Similarity-based clusters are representational units of visual working memory.

Area of Science:

• Cognitive Psychology
• Neuroscience
• Human Perception

Background:

• Visual working memory (VWM) is crucial for cognitive tasks.
• The organizational principles of VWM representations are not fully understood.
• Feature similarity's role in VWM organization requires further investigation.

Purpose of the Study:

• To determine if feature similarity-based clustering enhances VWM representational quality.
• To test the hypothesis that fewer clusters improve recall precision due to reduced memory load.
• To identify if clustering acts as a representational unit in VWM.

Main Methods:

• Six experiments involved participants remembering stimulus orientations or colors.
• Recall bias and precision were measured to assess clustering and representational quality.
• Stimulus similarity and number of items/clusters were manipulated across experiments.

Main Results:

• Consistent evidence showed similar items forming clusters.
• Recall precision increased as the number of clusters decreased, irrespective of item quantity.
• Clustering effects were observed with color and orientation stimuli but not with task-irrelevant features.
• Response errors correlated within clusters, not across them.

Conclusions:

• Similar items are organized into clusters within VWM.
• Reducing the number of clusters enhances individual item recall precision.
• Feature similarity-driven clustering represents a fundamental organizational principle in VWM.