Noisy Memory Generates Value in Changing Environments
Bibliographic record
Abstract
Experimental data suggest that episodic memory is involved in sequential value-based decision-making. By contrast, standard computational models of decision-making assume that prior reward outcomes are integrated into subjective values rather than remembered discretely. Previous work developed a minimal computational framework for sequential value-based decision-making that is based on noisy sampling of episodic memories, rather than calculating value. We called these agents "Imperfect Memory Programs" (IMPs) and showed how their single free parameter optimizes the trade-off between the magnitude of error and the complexity of imperfect recall. Here, we develop biologically plausible approximations to the IMPs with lossy agents (LIMPs) that maintain only 1 bit of reward memory for binary outcomes but fail to encode rewards with some probability. Both IMPs and LIMPs perform similarly to or better than a simple agent with perfect memory in multiple classic decision-making tasks and generate phenomenology that resembles value-based computations. We find that allowing different encoding probabilities for rewards and omissions improves performance further and allows to trade-off matching versus maximizing behavior, as well as flexible versus stable performance. Together, these results suggest that episodic agents can approximate value-based agents through capitalizing on realistic encoding and/or sampling noise. This suggests that mnemonic errors (1) can improve, rather than impair decision-making and (2) provide a plausible alternative explanation for some behavioral correlates of "value".
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.009 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.002 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".