Variability in the quality of visual working memory

Fougnie, Daryl; Suchow, Jordan W.; Alvarez, George A. · 2012 · OpenAlex-citations

DOI: 10.1038/ncomms2237

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper challenges the standard assumption in cognitive science that the quality of visual working memory representations is fixed for a given individual, determined solely by the allocation of a limited mental commodity. Instead, the authors demonstrate that there is substantial variability in the precision of working memory within an individual, even across items in a single trial. This finding contradicts both slot-based and resource-based models, which assume that memory fidelity is constant for remembered items and varies only based on how much "resource" is assigned to them. To investigate this, the researchers employed a behavioral task where participants memorized the colors of 1, 3, or 5 dots and later reported the color of a cued item. They analyzed the distribution of color errors, comparing a standard "fixed-precision" model against a new "variable-precision" model. The variable-precision model, which assumes that memory precision follows a distribution (specifically a truncated normal) rather than a single fixed value, provided a significantly better fit to the data, as evidenced by lower Akaike Information Criterion values. The variability in precision increased with set size, indicating that memory load exacerbates this instability. The authors conducted several control experiments to rule out alternative explanations for this variability. They demonstrated that the effect was not due to perceptual noise, eye movements, or differences in the memorability of specific colors or locations. Crucially, an experiment allowing participants to choose which item to report showed that participants could identify their best-remembered item with near-perfect accuracy. This confirmed that the variability exists within a single trial and is cognitively accessible, rather than resulting from fluctuations in attention or arousal across trials. Furthermore, tests for tradeoffs between items revealed no correlation in precision between different items in the same display, suggesting that the degradation of memory quality is independent across items rather than resulting from uneven resource allocation. The significance of these findings lies in the proposal of a new framework for understanding working memory limitations. The authors argue that existing models are incomplete because they ignore the stability of stored information. They propose that working memory limitations arise from a stochastic process of degradation, potentially driven by cortical noise, which plays out independently for each memory representation. This implies that visual working memory is stochastic not only in its content but also in its quality, requiring future models to account for the probabilistic survival of memory traces rather than just their initial allocation.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• cognitive capacity variation