Professional Documents
Culture Documents
&THEORY in the
STUDY OF
RELIGION
Justin L. Barrett
Centre for Anthropology & Mind, Institute of Cognitive & Evolutionary Anthropology,
University of Oxford, 64 Banbury Road, Oxford, OX2 6PN United Kingdom
justin.barrett@anthro.ox.ac.uk
Abstract
Boyer’s theory of counterintuitive cultural concept transmission claims that concepts that ideas
that violate naturally occurring intuitive knowledge structures enough to be attention-demand-
ing but not so much to undermine conceptual coherence have a transmission advantage over
other concepts (Boyer et al. 2001: 535-64). Because of the prominence of these counterintuitive
concepts in religious belief systems, Boyer’s theory features prominently in many cognitive treat-
ments of religion. Difficulties in identifying what are and are not counterintuitive concepts in
this technical sense, however, has made empirical treatment of Boyer’s theory irregular and dif-
ficult to evaluate. Further, inability to quantify just how counterintuitive a given concept is has
made ambiguous specifying where the alleged cognitive optimum lies. The present project
attempts to clarify Boyer’s theory and presents a formal system for coding and quantifying the
“counterintuitiveness” of a concept, and hence, facilitates empirical scrutiny of the theory.
Keywords
cognitive science of religion, religious concepts, counterintuitive, cultural transmission
1
The author thanks Pascal Boyer, Emma Cohen, Joe Henrich, Nicola Knight, Anders Lis-
dorf, Brian Malley, Bob McCauley, Tenelle Porter, and Rich Sosis for comments and suggestions.
This work was supported in part by a grant from the John Templeton Foundation.
© Koninklijke Brill NV, Leiden, 2008 DOI: 10.1163/157006808X371806
J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338 309
2
Boyer (2001, 2003) has identified other features of many religious concepts in addition to
their counterintuitiveness, and does not regard counterintuitiveness as either necessary or suffi-
cient for a concept to be regarded as “religious.”
310 J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338
For quite some time, cognitive developmental psychologists have talked about
intuitive knowledge, assumptions, or expectations of children. By “intuitive”
they refer to cognition that does not require or does not allow for conscious
inspection of the principles involved. Further, intuitive expectations are
assumed to be largely invariable across cultural contexts, arising through the
course of normal development through the interaction of human biology and
typically human natural and social environments. Whether some of this intu-
itive knowledge is innate or rather the product of learning remains an open
question for empirical inquiry and debate. But both the empiricists and nativ-
ists seem to agree that by early childhood, a certain set of intuitive knowledge
is present in children. For instance, there is little doubt that sometime in early
development, children acquire the intuitive expectation that bounded physical
objects cannot pass directly through other bounded physical objects. That
infants ‘know’ this property of physical objects does not entail that they can
consciously reflect on it or verbally articulate it. Rather, they know it intui-
tively in the sense that it is assumed without conscious reflection.
Recently, philosopher Robert McCauley (forthcoming) has usefully
described this intuitive knowledge as maturationally natural. That is, as a natu-
ral product of human maturation in ordinary human environments, certain
motor and conceptual competencies are acquired with great fluency and auto-
maticity. Walking is maturationally natural in this sense, as is knowing that
physical objects require support, or else they fall. Maturationally natural
cognition—or intuitive knowledge—does not require great conscious resources
but readily springs into action whenever needed. In addition to this automa-
ticity and fluency, maturationally natural competencies, according to McCau-
ley, are generally (but not always) marked by developing so early that by
adulthood we do not remember a time that we did not possess the compe-
tency. Further, acquiring these competencies occurs without special artefacts
or explicit tuition. Maturationally natural competencies show relatively little
variation within or across groups.
McCauley contrasts maturational naturalness with practiced naturalness.3
He notes that with the right sort of training or practice one can acquire auto-
maticity and fluency in motor and conceptual capacities that seem to rival that
which is characteristic of the maturationally natural. Consider the cognitive
skills of a chess master or the performance of a concert violinist. Nevertheless,
3
This labeling should not lead one to assume that maturationally natural competencies do
not require practice whereas practiced naturalness does. It isn’t the practice that distinguishes
these two types of natural cognition but their inevitability.
J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338 311
this sort of naturalness requires special artefacts (e.g., a chess set or violin),
typically involves explicit instruction, and varies considerably both within
groups (few become chess masters) and across groups (some cultures have
more chess masters than others). Cultural schemata and scripts (e.g., how one
orders food in a restaurant) may acquire this practiced naturalness in people.
I raise McCauley’s distinction because it emphasizes that what I refer to
here as intuitive cognition (what he would call maturationally natural) is not
the product of particular cultural conditions, but may be regarded as pan-
human. Hence, it plays a role in all thought and communication regardless of
cultural particulars. Further, intuitive cognition occurs early in development.
Hence, it plays a role in shaping thought and communication from the early
years of life.
This notion of intuitive cognition has been imported quite successfully into
explanations of recurrent cultural patterns. Simply put, ideas that match intu-
itive cognition are likely to appear widely within and across cultures. And why
not? Largely intuitive ideas are not only likely to spontaneously appear in
individual minds (in part because somewhere in our minds, their elements
already lurk), but also to spread readily from person to person (because, again,
somewhere in our minds, their elements already lurk). For instance, learning
about a previously unfamiliar animal that meets all intuitive expectations for
animals is easy because a host of presumed features need not be explicitly com-
municated. When explaining the features of a large South American rodent, I
need not mention that it moves in goal-directed ways, that it seeks nourish-
ment to survive, that it may die, that its parents and offspring will be of the
same species, that it has internal parts that serve the animal’s survival needs,
that it is composed of natural materials, that it exists continually in space and
time, that it cannot pass through solid walls, that it is subject to gravity, and
so forth. All of these intuitive expectations come for free.
The reason intuitive cognition plays such a powerful role in explaining var-
ious cultural phenomena is that whether or not an idea or practice is intuitive
(in the sense of being the product of maturationally natural systems) is not
anchored to a particular cultural context. Such considerations are not cultur-
ally variable. Intuitive cognition is regarded as part of basic human nature and
thereby can be appealed to for explaining cross-cultural recurrence.
4
At least, we have no evidence that boundaries on weight or size are specified beyond, per-
haps, the general sense that they should be ‘medium-size’ objects—not microscopic, and not on
the scale of planets or galaxies. In keeping with Boyer’s and others’ previous research in this area,
I reserve the term counterintuitive for ideas that violate maturationally natural expectations. Nev-
ertheless, just as one may acquire practiced naturalness, ideas may run counter to practiced
natural expectations. I term ideas that violate these practiced natural expectations countersche-
matic.
J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338 313
just a minor counterintuitive tweak are better, and those with too many coun-
terintuitive features are disastrous. Consider a statue that cries tears made of
pine sap whenever it reads someone’s mind and finds that the person intends
to do something that would please the statute only if the act happened exactly
ten years from next Friday at noon. Such a statue concept (if indeed it can be
called a concept) so greatly violates intuitive conceptual structure that it would
be hard to understand and communicate effectively without distortion. Fur-
thermore, its counterintuitive structure renders it cumbersome for generating
predictions, explanations, or inferences. A large, slightly-oxidized bronze
statue of a muscular bearded man with one hand raised and the other holding
a book that cries real tears does not present the same conceptual challenges for
its communication or for its ability to generate predictions, explanations, or
inferences.
To summarize, largely (but not entirely) intuitive concepts are “good” for
transmission, and massively counterintuitive ones are “bad” (but not impos-
sible). But this formulation raises a problem. How does one quantify counter-
intuitiveness? If Boyer has proposed an optimum, how many counterintuitive
properties must it incorporate? Can we be more precise than somewhere
between wholly intuitive and ridiculous? Below I attempt to clarify matters.
Before sketching what I take to be a means of more precisely specifying the
counterintuitive character of a given concept or idea, let me make an impor-
tant qualification. Boyer and others’ sense of what is and is not intuitive is
dependent on ongoing empirical and theoretical research in developmental
and cross-cultural psychology. What falls within or outside the body of intui-
tive, maturationally natural, pan-cultural knowledge continues to prompt
disagreement. For my analysis below, I appeal to what I take to be the likely
range of intuitive knowledge based on current relevant evidence, but I am
certain that as the relevant science progresses, this range will require revision.
What counts as intuitive knowledge is an empirical issue, however confident
we may be about most of what counts. Because the scope of what properly
counts as intuitive is a work in progress, gray cases in any classification scheme
will appear. One area of possible modification may be that fewer or more areas
of cognitive naturalness (in McCauley’s sense) may prove to be the product of
practiced naturalness rather than maturational naturalness.
I also restrict my discussion to objects (broadly construed), thereby avoid-
ing whether events (e.g., a collision, a sunrise, a competition, an examination),
substances (e.g., oxygen, water, metal), or abstractions (e.g., goodness, law,
brightness, a poem, an idea) might properly be considered intuitive or coun-
terintuitive in the technical sense that Boyer has developed. Indeed, we have
little reason to believe that abstract concepts activate pan-human cognitive
systems in the same way as objects or intentional agents, given that they do
314 J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338
not constitute a domain for causal cognition. Not surprisingly, then, Boyer’s
examples of counterintuitive concepts avoid abstractions or events and instead
focus on ‘persons’ (meaning humans); ‘artefacts’ or ‘tools’; and ‘plants’, ‘ani-
mals’, or ‘living things’ (Boyer 1994; 2001; Boyer et al. 2001: 535-64). Without
more relevant evidence, describing abstractions in terms of counterintuitiveness
would be inappropriate.
I propose six ‘steps’ for coding a concept, and determining the degree of
‘counterintuitiveness’ from that coding. This degree of counterintuitiveness
may be used to make a number of empirical predictions described near the
end of this essay.
I leave aside another sense in which something might be considered “coun-
terintuitive.” If an entity brings about counterintuitive events, it is “counter-
intuitive” in its causability but does not have any ontologically counterintuitive
properties. Consider a relic considered that heals those that touch it in compari-
son to a relic that is invisible. The first can be used to bring about an event that
may be counterintuitive (e.g. instantaneous restoration of sight), whereas the
second has a property that is counterintuitive in Boyer’s sense. It may be that
these two senses of being “counterintuitive” have different representational
properties and different transmissive potential. Because Boyer’s theory only
directly addresses the second sort (the invisible relic), I leave aside those objects
that are counterintuitive by virtue of causability and focus instead on ontologi-
cally counterintuitive entities for the present discussion and coding scheme.
5
It is worth noting that MCI concepts in folktales, films, and religions are not typically
understood as metaphors.
316 J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338
The Simplicity Principle also motivates a rule for the proposed coding
scheme that operates throughout the steps. Let’s call this rule the Simplicity
Rule. The Simplicity Rule states that when coding concepts, assume the simplest
(i.e., least counterintuitive) conceptual representation that captures the object’s
properties. Assume that the proper interpretation of an utterance or verbal
representation is the one that requires less suspension of intuitive expectations
and is, therefore, less effortful (Sperber and Wilson 1995).
The contexts I have in mind for coding public representations for counter-
intuitiveness are single communicative episodes such as a folktale, a conversa-
tion, a ritual, or on-line, real-time interactions with ideas or items. The unit of
analysis I am attempting to capture is the structure of a single individual’s
private representation of any given public representation. What is not intended
is, for instance, a tabulation of all of the different ways in which people gener-
ally use the term ‘God’ and what could be meant by it, as if the universal
semantic scope of the term is indicative of any given individual’s private rep-
resentation in any real-time episode. Coding how the Ghost of Christmas
Present in Dickens’ A Christmas Carol would likely be privately represented by
the audience members is the sort of problem imagined here, not coding for
how people use the term ‘ghost’ in the English speaking world generally. For
the sake of brevity, my examples below do not offer contexts that might help
to clarify (or muddy) the coding of likely private representations.
With these preliminary considerations in mind, please consider the six steps
for coding counterintuitives.
The basic level category is the level of object categorization that minimizes
differences within members of the category while maximizing differences
between categories (Rosch et al. 1976: 573-605). The basic level is typically
the category membership first learned for any given object in the course of
development and usually the first category to be linguistically marked.
Fortunately, an easy and reliable heuristic exists for identifying something’s
basic level. Simply answer the question, “In one word, what is it called?” The
one-word, common, first-learned label for an object is almost always its basic
level (in English). Examples include “apple,” “cat,” “chair,” “shoe,” “cup,”
“rock,” and “shadow.” More precisely, the shortest, one word, common label
for an object usually picks out its basic level category membership. Hence, in
identifying basic level category membership, a golden retriever is classed as a
dog, not as a golden retriever (a subordinate category) or as a mammal, an
animal, or a living thing (superordinate categories). A recliner is classed as a
J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338 317
For this discussion, objects may be thought of as falling in one or more of five
ontological categories: Spatial Entities, Solid Objects, Living Things that do not
appear to be self-propelled, Animates, and Persons. These are intuitive onto-
logical categories and do not necessarily map onto genuine ontological dis-
tinctions.
These five ontological categories arise from differential activation of five dif-
ferent intuitive expectation sets I am calling Spatiality, Physicality, Biology, Ani-
macy, and Mentality. These expectation sets are not ontological categories but,
as specified below, different combinations of their activation characterize intu-
itive ontological categories. Table One (see Appendices) summarizes these
intuitive expectation sets and their characteristic implicit assumptions.
By Spatiality, I mean those intuitions that govern objects, substances, and
other entities that might be located in space. I am hazarding that the primary
assumptions here are that all objects (and substances) have a specifiable loca-
tion in space and time. Further, Spatiality is distinguishable as a specific
domain of reasoning from object Physicality (described below), because some
things such as clouds and shadows (and perhaps minds, as discussed below)
are not intuitively regarded as having all the same properties as bounded phys-
ical objects but yet have spatial properties. A cloud is in one place. If you
divide a cloud you get two clouds, each with their own location. Spatiality
assumptions apply to clouds, other Spatial Entities, and all five categories of
object under discussion here. But clouds do not meet the intuitive assump-
tions I refer to as Physicality.
Physicality includes at least the intuitive expectations of cohesion, solidity,
continuity, and contact (Spelke 1990: 49-56; Spelke et al. 2007: 89-96). Cohe-
sion refers to objects moving as connected wholes. Solidity entails that solid
objects do not readily pass directly through each other or occupy the same
space and at the same time as each other. That objects have continuity means
6
It may be that the basic kind is a class you suspect is counterintuitive such as ‘ghost’ or ‘god.’
If so, an ontologically similar basic kind label may be selected or the step skipped and returned
to after further consideration using the subsequent steps.
318 J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338
that for them to move from point A to point B, they must traverse the inter-
mediate space (as opposed to teleporting). Finally, objects move based on con-
tact. For an object at rest to begin moving, it must be launched through
contact. For it to change course or stop abruptly, contact is necessary. This set
of intuitive expectations are sometimes called folk or naïve physics and appear
to be part of a human’s reasoning during infancy (Spelke 1990: 49-56; Spelke
et al. 2007: 89-96). We can add to these properties two perceptible features
presumed by these studies in early infancy: being visible and tangible.
Biology (more typically called folk biology) includes the intuitive expecta-
tions of growth and development (Rosengren et al 1991: 1302-20; Hatano et
al. 1994: 171-88), internal parts that sustain life or ‘vital force’ (Inagaki et al.
2002; Inagaki et al. 2006: 177-81; Keil 1992: 103-38), vulnerability to death
(meaning cessation of biological processes) (Slaughter 2005:179-86), and
reproduction of like kinds (Hatano et al. 1994: 171-88; Springer et al. 1989:
637-84). Further, Biology assumes composition of natural substances and not
manufactured materials (Simons et al. 1995: 129-63). All of these properties
arise out of a single, unseen, kind-specific “essence” that generates and accounts
for the physically observable features (Keil 1989).
Animacy adds to Biology the expectation that a thing is “self-propelled”
(Premack 1990: 1-16) or has “Force” (Leslie 1995: 121-149). That is, it can
act in or on its environment and not merely be acted upon. Being able to
move oneself from one location to another in space is “self-propelledness,” but
the concept also captures instances of animacy such as changing appearance
(e.g., by size or shape change or illumination), or by making noise (e.g., a cry
or utterance). Further, in animacy, this self-propelledness is perceived as con-
stituting goal-directed action; for instance, to communicate, avoiding some-
thing, threaten, or move to a particular location (not just moving aimlessly).
Mentality (often called folk psychology or Theory of Mind ) extends Animacy
to include assumptions that a thing’s activity is guided and shaped by percepts,
beliefs, desires, emotions, and perhaps personality. For instance, a mental
being will act to satisfy desires, satisfied desires result in positive emotions,
percepts inform beliefs about desired objects, and so forth.
Though some might disagree and wish to add to these five expectation sets,
collapse them (e.g., combining Animacy with Mentality, or Spatiality with
Physicality), or reduce their number (e.g., omit Biology), I suggest these five
because of their general support in the cognitive developmental literature,
their apparent independence in developmental course, and their general map-
ping onto what appear to be reasonable proper domains of the natural world
throughout human existence. Clear evidence for Physicality in humans exists
within the first few months of life and appears robust in adult non-human
primates. Hence, it has been dubbed part of ‘core knowledge’ (Spelke &
J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338 319
Universals and Spatiality but no other expectation set, its intuitive ontological
category is Spatial Entities. If a thing activates Universals, Spatiality, and Phys-
icality but no other expectation sets, its intuitive ontological category is Solid
Objects. The complete set of relationships among expectation set activation
and ontological category membership is represented in Table Two (see Appen-
dices). As intuitive, usually implicit categorizations, they do not necessarily
map onto scientific or philosophically defensible divides, or even how ordi-
nary people would categorize the same objects across all contexts. People treat
clouds and shadows as Spatial Entities—entities with specifiable locations and
even, perhaps, identifiable boundaries, but not Solidity. Solid Objects includes
both artefacts (e.g., chairs, shoes, pencils) and natural non-living objects (e.g.,
stones, icebergs). People classify trees, mushrooms, sea sponges, and many
other (apparently) non-moving biological kinds into Living Things. The cate-
gory Animates may typically include animals that appear to propel themselves
such as sharks, earthworms, crickets, and bunnies, but in some situations may
include non-animals such as complex machines. Persons captures human
beings, and perhaps (depending upon how they are conceptualized) some ani-
mals such as chimpanzees and the family dog. (Exactly which animals are
intuitively granted full-blown minds, albeit slightly different than humans’, is
unknown and likely quite variable.) Note that contrary to how I understand
Boyer’s use of the term ‘person’ (Boyer 1994; Boyer 2001) here ‘Person’ is not
synonymous with ‘human.’
Spatiality, Physicality, Biology, Animacy, and Mentality do not exhaust the
different sets of intuitive expectations applied to things in the world. Some
evidence exists that artefacts may be discriminated from other Solid Objects
by virtue of their activation of a set of functional expectations (Casler et al.
2007: 120-30; Keil 1995: 234-67). As this is an area of rapid development and
does not change my general analysis, I leave artefacts and natural non-living
objects together in one group. Further, I assume all entities are intuitively
subject to the set of intuitive expectations I call Universals.
As displayed in Table One, the relationship between expectation sets and
ontological categories is not strictly hierarchical. By virtue of activating the
same expectation sets, Solid Objects assume and extend the properties of Spa-
tial Entities, and Living Things also assume and extend the properties of Solid
Objects and Spatial Entities. But evidence suggests that things which are intu-
itively categorized as Animates do not necessarily activate the same expecta-
tions as Living Things. For instance, complex machines including computers
may be conceptualized using the kind of goal-directed or teleological agency
reasoning without assuming Biology. Indeed, this type of reasoning about
Animates arises during infancy and is evident in other species (Gergely et al.
2003: 287-92), presumably before folk biological reasoning emerges, suggest-
J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338 321
ing its independence from the Biology expectation set that marks Living
Things off from Solid Objects. Though the vast majority of Animates that we
represent (i.e., animals) do activate Biology, the differences in developmental
and probably evolutionary courses suggests the relative independence of the
Animacy and Biology expectation sets.
The observation that Persons and Animates (as opposed to humans) do not
automatically assume the biological and physical properties of animals is a
departure from previous discussions of counterintuitives (Boyer 1994). Nev-
ertheless, I find support for this departure in the recent research on the devel-
opment of Theory of Mind and teleological agency reasoning in children. We
now have a growing body of evidence that from infancy children distinguish
the causal properties relevant to agents and minds and those relevant to phys-
ical objects (Bloom 2004; Gergely et al. 2003: 287-92). For instance, Ani-
mates are presumed to be free from the contact principle that requires physical
objects to be contacted in order to begin moving (Spelke et al. 1995). Infants
expect objects that propel themselves in an apparently purposeful or goal-
directed fashion to continue to behave in a purposeful fashion even if violating
basic expectations of mechanical movement (Gergely et al. 2003: 287-92;
Gergely et al. 1995: 165-93; Leslie 1995: 121-149; Premack 1990: 1-16).
Importantly, these expectations may be triggered from infancy with stimuli
that do not resemble three-dimensional objects (but two-dimensional spots)
let alone human beings (Rochat et al. 1997: 537-61; Scholl et al. 2000: 299-
308). These findings suggest that reasoning about Animacy and Mentality
does not presume later-developing Biology—as spots are unlikely to be
construed as Living Things—or perhaps even standard object Physicality.
Suggestively Kuhlmeier, Bloom, and Wynn (2004: 95-103) have offered pre-
liminary experimental evidence suggesting that infants that demonstrate
awareness of the continuity principle for Solid Objects (part of the Physicality
expectation set) do not demonstrate awareness that human beings must move
according to the continuity principle. Together with research demonstrating
how readily preschool-aged children reason with disembodied imaginary
friends (Taylor 1999), these data hint that Mentality and perhaps even Ani-
macy expectations may operate without a solid, bounded physical object as
the target. For young children, minds do not need physical bodies, a sugges-
tion captured by the term ‘intuitive dualism’ (Bloom 2004). Though I find the
available data suggestive, I concede that what I am offering amounts to a
speculation in need of further, direct empirical support.
The proposed expectation sets are not a class-inclusion hierarchy, but
the sets are not wholly independent either. Some expectation sets automati-
cally activate others; to have one active in a conceptual structure is to have
another present. If a concept activates Biology, it likewise activates Physicality,
322 J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338
Once basic kind and ontological category have been determined, it can be
determined whether a concept, as used in the communicative context under
examination, includes properties from a ‘non-native’ set of intuitive expecta-
tions. As a Solid Object, a rock should (intuitively) only include expectations
generated by Spatiality and Object Physics. If the rock is attributed the ability
to reproduce, then folk Biology properties have been transferred. It would be
represented thus, BROCK with the superscript ‘B’ representing Biology. Simi-
larly, a tree that can verbally communicate has had Mentality transferred. It
would be coded, MTREE. A rock that eats passersby and talks would have both
Animacy and Mentality transferred. To represent both of these transferred
properties, + is used: A+MROCK.
Following the Simplicity Rule (that the least counterintuitive or “simplest”
representation should be assumed), a TREE that both listens empathetically
(a property from Mentality) and verbally communicates (another Mentality
property) would still be coded MTREE. The assumption is that the entire set
of Mentality expectations have been transferred, not just the one named men-
tal ability. What would it mean for a tree to be able to talk to you but not
think, remember, perceive, and so forth? Unless the concept explicitly
renounces other properties from the same set of expectations, we assume the
entire set has been transferred.7
7
Lindeman & Aarino (2006: 585-602) identify transferring intuitive expectations from one
causal domain to another as the hallmark of “paranormal” thought.
J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338 323
8
Steps Three and Four may be reversed as there is no necessary priority of transfers over
breaches. I thank Joe Henrich for helpful suggestions regarding the formalization of the coding
scheme.
9
I suggest capital letters to represent transfers but lowercase for breaches for two reasons.
First, such a strategy redundantly discriminates between transfers and breaches. Second, it also
symbolizes that, in the case of transfers, the whole expectation set has been transferred, but for
breaches only a single expectation from the designated set has been violated.
324 J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338
Unlike transfers, a breach does not necessarily presume that the entire intu-
itive expectation set has been violated. When hearing about an invisible rock,
people need not presume that it must be intangible as well. Consequently, an
invisible rock that can pass through other solid objects could be represented
with two Physicality breaches but presume all other Physicality expectations
such as being tangible, requiring contact to move, requiring support, and
moving as a continuous unit. The invisible rock that can pass through other
solid objects would be coded as ROCKp+p . Note, however, that with increas-
ing breaches of an expectation set, the Simplicity Principle is increasingly
likely to find a more parsimonious representation by assuming the entire set is
violated and hence, revise the ontological category membership (Step Two). A
brick that is intangible, invisible, and need not move as a coherent whole
begins to look like something other than a Solid Object, and so it seems
unlikely to really be anything we would normally regard as a brick. With that
many Physicality violations, I am unsure that any Physicality expectations
apply and so rather than code it as a rock with three breaches (ROCKp+p+p), I
may intuitively represent it as a substance or cloud that happens to have a
brick-like shape with no breaches or transfers (VAPOR) or as a vapor that has
rock-like boundaries, a transferred property from Physicality (PVAPOR).
As with transfers, multiple breaches from separate intuitive expectation sets
would be joined with +. For instance, a rock that is invisible (Physicality
breach) and ceases to exist on Wednesdays (Universals breach) would be coded,
ROCKp+u .
Combining Steps Three and Four, we have a method for coding even more
complex concepts that include both breaches and transfers. Consider an invis-
ible chair that eats underpants ACHAIRp . Or consider a singing rosebush that
experiences time backwards MBUSHu.
It is possible to conceptualize a transferred property that itself has been
breached. For instance, ‘a chair that swallows people only when it is not hun-
gry,’ would likely be understood as an inanimate Solid Object that has been
transferred a Biology property (nutritional needs and activity to satisfy those
needs) that have been violated (by acting contrary to the Biology assumption).
In cases such as this, I suggest using the same coding notation as with any
other breach/transfer combination. For instance, with BCHAIRb, the prefix ‘B’
signifies the transfer of Biology to the chair, and the suffix, lower-case ‘b’ signi-
fies a breach of that transferred Biology. Another example might be ‘a statue
that can hear your thoughts,’ coded as MSTATUEm. The statue has been trans-
ferred Mentality that has been breached (thoughts are not within the scope of
hearing). If necessary, use ‘+’ to represent multiple breaches of the same trans-
ferred expectation set. A ‘statue that can hear your thoughts and has x-ray
vision’ could be symbolized as: MSTATUEm+m.
J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338 325
The final step in this coding process is to simply count up the superscript let-
ters employed. A concept that can be represented with nothing more than the
basic level, all-caps word, scores zero on counterintuitiveness. Each breach or
transfer adds one point. Table Three (see Appendices) gives examples.
How might commonly-held folk concepts be treated under the proposed sys-
tem? Allow me two examples: ghosts (whether ancestors or spirits), and God.
To make these illustrative codings I am relying on my own folk knowledge of
these concepts in English-speaking North America to speculate about how a
representative individual might privately represent these public representa-
tions in a single representative episode. I do not regard my knowledge on these
matters as authoritative and acknowledge the potential for great individual
variation. I only offer these analyses for the sake of illustration and to raise
empirical questions. Recall, too, that in actual instances of coding, particular
communicative episodes should be considered such as a story about a ghost
and not the folk concept ‘ghost,’ or what a person says about God in conversa-
tion and neither the entire corpus of God-related folklore or theology, nor the
entire semantic scope of ‘God.’
First, let’s deal with ghosts. A complaint frequently raised against the MCI
theory as originally formulated has been that some very successful cultural
326 J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338
10
McCauley & Lawson previously made a similar point about the need for special cultural
scaffolding to support religious rituals that too greatly deviate from intuitive cognitive con-
straints (2002).
J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338 327
are that expectation sets are non-independent but also non-hierarchical, and
the Simplicity Principle. For a spirit/ghost concept these observations mean
that ghosts might not actually be intuitively represented as humans with mul-
tiple Physicality breaches and additional Biology breaches. Rather, because
Biology presumes natural intuitive Physicality, multiple Physicality breaches
automatically deactivate Biology expectations. Consequently, Biology expec-
tation breaches need not be represented in the conceptual structure. Perhaps
then ghosts are represented as, HUMANp+p+p instead of HUMANp+p+p+b.
Because Biology requires Physicality and Physicality has been breached, Biol-
ogy is not presumed relevant. The counterintuitiveness score of the ghost con-
cept drops from 4 or more to 3.
Further still, the Simplicity Principle suggests that when several breaches
from a single expectation set are violated, the human conceptual tendency
toward less complex representations will find a different representation that
captures the same properties but with fewer violations of intuitive expecta-
tions. It could be, then, that the three proposed breaches of Physicality (invis-
ibility, intangibility, and being able to pass through solid objects in violation
of Solidity) disqualify the ghost as a Solid Object outright. Perhaps ghosts and
spirits are represented as Spatial Entities with transferred Mentality. Indeed,
the idea of a ghost or spirit being the formless, shapeless agency of a human
that is recurrent in ethnography and comparative religion, suggests that spa-
tial-entity-with-mentality is true to how people represent ghosts and spirits.
For lack of a better basic level label, I use SUBSTANCE for the non-specified
Spatial Entities. In the coding scheme presented, a ghost might be, MSUB-
STANCE. Because Mentality presumes Animacy (to purposefully act in or on
the environment), coding the transfer of Animacy is unnecessary. The coun-
terintuitiveness score for ghost/spirit is then 1; the epitome of a minimally
counterintuitive concept. In different cultural contexts, however, especially
where ghosts motivate rituals and other collective behaviours because they are
part of the local religious system, ghosts may have additional counterintuitive
features joined to them. The prediction is that concepts with a counterintui-
tiveness score of more than 2 would likely require additional cultural scaffold-
ing (e.g., recurrent formalized teaching on the matter, development of
theology, rituals, etc.) to successfully transmit the concept. Even still, it may
be that in a single on-line episode or communicative act (e.g., a single story),
only one or two counterintuitive features would be represented (Barrett 1999:
325-339).
If, as suggested in the discussion of “intuitive dualism” above, Animacy
and Mentality do not presume object Physicality but only require a location
that can change or a spatial form that can otherwise act, a ghost is not a
Solid Object that has breaches of Physicality but, instead, a bodiless Person.
328 J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338
One modification to the MCI theory from its earlier form (Boyer 1994)
has been the addition of Inferential Potential (Boyer 2001). Inferential Poten-
tial refers to the ability for a concept or idea to readily generate explanations,
330 J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338
Predictions
One aim of this paper was to clarify the predictions of a MCI theory defensi-
ble in terms of current thought regarding early-developing pan-human con-
ceptual systems. Using the proposed coding scheme and re-analysis of the
MCI theory, I can make a number of preliminary predictions. Only the first is
J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338 331
critical for the general utility of the theory. If the first can be demonstrated to
be false, the MCI theory must be amended or rejected. I offer the subsequent
predictions to illustrate ways in which the theory could be further developed
and made useful.
I hope it is clear that the version of the MCI theory presented here does
not apply to concepts falling outside of the intuitive ontological categories
presented.
(1) First, following Boyer, I predict that MCI concepts, that is, ideas with
a score of 1 for counterintuitiveness, are more readily remembered and com-
municated faithfully than other concepts (all else being equal). For example, a
book that thinks would be better remembered and communicated than a
book about thought or a book that thinks and has babies.
(2) More than two breaches from the same expectation set will tend to get
re-represented in a simpler (i.e., less counterintuitive) form.
(3) A single transferred property from an expectation set will tend to
prompt the assumption that the entire set of properties is transferred.
(4) A concept with a score higher than three will be unlikely to possess on-
line inferential potential and will be discarded in favour of a simpler form. A
cut-off score of three is derived from an analysis of the counterintuitiveness of
folktales from Africa, the Indian Subcontinent, and Native America (Barrett
1997), and Lisdorf ’s analysis (2001) of Roman prodigies from the first and
second centuries B.C.E. Both projects found essentially no counterintuitive
concepts with more than two counterintuitive features (at least in a single nar-
rative context). For the same reason, I offer a related prediction that on-line
inferential potential peaks with a counterintuitiveness score of 1 or 2 and then
drops with increasing counterintuitiveness.
Boyer’s theory and my re-formulation make no distinction in terms of pro-
cessing burden or mnemonic advantage for ideas that are counterintuitive by
virtue of a breach versus counterintuitive by virtue of a transfer. Nevertheless,
such differences might be subjected to empirical investigation.
Conclusion
In sum, for the MCI theory to be fruitful in the study of cognition and culture
and to be rigorously tested and either modified or rejected based on that
testing, more precision and uniformity is needed regarding the scope and pre-
dictions of the theory. I hope that the presented sketch of the theoretical
grounds for the theory and the introduction of a formal coding and quantifi-
cation scheme contribute to this need for precision and uniformity. Using the
332 J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338
References
Barrett, Justin L. (1997). Anthropomorphism, intentional agents, and conceptualizing God.
Unpublished Ph.D. dissertation, Cornell University, Ithaca, New York.
——— (1998). Cognitive constraints on Hindu concepts of the divine. Journal for the Scientific
Study of Religion 37(4): 608-619.
——— (1999). Theological correctness: Cognitive constraint and the study of religion. Method
& Theory in the Study of Religion 11(4): 325-339.
——— (2000). Exploring the natural foundations of religion. Trends in Cognitive Sciences 4(1):
29-34.
——— (2004a). Counterfactuality in counterintuitive religious concepts. Brain & Behavioral
Sciences 27(6): 731-732.
——— (2004b). Why Would Anyone Believe in God? Walnut Creek, CA: AltaMira Press.
——— (in press). Why Santa Claus is not a god. Journal of Cognition and Culture.
Barrett, Justin L. and Frank. C. Keil (1996). Anthropomorphism and God concepts: Conceptu-
alizing a non-natural entity. Cognitive Psychology 31(3): 219-47.
Barrett, Justin L., Roxanne Moore Newman & Rebekah A. Richert (2003). When seeing does
not lead to believing: Children’s understanding of the importance of background knowl-
edge for interpreting visual displays. Journal of Cognition & Culture 3(1): 91-108.
Barrett, Justin L. & Melanie A. Nyhof (2001). Spreading non-natural concepts: The role of
intuitive conceptual structures in memory and transmission of cultural materials. Journal of
Cognition & Culture 1(1): 69-100.
Barrett, Justin L. & Brant VanOrman (1996). The effects of image use in worship on God con-
cepts. Journal of Psychology and Christianity 15(1): 38-45.
Bass, Jules & Arthur Rankin (directors) (1970). Santa Claus is Comin’ to Town. Rankin/Bass
Productions.
Bering, Jesse M. (2002). Intuitive conceptions of dead agents’ minds: The natural foundations of
afterlife beliefs as phenomenological boundary. Journal of Cognition & Culture 2(4): 263-308.
Bloom, Pascal (2004). Descartes’ Baby: How the Science of Child Development Explains What
Makes Us Human. New York: Basic Books.
Boyer, Pascal (1994). The Naturalness of Religious Ideas: A Cognitive Theory of Religion. Berkeley:
University of California Press.
——— (2001). Religion Explained: The Evolutionary Origins of Religious Thought. New York:
Basic Books.
——— (2003). Religious thought and behaviour as by-products of brain function. Trends in
Cognitive Sciences 7(3):119-24.
Boyer, Pascal & Charles Ramble (2001). Cognitive templates for religious concepts: Cross-
cultural evidence for recall of counter-intuitive representations. Cognitive Science 25(4):
535-64.
Casler, Krista & Deborah Kelemen (2007). Reasoning about artifacts at 24 months: The devel-
oping teleo-functional stance. Cognition 103(1): 120-30.
J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338 333
Cohen, Emma (2007). The Mind Possessed: The Cognition of Spirit Possession in an Afro-Brazilian
Religious Tradition. New York: Oxford University Press.
Gergely, Gyorgy, & Gergely Csibra (2003). Teleological reasoning in infancy: the naïve theory of
rational action. Trends in Cognitive Sciences 7(7): 287-92.
Gergely, Gyorgy, Zoltan Nadasdy, Gergely Csibra & Szilvia Biro (1995). Taking the intentional
stance at 12 months of age. Cognition 56(2): 165-93.
Gonce, Lauren O., Upal M. Afzal, D. Jason Slone & Ryan D. Tweney (2006). Role of Context
in the Recall of Counterintuitive Concepts. Journal of Cognition & Culture 6(3-4):
521-47.
Hatano, Giyoo and Kayako Inagaki (1994). Young children’s naïve theory of biology. Cognition
50(1-3): 171-88.
Hirschfeld, Lawrence A. (1996). Race in the Making: Cognition, Culture and the Child’s Construc-
tion of Human Kinds. Cambridge: MIT Press.
Inagaki, Kayoko & Giyoo Hatano (2006). Young children’s conception of the biological world.
Current Directions in Psychological Science 15(4): 177-81.
——— (2002). Young Children’s Naïve Thinking about the Biological World. New York: Psychol-
ogy Press.
Keil, Frank C. (1979). Semantic and Conceptual Development: An Ontological Perspective. Cam-
bridge: Harvard University Press.
——— (1989). Concepts, Kinds, and Cognitive Development. Cambridge, MA: MIT Press.
——— (1992). The origins of an autonomous biology. In M. R. Gunnar and M. Maratsos
(Eds.) Minnesota Symposium on Child Psychology, 103-38. Hillsdale, NJ: Earlbaum.
——— (1995). The growth of causal understandings of natural kinds. In D. Sperber,
D. Premack, & A. J. Premack (Eds.), Causal Cognition: A Multidisciplinary Debate, 234-67.
Oxford: Clarendon Press.
Kelly, Michael and Frank C. Keil (1985). The more things change . . . : Metaphorphoses and
conceptual structure. Cognitive Science 9(4): 403-16.
Kuhlmeier, Valerie A., Paul Bloom & Karen Wynn (2004). Do 5-month-old infants see humans
as material objects? Cognition 94(1): 95-103.
Leslie, A. (1995). A theory of agency. In D. Sperber, D. Premack, and A. J. Premack (Eds.),
Causal cognition: A Multidisciplinary Debate 121-149. Oxford: Clarendon Press.
Lindeman, Marjaana & Kia Aarnio (2006). Paranormal Beliefs: Their Dimensionality and Cor-
relates. European Journal of Personality 20(7): 585-602.
Lisdorf, Anders. (2001). The spread of non-natural concepts. Journal of Cognition and Culture
1(4): 151-74.
McCauley, Robert N. (forthcoming). The naturalness of religion and the unnaturalness of science.
McCauley, Robert N. and E. Thomas Lawson (2002). Bringing Ritual to Mind: Psychological Foun-
dations of Cultural Forms. Cambridge: Cambridge University Press.
Norenzayan, Ara, Scott Atran, Jason Faulkner & Mark Schaller (2006). Memory and Mystery:
The Cultural Selection of Minimally Counterintuitive Narratives. Cognitive Science 30(3):
531-53.
Premack, David (1990). The infant’s theory of self-propelled objects. Cognition 36(1): 1-16.
Pyysiäinen, Ilkka, Marjaana Lindeman & Timo Honkela (2003). Counterintuitiveness as the
hallmark of religiosity. Religion 33(4): 341-55.
Rochat, Philippe, Morgan Rachel & Malinda Carpenter (1997). Young infants’ sensitivity to
movement information specifying social causality. Cognitive Development 12(4): 537-61.
Rosch, Eleanor, Carolyn Mervis, Wayne D. Gray, David Johnson & Penny Boyes-Braem (1976).
Basic objects in natural categories. Cognitive Psychology 7(4): 573-605.
Rosengren, Karl S., Susan A. Gelman, Charles W. Kalish, & Michael McCormick (1991). As time
goes by: Children’s early understanding of growth. Child Development 62(6): 1302-20.
334 J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338
Rubin, David C. (1995). Memory in Oral Traditions: The Cognitive Psychology of Epic, Ballads,
and Counting-out Rhymes. New York: Oxford University Press.
Rumelhart, David E. (1977). Understanding and summarizing brief stories. In D. LaBerge &
S. J. Samuels (Eds.), Basic Processes in Reading: Perception and Comprehension. Hillsdale, NJ:
Lawrence Erlbaum Associates.
Scholl, Brian J. & Patrice D. Tremoulet (2000). Perceptual causality and animacy. Trends in
Cognitive Sciences 4(8): 299-308.
Schmidt, Stephen R. (1991). Can we have a distinctive theory of memory? Memory & Cognition
19(6): 523-42.
Simons, Daniel J. & Frank C. Keil (1995). As abstract to concrete shift in development of bio-
logical thought: The insides story. Cognition 56(2): 129-63.
Slaughter, Virginia (2005). Young children’s understanding of death. Australian Psychologist 40
(4):179-86.
Spelke, Elizabeth S. (1990). Principles of object perception. Cognitive Science 14(1): 49-56.
Spelke, Elizabeth S. & Katherine D. Kinzler (2007). Core knowledge. Developmental Science
10(1): 89-96.
Spelke, Elizabeth S., Ann Phillips & Amanda L. Woodward (1995). Infant’s knowledge of object
motion and human action. In D. Sperber, D. Premack, & A. J. Premack (Eds.), Causal
Cognition: A Multidisciplinary Debate, 44-78. New York: Oxford University Press.
Sperber, Dan (1996). Explaining Culture: A Naturalistic Approach. Oxford: Blackwell.
Sperber, Dan and Lawrence Hirschfeld (2004). The cognitive foundations of cultural stability
and diversity. Trends in Cognitive Sciences 8(1): 40-46.
Sperber, Dan & Deidre Wilson (1995). Relevance: Communication and Cognition. Oxford,
Blackwell.
Springer, Ken & Frank Keil (1989). On the development of biologically specific beliefs: The case
of inheritance. Child Development 60(3): 637-84.
Taylor, Marjorie (1999). Imaginary Companions and the Children Who Create Them. New York:
Oxford University Press.
Thorndyke, Perry W. (1977). Cognitive structures in comprehension and memory of narrative
discourse. Cognitive Psychology 9(1): 77-110.
Tremlin, E. Thomas (2006). Minds and Gods: The Cognitive Foundations of Religion. New York:
Oxford University Press.
Tweney, Ryan D., M. Afzal Upal, Lauren O. Gonce, D. Jason Slone & Katie Edwards (2006).
The creative structuring of counterintuitive worlds. Journal of Cognition & Culture 6(3-4):
483-98.
Upal, M. Afzal, Lauren Owsianiecki, D. Jason Slone, & Ryan Tweney (2007). Contextualizing
counterintuitiveness: How context affects comprehension and memorability of counterin-
tuitive concepts. Cognitive Science 31(1): 1-25.
Waddill, Paula J. & Michael A. McDaniel (1998). Distinctiveness effects in recall: Differential
processing or privileged retrieval? Memory & Cognition 26(1): 108-20.
Ward, Thomas B. (1994). Structured imagination: The role of category structure in exemplar
generation. Cognitive Psychology 27(1): 1-40.
——— (1995). What’s old about new ideas? In S. M. Smith, T. B. Ward & R. A. Finke (Eds.),
The Creative Cognition Approach, 157-178. Cambridge, MA: MIT Press.
Wellman, Henry, David Cross & Julanne Watson (2001). Meta-analysis of theory of mind devel-
opment: The truth about false-belief. Child Development 72(3): 655-84.
Whitehouse, Harvey (2004). Modes of Religiosity: A Cognitive Theory of Religious Transmission.
Walnut Creek, CA: AltaMira Press.
J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338 335
Appendices
Table One
Intuitive Expectation Set Properties Assumed a
Spatiality Specifiable location in space and time
Physicality Cohesion (move as connected whole)
Contact (physical contact required for launching
or changing direction of movement)
Continuity (movement is continuous in space)
Solidity (cannot pass through or be passed
through by other solid objects)
Tangibility
Visibility
Biology Growth & development
Like begets like
Natural composition
Nourishment needs and processes to satisfy those
needs (if animate, actively seeks to satisfy these
needs)
Parts serve the whole to sustain life
Vulnerability to injury & death (if animate, seeks
to avoid injury & death)
Kind-specific essence
Animacy Goals
“Self-propelled” (including moving in space,
changing appearance, emitting sounds, etc.)
Mentality Reflective & representational mental states (e.g.,
beliefs, desires) and standard relationships
among them and limitations of them (e.g.,
limited perceptual access)
Self-awareness (including emotions and epis-
temic states)
Understand language & communication
Universals Consistency (assumptions apply continuously;
past was like present, future will be like present)
Time (and hence, causation) is unidirectional
a
This summary of intuitive expectation sets and entailed implicit assumptions
may be expanded as more assumptions are discovered.
336 J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338
Table Two
Spatial Solid Living Animates Persons
Entities Objects Things
Expectation Sets:b
(with coding
abbreviation)
Universals (u) YES YES YES YES YES
Spatiality (s) YES YES YES YES YES
Physicality (o) NO YES YES YES YESc
Biology (b) NO NO YES --- ---
Animacy (a) NO NO NO YES YES
Mentality (m) NO NO NO NO YES
b
Table Two depicts expectation sets and their relationship to intuitive ontological
categories. ‘YES’ indicates that an ontological category (listed across the top) intu-
itively assumes a particular expectation set (listed in the left-hand column). ‘NO’
indicates that an ontological category does not allow a particular expectation set.
‘---’ indicates that the ontological category does not intuitively assume or disallow
an expectation set but may intuitively include members that do or do not activate
the expectation set.
c
It may be that minds are intuitively represented as having Spatiality without
Physicality, more akin to an unbounded substance, and are intuitively separable
from physical bodies as suggested by Bloom (2004). If so, a disembodied mind is
a Person and Persons need not automatically activate Physicality. Such a finding
would have implications for the discussion of ghost and God concepts toward the
end of this paper. A disembodied mind in a particular location would not, by
itself, be counterintuitive.
J. L. Barrett / Method and Theory in the Study of Religion 20 (2008) 308-338 337
Table Three
Concept (public representation) Coding Counterintuitiveness
Score
A ferret with four legs and fur FERRET 0
A bright green ferret FERRET 0
An invisible ferret FERRETp 1
A ferret that is invisible weekly (FERRETp)u 2
A ferret that is invisible and immortal FERRETp+b 2
Table Four
Simplicity Rule When coding concepts, assume the simplest (i.e., least
counterintuitive) conceptual representation that captures
the object’s properties.
Step 1 Identify the basic level membership. (Revision of
identification may be required by the Simplicity Rule
after considering Steps 3-6.)
Step 2 Identify the ontological category or categories. Candidate
categories include Persons, Animates, Living Things,
Solid Objects, and Spatial Entities. If the item in
question does not obviously fall into one or more of
these five categories (e.g., abstractions, events) do not
proceed. (Revision of categorization may be required by
the Simplicity Rule after considering Steps 3-6.)
Step 3 Code transfers as superscript prefixes with capital letters,
joined by + if necessary.
Step 4 Code breaches as superscript suffixes with lowercase
letters, joined by + if necessary.
Step 5 Code breaches within breaches with parentheses.
Step 6 Quantify counterintuitiveness by totalling the number of
symbolic letters.