Krechevsky on "Hypothesis versus chance in the pre-solution
period in sensory discrimination-learning"
In a
previous paper, it was pointed out that a re-examination of the data obtained from sensory
discrimination experiments necessitates the adoption of a new description of learning. It
was found that instead of considering the first part of 'learning' as consisting of
random, haphazard behavior, we must recognize that the animal, during that period, is
responding in an orderly, systematic manner. He is attempting various solutions and giving
them up when they fail, until he hits finally upon the 'correct' one. The present paper
presents part of the experimental evidence for such a thesis.
As originally
planned and carried out the experiment involved the setting up of two discrimination
habits in each animal - a visual and 'hurdle' habit - as well as a study of the general
problem of transfer of training. The present report will concern itself, however, only
with that part of the study relevant to this question of hypotheses. And for this purpose
the results of the 'hurdle' discrimination only will be considered.
In his Brain
Mechanisms and Intelligence, Lashley made the first suggestion of a possible
relationship between two often observed phenomena connected with the setting up of
discrimination habits by the white rat - the peculiar shape of the learning curve and the
tendency of the animal to form various position habits before mastering the problem.
'There are many indications,' he writes, 'that... in the discrimination box, responses to
position, to alteration, or to cues from the experimenter's movements usually precede the
reaction to light and represent attempted solutions that are within the rat's
customary range of activity....The form of the learning curve is the more significant when
considered in relation to such behavior...it suggests that the actual association is
formed very quickly and that both the practice preceding and the errors following are
irrelevant to the actual formation of the association.
In other
words, even when the 'learning curve' appears to show 'random' behavior the animal may be
responding in a wholly systematic manner. Lashley regrets, however, that 'there is no
present way to record such behavior objectively and I can present the description only as
an impression from the training of several hundred animals in these problems'. The data
from the present experiment have been examined in the light of Lashley's suggestion and an
attempt has been made to devise a method for the objective determination of the validity
of that suggestion.
See Spence's comments concerning a continuity account of
learning