Abstract: Memory-guided value-based decision-making refers to situations in which, when choosing among multiple options, individuals retrieve option-related episodic memories. Such retrieval shapes value estimation and, in turn, drives the decision process. Current work highlights a central role of episodic memory in value-based decision-making and addresses three core questions. (1) How distinct properties of episodic memory influence decision-making: flexible recombinative encoding, strategic selective retrieval, and memory decay and forgetting. (2) Computational modeling frameworks for episodic influences on value-based decisions: model-free reinforcement learning (MF–RL), which emphasizes habits and incremental learning; model-based reinforcement learning (MB–RL), which computes predictions by simulating an internal model of environmental structure; and episodic reinforcement learning (ERL), which posits that decisions are shaped by “reminding” and sampling a single past episodic event. (3) Neural circuit accounts of episodic memory-guided value-based decisions suggest that the core neural network underlying this process is centered on the interaction between the medial temporal lobe (MTL, including the hippocampus and surrounding structures) and the orbital and medial prefrontal cortex (OMPFC), with the striatum and basolateral amygdala contributing in complementary ways. Future research should focus on the dynamics and neural signatures of memory-based valuation, and on the cognitive and neural mechanisms by which false memories bias value-based decisions, including their similarities to and differences from true memories.

Key words: episodic memory, value-based decisions, underlying mechanism, cognitive model, neural circuits