Excerpt from Kenneth Spence's paper "Gradual versus sudden solution of
discrimination problems by chimpanzees"
Journal of Comparative Psychology, 1938, 213-224
Kohler, in his 1925 paper likewise considered sudden solutions in the learning of
discrimination problems. While he pointed out that discrimination learning is usually a
slow gradual process even in the anthropoids, he nevertheless claimed that learning is
probably quite different in cases in which the solution is sudden. The occurrence of such
sudden solutions is, moreover, evidence to him that the learning involves more than mere
associative processes or, to quote him: 'We do not well describe experiments of this type
by saying, as we usually do, that an animal in such a situation learns to connect certain
stimuli with certain reactions and that the connection is "stamped in". This
formulation of the process gives too much importance to the memory or association side of
the problem, and it neglects another side of it which may be even more important and more
difficult'(2). This neglected aspect of the problem, according to him, is the process by
which the sensory field becomes organized, or rather reorganized, during learning, and it
is this part that is responsible for the sudden solutions or insights in learning.
But are such instances of sudden jumps in the learning curves of animals evidence
either for the existence of an insightful factor in learning, or for the inadequacy of the
view that learning consists in the formation of stimulus response connections by a
repetitive process of reinforcement and non-reinforcement? In the opinion of the writer,
both parts of the question are, as yet, to be answered in the negative. To say, as the
Gestalt psychologists do, that such learning involves the development of insight is merely
to restate the problem in new terms. No satisfactory theoretical account of how such an
insight factor operates to produce such sudden learning in discrimination problems, or for
that matter any other type of problem, has ever been given by its Gestalt proponents. The
association theorists, of course, are in exactly the same position, for they have never
given a satisfactory theoretical account of this phenomenon. It does not logically follow,
however, that an adequate theory cannot be developed on association principles.
The writer recently presented a theory of the nature of discrimination learning, based
on association or conditioning principles, which was shown to be capable of explaining the
various phenomena known to be characteristic of the presolution period of learning (5).
Quite in contrast to the Gestalt configuration interpretation, this theory conceives
discrimination learning as a cumulative process of building up the strength of the
excitatory tendency of the positive stimulus cue (i.e., the tendency of this stimulus to
evoke the response of approaching it) by means of the successive reinforcements of the
response to it, as compared with the excitatory strength of the negative stimulus,
responses to which receive no reinforcements. Learning is completed when the difference
between excitatory strengths of the two cue stimuli is sufficiently large to offset always
any differences in strength which may exist between other aspects of the stimulus
situation that happen to be allied in their action with one or other of the cue stimuli;
for example, such differences as may exist between the excitatory strengths of the food
boxes on which the cue stimuli are placed. In the presentation of this theory, no
consideration was given to the problem of the occurrence of sudden solution in
discrimination learning.
Back to Kohler
Go to Krechevsky's comments about a non-continuity account of
learning