A theory of learning and transfer

Abstract

has pointed out an error in our article which we wish to correct.On page 248 of the December, 1938, issue of Psychometrika we stated that Spence assumes "that at the beginning of training the rat has tendencies to respond directionally (to the right or left) and relatively (to the larger or smaller) ..... We rejected Spence's definition and adopted a directional one when we found it possible to predict relational transfer on the basis of directional habits. This prediction seemed scientifically more interesting than to assume a relative tendency at the beginning and then deduce relative transfer later."The foregoing statement of Spence's view is erroneous. He assumes no relational tendencies. Instead, he assumes that an animal possesses four stimulus-response tendencies: to approach the right; to approach the left; to approach one of the two stimuli to be discriminated; and, to approach the other stimulus. From the assumption of these tendencies, none of which are relational, Spence deduces relational transfer.Spence's theory differs from ours in that he assumes four stimulus-response tendencies, whereas we assume only two, namely, tendencies to go to the right and to the left. Both theories agree in deducing relational transfer from asst~ptibns which involve no relational tendencies.We greatly regret this error and welcome the opportunity to correct it.

Conjunctive representations, the hippocampus, and contextual fear conditioning

“…The fundamental issue is to conceptualize how the organism represents a stimulus compound. Is it represented as a set of independent elements or features (e.g., Spence, 1936Spence, , 1937, or is it represented as a configuration (e.g., Gulliksen & Wolfle, 1938)?…”

The nature of the response in Simon discriminations by pigeons

“…The top half of Table 1 presents a configurational perspective (Gulliksen & Wolfle, 1938; see also Pearce, 1994) of pigeons' choice responding in the standard Simon task. The left column lists each left-right configuration of the R or G stimulus with the alternative W stimulus.…”

Transverse patterning in pigeons

“…All three explanations have a place for pair-wise interaction and so provide some basis for understanding transverse patterning. In a loop problem, a circle (literally, its afferent correlate) may be different in the presence of a square than in the presence of a triangle (the Hullian assumption); or it may be enriched by a pair-unique component that is different in the presence of a square than in the presence of a triangle; or, to borrow an idea from Gulliksen and Wolfle (1938), each ofthe three pairs offigures (circle-square, square-triangle, and triangle-eircle) may function as a separate configuration. The first two interpretations have in common the assumption that the animal chooses between unitary compounds, but the third is different and requires some elaboration.…”

“…The first two interpretations have in common the assumption that the animal chooses between unitary compounds, but the third is different and requires some elaboration. Gulliksen and Wolfle (1938) assumed that the two lateral arrangements of a single pair of stimuli in a simple simultaneous problem (e.g., A+B-and B-A+) function as separate configurations to which the animal responds by going to one or the other of the two positions (left or right) occupied by the stimuli (left to A+B-and right to A -B+). Although that assumption was subsequently shown by Nissen (1950) to be incorrect, there is in fact some persuasive evidence for the discrimination of the properties of pairs of stimuli as pairs (Bitterman, 1953), of which the most striking is perhaps provided by performance of rats in "two-situational" problems (Teas & Bitterman, 1952;Turbeville, Calvin, & Bitterman, 1952;Zeiler & Paul, 1965).…”