Value-based control over memory encoding and search

Memory is an adaptive cognitive function that enables individuals to use past experiences to guide future behavior. Prior research has demonstrated that goal-relevant, or rewarded, items are remembered more effectively, but the underlying mechanisms—ranging from automatic neuromodulation to strategic elaboration—remain debated. In this study, we investigated how different reward structures influence the temporal organization of free recall in mixed-item lists. Rewards were administered either during item encoding or were contingent on successful later recall. Our behavioral results reveal a striking mid-list primacy effect: when rewards are contingent on later recall, participants disproportionately initiate search at the onset of rewarded subsequences, effectively treating reward shifts as privileged access points. To account for these findings, we developed a novel hierarchical Bayesian implementation of the context maintenance and retrieval (CMR) model of free recall. By fitting the model to individual recall sequences, we adjudicated between competing accounts of memory enhancement. Our results demonstrate that motivated memory relies on two key mechanisms: the strengthening of context-to-item associations and the strategic creation of accessible event boundaries. These findings suggest the primacy effect involves the strategic reinstatement of a privileged context rather than solely the enhanced encoding of early list items. With regard to executive control and event segmentation theories, this suggests that participants exercise fine-grained control over their mental context, perhaps through varying mnemonic techniques. We propose that motivated memory is not merely a product of enhanced associative binding, but also involves active, structured (re)organization of experience.