Constructing value on the fly from episodic memory supports flexible decision-making

Value-based decisions often require people to flexibly draw on and combine information from past experiences to evaluate novel stimuli and adapt to changing demands. Although numerous studies have highlighted the involvement of episodic memory in value-based decision-making, whether and how a common episodic mechanism can support diverse decision-making scenarios remains poorly understood. Here, we introduce GENESIS-Value, a computational framework that constructs value on the fly by retrieving and integrating episodic memories, rather than relying on stable, precomputed value representations. The model combines generative stimulus representations with a flexible query-key-value retrieval mechanism, allowing the same memory store to support different valuation and choice strategies. Across five simulations, we demonstrate that GENESIS-Value can use episodic sampling to reconstruct the values of experienced stimuli, generalize to novel stimuli based on shared features, estimate feature values, guide both individual and sequential choices, and select among multiple offers involving different rewards and levels of risk. We further show that retrieval can be guided by stimulus features, temporal proximity, reminders, and reward outcomes, as well as combinations of these factors, enabling adaptation to different decision demands without requiring separate value functions. The model also captures qualitative behavioral signatures reported in empirical studies of episodic decision-making. Together, these findings establish a unified computational account of how episodic retrieval can support diverse forms of value-based inference and choice. Rather than replacing stable value representations, episodic retrieval may provide a complementary mechanism that allows decision-makers to flexibly address a variety of choice scenarios by reconstructing the value of available options on the fly.

Authors

Institutions

Publication Details

Journal
bioRxiv (Cold Spring Harbor Laboratory)
Published
2026-11-19
DOI
https://doi.org/10.64898/2026.11.18.767634
Primary Topic
Decision-Making and Behavioral Economics
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Constructing value on the fly from episodic memory supports flexible decision-making

Giovanni Pezzulo, Marco D’Alessandro
bioRxiv (Cold Spring Harbor Laboratory)
Decision-Making and Behavioral Economics
preprint

Constructing value on the fly from episodic memory supports flexible decision-making

Giovanni Pezzulo, Marco D’Alessandro
preprint en

Abstract

Value-based decisions often require people to flexibly draw on and combine information from past experiences to evaluate novel stimuli and adapt to changing demands. Although numerous studies have highlighted the involvement of episodic memory in value-based decision-making, whether and how a common episodic mechanism can support diverse decision-making scenarios remains poorly understood. Here, we introduce GENESIS-Value, a computational framework that constructs value on the fly by retrieving and integrating episodic memories, rather than relying on stable, precomputed value representations. The model combines generative stimulus representations with a flexible query-key-value retrieval mechanism, allowing the same memory store to support different valuation and choice strategies. Across five simulations, we demonstrate that GENESIS-Value can use episodic sampling to reconstruct the values of experienced stimuli, generalize to novel stimuli based on shared features, estimate feature values, guide both individual and sequential choices, and select among multiple offers involving different rewards and levels of risk. We further show that retrieval can be guided by stimulus features, temporal proximity, reminders, and reward outcomes, as well as combinations of these factors, enabling adaptation to different decision demands without requiring separate value functions. The model also captures qualitative behavioral signatures reported in empirical studies of episodic decision-making. Together, these findings establish a unified computational account of how episodic retrieval can support diverse forms of value-based inference and choice. Rather than replacing stable value representations, episodic retrieval may provide a complementary mechanism that allows decision-makers to flexibly address a variety of choice scenarios by reconstructing the value of available options on the fly.

bioRxiv (Cold Spring Harbor Laboratory)
National Research Council (IT)
Decision-Making and Behavioral Economics
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon