Effective connectivity of working memory performance: a DCM study of MEG data

Area of Science:

• Cognitive Neuroscience
• Neuroimaging
• Systems Neuroscience

Background:

• Visual working memory (WM) involves a network of frontal, parietal, and visual regions.
• The precise interaction dynamics, specifically feedforward versus feedback connections, remain unclear.
• Understanding these dynamics is crucial for debates on the roles of different brain regions in WM.

Purpose of the Study:

• To investigate network activity supporting WM using MEG data.
• To differentiate between feedforward and feedback connections in WM maintenance.
• To link network dynamics to specific types of behavioral errors in a WM task.

Main Methods:

• Acquired magnetoencephalography (MEG) data during a multi-item delayed estimation WM task.
• Employed computational modeling to classify behavioral responses (high accuracy, low accuracy, swap trials).
• Utilized dynamic causal modeling (DCM) to analyze effective connectivity within the WM network.

Main Results:

• Identified behaviorally dependent changes in effective connectivity between frontoparietal and visual areas.
• Observed disrupted signals in frontoparietal and frontooccipital networks correlating with error types.
• Found specific patterns of disrupted feedback and feedforward signals for low accuracy and swap errors during WM maintenance.

Conclusions:

• Results support a distributed model of WM, highlighting the role of visual regions in storage.
• Network configuration changes significantly impact memory-guided behavior.
• Specific disruptions in feedforward and feedback signaling underlie different WM error types.