Working Memory of Multi-Object Scenes in Primate Frontal Cortex

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Working memory enables complex reasoning, but the neural mechanisms for storing multiple objects are debated.
• Existing theories include discrete slots, dynamic switching, and weighted combinations of representations.

Purpose of the Study:

• To quantitatively test competing models of multi-object working memory against neural data.
• To determine the representational format used by the frontal cortex for simultaneous object storage.

Main Methods:

• Formalized theoretical hypotheses into quantitative predictions for neural activity.
• Recorded dense neural data from monkey dorsomedial frontal cortex and frontal eye field during a multi-object working memory task.
• Cross-validated model performance against neural recordings and behavioral data.

Main Results:

• A Gain model, representing population activity as weighted combinations of object responses, significantly outperformed Slot and Switching models.
• Trial-specific gain estimates correlated with behavioral errors and reaction times, reflecting memory fidelity.
• Findings were robust across independent datasets and varying spatial configurations.

Conclusions:

• The frontal cortex likely employs a weighted-sum representation for multi-object working memory.
• This study provides a rigorous framework for understanding neural representations in working memory.
• Identified a specific neural mechanism underlying the brain's ability to manage multiple items in memory.