Directed forgetting and working-memory load reduce outcome-selective Pavlovian-to-instrumental transfer in humans

Abstract In outcome-selective Pavlovian-to-instrumental transfer (PIT), reward-related cues selectively energize instrumental responding for associated outcomes. In humans, this effect has been attributed both to automatic associative processes and to cognitive inferences about which rewards are available in the presence of a cue. The present research examined whether outcome-selective PIT is influenced by manipulations that discourage such inferences. Across three experiments using a stock-market scenario with monetary rewards, participants learned Pavlovian cue–outcome relations and instrumental response–outcome contingencies before a transfer test. Experiment 1 established an outcome-selective PIT effect, which was substantially larger among participants reporting strategic use of the cues. Experiment 2 showed that instructing participants to forget the Pavlovian relations after an ostensible software error significantly reduced this effect. Experiment 3 showed that a concurrent working-memory load during the transfer test also attenuated PIT. However, a small effect persisted under the combined manipulations. Together, these findings indicate that outcome-selective PIT is modulated by cognitive reward expectancies and by beliefs about cue relevance. The residual transfer is not readily explained by experimenter demand, which here worked against cue-based responding. Cue reactivity therefore appears to rest on reward expectations that persist even when they are false and have been explicitly disavowed.