Serial learning vs. reward schedules: What is learned?

Abstract

Rats were employed in a runway to investigate whether different conditions for presenting reward events affect what is learned. One view is that, depending on how reward events are presented, animals may learn either simple associations or encode rules (rule view). Another view is that under all conditions for presenting reward events, animals employ the memory of prior reward events as signals for current reward events (memory view). To test these views, in each of three investigations reported here, rats were shifted to a single alternation series of rewarded and nonrewarded trials following various types of initial training. Concern was invested in whether in shift, immediate single alternation pattern responding would occur consisting of faster running on rewarded than on nonrewarded trials. All of the initial training conditions employed here were considered to be too minimal to support immediate single alternation pattern responding according to the rule view. This was true of some but not all conditions of initial training according to the memory view. Contrary to the rule view and consistent with the memory view, immediate single alternation pattern responding occurred following some types of minimal initial training but not others. Finally, we discussed the possibility that only the mechanisms within the memory view can similarly deal with all varieties of reward schedule phenomena which have previously been attributed to different, separate theories.