"Value" emerges from imperfect memory

Whereas computational models of value-based decision-making generally assume that past rewards are perfectly remembered, biological brains regularly forget, fail to encode, or misremember past events. Here, we ask how realistic memory retrieval errors would affect decision-making. We build a simple decision-making model that systematically misremembers the timing of past rewards but performs no other value computations. We call these agents "Imperfect Memory Programs" (IMPs) and their single free parameter optimizes the trade-off between the magnitude of error and the complexity of imperfect recall. Surprisingly, we found that IMPs perform better than a simple agent with perfect memory in multiple classic decision-making tasks. IMPs also generated multiple behavioral signatures of value-based decision-making without ever calculating value. These results suggest that mnemonic errors (1) can improve, rather than impair decision-making, and (2) provide a plausible alternative explanation for some behavioral correlates of "value".