The goal of the proposed research is to expand our knowledge of the elementary learning processes of Pavlovian conditioning and instrumental training. Prior work has characterized the associations that develop when organisms experience relations among events in their environment. Much of this work has employed two techniques that allow separate measurement of associations: the devaluation and transfer techniques. Recently this laboratory has used those techniques to assess the state of previously learned associations when the conditions that produced them no longer obtain. That work has provided the first clear documentation that associations remain relatively unaffected by procedures that break the relations that produced original learning, procedures like extinction, arranging random relations, or replacing one outcome by another. The proposed work would pursue the suggestion that this preservation of the associations depends on the learning at the level of behaviors. It would also explore the further possibility that the original associations not only remain intact but actually continue to contribute to behavior in a routine manner. Exploration of these possibilities would make use of the introduction of new stimulus events during extinction. In some experiments these new stimuli would originally be neutral but would then be tested to assess the properties that they acquired during extinction. In other experiments these new stimuli would be able to themselves evoke responses; the goal would be to see whether their doing so affects the course of extinction, as anticipated if response learning is involved. Another set of experiments would employ new compound techniques to test a widespread assumption of associative theories -- that all stimuli present when learning occurs change by the same absolute amount. If that proves wrong, it would force reevaluation of a broad class of models of learning. A final set of experiments would explore the effects of reestablishing the relation after extinction. These studies would importantly influence our thinking about elementary associative learning processes. They would illuminate the process of extinction, a process important both to clinical applications and to neural investigations.
