A resolution of the debate about associative and temporal learning.

For more than a century, there has been an extensive experimental literature on both associative and temporal learning. Associative learning is based on strength of associations between elements. In contrast, temporal learning is based on durations of intervals between time markers. The same procedures have often been used to examine the formation of associative bonds between elements and to examine the learning of durations between time markers. Although there is general agreement on the behavioral results, different computational models of associative and temporal learning have led to a lengthy debate regarding whether associations between elements or interval durations account for these results. The purpose of this article is to propose a resolution that requires the development and evaluation of a computational model of procedures that produce associative and/or temporal learning. Standard methods of goodness-of-fit, simplicity, and generality can be supplemented by Turing tests to determine the extent to which a computer algorithm can predict whether the behavior came from the animal or the model. A successful general model should help guide the development of specific alternative models.