Inference guides retroactive effects of arousal on memory

Aversive category conditioning has been shown to retroactively increase recognition of categorically related items, possibly via arousal-mediated reactivation. Schemas and incongruent, one-shot associations have been shown to guide reactivation, though in separate paradigms and through distinct neural pathways. Here, we designed two experiments to test the relationship between the contingency structure of aversive learning and retroactive memory enhancements. Both experiments included an initial encoding phase with items overlaid on scenes evoking a particular schema (e.g., the beach), with each item either congruent (e.g., a beach ball) or incongruent (e.g., a hair dryer) with that schema. In Experiment 1, the schema alone was subsequently associated with an aversive shock. Memory results showed improved recognition for items congruent, but not incongruent, with the schema that was later conditioned. In Experiment 2, the schema was subsequently associated with shock – while maintaining one scene as safe. Memory results showed improved high-confidence recognition for all items, both congruent and incongruent, that had been associated with the schema that was later conditioned. Individual differences in how strongly participants generalized threat across the entire schema – that is, whether they infer the individual scenes or the schema broadly to be the source of the aversion – were then predictive of the memory effects from Experiment 1. Possible neural mechanisms supporting inferential reasoning and memory reactivation underlying these behavioral effects are discussed.