Dissociating models of visual working memory by reaction-time distribution analysis.

Area of Science:

• Cognitive Psychology
• Neuroscience
• Human Memory Research

Background:

• Ongoing debate exists regarding the representational nature of visual working memory (VWM), contrasting discrete-slot models with flexible-resource models.
• A critical, yet under-explored, factor in this debate is the processing speed during accuracy assessments of stored information in VWM.

Purpose of the Study:

• To investigate contrasting predictions from discrete-slot and flexible-resource VWM models concerning decision times in change detection tasks.
• To analyze reaction time (RT) distributions to elucidate the underlying mechanisms of information processing within VWM across varying set sizes.

Main Methods:

• Employed change detection tasks with set sizes ranging from 2 to 8 items.
• Estimated ex-Gaussian parameters (μ, σ, τ) from reaction time distributions to model processing speed.
• Correlated individual capacity limits with the set size at which Gaussian parameters plateaued.

Main Results:

• Gaussian components (μ, σ) increased from set sizes 2 to 4, then remained constant for sizes 6 and 8.
• The exponential component (τ) increased significantly at set sizes exceeding estimated VWM capacity.
• Decision time for in-memory items remained constant, while time for exceeding capacity items increased with set size.

Conclusions:

• Findings support the discrete-slot model, indicating constant decision times for items within VWM capacity.
• Processing time increases when set size surpasses VWM capacity, aligning with discrete storage limitations.
• The study provides evidence that VWM capacity correlates with the point at which processing speed stabilizes.