"Configural" conditioning in discrete-trial bar pressing.

Three experiments are reported comparing responses to stimulus compounds and to their elements. Experiment 1 found no loss in responding to the elements of a repeatedly reinforced stimulus compound. Experiments 2 and 3 explicitly trained a discrimination between a compound and its elements and found successful learning whether the compound or the elements were reinforced. Addition of an explicit stimulus common to the compound and its elements retarded, rather than facilitated, the discrimination. Replacement of one element in a compound by another of similar intensity and disrupted performance. These findings rule out a number of interpretations of configuring, but permit an interpretation in terms of a unique stimulus element and limitations on the total of the compound. An elementary problem in the development of any theory of behavior is that of providing an account of the response to a stimulus complex in terms of the responses to the elements of that complex. If an organism exhibits a learned response to a stimulus compound, AB, how can we understand this response in terms of the learning to the individual elements, A and B? Most theories (e.g., Estes, 1950; Hull, 1943) assume that the response to the compound is some combination of the learning to the elements. Indeed, a common specific combination rule has been proposed: that the associative strength of the compound is simply the algebraic sum of the of the elements. Such combinatorial approaches have