At the turn of the 19th century, the Russian physiologist Ivan Pavlov had a problem. His special interest was the physiology of digestion, the biological mechanics that tied brain to gut, from the perception of food to a salivating gland. It didn't of course work the way he thought. The reflexive aspects of the digestive system, from gastric to salivary secretions was not hardwired like a kneejerk. It was plastic, malleable, and dependent upon an animal's vicarious experience with the world. The digestive process, at once so simple and reduceable to a cantilever of glands and muscles, was tethered to a complex and indecipherable brain. Pavlov surmised that the simple salivary response provided a doorway to the understanding of the labryinthine neural processes that modulated the aspects of not only such 'respondent' or classically conditioned behavior, but perhaps behavior in all its manifestations.
Through countless experiments on his hapless subjects , dogs, Pavlov cobbled together a theory of learning and the brain. It was, of course, wrong. But it was 'right' to be called a theory because it was testable, or more specifically, falsifiable.
Pavlov's work represented one of the first examples of a learning 'theory'. A learning theory is a series of interlocking conjectures about how experience influences behavior. These conjectures necessarily involved processes or mechanisms that make the model for behavior work. But with inference comes the means to test if it is true. Without the ability to test an inference, a scientific theory can never be. The mettle of a theory and whether we may even designate it a theory depends upon our ability to test the key inferences that make its predictions work. An 'effective' or robust theory in turn emerges when we can test it logically and empirically from every level, every perspective. This definition of a scientific theory, the work of the distinguished 20th century philosopher of science Karl Popper, was met by Pavlov's theory of learning, and with but one major exception, every theory of learning that was to follow. The path was laden with dead ends, misteps, and contentiousness, but progress began to be made.
Ivan Pavlov and happy, willing subject.
But progress demanded a special manner of thinking, that as with all the biological sciences, required a bit of generalization from knowledge of simpler but similar things. The basic processes of learning,like the essential aspects of ingestion, respiration, and life itself were assumed to be not dissimilar across species. Thus the root essentials of learning were presumably shared between species, and the dependent measure of learning, namely behavior, was like the shadings of light to an astronomer, the stuff that theories are made of. So learning theories met the logical demands of science, were derived root and branch from an 'ethological' (animal experiments) perspective, and because they emerged from an observation of non-verbal behavior unclouded by the metaphors of speech, were 'behavioristic' in nature.
The learning theories that would follow Pavlov, from Thorndike, Guthrie, Tolman, and Hull to the modern biological learning theories of Bolles, Toates, and Berridge were all of necessity behaviorisms, and were bound to the deductive method of hypothesis testing that drove the progress of the biological and physical sciences.
Yet what the popular and even adademic imagination knows of behaviorism is not a deductive and theoretical approach to learning that informs the way we look at ourselves and our behavior, but rather an inductive and atheoretical approach that aims to supplant our knowledge of self and behavior. This intellectual certitude, and perhaps arrogance was not a recipe for vigorous and fruitful debate, but for a bitter and contentious struggle that has marred and retarded the development of psychology to this day.
The Perversion of Behaviorism
It started innocently enough with a unique procedure and apparatus for the recording of behavior. A little box, fitted with a roll of paper and a depressable key could provide a cumulative record, that like a siesmograph, could record the frequency of activity as the animal pressed the key to 'work' for food. Adjust the timing of reward by altering the dependencies between reward and the number and/or interval between key strokes, and the data would come out differently, demonstrating how behavior is shaped by schedules or contingencies of reward or reinforcement. The psychologist B. F. Skinner took this raw data and ran with it, or as he called it, 'fled' from the laboratory. The ethological data of laboratory animals was replicated, with no small success, in the real workaday world of humans. The problem though is that when you move a level of description from animals to humans, humans will in turn interject the metaphors of common language, and metaphors have a habit of becoming an unfalsifiable 'reality'. So the problem Skinner faced was that the wholesale transfer of a behavioristic methodology to human affairs compromised behaviorism, since unfalsifiable mentalisms of need, drive, and desire got in the way of cause and effect. And it was the realities of cause and effect, the facts of behavior that ultimately mattered and were all that counted. Because the mentalisms that populated common speech were unfalsifiable, Skinner simply banished them, and the methodology of his brand of behaviorism, or a 'methodological' behaviorism became by virtue of its intrusion in common affairs, the behaviorism that is popularly known today in notoriety and fame.
B. F. Skinner: A philosopher for the birds?
By transposing behaviorism to human affairs, Skinner unwittingly changed the rules that made behaviorism a bonafide science. Removed from its ethological roots and the deductive principles that allowed it to test the boundaries of knowledge, behaviorism became merely guided by the dead hand of fact gathering. It was psychology as if imagined by Price-Waterhouse, a colorless tableau of ordered facts that only an accountant would call beautiful.
Confronted by the specious metaphorical realities of common sense or 'folk' psychology, Skinner chose to avoid them entire. But the vast majority of psychologists continue to use them, and vault a myriad mentalisms into an ever morphing kaleidoscope of unfalsifiable realities we call contemporary psychology.
The Clash of Antipodes
The supreme value of science is that it allows us to separate out words and their meaning by providing us with the ability to determine the literal from the aliterative, the metaphorical from the real. Human speech is profoundly metaphorical, and we often transpose words to others to suggest meanings they never had. Thus we speak of engine horse power, cool mints, and splitting headaches, yet because we know the nature of engines, mints, and headaches are able to see that their descriptions are metaphorical and not real.
However, the endeavor to limit the metaphors you use does not entail the ability to determine what is metaphorical and what is not. You can't after all have a witchhunt without knowing who the witches are. By decrying the use of inferred mentalistic forces, Skinner did not provide the tools to determine their metaphorical content; or in other words, whether they can be falsified or not. To do that requires a behaviorism, a theoretical behaviorism.
To illustrate this, let us use a Skinnerian example. In the language of Skinnerian or 'operant' conditioning, behavior occurs and is modulated in rate and form by its functional and/or temporal dependencies to reward or reinforcement. Press a key five times to get a reinforcer, and the resulting pattern of behavior is due to a FR 5 or fixed ratio schedule of reinforcement. .Similarly, wait ten minutes for the bus and your resulting behavior or lack of it is due to a FI 10 or fixed interval schedule.
On the other hand, variable schedules (VR) of reinfrcement occur when we don't quite know what step of our behavior will be followed by a reinforcer. The reinforcer may occur after any response or on average after any set series of responses, or all at any minute, any time. Strange thing though about variable schedules, they bring out the metaphor in all of us. Press a lever down five times for a reward and you are a bored pieceworker. Make the number of pulls variable and vary as well the size of the reward, and you are an energized and addicted slot player. Make the variance more fine grained, with every nuance of behavior followed by a randomized and important event, and like an artist, athlete, or surgeon engaging in the touch and go of their craft, behavior vaults to a higher consciousness, flow or peak experience, self-actualization, or intrinsic motivation.
A Skinnerian would see merely superfluity and the obfuscation of the facts of behavior, namely that variable schedules correlate with high rates of behavior. The rest was specious, unprovable, unnecessary. A humanistic or 'self determination' (SDT) theoriest would disagree, and say that such states were real, irreduceable and non-metaphorical things, that there are indeed unique flow states, intrinsic motivating processes, and peak experiences.
To a Skinnerian, learning is to be effective, to measure in economic terms behavior and its results, and by coordinating them just so making the economic measure of man. To a humanist or SDT theorist, learning is affective, a thing in itself, unfalsifiable, and unmeasureable in economic terms But between them both is the unbridgeable barrier of metaphor, made so because one party would abandon theory, and the other, misinterpret it.
Science represents the endless quest to separate the essential from the derived. The colors of the rainbow, the nucleotides of DNA, and the structure of matter itself represent the modulation and permutation of simple things. Each of the five senses make palettes of sound, color, and taste from the permutations of simple neural inputs, or rudimentary sensations of bitter and sweet, of red, green and blue, of simple vibrations of sound. The conjectures of science explained the five of them, and have dismissed the sixth. But the seventh and last sense has now been discovered. It is no less than learning itself.
Learning is more than an effective thing, it is an affective thing, and to understand this basic truth gives us comand of the most important and human sense of them all. Although Skinnerian behaviorists would ignore it, and humanistic psychologists deny it, how we learn is not less sensual than any of the senses. We know this from behaviorism, a true theoretical behaviorism, the fact that we are as sensitive to the problems or discrepancies of our world as our irises are to light, and our ears to vibrations of sound. Tune them just right and you have the real pleasure of fulfillment, satisfaction, and pride, but misunderstand or misapply them, and you have the pain of boredom, dissatisfaction, and despair.
Above the arguments is the light of scientific truth. Whether or not psychologists or behaviorists for that matter can learn from the behaviorisms of science only time will tell.
Author's Note: for more essays on the nature and implications of 'bio-behavioral' learning theory, please go to the 'Essential Mezmer' section of my website. For a much longer academic version of my argument, the following recent article (introduction copied below) by the neuro-psychologist Kent Berridge chronicles the progress of learning theory from its beginnings to its present form that recognizes the crucial importance of 'affect' in human learning. Please note that although Berridge notes the 'demise' of behaviorism, he is discussing the a-theoretical 'methodological' or 'radical' brand of behaviorism of B. F. Skinner, not the theoretical behaviorisms of Thorndike, Hull, Tolman, Bolles, Toates, etc. that are at the root of modern biological learning theories, including Berridge's own. One more note, the difference between a methodological and radical behaviorism lies in what each brand of behaivorism considers to be 'behavior' Methodological behaviorism concerns itself with primarily 'overt' responses that are subject to easy measurment, whereas radical behaviorism admits as well molecular or biological responses as an equally important subset of behavior. Both however eschew theory, which is in mine and Berridge's opinion, a serious error
Reward Learning: Reinforcement, Incentives, and Expectations.
Kent C. Berridge
(article is available entire in pdf format on Berridge's site, hyperlinked below)
...The radical behaviorist sense of reinforcement was the simplest, though not chronically the first version of the reinforcement concept. Reinforcement in a behaviorist sense as used by Skinner was merely a description of the relation between a change in a behavior caused by following it with a stimulus such as food. No explanation was offered. As Skinner put it,
The Law of Effect is no theory. It simply specifies a procedure for altering the probability of a chosen response. But when we try to say why reinforcement has this effect. Theories arise. Learning is said to take place because the reinforcement is pleasant, satisfying, tension reducing, and so on..(Skinner, 1950).
Skinner took a dim view of theoretical attempts to explain the 'why' of reinforcement. "Theories are fun. But it is possible that the most rapid progress toward an understanding of learning may be made by research which is not designed to test theories."
In the first half of the 20th century, such an explicitly antitheoretical account had two forms of appeal. First, it offered a purely objective alternative way of talking about behavior that could be claimed to escape the theoretical disagreements of early mentalistic psychology. Second, at the time Skinner wrote it seemed to many possible to believe that his hope for "the most rapid progress" might actually be true. One could be impressed in 1938, when Skinner's first major book appeared, by the clarity of the early studies of operant performance curves, of the smooth versus scalloped differences in their shapes produced by fixed ratio versus fixed interval reinforcement schedules, and so on. Such precision in behavioral description had never before been achieved, let alone brought under experimental control. One could still hope then that the second half of the 20th century would be filled with many powerful demonstrations of new and useful behaviorist principles for predicting and controlling behavior.
But the Skinnerian hope turned out to be unfulfilled. Very few new principles of use were produced by behaviorist studies over the next 50 years. However objective a behaviorist description might be, it generated very little for the predicting or control of behavior beyond the original law of effect, and a few straightforward applications of that law. And as Skinner himself acknowledged, the radical behaviorist concept of reinforcement offered nothing at all for understanding the 'why' of reward's effect on behavior. It had no theory and no explanatory power.
Today there remain a few calls for a return to atheoretical behaviorism (Staddon, 1999, Uttal, 2000). They reflect an almost aesthetic preference held by these authors for a purely atheoretical behavioral description, which avoids any inference about mediating processes that could turn out to be wrong, even at the cost of explanatory power. But contemporary behaviorists draw mostly on the old evidence for support, and simply reassert the original Skinnerian faith that good progress might be made in an atheoretical fashion.That faith, once plausible, now has nearly died away from psychology for lack of fruition. Most importantly to the demise of behaviorism has been a consensus in psychology that the behaviorist fear of theory and psychological constructs was misplaced. It has proven eminently possible to test theories grounded in cognitive, associative, and motivational hypotheses over the past 50 years. Many such theoretical constructs have been disproven, others confirmed, and still others modified on the basis of experimental results. Contemporary psychology and neuroscience of reward learning are both concerned almost entirely with accounts of why reward works. We will therefore turn to theoretical answers to the why of reward and reinforcement......