Effective visual working memory capacity: an emergent effect from the neural dynamics in an attractor network

Area of Science:

• Cognitive psychology
• Computational neuroscience
• Neuroscience

Background:

• Working memory capacity is crucial for cognition but its origins are debated.
• Visual saliency's role in working memory is a recent area of experimental investigation.
• Attractor networks model working memory by simulating sustained neural activity.

Purpose of the Study:

• Investigate the mechanisms of working memory capacity using a biophysically-realistic spiking neural network model.
• Incorporate experimental findings on visual saliency's effect on memory storage.
• Propose a new concept of effective working memory capacity (eWMC).

Main Methods:

• Developed a biophysically-realistic attractor network model with spiking neurons.
• Simulated working memory tasks incorporating visual saliency.
• Analyzed the interplay between encoding and maintenance processes.

Main Results:

• The model replicates the finding that visual saliency reduces the number of items stored in working memory.
• Visually salient items are retained at the expense of non-salient items.
• Working memory capacity depends on both item encoding and maintenance, influenced by neural competition and inhibition.

Conclusions:

• Working memory capacity is governed by encoding limitations and maintenance constraints.
• Visual saliency influences encoding by modulating neural excitation and competition.
• Introduced effective working memory capacity (eWMC) to describe task-dependent capacity limits.