Reward prediction errors create event boundaries in memory

Abstract

We remember when things change. Particularly salient are experiences where there is a change in rewards, eliciting reward prediction errors (RPEs). This feature of memory may be useful because it can help us find greater rewards and avoid lesser ones in the future. How do RPEs influence our memory of those experiences? One idea is that this signal directly enhances the encoding of memory. Another, not mutually exclusive, idea is that the RPE signals a deeper change in the environment, and leads to the mnemonic separation of subsequent experiences from what came before, thereby creating a new latent context and a more separate memory trace. We tested this in four experiments in which participants learned to predict rewards associated with a series of images within visually-distinct \"rooms.\" High magnitude RPEs indicated a change in the underlying distribution of rewards. To test whether these large RPEs created a new latent context, we first assessed recognition priming for sequential pairs that contained or did not contain a high-RPE event, as well as out-of-sequence pairs (Exp. 1: n=27 & Exp. 2: n=83). We found evidence of recognition priming for both sequential pair types, including the pair with the high-RPE event, indicating that the high-RPE event is bound to its predecessor in memory. Given that high-RPE events are themselves preferentially remembered (Rouhani et al, 2018), we next tested recognition priming for pairs that had one item in between them (i.e. the pairs were either across a high-RPE event or not), where none of the tested items were high-RPE items (Exp. 3: n=85). Here, sequential pairs across a high-RPE no longer showed recognition priming whereas pairs within the same latent reward state did, providing initial evidence for an RPE-modulated event boundary. We then investigated whether RPE event boundaries disrupt temporal memory of those events (Exp. 4). After reward learning, we asked participants to order and estimate the distance between two events that had either included a high-RPE event between them, or not. We found (n=49) and replicated (n=77) worse sequence memory for events across a high-RPE event. Altogether, these findings demonstrate that high-RPE events are both more strongly encoded and act as event boundaries that interrupt the sequential integration of events. We captured these effects in a variant of the Context Maintenance and Retrieval model (CMR; Polyn, Norman & Kahana, 2009), modified to incorporate RPEs into the encoding process.