1 (Spring)
9
10 B. F. SKINNER
found in its natural habitat. There is no If that is true, ethologists are equally
reason why, upon occasion, phylogenic guilty when, in studying natural behavior
behavior should not intrude in this way in the field, they make sure that there has
upon ontogenic. Certainly intrusions in been no chance for conditioning. Must we
the other direction are common enough. conclude that they cannot therefore be
Civilization shows the extent to which telling us anything important about
operant conditioning has suppressed behavior in the natura(environment?
phylogenic tendencies. Schwartz explains the success of ap-
Superstition. The effect of an acciden- plied behavior analysis by pointing to
tally contingent reinforcer offers some of other simplifications. The behavior of
the best evidence of the power of operant factory workers has been "captured"
conditioning, and possibly for that reason because the factory has eliminated other
it has been challenged-as, for example, influences-sociocultural rather than
by Staddon and Simmelhag (1971). The biological. But social behavior in the
behavior is said to drift toward world at large is certainly due to condi-
phylogenic forms. I am quite sure of my tioning, and if we are to understand it, we
original observation (Skinner, 1948). I must look to the basic processes.
have repeated it many times, often as a Sociobiology. Ethology has spawned a
sure-fire lecture demonstration. Deliver child which threatens to play Oedipus and
food every twenty seconds to a hungry kill its father. It has also been said
pigeon and it will soon exhibit a food- to threaten the experimental analysis
getting ritual of unpredictable of behavior. The term-with its roots
topography. I see no reason why there "bio-" and "socio-"-alludes to the roles
should not be a drift toward phylogenic played by genes in biology and society,
behavior. It would be something like the but skips over the individual. As I have
Breland Effect unopposed by operant pointed out (Skinner, 1981), selection is a
contingencies. causal mode, found only in living things,
"Misleading" Simplifications. In all which operates at three levels. Darwin
the experimental sciences it is a fun- revealed its role in natural selection, but
damental practice, when studying one Herbert Spencer had already pointed out,
process, to eliminate all others which may if none too clearly, a role in the behavior
affect the data. Chemists use pure of the individual and in the evolution of
substances for obvious reasons. Physicists cultural practices.
hold irrelevant variables constant. The ex- A recent issue of Science (Levin, 1982)
perimental space used in analyzing contains an interview with Ernst Mayr, a
behavior is as free as possible of distrac- leading figure in evolutionary theory and
ting influences including the releasers of the author of a new book called The
innate behavior. Barry Schwartz (1981) Growth of Biological Thought (Mayr,
has drawn a strange conclusion from this. 1982). In explaining why evolutionary
Operant conditioners, he says, "capture theory is misunderstood by physicists,
the behavior of pigeons and rats in Mayr neglects an important point about
laboratory environments by eliminating selection. As to the differences between
possible biological influences." He goes physical and living systems, he says,
on: "There isn't a process in a living organism
The experimental chamber generally seems to pre- that isn't completely consistent with any
vent the occurrence of behaviors like these; hence physical theory. Living organisms,
the claim that it reveals universal principles. One however, differ from inanimate matter by
must wonder, however, about whether any situation the degree of the complexity of their
which prevents the occurrence of behaviors as systems and by the possession of a genetic
powerful as these is not fundamentally distorting
our understanding of the principles of behavior. It program." Complexity itself is not a dif-
seems that if the conditioning chamber in fact ference in kind, nor was the "organiza-
prevents these sorts of species-typical behavior pat- tion" with which biologists, at an earlier
terns, it cannot be telling us anything very important date, usually defined an organism. The
about the control of behavior in the natural environ-
ment. "genetic program" points, though not
CAN EAB RESCUE PSYCHOLOGY? 13
directly, to the real difference: Organisms becomes the kind of feature eliciting or
differ from physical things because they releasing such a response. An article in the
show selection by consequenes. current Journal of the Experimental
In Sociobiology (1975), E. 0. Wilson Analysis of Behavior (Buzsaki, 1982)
points to certain features common to argues that some instances of Pavlov's
natural selection, operant conditioning, "orienting reflex" may be examples. The
and the evolution of cultures, and at- fact that the response to the key may ac-
tributes them all to genes. Genes no doubt tually reduce frequency of reinforcement
explain behavior which is due to natural should occasion no surprise.
selection, and they are also responsible Comments by two reviewers of
for operant conditioning as a process, but Autoshaping and Conditioning Theory by
once that process has evolved, a different Locurto, Terrace, and Gibbon (1981) are
kind of selection accounts for the relevant. In Contemporary Psychology
behavior of the individual and the evolu- (1981), Barry Schwartz writes:
tion of cultural practices. The key is lit, and then food is delivered. Pro-
Autoshaping. I studied another process cedurally, this is a mundane example of classical
said to threaten an operant analysis in the conditioning, with the key as a CS and food as a US.
late 'forties and tried to get a graduate But what is the classically conditioned response? It is
student to take it up for her thesis in 1950. not salivation, or an eye blink; it is a peck at the key.
The classically conditioned response is, or seems to
In my experiment, a spot of light moved be, what used to be viewed as a voluntary response,
across a screen and when it reached one not a reflex. What is going on? Is the key peck
edge, a food magazine operated. The voluntary or reflexive? Is autoshaping classical or in-
pigeon began to peck the spot as if it were strumental? Is there something wrong with our
distinction between the two conditioning processes?
driving it across the screen. Epstein and I
(1980) recently confirmed this result, There is nothing wrong except
although it is not clear that the pigeon is Schwartz's analysis. An operant cannot
driving the spot; it may be merely follow- be identified by topography alone; the
ing it. In the middle 'fifties, W. H. Morse controlling variables must be specified.
and I were curious about the great When several different variables are
variability in extinction following con- operative, as in verbal behavior, a struc-
tinuous reinforcement. After a given tural or formalistic approach is especially
number of reinforcements, some pigeons troublesome, as linguists are learning to
would emit many hundreds of responses their sorrow. Pecking a key is an operant
and others only a few. We thought the when it is primarily due to a particular
difference might be due to the fact that history of reinforcement. It is a released
some pigeons often missed the key or innate response when the lighting of a key
pecked too lightly to operate it and were is followed by the presentation of food, as
therefore actually on an intermittent in the autoshaping procedure.
schedule. We made a very sensitive key Schwartz draws another suppqsedly
and evoked a clear-cut exploratory threatening conclusion:
response with the method Brown and What autoshaping suggested was that pecking
Jenkins (1968) later called autoshaping. might indeed be special-peculiar to pigeons (and
We spoke of it as "conditioning a hot perhaps other birds) in feeding situations. In con-
key." (Incidentally, we got our answer, sequence, it raised the serious possibility that the
though we never published it. If you make massive accumulation of empirical generalizations
about the determinants of pigeons' pecking might
sure that all responses are reinforced, you not be applicable to all the instrumental behavior of
can reinforce many thousands of times all organisms. Instead, these generalizations might
and still get fewer than a hundred only be true of pigeons-or of organisms in situa-
responses in extinction.) Organisms tions in which the required instrumental response
was biologically related to the reinforcer.
presumably possess a repertoire of innate
behavior with which unusual features of But pigeons can press levers and rats
the environment are explored. Through a can peck keys and will do so under ap-
kind of Pavlovian conditioning, a key propriate contingencies of reinforcement.
which lights up before food is delivered As I pointed out in a recent paper in the
14 B. F. SKINNER
Behavior Analyst (Skinner, 1980), there the time which elapses between behavior
are several kinds of pigeon pecks, and and consequence. In operant condition-
they are not all concerned with ingestion. ing, a reinforcing consequence must be
Ferster and I explicitly acknowledged the closely contingent upon behavior. If it
ethological sources of the pecking were not, all intervening behavior would
response we studied. also be reinforced and chaos would
Schwartz continues: follow. Yet in the Garcia Effect a tenden-
Because autoshaping involves a commonly studied
cy to eat a particular food is affected by
behavior, in a commonly studied situation, the consequences occurring many hours later.
autoshaping phenomenon implies not only that an The result has obvious survival value in
organism's biology might contribute to how and protecting organisms from the further in-
what it learns, but also that the said biology has been gestion of poisons or highly indigestible
contributing all along, in studies that were presumed foodstuffs. Presumably the punishing
to have purged biology as a significant variable.
Because of this, autoshaping is a dual threat to tradi- consequence would affect the eating of
tional learning theory. It is a threat because it sug- any other unusual foodstuff at the same
gests, as does taste aversion, that learning theory time or during the interval but not other
must take biology seriously. And it is a threat kinds of behavior. There is little chance
because it suggests that learning theory has been
misunderstanding its own experiments. for confusion, because it is a special con-
sequence of ingestion. If other kinds of
But who are these people who believe deferred punitive consequences had a
that they have purged the behavior of an comparable effect, it would be felt by all
organism of biology as a significant intervening behavior. There is nothing in
variable? And what has been the Garcia Effect that contradicts any
misunderstood? part of an operant analysis or throws into
In a review of the same book in Science question any established facts. The con-
(1981) Peter Killeen says that "in 1968 sequence is punishing rather than
Brown and Jenkins demonstrated that positively reinforcing and seems to work
Pavlovian contingencies (pairing a key exactly as I describe punishment in
light with food in a standard experimental Science and Human Behavior. Through
chamber) yielded faster conditioning of Pavlovian conditioning, stimuli arising
the pigeon's key pecks than did tradi- from a situation in which behavior has
tional hand-shaping procedures." His been punished become aversive, and any
next sentence begins, "As if this were not behavior resulting in their reduction or
bad enough . . ." How bad it is depends removal is reinforced as escape or
on who does the shaping. Pavlovian con- avoidance.
ditioning is certainly slower than operant Probability of Reinforcement. In an
conditioning; I know of no instance in operant chamber the organism is in con-
which one pairing has ever been shown to tact with the contingencies only at the mo-
be effective (Pavlov's record-breaking ment of reinforcement. Ferster and I
dog showed a small effect after five pair- designed much of our research to show
ings of tone and food), but, as I reported that schedules have their appropriate ef-
nearly fifty years ago, a single reinforce- fects by virtue of the stimuli present at
ment of pressing a lever may be followed just that time-stimuli generated in part
by a sizeable extinction curve. I dare say by the organism's recent behavior.
the same thing can be shown for pecking a Several writers have recently implied that
key. Killeen also says that the work on organisms may be sensitive to an increase
autoshaping means that the discipline is in the mere probability of reinforcement
moving close to the biological bases of when no reinforcer is immediately con-
behavior, "a position it was a mistake tingent upon a response. I do not think
ever to have left." Again, I should like to that the possibility of a conditioned rein-
know who has left it. forcer has been satisfactorily eliminated
The Garcia Effect. Many years ago as an explanation, but I will rest my case
taste aversion was known as "stomach on the following experiment, which takes
memory." The unusual thing about it is advantage of the fact that the role of a
CAN EAB RESCUE PSYCHOLOGY? 15
search for an escape hole. A lever connected to an Philosophers, political scientists, economists, and
invisible opening, if touched accidentally, will per- others who once dismissed behaviorism as rat
mit escape. As the experiment is repeated several psychology are now seriously considering its im-
times, the animal-rat, mouse, hamster, monkey or plications. The journal Behaviorism, with its large
otherwise-will take less and less time to find the international board of editors, now in its 10th year,
solution of escape by touching the lever. Ultimately has become an important forum.
the animal becomes most proficient. I myself am most concerned with the possible
relevance of a behavioral analysis to the problems of
A passage like that is consoling because the world today.... If there are solutions to those
it makes one realize how far some of the problems, I believe that they will be found in the
critics of an operant analysis are from kind of understanding to which an experimental
understanding it. analysis of human behavior points.
So-called objections to operant theory The experimental analysis of behavior
need not detain us. There is work to be is alive and well. Psychology needs it.
done. My own contribution to that issue
of Psychology Today read in part as REFERENCES
follows:
I am inclined to rank progress in basic laboratory Buzsaki, Gyorgy. The "Where is it?" re-
analysis first. With the aid of miniaturized control- flex: Autoshaping the orienting response. Jour-
ling equipment and computers, behavior is now nal of the Experimental Analysis of Behavior,
observed and measured with increasing precision in 1982,37, 461-484.
operant laboratories throughout the world. Reper- Breland, K. & Breland, M. The misbehavior
toires of behavior are being studied which have a of organisms. American Psychologist, 1961, 16,
much greater breadth and complexity. It is still a 681.
hallmark of the operant-conditioning method . . . Brown, P. L. & Jenkins, H. M. Autoshaping
that the results may be formulated in centimeters, the pigeon's keypeck. Journal of the Experimen-
grams, and seconds rather than in the nonphysical talAnalysis of Behavior, 1968, 11, 1-8.
dimensions of mental life. Chomsky, N. Rules and Representations. Colum-
These advances have greatly increased the extent bia University Press, 1980.
to which the terms and principles drawn from an ex- Epstein, R. & Skinner, B. F. Resurgence of
perimental analysis can be used in interpreting responding after the cessation of response-
behavior in the world at large. Interpretation has not independent reinforcement. Proceedings of the
been well analyzed by scientific methodologists, and National Academy of Sciences, 1980, 77, No. 10,
it has been widely misunderstood by critics of the 6251-6253.
operant field. Among the processes which have been Herrnstein, R. J. The evolution of be-
submitted to more careful analysis and interpreta- haviorism. American Psychologist, August,
tion are many that have been attributed to . .. con- 1977.
cept formation, creativity, and decision-making. A Jenkins, H. M. & Moore, B. R. The form of
number of these are being clarified as an operant the auto-shaped response with food or water
analysis, particularly of verbal behavior, is better reinforcers. Journal of the Experimental
understood. Some behavior is contingency-shaped; Analysis of Behavior, 1973, 20, 163-181.
it has been selected by reinforcing consequences in Killeen, P. A challenge to learning
the past. [Other behavior may consist ofl imitating theory. Science, 1981, 214, 548.
the behavior of, or following the advice of, another Lewin, R. Biology is not postage stamp
person whose behavior has already been selected by collecting. Science, 1982, 216, 718-720.
its consequences. This distinction between rule- Locurto, C. M., Terrace, H. S., & Gibbon, J.
directed and contingency-shaped behavior is only Autoshaping and Conditioning Theory.
one example of a new approach to the analysis of so- Academic Press, 1981.
called cognitive processes. Mayr, E. The Growth of Biological Thought.
One advantage in relating behavior directly to en- Harvard University Press, 1982.
vironmental conditions is that one can then move Peterson, N. Control of behavior by pre-
directly to technological control. An experimental sentation of an imprinted stimulus. Science,
analysis points to the conditions which must be 1960, 132, 1395-1396.
changed to bring about changes in behavior for Schwartz, B. In Pursuit of B. F. Skinner.
practical purposes. Swarthmore College Bulletin, March, 1981.
Interest in the experimental analysis of behavior Schwartz, B. Autoshaping: Driving toward a
and its use in interpretation and practical control has psychology of learning. Contemporary
spread rapidly throughout the world during the last Psychology, 1981, 26, 823-825.
15 years. Associations have been organized and an- Skinner, B. F. "Superstition" in the pigeon.
nual conferences held in the United States, Latin Journal of Experimental Psychology, 1948, 38,
America, Europe, Israel, Japan, and elsewhere. The 168-172.
Association for Behavior Analysis, an international Skinner, B. F. The shaping of phylo-
organization, attracts new members each year and genic behavior. Acta Neurobiologiae Experimen-
its programs show an increasing scope. talis, 1975, 35, 409-415.
CAN EAB RESCUE PSYCHOLOGY? 17
Skinner, B. F. Why I am not a cogni- its implications for the principles of adaptive
tivepsychologist.Behaviorism, 1977,5, 1-10. behavior. Psychological Review, 1971, 73, No. 1,
Skinner, B. F. The species-specific be- 3-43.
havior of ethologists. The Behavior Analyst, Wade, Nicholas. New York Times, April 30,
1980, 3, 51. 1982.
Skinner, B. F. Selection by consequences. Wilson, E. 0. Sociobiology. Cambridge, Massa-
Science, 1981, 213, 501-504. chusetts: The Belknap Press of Harvard Univer-
Staddon, J. E. R. & Simmelhag, V. L. The sity Press, 1975.
"superstition" experiment: A reexamination of