Limb apraxia and active inference in the visuomotor pathways.

This study presents a unified neurocomputational account of apraxia within the visuomotor system using the active inference framework. In this framework, the brain encodes a generative model of the causes of sensory observations and continually reduces the surprise (or, more formally, variational free energy) associated with these observations. Our goal is to illustrate how the neuroanatomical and neurofunctional organization of the visuomotor system are emergent consequences of this minimization under the appropriate generative model, while apraxia arises from incorrect parameterizations of that model. Our active inference model of the visuomotor pathways provides a unifying perspective on key findings in the neuropsychological literature. By virtually simulating impairments in the model, we explain key aspects of five neuropsychological syndromes impacting cognitive aspects of motor behavior. This study therefore, provides a novel hypothesis about the neurocomputational basis of these pathologies and offers a formal quantitative approach to support clinical research.