A biophysical model of multiple-item working memory: a computational and neuroimaging study

Area of Science:

• Neuroscience
• Computational Neuroscience
• Cognitive Science

Background:

• Biophysically based computational models explain single-item working memory maintenance via persistent neural activity.
• Existing models do not fully account for the human capacity to store multiple items in working memory.

Purpose of the Study:

• To extend computational models of working memory to accommodate multiple items.
• To investigate the role of developmental cellular mechanisms in enhancing working memory capacity.

Main Methods:

• Implemented cellular mechanisms associated with working memory development, including increased synaptic strength and improved neural response contrast/specificity.
• Developed a computational neural network model capable of storing multiple items through sustained neural activity.
• Utilized functional magnetic resonance imaging (fMRI) to compare model predictions with human brain activity.

Main Results:

• The computational model successfully demonstrated the storage of multiple items via sustained neural activity.
• The model's predicted information-activity curve aligned with empirical data from the human posterior parietal cortex.
• Findings support the role of developmental mechanisms in scaling working memory capacity.

Conclusions:

• Developmental cellular mechanisms are sufficient to enable neural networks to store multiple working memory items.
• The model provides a biologically plausible explanation for human working memory capacity limitations.
• The study validates computational approaches for understanding brain function in working memory.