model were implemented in a set of computer simulations,
demonstrating a gradual evolutionary trajectory, from simple
associative learning, to chaining, to seldom-reinforced continu-
ous learning [in which a network model of the environment is
constructed (84)], to complex hierarchical sequential learning
that can support advanced cognitive abilities of the kind needed
EVOLUTION
for language acquisition and for creativity (85, 87). For the latter,
our modeling framework had to go well beyond chaining through
second-order conditioning (84, 88). The success of a computer
program, originally developed to simulate the behavior of ani-
mals learning to forage for food in structured environments (85),
in reproducing a range of findings in human language (86) sug-
gests that the model may be useful in the study of cognitive
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
evolution. Thus far the model’s implementation has been lim-
ited, for simplicity, to an unsupervised learning mode with a
learning phase, and then a test phase during which the learners
act based on what they have learned. Its extension to accom-
modate iterated cycles of learning and action, which is necessary
to capture the learning of behavioral contingencies through trial
and error, is straightforward. Detailed pseudocode for our model
is found in the supplementary material of refs. 85 and 86.
The model is based on coevolving mechanisms of learning and
data acquisition that jointly construct a complex network that
represents the environment and is used for computing adaptive
responses to challenges in the environment. In particular, the
network is used for search, prediction, decision making, and
generating behavioral sequences (including language utterances,
when applicable). The extent to which learners’ use of the net-
work produced adaptive behaviors was measured in our imple-
mentation by foraging success [in the context of animal foraging
(84, 85, 87)] and by a set of language performance scores [in the
context of language learning (86)]. Although the production of
adaptive behaviors depends on the structure of the network, this
structure is not directly coded by genes and therefore cannot
evolve directly. The components that may evolve over genera-
tions are the parameters of the learning and data-acquisition
mechanisms that construct the network through interaction
with the environment, and whose coordinated action, as we show
below, is critical for building the network appropriately. [This
coevolution is very much in the spirit of the notions of con-
structive development and reciprocal causation in the recently
proposed “extended evolutionary synthesis” (89).]
Constructing a Network. We now briefly sketch the main principles
that govern how the network is constructed. (This technical de-
scription may become clearer and more intuitive after reading
the simplified example outlined in the next subsection and il-
lustrated by Fig. 1). We assume that what data are acquired by
the learner is determined by its “data-acquisition mechanisms”:
the collection of sensory, attentional, and motivational mecha-
nisms that direct the learner to process and acquire whatever is
deemed relevant. These mechanisms [also referred to as “input
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7917
mechanisms” (33)] determine the content and the distribution of
different data items in the input. For simplicity, we assume that
the input takes the form of strings of symbols (i.e., linear se-
quences of discrete items), which are then processed through a
limited working-memory buffer, similar to the “phonological loop”
in humans (90, 91), and tested for familiar segments and statistical
regularities among their components. This is done by the learning
mechanisms in a sequence of steps.
A data sequence is scanned for subsequences that recur within
it and for previously learned subsequences, and is segmented
accordingly. This results in a series of chunks, which are either
previously known or are incorporated at this point into a network
of nodes that represents the world as it has been learned so far
(nodes stand for objects or other meaningful units; the links in
the network represent their association in time and space).
Weights are assigned to the nodes and to the links to reflect their
frequency of occurrence: links between nodes are established
whenever two nodes follow one another in the input. The weight
of a node or a link is increased whenever it is encountered in the
input; the weight also decreases with time if it is not encoun-
tered. This process ensures that only those units and relations
that are potentially meaningful are retained in memory, and
spurious occurrences are forgotten. If a node’s or link’s weight
increases above a fixation threshold, its decay becomes highly
improbable. The probability that a data item is learned is thus
determined by how frequently it is encountered in the data, and
A 756483617569813675628136
756483617569813675628136
756483617569813675628136
756 48361
98 136
28
B 756483617564836175648361
756981367569813675698136
756281367562813675628136
75648361 6 75628136
75698136
Fig. 1. Data input in the form of three strings, and the network that is
constructed as a result of acquiring and processing this input using the
learning mechanisms and parameter set described in the text. (A) Each data
string of 24 characters is composed of three nonidentical subsequences of
eight characters that share some common segments (highlighted using the
same shade of gray). The three strings are identical in this case, so labeling
each subsequence of eight characters as A, B, and C, respectively, allows
describing the structure of the input as ABC ABC ABC. (B) The same input as
in A is distributed differently over time, which can be described in short as
AAA BBB CCC. This input leads to a completely different network structure
due to fixation of A, B, and C as long eight-character chunks. The weights of
the nodes and the links of the networks are not shown in the figure, but all
of them exceed the fixation threshold of 1.0, as the weight-increase pa-
rameter was set to 0.4 per occurrence.
7918 | www.pnas.org/cgi/doi/10.1073/pnas.1620742114
by the parameters of weight increase and decrease. These pa-
rameters create a window for learning, during which data can be
either retained or discarded from the network.
We assume that if a data sequence reaches the threshold weight
for memory fixation, it remains in memory and is not segmented
any further. An intuitive example from language learning is a word
such as “backpack,” which would be fixed in memory if it were
heard repeatedly without prior exposure to instances of “back” or
“pack” (not even within other sequences, such as “on my back” or
“in the pack”). If “back” or “pack” are heard often, then their
partial commonality with “backpack” would result in “backpack”
being segmented into “back” and “pack” (with a directed link
between them; i.e., back→pack). The fixation of long sequences
may have a positive or a negative impact on a learner’s success, as
discussed below and in refs. 81 and 87. Note that the fixation of
“backpack” does not prevent the formation of separate nodes for
“back” and “pack” following later observations.
In addition to breaking up segments to form smaller segments,
a node can be formed by the concatenation of smaller segments
after they are repeatedly observed in succession. Thus, nodes can
be formed “top-down” directly from the raw input by segmen-
tation, or “bottom-up” through concatenation of previously
learned units, creating a hierarchical structure, with potentially
multiple hierarchies that can be perceived as “sequences of
shorter sequences” (see refs. 85 and 86 for more details). In both
cases, the effects that memory parameters have on learning
amount to a test of statistical significance: natural and mean-
ingful patterns are likely to recur and thus pass the test, whereas
spurious patterns decay and are forgotten.
A Simplified Example. To better understand the process of data
segmentation and network construction, a simplified example is
illustrated in Fig. 1A. This example shows a network that is
constructed as the result of acquiring three specific strings of
data, under the assumption that the weight-increase parameter is
0.4, the fixation threshold is 1.0 (which means that a data item
reaches fixation after three successive observations, because 3 ×
0.4 > 1), and the weight-decrease (decay) parameter is 0.01 (i.e.,
the weight of a data item that is not yet fixated decreases
by 0.01 with each symbol that enters the input). It is also as-
sumed that the working-memory buffer can accommodate up to
24 symbols (which is the length of one data string in the example
in Fig. 1A), and that the strings in this illustration are separated
by 30 additional (irrelevant) characters that prevent parts of any
two of these strings from being processed simultaneously in the
memory buffer. The figure demonstrates that repeated se-
quences within each string (highlighted by shades of gray for
clarity) are segmented based on their similarity and become the
data units that form the nodes of the network. Directed links
represent past association between these units; thus, they rep-
resent statistical regularities of the environment. For example,
98 always follows 756 and precedes 136, whereas 756 leads to
48361, 98, and 28 with equal probability. Despite the simplicity of
this network, we can already observe that 98 and 28 have a
similar link structure: both are preceded by 756 and followed by
136. In our earlier work (85, 86), we showed how such similarity
in link structure can be used for generalization, for the con-
struction of hierarchical representations, and for creativity (87).
We stressed earlier that, according to our model, the co-
ordinated action of learning and data acquisition mechanisms
and their evolution in response to typical input characteristics
are critical for building an effective network. This point is illus-
trated by Fig. 1B, where the same data as in Fig. 1A are now
distributed differently, leading to a radically different network
representation (although the learning parameters and the
working-memory buffer size remain the same). The distribution
of the data input in Fig. 1B leads to the fixation of large idio-
syncratic data sequences and to poor link structure, which may
hamper further learning and generalization (81, 87). For exam-
ple, no generalization can now be drawn for 98 and 28 because
each of them is “locked” within another segment. Recognizing
Lotem et al.

segments in novel input becomes more difficult (i.e., is less
likely) if the memory representation is based on large idiosyn-
cratic units. The learner is then less likely to place novel data in
context and to perform further segmentation.
Note that if we change only the weight-increase parameter
from 0.4 to 0.3, then the data input of Fig. 1B would result in
exactly the same network as in Fig. 1A and all of the problems
that we have just described would disappear. This is because the
segment “75648361” would not reach fixation after the first data
string and would not decay completely before the second string is
acquired, so it would be segmented when 756 is encountered in
the second string and again in the third. A similar result can be
achieved if we extend the working-memory buffer to include the
beginning of the next string (so that the fourth occurrence of
“756” can split the “75648361” and so on). This example shows
how different combinations of data distribution and memory
parameters can generate quite different or quite similar net-
works. It also demonstrates how relatively small modifications to
the learning parameters or to the distribution of data input can
lead to major changes in the network. As we discussed elsewhere
(82, 86), it is important to bear in mind that not only may the
learning parameters vary across individuals or species, but they
may also evolve to differ across different sensory modalities (or
different learning mechanisms) to better respond to the different
distributions of data types in nature. Similarly, the learning pa-
rameters may also be modulated by physiological and emotional
state, giving higher increase in memory weight to important but
relatively rare observations (82).
The learning mechanism described so far is sensitive to the
order in which elements appear in the data. For example,
756 and 576 are viewed as different data sequences. This may be
important for some data types, such as the sequence of actions of
a particular hunting technique, the phrases in a birdsong, or
human speech. But for some other types of data it may be suf-
ficient (or even better) to classify two sequences as the same by
merely recognizing some of their similar components. For exam-
ple, two instances of the same salad in a salad bar or a stand of
mango trees in the forest may be recognized based on a combi-
nation of stimuli, ignoring their exact serial order (which may
actually vary across instances). It is therefore possible that the
learning mechanisms may also differ in the set of parameters that
determine how sensitive they are to the exact serial order of data
items. Clearly, a change in these parameters can also influence the
segmentation process and the structure of the network, an issue
that will become relevant again when we discuss language and
tool-making, for which serial order is critically important.
Finally, as explained earlier, our model does not pretend to
capture cognitive mechanisms at the neuronal level. The nodes
and the links in our network do not correspond to neurons and
synapses. Nevertheless, the processes described in our model at
the computational level can be realized by neuronal structures
and activities, and a representation of the proposed network may
exist in the brain. We can assume that the neuronal structures
and brain circuits that realize the network are ultimately affected
by constraints of size and morphology that are at least partly
determined genetically. That is, adaptive changes in the data
acquisition and the learning mechanisms that can potentially
lead to the construction of an extensive network in the acoustic
domain, for example, may be subject to physical constraints that
are also genetically determined. Over generations, genetic vari-
ants that are better in relaxing these physical constraints and in
meeting the demand for larger or more appropriate neuronal
structures will be favored by selection. This view of brain evo-
lution is consistent with the “Baldwin effect” view (92, 93),
according to which genes may be selected based on how well they
support adaptive plastic processes, such as learning. Using this
approach to address the question of how culture shapes the
evolution of the brain would imply that culture exerts selective
pressure that shapes learning and data-acquisition parameters,
which in turn shape the structure of the constructed network.
Consequently, over evolutionary time scales, brain anatomy may
Lotem et al.
COLLOQUIUM
PAPER
be selected to better accommodate the physical requirements of
the constructed network. In the next two sections we consider
how this may have happened in the case of human language and
stone-tool production, reexamining in this light the two problems
discussed earlier regarding memory constraints and social-learning
mechanisms.
The Case of Human Language and Memory Constraints
Although the question of how language has evolved is in itself
the focus of extensive research (e.g., refs. 94–96), here we focus
on a more specific question: Given that language has evolved,
how can it shape the evolution of cognition? According to the
process-level approach described above, we should address this
question in terms of how the need to learn a language, or to use
it, selects for possible changes in data-acquisition or learning
mechanisms. The first expected change, which is quite obvious, is
in the data-acquisition mechanisms: we would expect that at-
tention to human speech, as well as to human gaze and gestures
that can help to learn the meaning of spoken words, would be-
come even more important than before. Indeed, these manifes-
tations of social attention are very typical of human infants and
young children (97, 98); impairments in such social attention
skills are known to lead to problems in language learning, as in
the case of autism (e.g., ref. 99). Perhaps the most significant
expected impact of changes to the data distribution is on the
EVOLUTION
segmentation process, and consequently on the construction of
the network: because we expect the data-acquisition and learning
parameters to coevolve, the evolution of language should also
affect the memory parameters of the learning mechanisms. Note
that this consideration takes us back to the problem of language
and memory constraints discussed earlier in the paper. It is
PSYCHOLOGICAL AND
highly unlikely that the memory and learning parameters that
COGNITIVE SCIENCES
evolved before language existed were best suited for processing
linguistic data. Although certain plastic adjustment of these
memory parameters on the basis of the learner’s individual ex-
perience cannot be ruled out, it is unlikely that adaptation to
language learning and use over hundreds of generations did not
also play a role in shaping the genetic basis of these parameters’
values. This claim is supported by the known genetic heritability
component in various types of memory (60–63). The question to
address next is how these parameters evolved as a result of
language evolution.
Intuitively, one would expect the challenge of language ac-
quisition to require and to select for a better working memory,
leading to a view of a memory limit as a constraint rather than as
an adaptation (see problem 2, above). However, according to our
model, there are at least two reasons why a limited working
memory may actually be adaptive. First, as explained earlier, the
parameters of weight increase and decrease create a window for
learning that serves as a test of statistical significance: natural
and meaningful patterns are likely to recur and thus to pass the
test, whereas spurious patterns decay and are forgotten.
According to this view, the reason that it is typically difficult to
learn a novel input from a single encounter is that the mecha-
nism of learning has evolved to expect more evidence before
deciding whether an item should be learned or ignored. The
evolution of learning parameters that allow data items to reach
fixation in memory after a single encounter should be possible.
There are in fact examples for such “one-trial learning,” that,
interestingly, occurs when rapid learning seems to be adaptive, as
in the context of fear learning and enemy recognition (100, 101)
or in the case of word “fast-mapping” in young children (102).
However, in the case of large quantities of sequential data, where
all items may be equally important, proximity of repeated occur-
rences in time and frequency of recurrence are the best first-resort
tests of meaningfulness. The selective pressure of language in the
direction of smaller buffer sizes and moderate fixation rates may
explain why people are not better at memorizing sequential data
verbatim (58), which would require larger buffer sizes and more
rapid fixation.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7919
 Second, the limited memory buffer can be viewed as repre-
senting an adaptive trade-off between memory and computation.
[Recall our earlier discussion of Christiansen and Chater’s
memory bottleneck (58).] We assume that a larger buffer that
can accommodate more data can evolve, but whether or not it
will be adaptive depends on the kind of computations needed to
process the data in the buffer. In the case of linguistic input,
serial order is important; the data must be segmented into words
or chunks based on recurring segments. This means that the
learner has to search and compare all possible chunks within the
buffer, and to find possible matches between these chunks and
those represented already as nodes in the network (so as to put
the incoming data in context). The number of possible chunks in
a linear sequence grows as the square of the length of the se-
quence [there are N(N − 1)/2 possible chunks in a linear se-
quence of N data items]. Thus, in a relatively small buffer of
5 data items, the learner has to find and compare only 10 possi-
ble chunks, whereas in larger buffers of, say 10 or 15 items, the
learner has to find and compare 45 or 105 possible chunks, re-
spectively. Therefore, increasing the buffer size leads to consid-
erable computational cost that may not be justifiable. Depending
on the number of items that comprise a typical meaningful chunk
in the language, it might be better to process a sequence of data
items by gradually scanning it with a small buffer of 5 items
rather than with a large buffer of 15 items.
The computational burden is likely to be much smaller if data
input does not need to be segmented accurately, but merely
recognized and classified based on some characteristic features.
As we mentioned earlier, recognizing a particular salad in a salad
bar or a typical fruit tree in a forest may not require paying
attention to the exact serial order of the data. In this case, a
larger memory buffer may not lead to such a sharp increase in
computation. A simple illustration of this phenomenon is given
by the three nonsegmented sentences presented in Fig. 2. En-
glish speakers who cannot read Hebrew or Japanese would most
certainly try to segment (almost automatically and subcon-
sciously) the first sentence that is in English but not the next two
sentences that are in Hebrew and Japanese. Those would be
classified quickly as “Gibberish in a foreign language” or, more
specifically, as “two sentences in Hebrew and Japanese that I
can’t read” (based on some distinctive features of Hebrew and
Japanese letters). Clearly, readers who know Hebrew or Japa-
nese will segment those sentences automatically. The point of
this example is to demonstrate that the “ecological” context and
the cultural background of a learner can determine the level of
computation applied to processing incoming data.
Some level of data segmentation was clearly required even
before the evolution of language; for example, when animals
forage for food in structured environments (85) or need to learn
or interpret observed behavioral sequences (103). However, the
evolution of language almost certainly increased the proportion
of data input that must be accurately segmented, thereby im-
posing more significant computational requirements for a given
memory buffer size. Individuals with a genetic predisposition to
use a smaller buffer may have been selected, which may explain
why the memory bottleneck is indeed so small. This hypothesis
predicts that some of the humans’ close relatives that do not
possess language may be endowed with a larger working-memory
buffer. Indeed, a notable study on working memory for numerals
in chimpanzees (104) shows that young chimpanzees have a
better capacity for numerical recollection than human adults.
Fig. 2. The sentence: “The number of possible chunks in a linear sequence =
N(N − 1)/2” written in a nonsegmented form in three different languages,
English, Hebrew, and Japanese. See the explanation in the main text.
7920 | www.pnas.org/cgi/doi/10.1073/pnas.1620742114
This ability may be useful for fast recognition of objects or
structures in the field that does not depend on serial order. It
may also improve rote learning of recent actions that might be
helpful in systematically searching for food without returning to
a place that was just visited. Interestingly, exceptional ability to
retain in memory an accurate, detailed image of a complex scene
or pattern (also known as “eidetic imagery”) is more common
among young children (105) and autistic savants (106). Such an
ability may facilitate rote learning at the expense of effective
segmentation and network representation (81). Regardless of
whether or not we understand this phenomenon correctly, it
clearly shows that having a larger working-memory buffer is bi-
ologically feasible and that genetic variants that possess this
ability are already present in human populations. The fact that
they do not spread and become the norm suggests that a small
memory bottleneck is somehow more adaptive.
The Case of Tool-Making and the Evolution of Social-
Learning Mechanisms
Cultural transmission of tool-making techniques depends on
social-learning mechanisms. Whereas learning some advanced
techniques may involve teaching and verbal instruction (107), the
ability to make stone tools probably depended, initially at least,
on social-learning mechanisms of the type needed to facilitate
imitation or emulation (108–110). How, then, could the evolu-
tion of culturally transmitted techniques for making tools (i.e.,
the culture of “tool-making”) affect the genetic evolution of such
learning mechanisms? This question is a specific instance of the
more general questions addressed earlier (Problem 1, above) of
how using learning mechanisms for social functions shapes their
evolution. To answer this question, we should first consider how
imitation or emulation works. Here we try to explain it in terms
of our model. We assume that for imitation, the coupling be-
tween perception and action is developed through experience
(see also ref. 2). That is, when an individual repeatedly observes
its own actions, it gradually—and quite automatically—asso-
ciates the perception and the motor experience of those actions.
Eventually, seeing another individual performing those actions
activates the observer’s representation of the motor experience
of performing those actions [because the observed actions are
perceived as being similar to the (perceptual representation of)
the individual’s own actions, which are already coupled with the
relevant motor experience]. Thus, the first expected effect, which
may be described in terms of data-acquisition mechanisms (33,
82), is an increase in attention to the behavioral patterns of other
individuals (in imitation) or to the outcomes of their actions (in
emulation). For imitation, it is also important to acquire much
information on self-actions to create the coupling between the
perception of these actions and their motor experience.
The next question is how to organize the acquired data in
memory. In our framework, this amounts to asking how the
network should be constructed. In the case of imitation, there
seem to be two possibilities. One is to represent long sequences
of observed actions or sensory experiences as large chunks: exact
copies of entire sequences that can then be executed. The other
possibility is to segment the observed behavior or sensory expe-
rience into smaller basic units, just as in the process of language
acquisition, and then compose them again into larger sequences
in the production process. The first possibility of exact imitation
fits the notion of specialized imitation ability that allows copying
and executing complex behaviors accurately and almost auto-
matically. However, this approach leads to three problems. First,
the expected effect of exact imitation on the evolution of
learning is that both the “working-memory buffer” and the
weight-increase parameter should increase in size. The working-
memory buffer should be large enough to capture the long
sequences of observed behaviors, and the weight-increase pa-
rameter should be high enough to allow rapid fixation in mem-
ory. This prediction does not seem to hold. As discussed earlier,
the human working-memory buffer is typically small, and com-
plex patterns require repeated encounters to be learned. The
Lotem et al.

second problem with exact imitation of novel sequences is
whether or not it can work within the framework of the asso-
ciative account. To create the coupling between perception and
action, the learner must also produce the action, and this cannot
be done before successful imitation (because the perception of a
unique novel sequence does not yet have a match in motor
representation that can be executed). Finally, the learning of
long fixed-action sequences reduces flexibility. It would require,
for example, that the initial raw stone from which a tool is to be
produced be nearly identical to the raw stone that was used in
the original learned sequence, a highly unlikely occurrence.
The alternative possibility, which involves sequence segmen-
tation and reassembly, is consistent with the associative account
and is also feasible. The learner first explores and practices a
large repertoire of simple behavioral actions, thereby creating
the necessary coupling between perception and action in a large
repertoire of basic behavioral units. It can then concatenate
these basic units in many possible ways; for the purpose of imi-
tation, it can concatenate them to gradually match the complex
behavior demonstrated by other individuals. The demonstrated
behavior is also segmented into familiar units, which can then be
associated with familiar actions, which helps in producing an
imitation. This scenario is quite consistent with mounting evi-
dence and recent views of experience-based imitation and em-
ulation (110, 111). It also suggests that similar processes are
involved both in language learning and in complex imitation,
which is in line with recent views according to which tool-making
possibly preadapted the brain to language learning (27).
Finally, our process-level approach may also help to explain
recent new studies linking neuroanatomical changes in the brain
to Paleolithic tool-making ability. These studies found that
the acquisition of tool-making abilities by experimental subjects
involved specific structural changes in the brain (27) and that
these structures and regions in the brain are more developed in
humans than in chimpanzees (28). This evidence for a short-term
plastic response colocalized with structures that underwent re-
cent evolutionary change strongly suggests a process akin to the
Baldwin effect, in which genetic variants are selected based on
how well they support the required plastic changes (92, 93). It is
yet to be explained, however, how the observed plastic changes
improve tool-making abilities. As we suggested earlier, in our
view, such neuroanatomical changes have to do with the path-
ways and the representational systems that are recruited to serve
the construction of the network. The result should be a rich
network that represents sensory and perceptual experiences of
1. Heyes C, Pearce JM (2015) Not-so-social learning strategies. Proc Biol Sci 282:20141709.
2. Cook R, Bird G, Catmur C, Press C, Heyes C (2014) Mirror neurons: From origin to
function. Behav Brain Sci 37:177–192.
3. Leadbeater E (2015) What evolves in the evolution of social learning? J Zool (Lond) 295:
4–11.
4. Iacoboni M, et al. (1999) Cortical mechanisms of human imitation. Science 286:
2526–2528.
5. Rizzolatti G, Fogassi L, Gallese V (2001) Neurophysiological mechanisms underlying
the understanding and imitation of action. Nat Rev Neurosci 2:661–670.
6. Leslie AM (1987) Pretense and representation: The origins of “theory of mind.”
Psychol Rev 94:412–426.
7. Baron-Cohen S (2000) Theory of mind and autism: A fifteen year review. Understanding
Other Minds, eds Baron-Cohen S, Tagar-Flusberg H, Cohen DJ (Oxford Univ Press,
Oxford), Vol A, pp 3–20.
8. Cosmides L, Tooby J, Fiddick L, Bryant GA (2005) Detecting cheaters. Trends Cogn Sci
9:505–506, author reply 508–510.
9. Chomsky N (1965) Aspects of the Theory of Syntax (MIT Press, Cambridge, MA).
10. Tooby J, Cosmides L, Barrett HC (2005) Resolving the debate on innate ideas. The Innate
Mind: Structure and Content, eds Carruthers P, Laurence S, Stich S (Oxford Univ Press,
New York), pp 305–337.
11. Anderson ML (2010) Neural reuse: A fundamental organizational principle of the
brain. Behav Brain Sci 33:245–266, discussion 266–313.
12. Bates E (1993) Modularity, domain specificity and the development of language.
Discuss Neurosci 10:136–148.
13. Scott-Phillips TC, Laland KN, Shuker DM, Dickins TE, West SA (2014) The niche con-
struction perspective: A critical appraisal. Evolution 68:1231–1243.
14. Laland KN, Odling-Smee FJ, Feldman MW (1999) Evolutionary consequences of
niche construction and their implications for ecology. Proc Natl Acad Sci USA 96:
10242–10247.
Lotem et al.
COLLOQUIUM
PAPER
various hand movements, stone tools in various stages of com-
pletion, and the association of all these images and segments in
time and space. According to our model, having a rich, well-
segmented, and well-connected network helps to put new ob-
servations in context and to produce effective actions (81, 82,
87). Moreover, the ability to create a well-segmented and well-
connected network likely depends on appropriate settings of the
parameters of the data acquisition and the learning mechanisms
that govern the dynamic process of network construction. Thus,
the fine-tuning of these mechanisms by natural selection to
produce the most effective network for the purpose of learning
to make tools would be precisely the manner in which the culture
of tool-making shapes the evolution of cognition.
Conclusions
In this paper we embraced the view that cognitive mechanisms
have evolved to accommodate—among other tasks—the rela-
tively new challenges of learning cultural constructs, such as
language and tool-making techniques or, simply put, that culture
shaped cognition. We claim, however, that to study how such
cultural constructs shape cognitive evolution, a computationally
explicit process-level mechanistic model of learning may be re-
quired. We described such a model, one that is based on coevolving
mechanisms of learning and data acquisition that jointly construct a
complex network, capable of supporting a range of cognitive abil-
EVOLUTION
ities. The effect of culture on cognitive evolution is captured
through small modifications of these coevolving learning and data-
acquisition mechanisms, whose coordinated action improves the
network’s ability to support the learning processes that are involved
in cultural phenomena, such as language or tool-making. Finally, we
proposed that culture exerts selective pressure that shapes learning
PSYCHOLOGICAL AND
COGNITIVE SCIENCES
and data acquisition parameters, which in turn shape the structure
of the representation network, so that over evolutionary time scales,
brain anatomy may be selected to better accommodate the physical
requirements of the learned processes and representations.
ACKNOWLEDGMENTS. We thank the organizers and funders of the Arthur
M. Sackler Colloquium on “The Extension of Biology Through Culture.” We
also thank two anonymous reviewers for highly constructive comments.
Funding was provided by the Israel Science Foundation Grant 871/15 (to
A.L.) and by NSF Grant CCF-1214844, Air Force Office of Scientific Research
Grant FA9550-12-1-0040, and Army Research Office Grant W911NF-14-1-0017
(to J.Y.H.). O.K. was supported by the John Templeton Foundation Grant ID
47981 and by the Stanford Center for Computational, Evolutionary, and
Human Genomics.
15. Odling-Smee FJ, Laland KN, Feldman MW (2003) Niche Construction: The Neglected
Process in Evolution (Princeton Univ Press, Princeton, NJ).
16. Iriki A, Taoka M (2012) Triadic (ecological, neural, cognitive) niche construction: A
scenario of human brain evolution extrapolating tool use and language from the
control of reaching actions. Philos Trans R Soc Lond B Biol Sci 367:10–23.
17. Laland KN, Hoppitt W (2003) Do animals have culture? Evol Anthropol 12:
150–159.
18. Laland KN, Janik VM (2006) The animal cultures debate. Trends Ecol Evol 21:542–547.
19. Stout D, Hecht EE (2017) Evolutionary neuroscience of cumulative culture. Proc Natl
Acad Sci USA 114:7861–7868.
20. Solan Z (2005) Unsupervised learning of natural languages. Proc Natl Acad Sci USA
102:11629–11634.
21. Laland KN, Odling-Smee J, Myles S (2010) How culture shaped the human genome:
Bringing genetics and the human sciences together. Nat Rev Genet 11:137–148.
22. Williamson SH, et al. (2007) Localizing recent adaptive evolution in the human ge-
nome. PLoS Genet 3:e90.
23. de Magalhães JP, Matsuda A (2012) Genome-wide patterns of genetic distances
reveal candidate loci contributing to human population-specific traits. Ann Hum
Genet 76:142–158.
24. Somel M, Liu X, Khaitovich P (2013) Human brain evolution: Transcripts, metabolites
and their regulators. Nat Rev Neurosci 14:112–127.
25. Cáceres M, et al. (2003) Elevated gene expression levels distinguish human from non-
human primate brains. Proc Natl Acad Sci USA 100:13030–13035.
26. Somel M, Rohlfs R, Liu X (2014) Transcriptomic insights into human brain evolution:
Acceleration, neutrality, heterochrony. Curr Opin Genet Dev 29:110–119.
27. Hecht EE, et al. (2015) Acquisition of Paleolithic toolmaking abilities involves
structural remodeling to inferior frontoparietal regions. Brain Struct Funct 220:
2315–2331.
PNAS | July 25, 2017 | vol. 114 | no. 30 | 7921
28. Hecht EE, Gutman DA, Bradley BA, Preuss TM, Stout D (2015) Virtual dissection and
comparative connectivity of the superior longitudinal fasciculus in chimpanzees and
humans. Neuroimage 108:124–137.
29. Creanza N, Fogarty L, Feldman MW (2016) Cultural niche construction of repertoire
size and learning strategies in songbirds. Evol Ecol 30:285–305.
30. Shettleworth SJ (2010) Cognition, Evolution, and Behavior (Oxford Univ Press, New York).
31. Heyes C (2010) Where do mirror neurons come from? Neurosci Biobehav Rev 34:575–583.
32. Heyes C (2016) Homo imitans? Seven reasons why imitation couldn’t possibly be
associative. Philos Trans R Soc Lond B Biol Sci 371:20150069.
33. Heyes C (2012) What’s social about social learning? J Comp Psychol 126:193–202.
34. Laland KN (2004) Social learning strategies. Learn Behav 32:4–14.
35. Catmur C, Press C, Cook R, Bird G, Heyes C (2014) Authors’ response: Mirror neurons:
Tests and testability. Behav Brain Sci 37:221–241.
36. Felsenstein J (1983) Parsimony in systematics: Biological and statistical issues. Annu
Rev Ecol Syst 14:313–333.
37. Lotem A (1993) Secondary sexual ornaments as signals: The handicap approach and
three potential problems. Etologia 3:209–218.
38. Mery F, Belay AT, So AK, Sokolowski MB, Kawecki TJ (2007) Natural polymorphism
affecting learning and memory in Drosophila. Proc Natl Acad Sci USA 104:13051–13055.
39. Dunlap AS, Stephens DW (2014) Experimental evolution of prepared learning. Proc
Natl Acad Sci USA 111:11750–11755.
40. Garcia J, Kimeldorf DJ, Koelling RA (1955) Conditioned aversion to saccharin re-
sulting from exposure to gamma radiation. Science 122:157–158.
41. Finkel D, Pedersen NL, McGue M, McClearn GE (1995) Heritability of cognitive abilities
in adult twins: comparison of Minnesota and Swedish data. Behav Genet 25:421–431.
42. Plomin R, Spinath FM (2002) Genetics and general cognitive ability (g). Trends Cogn
Sci 6:169–176.
43. Briley DA, Tucker-Drob EM (2013) Explaining the increasing heritability of cognitive
ability across development: A meta-analysis of longitudinal twin and adoption
studies. Psychol Sci 24:1704–1713.
44. Pearson-Fuhrhop KM, Minton B, Acevedo D, Shahbaba B, Cramer SC (2013) Genetic
variation in the human brain dopamine system influences motor learning and its
modulation by L-Dopa. PLoS One 8:e61197.
45. Mery F (2013) Natural variation in learning and memory. Curr Opin Neurobiol 23:52–56.
46. Stephens DW, Krebs JR (1986) Foraging Theory (Princeton Univ Press, Princeton, NJ).
47. Leadbeater E, Dawson EH (2017) A social insect perspective on the evolution of social
learning mechanisms. Proc Natl Acad Sci USA 114:7838–7845.
48. Lotem A, Kolodny O (2014) Reconciling genetic evolution and the associative
learning account of mirror neurons through data-acquisition mechanisms. Behav
Brain Sci 37:210–211.
49. Thompson B, Kirby S, Smith K (2016) Culture shapes the evolution of cognition. Proc
Natl Acad Sci USA 113:4530–4535.
50. Christiansen MH, Chater N (2008) Language as shaped by the brain. Behav Brain Sci
31:489–508, discussion 509–558.
51. Barbujani G, Sokal RR (1990) Zones of sharp genetic change in Europe are also lin-
guistic boundaries. Proc Natl Acad Sci USA 87:1816–1819.
52. Laland KN (2017) The origins of language in teaching. Psychon Bull Rev 24:225–231.
53. Belfer‐Cohen A, Goren‐Inbar N (1994) Cognition and communication in the Le-
vantine Lower Palaeolithic. World Archaeol 26:144–157.
54. d’Errico F, et al. (2003) Archaeological evidence for the emergence of language, sym-
bolism, and music—An alternative multidisciplinary perspective. J World Prehist 17:1–70.
55. Mellars P (2006) Why did modern human populations disperse from Africa ca.
60,000 years ago? A new model. Proc Natl Acad Sci USA 103:9381–9386.
56. Chater N, Reali F, Christiansen MH (2009) Restrictions on biological adaptation in
language evolution. Proc Natl Acad Sci USA 106:1015–1020.
57. Chater N, Christiansen MH (2010) Language acquisition meets language evolution.
Cogn Sci 34:1131–1157.
58. Christiansen MH, Chater N (2016) The Now-or-Never bottleneck: A fundamental
constraint on language. Behav Brain Sci 39:e62.
59. Lotem A, Kolodny O, Halpern JY, Onnis L, Edelman S (2016) The bottleneck may be
the solution, not the problem. Behav Brain Sci 39:e83.
60. Mueller ST, Krawitz A (2009) Reconsidering the two-second decay hypothesis in
verbal working memory. J Math Psychol 53:14–25.
61. Cui J, Gao D, Chen Y, Zou X, Wang Y (2010) Working memory in early-school-age
children with Asperger’s syndrome. J Autism Dev Disord 40:958–967.
62. Blokland GAM, et al. (2011) Heritability of working memory brain activation.
J Neurosci 31:10882–10890.
63. Vogler C, et al. (2014) Substantial SNP-based heritability estimates for working
memory performance. Transl Psychiatry 4:e438.
64. Chater N, Christiansen MH (2016) Squeezing through the Now-or-Never bottleneck: Re-
connecting language processing, acquisition, change, and structure. Behav Brain Sci 39:e91.
65. McNamara JM, Houston AI (2009) Integrating function and mechanism. Trends Ecol
Evol 24:670–675.
66. Fawcett TW, Hamblin S, Giraldeau LA (2013) Exposing the behavioral gambit: The
evolution of learning and decision rules. Behav Ecol 24:2–11.
67. Kacelnik A, Bateson M (1997) Risk-sensitivity: Crossroads for theories of decision-
making. Trends Cogn Sci 1:304–309.
68. van den Berg P, Weissing FJ (2015) The importance of mechanisms for the evolution
of cooperation. Proc Biol Sci 282:20151382.
69. Trimmer PC, McNamara JM, Houston AI, Marshall JA (2012) Does natural selection
favour the Rescorla-Wagner rule? J Theor Biol 302:39–52.
70. Mcnamara JM, Trimmer PC, Houston AI (2012) The ecological rationality of state-
dependent valuation. Psychol Rev 119:114–119.
71. Arbilly M, Motro U, Feldman MW, Lotem A (2010) Co-evolution of learning com-
plexity and social foraging strategies. J Theor Biol 267:573–581.
7922 | www.pnas.org/cgi/doi/10.1073/pnas.1620742114
72. Lange A, Dukas R (2009) Bayesian approximations and extensions: Optimal decisions
for small brains and possibly big ones too. J Theor Biol 259:503–516.
73. Katsnelson E, Motro U, Feldman MW, Lotem A (2011) Evolution of learned strategy
choice in a frequency-dependent game. Proc Biol Sci 279:1176–1184.
74. Hamblin S, Giraldeau L-A (2009) Finding the evolutionarily stable learning rule for
frequency-dependent foraging. Anim Behav 78:1343–1350.
75. Nilsson DE, Pelger S (1994) A pessimistic estimate of the time required for an eye to
evolve. Proc Biol Sci 256:53–58.
76. Schmidhuber J (2015) Deep learning in neural networks: An overview. Neural Netw
61:85–117.
77. LeCun Y, Bengio Y, Hinton G (2015) Deep learning. Nature 521:436–444.
78. Edelman S (2016) The minority report: Some common assumptions to reconsider in
the modelling of the brain and behaviour. J Exp Theor Artif Intell 28:751–776.
79. Mnih V, et al. (2015) Human-level control through deep reinforcement learning.
Nature 518:529–533.
80. Gu S, Lillicrap T, Sutskever I, Levine S (2016) Continuous deep Q-learning with model-
based acceleration. Proceedings of The 33rd International Conference on Machine
Learning, eds Balcan MF, Weinberger KQ (Proceedings of Machine Learning Re-
search, New York), Vol 48, pp 2829–2838.
81. Lotem A, Halpern JY (2008) A data-acquisition model for learning and cognitive
development and its implications for autism. Cornell University Computing and In-
formation Science Technical Reports. Available at https://ecommons.cornell.edu/
handle/1813/10178. Accessed May 11, 2017.
82. Lotem A, Halpern JY (2012) Coevolution of learning and data-acquisition mecha-
nisms: A model for cognitive evolution. Philos Trans R Soc Lond B Biol Sci 367:
2686–2694.
83. Goldstein MH, et al. (2010) General cognitive principles for learning structure in time
and space. Trends Cogn Sci 14:249–258.
84. Kolodny O, Edelman S, Lotem A (2014) The evolution of continuous learning of the
structure of the environment. J R Soc Interface 11:20131091.
85. Kolodny O, Edelman S, Lotem A (2015) Evolution of protolinguistic abilities as a by-
product of learning to forage in structured environments. Proc Biol Sci 282:
20150353.
86. Kolodny O, Lotem A, Edelman S (2015) Learning a generative probabilistic grammar
of experience: A process-level model of language acquisition. Cogn Sci 39:227–267.
87. Kolodny O, Edelman S, Lotem A (2015) Evolved to adapt: A computational approach
to animal innovation and creativity. Curr Zool 61:350–367.
88. Enquist M, Lind J, Ghirlanda S (2016) The power of associative learning and the
ontogeny of optimal behaviour. R Soc Open Sci 3:160734.
89. Laland KN, et al. (2015) The extended evolutionary synthesis: Its structure, assump-
tions and predictions. Proc Biol Sci 282:20151019.
90. Baddeley A, Gathercole S, Papagno C (1998) The phonological loop as a language
learning device. Psychol Rev 105:158–173.
91. Burgess N, Hitch GJ (1999) Memory for serial order: A network model of the pho-
nological loop and its timing. Psychol Rev 106:551–581.
92. Baldwin JM (1896) A new factor in evolution. Am Nat 30:441–451.
93. Weber BH, Depew DJ (2003) Evolution and Learning: The Baldwin Effect Reconsidered
(MIT Press, Cambridge, MA).
94. Dunbar R (1998) Grooming, Gossip, and the Evolution of Language (Harvard Univ
Press, Cambridge, MA).
95. Pinker S (2003) Language as an adaptation to the cognitive niche. Stud Evol Lang 3:
16–37.
96. Premack D (1985) “Gavagai!” or the future history of the animal language contro-
versy. Cognition 19:207–296.
97. Butterworth G, Jarrett N (1991) What minds have in common is space: Spatial
mechanisms serving joint visual attention in infancy. Br J Dev Psychol 9:55–72.
98. Scaife M, Bruner JS (1975) The capacity for joint visual attention in the infant. Nature
253:265–266.
99. Klin A, Lin DJ, Gorrindo P, Ramsay G, Jones W (2009) Two-year-olds with autism
orient to non-social contingencies rather than biological motion. Nature 459:
257–261.
100. Curio E, Ernst U, Vieth W (1978) Cultural transmission of enemy recognition: One
function of mobbing. Science 202:899–901.
101. Dunsmoor JE, Murty VP, Davachi L, Phelps EA (2015) Emotional learning selectively
and retroactively strengthens memories for related events. Nature 520:345–348.
102. Medina TN, Snedeker J, Trueswell JC, Gleitman LR (2011) How words can and cannot
be learned by observation. Proc Natl Acad Sci USA 108:9014–9019.
103. Byrne RW (1999) Imitation without intentionality. Using string parsing to copy the
organization of behaviour. Anim Cogn 2:63–72.
104. Inoue S, Matsuzawa T (2007) Working memory of numerals in chimpanzees. Curr
Biol 17:R1004–R1005.
105. Conway ARA, Jarrold C, Kane MJ, Miyake A, Towse JN (2007) Variation in working
memory: An introduction. Variation in Working Memory, eds Conway ARA, Jarrold C,
Kane MJ, Miyake A, Towse JN (Oxford Univ Press, Oxford), pp 3–17.
106. Snyder AW, Mitchell DJ (1999) Is integer arithmetic fundamental to mental pro-
cessing?: The mind’s secret arithmetic. Proc Biol Sci 266:587–592.
107. Morgan TJH, et al. (2015) Experimental evidence for the co-evolution of hominin
tool-making teaching and language. Nat Commun 6:6029.
108. Whiten A (2000) Primate culture and social learning. Cogn Sci 24:477–508.
109. Heyes CM, Galef BG, Jr (1996) Social Learning in Animals: The Roots of Culture
(Academic, San Diego).
110. Galef BG (2015) Laboratory studies of imitation/field studies of tradition: Towards a
synthesis in animal social learning. Behav Processes 112:114–119.
111. Truskanov N, Lotem A (2017) Trial-and-error copying of demonstrated actions re-
veals how fledglings learn to ‘imitate’ their mothers. Proc Biol Sci 284:20162744.
Lotem et al.
 NEWS FEATURE
NEWS FEATURE
Can animal culture drive evolution?
Once the purview of humans, culture has been observed in all sorts of animals. But are these
behaviors merely ephemeral fads or can they shape the genes and traits of
future generations?
Carolyn Beans, Science Writer
In Antarctic waters, a group of killer whales makes a
wave big enough to knock a seal from its ice floe.
Meanwhile, in the North Atlantic, another killer whale
group blows bubbles and flashes white bellies to
herd a school of herrings into a ball. And in the Crozet
Archipelago in the Southern Ocean, still another group
charges at seals on a beach, grasps the prey with their
teeth, and then backs into the water (1). Some re-
searchers see these as more than curious behaviors
or YouTube photo ops: they see cultural mores—
introduced into populations and passed to future gener-
ations—that can actually affect animals’ fitness.
Killer whales, also known as orcas (Orcinus orca), have
a geographic range stretching from the Antarctic to the
Arctic. As a species, their diet includes birds, fish, mam-
mals, and reptiles. But as individuals, they typically fall
into groups with highly specialized diets and hunting
traditions passed down over generations. Increasingly,
scientists refer to these learned feeding strategies as cul-
ture, roughly defined as information that affects behavior
and is passed among individuals and across generations
through social learning, such as teaching or imitation (2).
Killer whales are divided into groups known as ecotypes, with highly specialized
diets and hunting traditions passed down over generations. Here, a
mammal-eating ecotype in the North Pacific hunts seal. Photograph by David
Ellifrit, courtesy of Center for Whale Research.
7734–7737 | PNAS | July 25, 2017 | vol. 114 | no. 30
Scientists once placed culture squarely in the human
domain. But discoveries in recent decades suggest that
a wide range of cultural practices—from foraging tactics
and vocal displays to habitat use and play—may influ-
ence the lives of other animals as well (3). Studies at-
tribute additional orca behaviors, such as migration
routes and song repertoires, to culture (4). Other re-
search suggests that a finch’s song (5), a chimpanzee’s
nut cracking (3), and a guppy’s foraging route (6) are all
manifestations of culture. Between 2012 and 2014, over
100 research groups published work on animal culture
covering 66 species, according to a recent review (7).
Now, scientists are exploring whether culture may
shape not only the lives of nonhuman animals but the
evolution of a species. “Culture affects animals’ lives and
their survival and their fitness,” says the review’s (7) coau-
thor, behavioral scientist Andrew Whiten of the University
of St Andrews in Scotland. “We’ve learned that’s the case
to an extent that could hardly have been appreciated half a
century ago.” Based on work in whales, dolphins, and
birds, some researchers contend that animal culture is likely
a common mechanism underlying animal evolution. But
testing this hypothesis remains a monumental challenge.
Riding a Cultural Wave
Animal populations essentially have two streams of in-
formation, genetic and cultural, explains ecologist and
whale researcher Hal Whitehead of Dalhousie University in
Canada. In the case of the cultural stream, he says, “things
are being learned, sometimes from the mother, possibly
from the father, as well as from peers and unrelated
adults.” Whitehead and others want to understand how
these streams interact. Lactose tolerance in humans is a
classic example. Studies suggest that adult production of
lactase—the enzyme necessary for digesting the sugar
lactose in milk—coevolved with the cultural practice of
dairy farming in Europe in the last 10,000 years (8).
Showing that culture can influence the distribution
of genes in an animal population would confirm its role
as an evolutionary driver, and Whitehead believes he
may have found evidence for exactly that. In the 1990s,
Whitehead observed that matrilineal whale species—
whose daughters stick with their mothers for life—have
low genetic diversity of mitochondrial DNA (9). He
coined the term “cultural hitchhiking” to explain how
this pattern might emerge. In these species, cultures
www.pnas.org/cgi/doi/10.1073/pnas.1709475114
are passed from mothers to offspring. If a cultural
behavior increases a descendant’s chances of sur-
vival and reproduction, then this behavior would
persist and become more common in the population.
The maternal line’s particular mitochondrial DNA hap-
lotype, which also passes directly from mother to off-
spring, would simultaneously become more common.
“The culture is driving and the gene is riding along,”
says Whitehead. “There is no particular functional
linkage between them.” Whitehead demonstrated
through computer models that cultural hitchhiking is a
plausible explanation for reduced genetic diversity in
matrilineal whale species.
Cultural hitchhiking, it seems, is also at work in a
population of bottlenose dolphins in western Shark
Bay, Western Australia, according to research by evo-
lutionary geneticist Michael Krützen of the University of
Zurich (10). In this population, some dolphins carry
sponges on their rostrums, most likely for protection as
they probe the rough seafloor for fish that they other-
wise couldn’t reach (11). This behavior is passed from
mothers to offspring through social learning and all
“sponging” dolphins in the population share the same
mitochondrial haplotype. Because the sponging dol-
phins primarily inhabit a deep channel where the
sponges occur, this culture appears to affect the fine-
scale geographic distribution of the mitochondrial
genes. “What is really exciting here is that the cultural
practice of sponging has led to a change in the genetic For birds in the tanager family, like this magpie tanager (Cissopis leveriana) in
make-up of the population when you look at mito- Brazil’s Itatiaia National Park, song is a cultural trait that must be learned. Song
chondrial DNA,” says Krützen. evolves faster in this family than in the ovenbird family, whose species have innate
song. Image courtesy of Daniel J. Field (University of Bath, Bath, United Kingdom).

An Evolutionary Force
Longstanding ecological and evolutionary theories
suggest that culture could also more directly affect the
evolution of traits, and even the making of species.
Animal populations evolve through natural selection
when a heritable trait, like beak size or fur color, varies
and different versions of the trait allow some individuals
to survive and reproduce more than others.
Animal culture has the potential to affect this process
in a number of ways, says Whiten. For one, cultural in-
novations, such as tools or predator-avoidance tactics,
could increase an animal’s survival and reproduction,
buffering them against some selection pressures. But
culture could also enable animals to colonize regions
they otherwise couldn’t, exposing them to new selec-
tion pressures, such as novel temperatures, predators,
or food sources. And culture could generate selection
for animals to be better suited to a cultural behavior
through physical changes, such as stronger arms for
more powerful hammering, or cognitive ones, such as
the ability to learn tool use by mirroring others. “And
that, of course, may affect the evolution of the brain to
match,” says Whiten. Furthermore, cultural differences,
such as birdsong or migration patterns, could prevent
groups from mating together, which could help main-
tain or even generate new species.
Or anyway, those are the working theories. Finding
definitive evidence is a tricky prospect, though recent
research in whales and birds offers some substantive
support. Scientists refer to the many orca groups with
distinct hunting strategies as ecotypes, subsets of a
species that occupy unique ecological niches. New
genomics technologies allow researchers to search for
evolutionary consequences of these various hunting
cultures. “We came into the genomics era and really
wanted to see whether these cultural traditions in killer
whales led to enough of a long-term selection pressure
that you would actually see changes in the genome,”
says evolutionary biologist Andrew Foote of Bangor
University in the United Kingdom.
Foote and colleagues sequenced the genomes of
48 orcas across 5 ecotypes to identify whether the
groups were truly genetically isolated, and whether their
different cultures were associated with unique genomic
changes (1). The sample included one mammal-eating
and one salmon-eating ecotype from the North Pacific,
and one mammal-eating, one penguin-eating, and one
Antarctic toothfish-eating ecotype from the Antarctic.
The researchers found that the groups were genetically
distinct. “What is really surprising is just how differenti-
ated the ones that live in the same area are,” says Foote.
“The two North Pacific ones are really different genet-
ically even though there is overlap in their range.”
Foote estimated that these ecotypes began di-
verging within the last 250,000 years. He traced some
of the genetic differences among groups to gene
variants possibly associated with adaptation to the
hunting traditions of each ecotype, and the unique

Beans PNAS | July 25, 2017 | vol. 114 | no. 30 | 7735

 geographic regions those ecotypes colonized. For
example, the two mammal-eating ecotypes were
each associated with gene variants that play key roles
in regulating the metabolism of methionine, an es-
sential amino acid that mammal eaters consume in a
boom–bust cycle with influxes following each kill. And
the ecotypes that live in the extreme cold of the
Antarctic were associated with gene variants involved
in the development of adipose tissue, which could
protect individuals from the frigid climate.
Foote doesn’t believe the cultural barrier between
orca ecotypes is long-lasting enough to divide groups
into different species altogether. “They probably radi-
ate and collapse and radiate and collapse,” he says.
“Maybe one or two might escape that process and go
on to become fully fledged species. But looking at what
we know now, knowing that they have a relatively re-
cent common ancestor, it would suggest that [splitting
into species] never happened in the past.”
Speciation and Song
Birdsong, often used to identify mates, offers another
robust means for probing culture’s impact on specia-
tion and animal evolution. For some bird species, song
is innate. For others, it’s essentially cultural, a trait that
must be learned. In both cases, song evolves over time
“My inkling is that rapid evolution of birdsong could
contribute to speciation.”
—Nicholas Mason
as it passes through generations. Evolutionary biologist
Elizabeth Derryberry of the University of Tennessee,
Knoxville, and Nicholas Mason, a doctoral candidate in
ecology and evolutionary biology at Cornell University,
studied two families of birds: the tanagers (Thraupidae)
that learn song and the ovenbirds (Furnariidae) that are
innate singers (12). What they found suggests that
culture could play a sizeable role.
Derryberry and Mason analyzed nearly 4,500 song
recordings across nearly 600 species within these
families. For each recording, they measured eight
vocal characters, including maximum volume, range
of pitch, and length. By studying differences in these
characters between species in the same family, the
researchers estimated how quickly song evolved in
different branches of the family tree. If song differed
greatly between closely related species, for example,
that would suggest a fast rate of song evolution. For
each family, they merged this song dataset with a
genetic one that showed rates of speciation; the idea
was to identify any connection between the rates of
song change and species divisions.
Derryberry and Mason found that when the rate of
song evolution sped up in a branch of a family tree, so
too did the rate of speciation. But song evolved
1.4 times faster in the tanager family, with cultural
transmission of song, than in the ovenbird family, with
7736 | www.pnas.org/cgi/doi/10.1073/pnas.1709475114
innate song. Culture, therefore, might actually ramp up
the pace of speciation.
Derryberry and Mason (12) acknowledge that they
don’t know whether bird song evolution drives speciation
or vice versa. In one scenario, bird song could diverge first,
which would prevent individuals with different songs from
mating together, setting their lineages on the path to be-
coming distinct species. Alternatively, the species could
diverge first by some other mechanism, which would cre-
ate strong natural selection for song divergence to follow. If
song evolution comes first, then the faster bird song
evolves, the more rapidly species diverge. “My inkling is
that rapid evolution of birdsong could contribute to spe-
ciation,” says Mason. “At the scale we are looking at, we
look at patterns, so interpreting process becomes tricky.”
A separate, long-term study may offer insight into
the speciation cause and effect. For four decades,
evolutionary biologists Peter and Rosemary Grant of
Princeton University carefully tracked the survival, mat-
ing, and reproduction success of about 12,000 individ-
ual birds in species of Darwin’s finches on the island of
Daphne Major in the Galápagos (5). In these species,
which belong to the tanager family included in Mason’s
study, offspring learn song from their fathers.
In 1981, a male bird from a nearby island arrived on
Daphne Major singing a song the Grants had never
heard. Genetic analyses (conducted decades later with
microsatellite data) suggested that it was possibly a hy-
brid of the medium ground finch (Geospiza fortis) and
the cactus finch (Geospiza scandens), two species that
were also found on Daphne Major. But this bird, which
the Grants call “Big Bird,” was much larger than the
parent species. Big Bird survived for 13 years in his new
home and found six mates: the first three hybrids like
himself, the last three all medium ground finches. To-
gether with one of the medium ground finches, Big Bird
produced offspring that bred only with one another,
resulting in the beginnings of an incipient species that
the Grants have now followed through six generations.
The bird’s unique song passed on through genera-
tions helped members of his lineage recognize one an-
other as potential mates. “It’s very important that it’s had
cultural transmission of song,” says Rosemary Grant.
There is no agreed upon standard for how many gener-
ations a lineage must remain reproductively isolated be-
fore it can be called a new species, so the Grants maintain
only that the Big Bird lineage is a species in the making.
Many Unknowns
Despite such findings, well-documented examples of
animal culture influencing evolution remain rare, even in
humans’ closest relatives, primates. There’s little doubt
that cultural differences and social learning are impor-
tant to primates’ lives. But can such behaviors have an
evolutionary impact?
Whiten cites work supporting an evolutionary ver-
sion of the “cultural intelligence hypothesis” in pri-
mates: the idea that species with culturally rich
communities will experience selection to enhance the
cognitive abilities that support social learning, which
would in turn require a larger brain capable of pro-
cessing learning and storing learned information. This
Beans
larger brain could possibly result in increased overall
intelligence. “You may get into a feedback cycle here;
as you become more cultural, that selection pressure
on the brain and cultural capacities then make you
able to become more cultural, which in turn selects for
greater brain size,” says Whiten.
Indeed, a recent study by Kevin Laland, of the Uni-
versity of St Andrews, and colleagues found that in
primate species, reliance on social learning is positively
correlated with brain volume, as well as social group
size and lifespan (13). But the authors acknowledge that
they cannot determine whether selection on social
learning actually caused the evolution of larger brain
size. It’s also possible that larger brains evolved first for
some other purpose, and then cultural advances made
possible by enhanced cognitive abilities followed.
Whitehead wouldn’t be surprised if researchers find
support for gene-culture evolution in primates. “There is
evidence that some cultural elements in chimpanzees
have remarkable stability, and stability is a prerequisite for
culture having a major effect on genetics.” To find that
effect, he says researchers might test, for example,
whether chimpanzee populations with a culture of using
stones as tools also carry gene variants that enhance their
abilities to use those tools, such as greater hand–eye co-
ordination or muscle strength in parts of their bodies. But
the field is still new and such proposals remain theoretical.
The paucity of examples could also indicate that animal
culture is not quite influential or stable enough to routinely
have an impact on evolution, says cultural evolutionist
Peter Richerson of the University of California, Davis, who is
president of the recently founded Cultural Evolution So-
ciety. “There [are] just more targets in the case of humans
than in the case of other culture-bearing animals,” he says.
“That doesn’t mean that we won’t find a lot of examples—
I expect we will in the long run. But it still ought to be more
spectacular in humans than in other animals.”
Even in humans, thus far there are only a few ex-
amples in which potential genetic changes have been
conclusively linked to cultural variation. “The strong
cases are where we see independent cultural evolution:
dairy farming cropping up multiple places and the
same genes popping up,” says Foote. “That’s missing
in most tentative cases.”
One of the biggest challenges with animal studies is
determining whether genetic differences between pop-
ulations are really a response to culture or merely a sig-
nature of genetic drift: chance fluctuations in the
frequencies of gene variants over time. “It’s not an easy
task. You really need to know something about the de-
mographic history of your species,” says Krützen. He calls
Foote’s orca ecotypes study amazing, in part because it at
least partially disentangled genetic drift from culturally
driven natural selection. Ideally, Krützen would like to see
evidence that this same gene is under selection in many
different species all experiencing a similar selection
pressure. Indeed, Foote notes that adipose tissue gene
variants are also under selection in the polar bear when
compared with the brown bear, which may similarly help
buffer this Arctic animal against a frigid climate (14).
Even once scientists identify genes that may be
under selection, they still must determine whether the
genes connect back to culture. “One of the difficulties
is that there aren’t that many genes that we actually
know exactly what they do, even in humans, even less in
other species, and certainly in nonmodel species like
the whales,” says Whitehead. The conventional strat-
egy for definitively determining a gene–phenotype
connection entails experimentally altering, for example,
a gene in orcas related to adipose tissue, and then re-
cording the effects. “But, of course, we can’t do this,”
Krützen says. He says that instead, scientists often at-
tempt to assign function to a gene by comparing it to
similar genes of known function in humans. This ap-
proach may work well for studying close animal rela-
tives like primates. “The farther you go away from
humans,” he says, “the harder this gets.”
Even while recognizing the research limitations, Krüt-
zen remains undeterred. “I’m convinced that as time goes
by there will be more studies finding more evidence of
genetic change based on culture,” he says. Foote is less
certain. “I think a lot of it comes under ‘untested and
unknown’,” he says. “And we have to keep an open mind
as to what the alternative hypotheses are.”

1 Foote AD, et al. (2016) Genome-culture coevolution promotes rapid divergence of killer whale ecotypes. Nat Commun 7:11693.
2 Whiten A, Ayala FJ, Feldman MW, Laland KN (2017) The extension of biology through culture. Proc Natl Acad Sci USA 114:7775–7781.
3 Whiten A (2017) Culture extends the scope of evolutionary biology in the great apes. Proc Natl Acad Sci USA 114:7790–7797.
4 Whitehead H (2017) Gene–culture coevolution in whales and dolphins. Proc Natl Acad Sci USA 114:7814–7821.
5 Grant PR, Grant BR (2014) 40 Years of Evolution: Darwin’s Finches on Daphne Major Island (Princeton Univ Press, Princeton, NJ).
6 Laland KN, Williams K (1997) Shoaling generates social learning of foraging information in guppies. Anim Behav 53:1161–1169.
7 Galef BG, Whiten A (2017) The comparative psychology of social learning. APA Handbook of Comparative Psychology, ed Call J
(American Psychological Association, Washington, DC), pp 411–439.
8 Itan Y, Powell A, Beaumont MA, Burger J, Thomas MG (2009) The origins of lactase persistence in Europe. PLOS Comput Biol
5:e1000491.
9 Whitehead H (2003) Sperm Whales: Social Evolution in the Ocean (The Univ of Chicago Press, Chicago).
10 Kopps AM, et al. (2014) Cultural transmission of tool use combined with habitat specializations leads to fine-scale genetic structure in
bottlenose dolphins. Proc Biol Sci 281:20133245.
11 Krützen M, et al. (2014) Cultural transmission of tool use by Indo-Pacific bottlenose dolphins (Tursiops sp.) provides access to a novel
foraging niche. Proc Biol Sci 281:20140374.
12 Mason NA, et al. (2017) Song evolution, speciation, and vocal learning in passerine birds. Evolution 71:786–796.
13 Street SE, Navarrete AF, Reader SM, Laland KN (2017) Coevolution of cultural intelligence, extended life history, sociality, and brain
size in primates. Proc Natl Acad Sci USA 114:7908–7914.
14 Liu S, et al. (2014) Population genomics reveal recent speciation and rapid evolutionary adaptation in polar bears. Cell 157:785–794.

Beans PNAS | July 25, 2017 | vol. 114 | no. 30 | 7737

The Extension of Biology Through Culture
November 16–17, 2016
Arnold and Mabel Beckman Center, Irvine, CA
Organized by Andrew Whiten, Francisco J. Ayala, Marcus W. Feldman, and Kevin N. Laland

Program

Wednesday, November 16

Session I

Welcome Remarks
Marcus W. Feldman, Stanford University

Evolution and revolution in cetacean vocal culture: Lessons from humpback

whale song
Ellen C. Garland, University of St. Andrews

Gene–culture coevolution in whales and dolphins

Hal Whitehead, Dalhousie University

Cultural legacies: Unpacking the intergenerational transmission of information

in birds
Lucy M. Aplin, University of Oxford

What evolves in the evolution of social learning? A social insect perspective

Ellouise Leadbeater, Queen Mary University of London

Session II

Can culture reshape the evolution of learning and how?

Arnon Lotem, Tel Aviv University

What long-term field studies reveal of primate traditions

Susan E. Perry, University of California, Los Angeles

Can we identify a primate signature in social learning?

Dorothy M. Fragaszy, University of Georgia

The evolution of primate intelligence

Kevin N. Laland, University of St. Andrews
 Distinctive Voices Public Lecture

How animal cultures extend the scope of biology: Tradition and learning from apes to
whales to bees
Andrew Whiten, University of St. Andrews

Thursday, November 17

Session III

Skill learning, neuroplasticity, and exaptation in the evolution of human tool-making

and language
Dietrich Stout, Emory University

The role of cultural innovations, learning processes, and ecological dynamics in shaping
Middle Stone Age cultural adaptations
Francesco d’Errico, University of Bordeaux

The ontogenetic foundations of cumulative cultural transmission

Cristine H. Legare, The University of Texas at Austin

“I don’t know”: Ignorance and question-asking as engines for cognitive development

Paul L. Harris, Harvard University

Session IV

Childhood as simulated annealing: How wide hypothesis exploration in an extended

childhood contributes to cultural learning
Alison Gopnik, University of California, Berkeley

How language shapes the nature of cultural inheritance

Susan A. Gelman, University of Michigan

Big data, cultural macroevolution, and the prospects for an evolutionary science of
human history
Russell D. Gray, Max Planck Institute for the Science of Human History

Ongoing prospects for a unified science of cultural evolution

Alex Mesoudi, University of Exeter

Concluding Remarks
Francisco J. Ayala, University of California, Irvine

Presentations from this colloquium can be viewed at

http://www.nasonline.org/Extension_of_Biology_Through_Culture
This page left intentionally blank
This page left intentionally blank
National Academy of Sciences
Sackler Colloquium Series

The Arthur M. Sackler Colloquia of the National Academy of Sciences

address scientific topics of broad and current interest, cutting across the
boundaries of traditional disciplines. Each year, three or four such collo-
quia are scheduled, typically 2 days in length and international in scope.
Each colloquium is organized by a member of the Academy, often with
the assistance of an organizing committee, and features presentations
by leading scientists in the field and discussions with 100 or more
researchers with an interest in the topic. Colloquium presentations are
recorded and posted on the Sackler Colloquia website
www.nasonline.org/programs/sackler-colloquia and the Sackler
YouTube Channel www.youtube.com/sacklercolloquia. Many colloquia
also result in the publication of scholarly papers, usually published as a
collection in the Proceedings of the National Academy of Sciences
(PNAS). These colloquia are made possible by a generous gift from
Mrs. Jill Sackler, in memory of her husband, Arthur M. Sackler.
National Academy of Sciences
500 Fifth Street, NW
Washington, DC 20001