Professional Documents
Culture Documents
Selection by Consequences - Skinner
Selection by Consequences - Skinner
Selection by consequences
B. F. Skinner
Department of Psychology and Social Relations,
Harvard University, Cambridge, Mass. 02138
Abstract: Human behavior is the joint product of (i) contingencies of survival responsible for natural selection, and (ii) contingencies
of reinforcement responsible for the repertoires of individuals, including (iii) the special contingencies maintained by an evolved
social environment. Selection by consequences is a causal mode found only in living things, or in machines made by living things. It
was first recognized in natural selection: Reproduction, a first consequence, led to the evolution of cells, organs, and organisms
reproducing themselves under increasingly diverse conditions. The behavior functioned well, however, only under conditions
similar to those under which it was selected.
Reproduction under a wider range of consequences became possible with the evolution of processes through which organisms
acquired behavior appropriate to novel environments. One of these, operant conditioning, is a second kind of selection by
consequences: New responses could be strengthened by events which followed them. When the selecting consequences are the
same, operant conditioning and natural selection work together redundantly. But because a species which quickly acquires behavior
appropriate to an environment has less need for an innate repertoire, operant conditioning could replace as well as supplement the
natural selection of behavior.
Social behavior is within easy range of natural selection, because other members are one of the most stable features of the
environment of a species. The human species presumably became more social when its vocal musculature came under operant
control. Verbal behavior greatly increased the importance of a third kind of selection by consequences, the evolution of social
environments or cultures. The effect on the group, and not the reinforcing consequences for individual members, is responsible for
the evolution of culture.
Keywords: behaviorism; consequentialism; culture; evolution; law of effect; learning; natural selection; operant conditioning;
reinforcement contingencies; social environment; verbal behavior
The history of human behavior, if we may take it to begin could be strengthened ("reinforced") by events which
with the origin of life on earth, is possibly exceeded in immediately followed them.
scope only by the history of the universe. Like astrono-
mer and cosmologist, the historian proceeds only by
reconstructing what may have happened rather than by A second kind of selection
reviewing recorded facts. The story presumably began,
not with a big bang, but with that extraordinary moment Operant conditioning is a second kind of selection by
when a molecule came into existence which had the consequences. It must have evolved in parallel with two
power to reproduce itself. It was then that selection by other products of the same contingencies of natural selec-
consequences made its appearance as a causal mode. tion - a susceptibility to reinforcement by certain kinds of
Reproduction was itself a first consequence, and it led, consequences and a supply of behavior less specifically
through natural selection, to the evolution of cells, committed to eliciting or releasing stimuli. (Most oper-
organs, and organisms which reproduced themselves ants are selected from behavior which has little or no
under increasingly diverse conditions. relation to such stimuli.)
What we call behavior evolved as a set of functions When the selecting consequences are the same, oper-
furthering the interchange between organism and en- ant conditioning and natural selection work together
vironment. In a fairly stable world it could be as much a redundantly. For example, the behavior of a duckling in
part of the genetic endowment of a species as digestion, following its mother is apparently the product not only of
respiration, or any other biological function. The involve- natural selection (ducklings tend to move in the direction
ment with the environment, however, imposed limita- of large moving objects) but also of an evolved susceptibil-
tions. The behavior functioned well only under condi- ity to reinforcement by proximity to such an object, as
tions fairly similar to those under which it was selected. Peterson (1960) has shown. The common consequence is
Reproduction under a much wider range of conditions that the duckling stays near its mother. (Imprinting is a
became possible with the evolution of two processes different process, close to respondent conditioning.)
through which individual organisms acquired behavior Since a species which quickly acquires behavior appro-
appropriate to novel environments. Through respondent priate to a given environment has less need for an innate
(Pavlovian) conditioning, responses prepared in advance repertoire, operant conditioning could not only supple-
by natural selection could come under the control of new ment the natural selection of behavior, it could replace it.
stimuli. Through operant conditioning, new responses There were advantages favoring such a change. When
that characterize cultural evolutionary processes. The other, successful changes are selected in an anticipatory fashion.
selection of behavioral innovations in individual repertories is Genetic alterations are not decided in advance by environmen-
far more common. In fact, group selection is merely the limiting tal pressures; the strengthening or weakening of behavior is not
case of individual selection in which the consequences are carried out in anticipation of its outcomes; social structures are
catastrophic for all group members. Sociocultural systems usu- not established with the intent of effecting social change. Where
ally change well before catastrophic consequences lead to group there is an apparent anticipation of successful change, this is
extinction. One has merely to cast an eye at the rapid pace of owing to biases (of the form of responding, for example) or to
changes in technology and domestic life to get the point. Auto- "rules" that have themselves been selected in retrospective
mobiles and electric lights were not selected for as a conse- fashion. In fact, much of Skinner's argument was anticipated in a
quence of their contribution to group survival (cf. "contingen- provocative article by Donald T. Campbell (1960), concerning
cies . . . promoting behavior which contributes to the survival variation and selective retention in cognitive and creative
of the group") but because they constituted reinforcements for behavior.
specific individuals whose behavior was thereby shaped. While we recognize the retrospective action of the conse-
When we say that behavior has been selected for as a result of quences of genetic or behavioral or social variation, we must also
its favorable consequences for a group, we can only mean that it note that these changes each incorporate mechanisms for sta-
has had favorable consequences for some or all members of the bility that oppose further change. In the case of evolution, this is
group sufficient to outweigh its adverse effects on some or all of perhaps a trivial point; a morphological change is permanent
the members. The cumulative shaping of individual behavior is until modified by a further, presumably random, successful
precisely what cultural evolution is all about. Of course, these mutation. The change is "stored" genetically. Social change is
behaviors are interrelated and in conjunction with various "codified" and transmitted as law or tradition (Tevye's sons in
environmental and social feedback processes possess systemic Fiddler on the Roof), and the retrospective origins of the change
properties that are the logico-empirical basis for the concepts of are readily forgotten by those who are controlled by social
society, culture, and sociocultural systems. For an an- institutions. But how is behavioral change "fixed" and used by
thropological behaviorist, events on the sociocultural level are individual organisms? Skinner is, as ever, silent on this ques-
necessarily abstractions (concrete and real) derived from the tion. The functional definition of reinforcement as a retro-
observation of behavioral changes in individuals, and the evolu- spective process cannot at the same time explain the action of
tion of sociocultural systems is necessarily the evolution of such reinforcement in fixing behavior.
behavior. Skinner suggests that the reinforceability of behavior is itself a
Thus human behavioral repertories consist overwhelmingly consequence of evolution, because adaptiveness of behavior
of operantly conditioned responses that are at the same time enhances survival value. It is equally reasonable to suppose that
culturally conditioned responses, that is, responses shaped in animals evolved mechanisms for rendering the selected behav-
conformity with culturally determined reinforcement schedules ior resistant to change. The physiological forms of such mecha-
and contingencies. Therefore Skinner's claim that "the con- nisms still need to be identified, but students of animal learning
tingencies of selection at the three levels are quite different" is seem generally to agree that "associations" are formed either
incorrect in the human case as regards levels ii and iii. In the between the behavior and its consequences (instrumental con-
human case the contingencies of selection are not random but ditioning) or between an initial signal and a primary or uncondi-
occur in conformity with programs encoded primarily in the tioned stimulus that elicits a response (classical conditioning).
brains (or other neural pathways as distinguished from the The nature of association as a psychological concept has been the
genes) of enculturated individuals (and not, as Skinner pro- object of intense study in recent years. Much current evidence
poses, merely in "documents, artifacts, and other products suggests anticipatory mechanisms both in instrumental learning
o f . . . behavior"). and in classical conditioning. For example Breland and Breland
Behaviorist principles can tell us how these individuals shape (1961) showed in their study of the "misbehavior" of organisms
each other's behavior, but they cannot tell us what behavior that the instrumental response would "drift" toward the con-
they will shape. Skinner's criterion for separating levels ii and iii summatory response required by the reinforcer that follows the
in the human case obscures this problem and deters fruitful response. Likewise, in autoshaping (a classical conditioning
collaboration between materialist, behaviorist, nomothetic an- procedure) the responses elicited by the conditioned stimulus
thropologists and like-minded psychologists. take a form that is appropriate for consumption of the reinforcer
(Jenkins & Moore 1973). In fact, classical conditioning is gener-
Practitioners of the science of culture need to know more from ally viewed as the development of an anticipatory process.
psychologists than the general laws of operant behavior. In Skinner has never claimed that the change of behavior that is
order to predict or retrodict favored or unfavored innovations in part of classical conditioning results from selection by reinforce-
cultural repertories and hence to understand the divergent and ment following the conditioned response.
convergent (not merely unilinear) trajectories of sociocultural
evolution, we need to be able to measure cost-benefit conse- Many other findings suggest the importance of anticipatory
quences as "currencies" relevant to the biologically determined mechanisms in learning. The "blocking" of conditioning when a
discriminative stimuli and biologically determined reinforccrs second CS is redundantly added to an established CS (Kamin
that underlie operant conditioning in the human case (i.e. 1969) is usually explained by the fact that the UCS can be
innate biopsychological drives, needs, instincts, etc.). Without anticipated from the latter; only when the outcome of the trial is
such knowledge we cannot specify the consequences of behav- "surprising" does the second CS gain control of the conditioned
ioral innovations and hence cannot operationalize the principle response. Likewise, the "value" of a reinforcer can be enhanced
of selection by consequences. or reduced "off baseline," and this will subsequently be re-
flected in the performance of the acquired response that pre-
cedes the reinforcer (see Adams & Dickinson 1981).
On the stabilization of behavioral selection Similarfindingssupport anticipatory processes in the produc-
tion of established behavior; in particular, animals seem to
Werner K. Honig anticipate temporal durations and delays (see Honig 1981, for a
Department of Psychology, Dalhousie University, Halifax, N.S., review). In "short-term" memory procedures, performance
Canada B3H 4J1 based upon an initial stimulus in a trial is markedly enhanced
The processes of change described by Skinner in "Conse- when different outcomes can be expected following different
quences" are clearly 'retrospective'; neither successful nor un- initial stimuli (Peterson, Wheeler & Trapold 1980). Evolution
Skinner has insightfully argued), and if selection-by-conse- in which they make contact with the environment, and in the ways in
quence mechanisms provide the modern explanation for intel- which they act upon the environment, what remains of their behavior
ligence and purpose (as Skinner affirms), and if such mecha- shows astonishingly similar properties. (Skinner 1959)
nisms need not be seen as denying the existence of the pheno- In addressing language and biological predispositions for
mena they explain (as many modern mechanistic mentalists learning Skinner appears to have accommodated the concerns of
maintain), then on what grounds do we dismiss the possibility many biopsychologists. But what about Skinner's treatment of
that evolving species are not in some sense individuals (Ghiselin the brain? Many have accused Skinner of treating the brain as a
1981) of limited but genuine intelligence and purpose (Schull, in black box and simply ignoring it. This, of course, is not true.
preparation)? The very question is enough to drive one back to Rather, Skinner has asserted that the brain should be studied in
behaviorism. much the same way that all other events are studied - by
examining observables. Skinner therefore soundly rejects the
notion of the "conceptual nervous system." He argues that a
Bridges from behaviorism to biopsychology conceptual nervous system cannot be used to explain the behav-
ior from which it is inferred (Skinner 1974, p. 213). Many
Paul R. Solomon behavioral neuroscientists would disagree with this statement.
Department of Psychology, Williams College, Williamstown, Mass. 01267
The conceptual nervous systems generated to help explain
sensory (Helmholtz 1852), motor (Sherrington 1947), and asso-
It would be difficult to overstate the magnitude of Skinner's ciative (Pavlov 1927) processes were all inferred from behavior,
contribution to psychology and related disciplines. Many of the and each has served a valuable heuristic role in subsequent work
approaches to the study of behavior that we take for granted can on the neural aspects of behavior. To deny the value of such
be traced to his initiative. Yet Skinner has not been without his models is to dismiss the basis for substantial progress in the
critics. On the one hand there are those who express moral or neurosciences.
philosophical outrage at some of his ideas about the control of Skinner also rejects the notion that behavior can be com-
behavior. At a more concrete level, there are those who have pletely understood by understanding the brain and nervous
criticized specific aspects of Skinner's views. Some of the most system. Instead, he has argued that the goal of neuroscience (he
severe criticisms have come from biopsychologists and neuro- calls it the promise of physiology) is to describe how the nervous
scientists. Skinner has often been criticized for his treatment of system mediates the contingencies between a discriminative
the biological aspects of behavior. This is especially true con- stimulus, a response, and a reinforcer. In doing this, Skinner has
cerning his treatment of language and the predispositions for argued that it is behavior that neuroscientists must strive to
learning, and his general views on the brain. And although some explain. It is this adherence to the notion that behavior is the
of the tensions between radical behaviorism, biopsychology, phenomenon that neuroscientists must consider focal that forms
and neuroscience will not be easily resolved, in "Conse- Skinner's most valuable contribution to the study of the brain.
quences," Skinner appears to be building some bridges. Through this argument he has suggested that a theory of brain
One of the most stinging criticisms of Skinner's work concerns function can be meaningful only if it is posited in the context of
his view on language. The major thrust of this criticism is that behavior.
Skinner has ignored the biological aspects of language. A second Consider, for example, how neuroscientists study memory.
point of contention between Skinner and his critics concerns the One approach is to understand the anatomy, physiology, and
apparent incompatibility between Skinner's views on the ac- pharmacology of the nervous system with the hope of eventually
quisition of operantly conditioned responses and the work understanding how various neuronal processes combine to code
showing that certain associations are more easily learned than memory. The goal here is to uncover mechanisms of plasticity
others (e.g. Garcia, McGowan & Green 1972). Here Skinner has that might be responsible for coding memory (see, for example,
been criticized for ignoring biological predispositions which Swanson, Teyler & Thompson 1982). As those working in this
either facilitate or hinder learning. A third criticism leveled by field recognize, much of the work using this approach is both
biopsychologists and neuroscientists concerns Skinner's treat- elegant and important. Yet because it divorces itself from
ment of the brain. To some, Skinner's assertion that it is not behavior, I doubt that Skinner would favor it. Skinner's ap-
useful to study and model the nervous system through inference proach is better reflected in the model systems approach to
from behavior ignores many of the insights obtained about studying neural mechanisms of learning and memory (see Kan-
brain-behavior relationships using just this strategy. del 1976; Thompson 1976). Here the beginning point, the
Skinner has not ignored his critics. In "Consequences' he incontrovertible evidence which must always be considered, is
both revitalizes earlier arguments and marshals new ones to the behavior of the organism. All attempts to understand possi-
answer these criticisms. Yet the tone of this article seems more ble mechanisms of memory must stem from and be consistent
compromising than that of earlier treatments. In discussing with a particular learned behavior. This seemingly simple point
language, Skinner seems to have modified his earlier views that many of us studying brain-behavior relationships now take
somewhat. In considering the importance of language he notes for granted has its foundations in the work of Skinner and other
that "vocal behavior can be modified through operant condition- behaviorists.
ing, but apparently only with respect to the occasions upon In "Selection by Consequences" Skinner has built bridges
which [the various kinds of vocal behavior] occur or their rate of between the often separated approaches of behaviorists and
occurrence." With this statement, it appears that Skinner has biopsychologists. This is especially true in his treatment of
made room for the biological aspects of language, while at the language and biological constraints. There still appears to be
same time providing an important role for operant conditioning. some tension between radical behaviorism and the neuro-
In "Consequences" Skinner also attempts to accommodate science community concerning the relationship between brain
the compelling data concerning biological predispositions for and behavior. Yet behavioral neuroscientists in particular
learning. For example, he notes that a duck may follow a large should take note of Skinner's basic contention: Behavior is the
moving object (in this case its mother) both because it has been starting point for studying brain function, and any final under-
selected to do so and because of "an evolved susceptibility to standing of brain function must be commensurate with the
reinforcement by proximity to such an object. ' This is a far cry behavior it is evoked to explain.
from Skinner's earlier statement:
Pigeon, rat, monkey, which is which? It doesn't matter. Of course, ACKNOWLEDGMENT
these species have behavioral repertoires which are as different as I am grateful to Andrew Crider for helpful comments on an earlier
their anatomies. But once you have allowed for differences in the ways version of this paper.
Selection misconstrued learning and the impact of learning on evolution than are
suggested by Skinner's straightforward analogy. What things
Stephen C. Stearns are easy to learn, what things are hard to learn, and why did that
Biological Laboratories, Reed College, Portland, Oreg. 97202 distinction evolve? Skinner passes that essential question by.
Skinner's final notable point is that selection processes are
Skinner makes at least three claims that attract the attention of blind mechanisms, and that too much attention has been paid to
an evolutionary biologist: (a) that conditioning works on indi- the object selected and not enough to the process of selection. I
vidual behavior in a manner strictly analogous to the operation completely agree with his denial that organisms can be con-
of natural selection in evolution; (b) that this analogy sheds light strued as initiating the process of adaptation, and with his plea
on the evolution of learning and on the impact of learning on that more attention be paid to the process of selection, but I
evolution; and (c) that more attention needs to be paid to the cannot countenance his decision to treat the organism as a black
process of selection and less to the objects selected. How far do box and to concentrate solely on the external circumstances
these insights take us, and are they new? affecting the organism. We must understand enough of the
Lewontin (1970) stated quite clearly that as long as things vary internal structure of organisms to explain how they interact with
in a heritable manner and undergo differential survival and their external circumstances. That internal structure can make a
reproduction, those things will evolve, whether they be orga- critical difference to the predictions one would make both about
nisms, ideas, or cultures. Dawkins (1976) has taken the conse- the direction of evolution and about the development of
quences of that idea much farther, and it is now a familiar, if not behavior.
universally accepted, part of evolutionary biology. Thus Skin- In summary, Skinner's analogy between evolution and learn-
ner's first claim is not new. ing is apt but hardly new, and his picture of evolution errs in its
I question his repeated presentation of natural selection as details. The questions Skinner states explicitly do not seem
designing traits for the benefit of species. Skinner seems to be likely to lead to productive new lines of research, but the
unaware of the controversy over group selection, individual questions he states implicitly are loaded with significance: How
selection, and gene selection that has occupied the attention of does learning evolve? Once evolved, what implications does
evolutionary biologists for the last 20 years. It has considerably learning have for subsequent evolution? Does it uncouple phe-
clarified the way we think about natural selection (see Maynard notype and genotype, as suggested by Huxley (1942)? What
Smith 1964; Price 1972; Wade 1978; Williams 1966). Although things are hard to learn and what things are easy to learn?
the debate is not over, it seems safe to say that only under special Biology and psychology can interact productively in these areas
circumstances could one conclude that a trait had evolved for (see Kamil & Sargent 1981). Skinner enters a final plea for an
the benefit of the species. In nearly all cases, traits have evolved objective assessment of the causes of human behavior. Although
because they increased the fitness of genes, individual organ- I do not think his scheme of selection by consequences is
isms, or both. There is some persuasive evidence that certain adequate to represent causation, I do agree that we must
behavioral traits have evolved because they increased the in- understand the biological basis of human behavior and the limits
clusive fitness of individuals, that is, fitness gained through it sets - if any - on what can be learned. That knowledge would
relatives (Hamilton 1972; Sherman 1977), and some evidence make it easier to solve problems caused by our own behavior.
indicates that certain traits, such as female-biased sex ratios
(Colwell 1981; Hamilton 1967) and recombination itself (May-
nard Smith 1978), have evolved in part through group selection.
However, biologists continue to produce alternative explana-
tions based on individual selection or gene selection (e.g. Selection by consequences: A universal
Felsenstein 1974; Hamilton 1980; Rice 1983) for even those causal mode?
traits, such as recombination, for which the evidence is most
consistent with group selection. Skinner also seriously under- William Timberlake
estimates the rapidity of evolutionary change. Significant Psychology Department, Indiana University, Blocmington, Ind. 47405
changes can occur in ecological time, on the order of tens of For some time it has been customary to attribute human behav-
generations (Kettlewell 1961; Stearns 1983). Skinner's view of ior to a combination of the effects of biology, individual learning,
natural selection is badly out of date. and society. Though it is clear that these determinants are
Skinner's most constructive point is to call attention to the related, experience has taught most of us to consider them
interaction of selection and learning. I particularly appreciated separately rather than to ignore divisions between disciplines
his description of how the evolution of learning could lead to and suffer the cross-fire of individuals defending their turf.
nonadaptive or even to maladaptive behavior. However, he Skinner has frequently crossed such disciplinary boundaries, in
again falls disappointingly short of the sort of analysis that would the present case by arguing that selection by consequences is
be suggested by recent developments in evolutionary biology. the key causal mechanism in all three areas. Though I admire his
Learning is a sophisticated form of developmental plasticity. bravery and have sympathy for his goal of integration, I think the
Ashby (1956) has suggested that plasticity in some traits may resemblances Skinner sees are overstated and his emphasis on a
buffer other traits against the force of selection, and Caswell common causal mechanism interferes with rather than facili-
(1983) has developed this idea in the context of the evolution of tates analysis of the relations among natural selection, individual
reproductive traits. Charnov and Bull (1977) have argued that learning, and culture.
labile sex determination is favored by natural selection when an What Skinner calls attention to is that changes in evolution
individual's fitness - as male or female - is strongly influenced (new species), individual learning (new behaviors), and culture
by environmental conditions and when the individual has little (new societies) typically occur in the context of some alteration
control over which environment it will experience. Stearns and of the environment. On the basis of this resemblance Skinner
Crandall (1983) have extended this argument to traits and argues that the changes are based on the common causal mecha-
environments that can vary continuously. These analyses of nism of selection by consequences. The fundamental problem
nonbehavioral traits provide a prototype for a biologically based
with this argument is that a consequence is by definition an
analysis of the evolution of learning. Cavalli-Sforzaand Feldman
(1981) have recently published an extensive discussion of the outcome that follows necessarily from a set of conditions. By this
interaction between genetic and cultural evolution. In all these definition only operant learning can be a product of selection by
biological discussions there are many more concrete points for consequences; examples of natural selection and cultural evolu-
departure and suggestive hypotheses about the evolution of tion do not follow from any set of prior conditions that can be
specified. I develop and add to this argument below. Since
have poorer vision than birds, but also a poorer muscular system
than other mammals of similar size. In fact, humans would not Author's Response
be able to defend themselves against predators if they were not
equipped with more efficient brains. The development of the
large cerebral hemispheres with their associative cortex com-
pensated for the deficiencies of other systems of the body, and Some consequences of selection
not only resulted in the survival of humans but also secured
them the highest position among living beings.
B. F. Skinner
Let me turn now to individual behavior consisting, for the
Department of Psychology and Social Relations, Harvard University,
most part, of operant behavior (using Skinner's terminology).
Cambridge, Mass. 02138
Does the development of operant behavior in an individual
resemble the evolution of life over millions of years? It has been Why has the role of selection by consequences appeared
asserted that the successive stages of ontogenetic development so late in the history of human thought? It is a principle
roughly approximate the successive stages of phylogenetic de- found only in living things and that is no doubt relevant,
velopment (Gould 1977). This process occurs mostly during the but people have been interested in living things as long as
prenatal period and continues for some limited time after birth. in nonliving. A more likely explanation is that the effects
Then operant behavior starts to develop in response to environ- of selection are somewhat delayed. We see the products
mental conditions.
The development of operant behavior and the evolutionary of selection but only some time after we have seen the
process are similar in that they are both based on selection by selection itself. The difference in time may have led to the
consequences and both become gradually more and more com- search for current surrogates. We look for operative
plex. But there are also big differences between them. By features in the product rather than at the selective events
contrast with natural selection, (1) operant behavior is an active responsible for them.
process capable of producing permanent changes in inborn Purpose is such a current surrogate of past conse-
reactions and complicating their patterns; (2) its development quences. Cognitive psychologists speak of operant behav-
seems to be quite a straight-line process unless slowed by ior as goal-directed behavior. Goal-directedness is a cur-
adverse conditions (including aging); and (3) it is not genetically rent property which replaces a history of reinforcing
transferable to later generations.
But there is still another important difference between natu- consequences. (The word operant alludes to an observed
ral selection and operant behavior. Although in both cases property of behavior - namely, the effect on the environ-
selection by consequences is the basic principle, the conse- ment. Whether the effect changes behavior is not men-
quences are different in each case. In natural selection the tioned.) The intentionalism of modern philosophy also
consequence is survival. But is it also that in operant behavior? springs from a search for a current property of behavior as
Let us take, for instance, feeding behavior. It has been reported a surrogate for history.
that 6-7-day-old rat pups prefer nonnutritional 0.1% saccharin The temporal subtleties of selection could in turn be
solution in water to 2.8% lactose solution corresponding to the responsible for the invention and subsequent flourishing
sweetness of mother's milk (Jacobs & Sharma 1969). Adult rats, development of a simpler cause, following the push-push
even when hungry, prefer nonnutritive 0.25% saccharin solu-
tion to nutritive 3% glucose solution (Valenstein 1967). Other causality of daily life. Thus, life on earth has simply been
experiments have shown that hungry rats, offered a choice created and behavior is simply intended, chosen, and
between food and intracranial self-stimulation, prefer to self- willed.
stimulate, although this leads to death from starvation (Routten- The reasons why selection by consequences was so long
berg & Lindy 1965). Excellent examples of behavior contrary to neglected are probably the reasons why it is still so badly
survival are also provided by drug addiction and dangerous misunderstood.
sports. It seems, then, that the consequences of operant behav-
ior must be not so much survival as sensory gratification. It can
be supposed that what is called "reinforcement" in operant I am sorry that in the four or five thousand words that
behavior is sensory satisfaction or, in other words, improvement were available to me when I wrote "Consequences" I
in sensory state resulting from the presence of unconditioned have not covered the field of natural selection to Barlow's
stimuli in approach behavior, or from the absence of uncondi- satisfaction. I am also sorry that he appears not to be
tioned stimuli in avoidance behavior (Wyrwicka 1975; 1980). aware of the extent of current research on operant condi-
So far, there is no objective and direct evidence that improve- tioning. I am happy that he agrees with me on the
ment in sensory state is the main causal factor in operant
behavior. Still, can we be sure that animals living in their natural evolution of culture, but he seems to miss its relevance to
environment do not care about the taste of food and eat only in the question of who is to decide what is good behavior.
order to survive? Or that they mate only in order to produce So far as the point of "Consequences" is concerned, it
progeny (that way securing the survival of the species), and not does not matter in the least whether any of the behavior
in order to get sensory satisfaction from mating? Of course, Barlow mentions is the product of natural selection,
there exist behaviors where survival is in stake. These include operant conditioning, the evolution of cultural practices,
fights with competitors for territory, food, or mates. But is or any combination thereof. The same issues arise: the
survival the real "purpose" of the fight? It may be so, but on need to abandon the concept of a creator, purpose,
condition that survival means experiencing sensory gratifica-
essences like life, mind, and Zeitgeist as contemporary
tion.
surrogates of histories of selection, and values. I repeat:
If the above supposition is correct, this means that the All these issues demand attention regardless of whether
survival of the species is secured only when sensory satisfaction
obtained from operant behavior goes together with survival. On the consequences are found in natural selection, operant
the other hand, in cases in which it works against survival, the conditioning, or the evolution of cultures. Although there
whole species can perish. This especially applies to humans, is lively controversy at all three levels, the basic notion of
who have developed such a variety of means to provide sensory selection by consequences survives and raises the ques-
satisfaction. tions I addressed.
ing our environment), but we do so because we have been the discovery of new practices and their transmission to
exposed to contingencies arranged by the social environ- other (especially younger) members of a group. But one
ment we call our culture. may still identify variations (new practices), reproduction
Organisms avoid self-destructive behavior without (the transmission to others), and selection through conse-
foreseeing the consequences, but, as I explain in "Prob- quences, whether for the individual or the group (and
lem Solving," people no doubt do so more effectively that last phrase distinguishes between the evolution of
when they have analyzed the contingencies and, in that cultural practices and the evolution of cultures). For
sense, have foreseen the consequences. example, the use of a new food or a new way of planting or
I agree that any dream of gaining complete control of storing it will be transmitted to other members of a group
the environment is "impossible," but from what we learn because of its reinforcing consequences for individuals,
when the environment is reasonably well controlled, we who are thereby more likely to escape hunger. The group
can at least interpret what is happening under more may then compete more successfully with another group
chaotic conditions. (say, for available land). In neither case is a special genetic
trait at work, as in kin selection. The evolved process of
I thank Dawkins for his refreshingly helpful commen- operant conditioning is a sufficient explanation.
tary and confine my remarks to questions he asks about I do not agree that respondent and operant condition-
levels ii and iii. ing are best regarded as "simply different procedures for
I do not know whether an "animal sets up a simulation studying behavioral change." As Ferster and I pointed
in its head of the various actions that it might pursue and, out in Schedules of Reinforcement (Ferster & Skinner
importantly, their probable consequences," but people 1957), a term like "conditioning" or "extinction" is tradi-
do something much like that when they examine prevail- tionally used to refer to two very different things: (1) the
ing contingencies and construct rules to be followed to role of the experimenter or the environment in bringing
respond to them effectively (see "Problem Solving"). about a change, and (2) the resulting change in the
Dawkins's suggestion that displacement activities at level organism. Donahue seems to add a third, "procedures for
ii may have the effect of mutations at level i throws light studying behavioral change." We are concerned here
on creativity - another chestnut in evolutionary theory with behavioral processes as they must have existed
closely related to this paper. Creative artists know how to before anyone promoted them or studied them. Whether
create mutations from which they then select those that there is a neurological principle common to respondent
are beautiful in the sense of reinforcing to them, greatly and operant conditioning is a question that will presum-
increasing the chances that their work will be original. ably be answered by neurologists; the two types of condi-
My only trouble with Dawkins's suggestion is that dis- tioning are still clearly distinguished by the contingencies
placement activities tend to be stereotyped, but - who under which they occur.
knows? - mutations may be, too.
There is clearly a question about what exactly is being I certainly do not claim, as Gamble implies, that
selected and what are the selecting consequences. With- experimental behavior analysts were the first to suggest a
in a given group, the answer seems to be practices - parallel between Darwinian selection and "trial and error
better ways of hunting, gathering, growing, making tools, learning," but I contend that the experimental analysis of
and so on. The practices are transmitted from generation behavior is by far the most detailed examination of the
to generation when those who acquire them under the contingencies of selection responsible for the behavior of
contingencies arranged by one generation become the the individual. I also believe that selection at level iii does
transmitters for the next. There is no competition be- not require a process different from natural selection or
tween cultures, no Social Darwinism, in such a formula- operant conditioning. I certainly did not mean to suggest
tion. But cultures as a whole have also come into exis- that very much has been done in those fields by operant
tence and perished. As I point out in my replies to Harris conditioners. Indeed, I regret that more has not been
and Maynard Smith, the evolution of cultural practices is done.
like the evolution of heart, stomach, eye, ear, fin, leg, I do not see the relevance of Gamble's comments on
wing, and so on. The evolution of cultures is like the one-trial learning. As I showed more than 50 years ago
evolution of species, each of which may have a particular (Skinner 1932), an operant like pressing a lever is easily
kind of heart, stomach . . . , and so on. It is clear that conditioned by one reinforcement. I do not suppose
cultural practices do not evolve because of successful Gamble means that a complex bit of phylogenic behavior
competition between cultures, except where the prac- (say, building a nest) once occurred in that form as a
tices have to do with conflicts between cultures - for variation and was selected by its consequences. It must
example, the invention of more powerful weapons. A have been the end result of a long process of shaping. I
culture which strengthens itself by developing new meth- have reviewed a few established geological processes
ods of agriculture, new social systems, and so on is more which could have supplied conditions for a gradual ap-
likely to compete successfully with another culture, but proach to complex phylogenic behavior (Skinner 1975).
the practices themselves evolve because of contributions
to the group that would also prevail if there were no I do not treat culture, as Ghiselin claims, "as if it were
competition with other groups.
identical with verbal behavior." I said that verbal behav-
ior (which I had just discussed) greatly increased the
Donahue wonders whether cultural evolution, or the importance of a third kind of selection by consequences.
evolution of cultural practices, is a different kind of Other ways in which new forms of behavior are transmit-
selection. I think it is, although I see in it no new ted to new members of a group include imitation and
behavioral process. I think operant conditioning explains modeling. I would define a culture as "a mass noun " as a