Logo IMG


Behaviorism at 100

Over its second 50 years, the study of behavior evolved to become a discipline, behaviorology, independent of psychology

Stephen Ledoux

Scientific Developments

Highlighting three of the many areas of experimental research can indicate the range of important findings discovered in the past 50 years. These three areas are schedules of reinforcement, recombination of repertoires and equivalence relations.

As Skinner relates in his 1957 article, reinforcers are postcedent stimuli whose occurrence produces increases in the frequency of the behaviors that they follow, and schedules of reinforcement are the patterns of intermittently occurring reinforcers. These schedules are defined in terms of either the number of responses since the last reinforcer (called ratio schedules) or the amount of time since the last reinforcer (called interval schedules). The values of either type can be fixed or variable, thereby defining the four fundamental intermittent schedules of reinforcement: fixed ratio (FR), variable ratio (VR), fixed interval (FI), and variable interval (VI). Researchers often combine or otherwise rearrange the elements of these basic schedules to study a range of more complex schedules.

Outside the laboratory VR schedules are common. They produce relatively rapid and steady response patterns. Purveyors of games of chance had intuitively arranged VR schedules that control the behavior of their players centuries before science discovered and analyzed this schedule. VR schedule effects (not the “gambling habits” of inner agents) are responsible for much citizen wealth reduction.

Schedule research has repeatedly led to several general conclusions, including these three: Many features of behavior emerge as the effects of particular reinforcement schedules. Schedules with only subtle differences often produce distinctly different response patterns. And the direct effects of schedules of reinforcement reduce a wide range of putative inner-agent emotional and motivational causes of behavior to misleading redundancies.

Highlighted next is the experimental research concerning recombination of repertoires, with important implications particularly for scientific, engineering and educational problem-solving behavior. In the 1980s Robert Epstein and Skinner coordinated some studies at Harvard called the Columban Simulation Project in which pigeon behaviors that were functionally related to explicit variables simulated complex human behaviors. Some of these complex behaviors concerned novel behavior, symbolic communication, and the use of memoranda and tools. The result of this research was a more objective explication of complex human behaviors. The same kinds of common contingencies known to be producing the pigeon-simulated behaviors were at work with the human behaviors.

The pigeon simulations began with analysis to establish the minimum repertoire components likely needed to produce a complex behavior when a challenge situation confronts the organism. Then, for each pigeon subject, after conditioning each required repertoire component (in isolation from other components, to avoid confounded results), the experimenters placed each pigeon in the challenge situation. The researchers found that, for different problematic tasks, once all the necessary component repertoire parts had been conditioned, then (and only then) the challenge situations evoked successful responses appropriately combining the trained repertoire components.

For example, many proud parents have watched as their child, too short to get a cookie from a jar atop a table and having never faced this situation before, looks around and, spotting a chair, moves it over to the table, climbs on it and retrieves a cookie from the jar, putatively due to something called insight. In experimenting to discover the variables involved in this situation, the researchers came upon three pigeon response classes using boxes and toy bananas. They conditioned the birds with no banana present to push a box around the chamber toward a target spot and, separately, to climb on a stationary box and, still separately, with no box or target spot present, to peck a toy banana within normal reach. These response classes approximated the components of the child’s cookie retrieval behavior. Finally, they placed each bird in a chamber with a box to one side and a toy banana suspended from the ceiling, a challenge situation that had never confronted the birds before. With some apparent confusion and sighting, like the child’s behavior, the birds pushed the box under the banana, climbed on the box and pecked the banana. Does this mean these birds showed “insight?” Was the child’s behavior due to insight, or was the child’s behavior also an example of previously conditioned repertoires combining under a novel circumstance? We do not usually observe children closely enough to track the conditioning of various repertoire components, but parsimony still requires accepting that the challenge-meeting responses are not a function of supposed higher mental processes, for pigeons or humans. Rather, they are a function both of the organism’s history having included the conditioning of relevant repertoire parts and of the current evocative control in the new pattern of related multiple stimuli in the challenge situation.

That line of research benefits the analysis of problem solving as well as enhances the justifications for multi-disciplinary education in science and engineering training curricula. As the range of an individual’s conditioned repertoire of behavior expands, so does the likelihood that needed parts are available to combine successfully in new circumstances for which no explicit response has previously been directly conditioned.

Apparently related to recombination of repertoires in ways that are still being explored, stimulus equivalence is the remaining experimental research area highlighted here. After explicitly conditioning some functional relations between environmental antecedent or postcedent stimuli and responses, the number of related behavior-controlling functional relations that we can successfully detect is greater than the number originally involved in the explicit conditioning. Researchers in this area have come to call these explicitly and implicitly conditioned relations equivalence relations.

Equivalence relations can transpire in fairly simple circumstances. For example, to train a cloakroom employee, we might first reinforce a trainee such that when shown a regular customer, Ms. Minkowner, and then shown a group of coats, including her pink mink coat, the Ms. Minkowner stimulus reliably evokes the trainee’s response of picking up her pink mink coat. Then we reinforce the trainee such that when shown the pink mink coat and several different coat-hanging cubicles, the pink mink coat reliably evokes the trainee’s depositing that coat in a particular cubicle, say, number seven. With no further training, we find that Ms. Minkowner’s appearance at the counter reliably evokes the trainee’s movement to cubicle number seven from which the trainee retrieves the pink mink coat.

Beyond such simplistic examples, researchers in this area have demonstrated the phenomena in far more complex circumstances. Using, for example, 6 sets of 3 stimuli each, explicit conditioning of a particular 15 environment-behavior functional relations turns out implicitly to condition an additional 75 behavior-evoking functional relations. In this instance, conditioning 15 particular relations can produce a total of 90 testable relations!

The implications of equivalence phenomena for a science-based revolution in, say, education can be substantial. More careful arrangements of what we would scientifically call educational conditioning programs can economize by explicitly conditioning only certain evocative functional relations, relevant to the subject matter, in ways that virtually guarantee the implicit conditioning of many other possible and relevant relations evocable by the same broad set of stimuli.

Although physiological research has yet to elucidate how the cellular and molecular mechanisms of respondent and operant conditioning work and contribute to equivalence relations, most researchers credit natural selection with the production of bodies that these processes can change in varying degrees. For example, if their genes happened to include variations that produced neural structures enabling the mediation of even a small extension of equivalence relations, then proto-species members could benefit from the survival and reproductive advantages conferred by the sort of “intelligence” that these emergent “abilities” imply. Over millions of years, the accumulation of such selected variations would result in genetically produced nervous-system structures of increasingly sophisticated potential. As a result, humans today inherit neural structures that generally mediate a relatively extensive range of equivalence relation phenomena.

Beyond experimental research, the past 50 years have seen an explosion of studies applying natural philosophy and science to practical problems. Touching on two applied-research areas, Project Follow Through in education and the refining of best practices for work with autistic children, barely suggests the extensive range of such concerns.

Project Follow Through was the most extensive and expensive federally funded educational experiment in U.S. history. It looked at how the outcomes of children taught with a range of instructional models, sponsored on voluntary district-wide bases, compared with the outcomes from children whose school districts across the country had not adopted any particular model.

The results led to a major observation: Although some models produced poorer outcomes than those of the control group, others produced consistently better outcomes, particularly the Direct Instruction and Behavior Analysis models. These successful models were explicitly based on the application of the principles and concepts of the natural science of behavior. This research had predictably revealed some science-based instructional approaches that work in education.

However, this revelation of best practices for regular education is widely ignored. Although the results of Project Follow Through focused mainly on student outcomes from the first several years of the project, the funding of various of its models continued for many years. Unfortunately, this funding was not limited to the models that produced improved student outcomes. C. L. Watkins concludes that suggestions to solve the problems of education include attempts to “change just about every structural and functional aspect of education except how children are taught.” Sadly, this indicates not only some blind respect for ineffective methods but also some persistence of the discredited notion that behaviorological laws are largely irrelevant to normal humans.

In the other applied-research example, best practices for work with autistic children have achieved greater recognition than best practices for regular education. Most of the research initially applying core behaviorological principles and concepts to a wide range of practical concerns, including interventions for autistic children, occurred before behaviorology emerged as an independent discipline. Consequently, many people refer to behaviorological practices with the term applied behavior analysis (ABA). The success of the ABA autism-related practices has made them the preferred intervention, especially for children diagnosed at a young age. For example, in 1999 the New York State Department of Health completed a multiyear project to evaluate the research literature on the numerous types of available autism treatments so as to make intervention recommendations based on scientific evidence of safety and efficacy. In its final report, the only intervention for autism that the department could fully recommend was ABA.

comments powered by Disqus


Subscribe to American Scientist