Interactions between visual working memory representations

Bae, Gi-Yeul; Luck, Steven J. · 2017 · Crossref

DOI: 10.3758/s13414-017-1404-8

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 investigates whether visual working memory (WM) representations of distinct objects are maintained independently or interact with one another. While many WM models assume items are stored as independent units, prior evidence suggests perceptual grouping and similarity influence performance. To isolate interactions between memory representations from perceptual confounds, the authors used a sequential presentation paradigm where observers reproduced two orientations after a delay. This design ensured that any interaction affecting the first item’s memory must occur within WM, not during initial perception. The researchers tested four theoretical models: independent representations, grouping/chunking, ensemble representation, and relational representation. The relational model predicted that similar orientations would repel each other (coded relationally to a reference point), while dissimilar orientations would attract. The study also examined how attentional priority modulates these interactions. In Experiment 1, 16 participants viewed two sequentially presented teardrop shapes with varying orientation differences (22.5° to 157.5°) and reported each orientation using a mouse-controlled response ring. The order of report was randomized to control for recency effects. Results demonstrated that WM representations are not independent but interact based on similarity and attention. When orientations were similar (<90° apart), reported values were repelled away from each other. When orientations were dissimilar (>90° apart), they were attracted toward each other. Crucially, these biases affected the first-presented item, confirming the interaction occurred in memory rather than perception. Furthermore, the bias was stronger for the second target than the first, suggesting the first item served as a reference point. Subsequent experiments (summarized in the abstract and introduction) revealed that attentional priority modulates these interactions: high-priority items were largely protected from distortion by low-priority items, whereas low-priority items were strongly influenced by high-priority ones. Additionally, errors in reporting the two items were positively correlated under certain conditions, supporting the grouping/chunking model’s prediction that items can be merged into a single complex representation. These findings challenge the assumption of independence in visual working memory. The results support a relational representation model where items influence each other based on similarity, with attention serving as a protective mechanism for high-priority information. This implies that WM is not a static store of independent slots but a dynamic system where representations are coded relative to one another and subject to attentional weighting. The study provides critical evidence for understanding the neural and cognitive mechanisms underlying how multiple objects are maintained in memory.

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.

• working memory