Part V. Learning to do — Learning to know

Abstract

Learning theorists have a long tradition of disagreement regarding even the most fundamental assumptions underlying their work. The answer to so seemingly simple a question as, “What is learned?” is likely to vary widely among different individuals. It is thus, not surprising that more complex questions evoke even greater disagreement. If consensus does not exist regarding so fundamental a matter as NJ,?& is learned in a learning experiment, it is obvious that questions regarding how it is learned may be answered in many different, often quite incompatible ways. One of the reasons why agreement regarding what is learned has eluded attainment by learning theorists is that the term “learning” is applied to so wide a range of different phenomena, even by laymen. A person is said to “learn to swim”, “learn a poem”, “learn history”, etc. Someone who has learned “to swim” will not drown when thrown into deep water; one who has learned “a poem” may recite it when asked to do so (if motivated to comply); someone who has learned “history” may explain why the Roman Empire finally collapsed (if motivated to do so). Learning to swim enables swimming behavior to be performed when necessary or desirable. Learning a poem enables behavior to be performed also, but the behavior may transcend mere rote repetition of sounds, since the sounds are symbols that have meaning. Learning history implies even more comprehension and less memorization of pat verbal responses. In general, there are things that people learn fo do and there are things that people learn ubout. Sometimes these are difficult to distinguish. Hull’s rats learned to make responses to stimuli, for example, ?o go down a runway, turn left, and enter a goal-box. The fact that a similar sequence of behavior previously was followed by the consumption of food in the goal-box was assumed to provide the associatigenic force for this S-R connection. Tolman’s rats, on the other hand, learned zh& the food was to be found in the left goal-box; the role of the food being there previously was to emphasize this S-S relationship. But Hull’s rats and Tolman’s rats must actually have learned the same thing, rather than two seemingly different things, since what they did was identical. The difference in question, obviously, was not between what Hull’s and Tolman’s rats learned, but between Hull’s and Tolman’s interpretation of their learning. Hull found it useful to assume that the rat learned to make responses; Tolman preferred to believe that the rat learned knowledge. The rat of course, simply learned. Levey and Martin have proposed that humans (either as subjects in conditioning experiments or clients in behavior therapy) also learn in two basically different ways. They learn “evaluations” 6f isolated stimuli in the environment, simply as a result of temporal contiguity between these stimuli and their positive or negative evaluative consequences. They use the term “consequence” solely in the temporal sense, implying nothing causal between the stimulus and its evaluative consequence. Evaluation learning is called “conditioning”, and it is in no way cognitive. The behavior it fosters is to some extent hardwired, since things that are positively evaluated usually are approached and in some fashion engaged, while those that are negatively evaluated are avoided or rejected. In addition to this kind of learning, humans also acquire cognitive summaries regarding sequences of events in the environment. This kind of learning is more complex and both phyloand ontogenetically higher than consequence learning, although it may have evolved out of it. The cognitive summary permits highly adaptive predictions of future events. It enables short-circuiting of complex routes, both spatial and conceptual. It makes it possible for problems to be solved