The computational neurology of movement under active inference

Area of Science:

• Computational Neurology
• Theoretical Neurobiology
• Clinical Neurology

Background:

• The brain functions as an active inference system, generating predictions about sensory data.
• Neural dynamics involve creating actions to match sensations with predictions and updating predictions when mismatches occur.

Purpose of the Study:

• To propose a computational neurology of movement integrating theoretical and clinical perspectives.
• To simulate neurological conditions and analyze motor control deficits using a computational model.

Main Methods:

• Framing brain function as active inference with a generative model.
• Simulating an upper limb coordination task to examine motor system behavior.
• Introducing computational lesions to model neurological deficits.

Main Results:

• Tendon reflexes naturally emerge from the proposed generative model.
• Simulated perturbations induce hyperreflexia and pendular reflexes, mimicking corticospinal and cerebellar lesions.
• Computational lesions replicate hypokinesia and coordination deficits.

Conclusions:

• The active inference framework provides a unified model for understanding normal and disordered movement.
• This approach offers a novel perspective on classic neurological dichotomies and motor control architecture.