Reward selectively preserves relational structure in memory

Abstract

Memories tend to form networks that reflect relationships between experiences. These networks, or cognitive graphs, have been used to study how structural features of memory can bias decisions — analogous to how reward shapes choices by prioritizing reward-associated items in memory. However, how reward interacts with the structure of existing cognitive graphs to influence subsequent value-based decisions remains unclear. Across multiple experiments, participants learned a community-structured cognitive graph and made value-based decisions, with some experiments including graph reconstruction and or/reward learning. Results indicated a limited retroactive spread of reward, contingent on the presence of a retrievable representation of either the entire cognitive graph or reward-item associations. Reward also shaped memory representations by preserving the relational structure of directly associated items to guide value-based decisions. These findings suggest that reward acts as an organizing force within memory representations — selectively preserving relevant associations to support adaptive, value-based decision making, and most prominently when these associations are retrievable.