Myopic control of neural dynamics

Area of Science:

• Neuroscience
• Control Theory
• Computational Neuroscience

Background:

• Current control schemes for neural systems do not adequately address their inherent variability and real-time dynamics.
• Effective manipulation requires controllers that adapt to the current neural state, independent of past trajectories.

Purpose of the Study:

• To propose a novel "myopic" controller within a nonlinear state-space feedback framework for manipulating neural dynamics.
• To demonstrate the controller's ability to transform one dynamical system into another, overcoming limitations of existing methods.

Main Methods:

• Formulation of a "myopic" controller using a novel variant of a model reference control cost.
• Implementation of a short-sighted control strategy targeting a single time step into the future.
• Application of the controller to two distinct neuroscience examples: cognitive processes and neurological disorders.

Main Results:

• The myopic controller successfully transformed a winner-take-all decision-making system into a neural integrator, probing the dynamics-behavior link.
• An unhealthy motor-like system with a beta-oscillation attractor was effectively controlled to function as a healthy motor system.

Conclusions:

• The proposed myopic controller offers a flexible and computationally efficient alternative for manipulating neural dynamics.
• This approach has broad utility in neuroscience research for understanding cognition and in clinical settings for treating neurological disorders.