Convergent neural dynamical systems for task control in artificial networks and human brains

Area of Science:

• Cognitive neuroscience
• Computational neuroscience
• Neuroimaging analysis

Background:

• Task switching is crucial for adaptive cognition but its underlying mechanisms are debated.
• Key debate: task switching influenced by interference or active preparation?

Purpose of the Study:

• To investigate the neural mechanisms of task switching.
• To reconcile the roles of interference and active preparation in cognitive control.

Main Methods:

• Modeled task representations in human electroencephalographic (EEG) recordings.
• Applied linear dynamical systems and control theoretic analyses.
• Compared neural data with task-optimized neural networks.

Main Results:

• Identified neural signatures of active preparation during task switching.
• Demonstrated how these signatures reconcile interference and control.
• Showcased a novel paradigm for analyzing human neuroimaging data.

Conclusions:

• Neural dynamics reveal active preparation in task switching.
• This reconciles long-standing debates on cognitive control mechanisms.
• Offers a new framework for neuroimaging analysis of cognitive control.