Representing Context and Priority in Working Memory.

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience

Background:

• Working memory (WM) prioritization is crucial for cognitive control.
• Neural representations may differ based on the nature of information (e.g., stable context vs. dynamic priority).

Purpose of the Study:

• To investigate the neural mechanisms underlying working memory prioritization using computational modeling and empirical data.
• To differentiate how trial context and priority are neurally encoded.

Main Methods:

• Simulated the double serial retrocuing task using recurrent neural networks.
• Applied principal component analysis for visualization of neural dynamics.
• Utilized dimensionality reduction and multiclass decoding on fMRI and EEG data.

Main Results:

• Recurrent neural networks distinguished between stable ordinal context (orthogonal subspaces) and dynamic priority (strata within subspaces).
• Empirical fMRI and EEG data showed distinct neural representations for priority and context in the dorsal visual stream.
• Behavioral performance correlated with priority coding variability, but not context coding variability.

Conclusions:

• Neural mechanisms for representing stable context and dynamic priority in working memory differ.
• These findings provide insights into the neural basis of cognitive control and information prioritization in the brain.