Response-independent outcome presentations dissociate stimulus and value based choice

Abstract

A stimulus that predicts the delivery of a specific food outcome can bias performance towards instrumental actions that earn that same outcome in a phenomenon known as specific Pavlovian-instrumental transfer (PIT). The precise mechanism by which the specific instrumental action is selected under these circumstances has remained elusive. The present set of experiments explored whether treatments that undermine the response-outcome (R-O) association also affect the expression of specific PIT. Consistent with previous work, in Experiment 1 we showed that specific PIT remains intact after an instrumental degradation treatment that attempted to undermine R-O associations. However, we additionally demonstrated that outcome-devaluation sensitivity also persisted after degradation, suggesting that R-O associations were impervious to the degradation treatment, and precluding any conclusions about the necessity of R-O associations for specific PIT expression. Nevertheless, given the two-lever two-outcome design of this experiment it is possible that R-O associations were indeed undermined by degradation and that the devaluation effect was driven by distinct, incidental Pavlovian lever-outcome associations. To nullify the obscuring effects of these incidental Pavlovian associations, we used a bidirectional lever for instrumental conditioning that could be pushed to the left or the right for distinct outcomes. In Experiment 2 we demonstrated that specific PIT could be observed on this bidirectional manipulandum whether the subjects were hungry or sated, consistent with the literature. The critical third Experiment used an identical design to Experiment 1 except that the two instrumental responses were made on the single bidirectional manipulanda. Here, specific PIT was intact after instrumental degradation and, crucially, we saw no evidence of outcome devaluation sensitivity in these same subjects, suggesting that the R-O associations were weakened or undermined by this treatment. We conclude that the expression of specific PIT is resistant to treatments that undermine R-O associations and disrupt value based choice, and discuss how these findings contribute to our understanding of the associative framework supporting behavioral control.