Dynamic Shifts of Limited Working Memory Resources in Human Vision

Bays, Paul M.; Husain, Masud · 2008 · OpenAlex-citations

DOI: 10.1126/science.1158023

archive: archived pipeline: cataloged verified

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

Summary

This study challenges the dominant "item-limit" model of visual working memory, which posits that humans can hold only a fixed number of items (typically three or four) in memory. Instead, Bays and Husain propose that visual memory capacity is a limited resource that is dynamically shared among all items in a scene. The precision with which an item is remembered depends on the proportion of resources allocated to it, a relationship governed by a simple power law. This resource allocation is flexible and biased by selective attention and upcoming eye movements. The researchers conducted experiments with 32 subjects who were asked to recall the location or orientation of visual items (colored squares or arrows) after a brief blanking period. In the primary experiment, subjects viewed displays containing one to six items. They either maintained fixation or made a saccade to one of the items before the display was blanked. Subjects then reported the direction of displacement or rotation for a single probed item. A second experiment involved subjects fixating on five items in sequence, with memory tested after the final saccade. Precision was quantified by fitting cumulative Gaussian distributions to response data, allowing the authors to measure the variability in stored representations. The results demonstrated that memory precision decreased gradually as the number of items increased, even from one to two items, contradicting the sharp drop-off predicted by fixed item-limit models. The data fit a power-law relationship ($P \propto R^{0.74}$), where precision ($P$) is a function of the resources ($R$) allocated per item. Crucially, the allocation of these resources was not uniform. Items targeted by upcoming saccades or cued by attention were remembered with significantly greater precision than non-targets. In the sequential fixation task, the majority of memory resources (56–61%) were allocated to the target of the next saccade rather than the currently fixated item. This indicates that high-resolution memory for a fixated item degrades as resources shift to the next target. These findings imply that visual working memory does not have a hard upper limit on the number of items stored. Instead, all items are stored with varying degrees of precision based on available resources. The study reconciles previous conflicting results by showing that "supra-threshold" change detection tasks (used in earlier studies) mask the gradual decline in precision, creating the illusion of a fixed capacity limit. The work establishes that attention and eye movements act as mechanisms for dynamically redistributing limited memory resources, prioritizing behaviorally relevant information while maintaining a degraded representation of other scene elements.

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.

• attention
• capacity resource theory