Multimodal Goal Representations and Feedback in Hierarchical Motor Control

摘要: The capabilities of human motor behavior build on the integration multiple sensory modalities in goal representation and feedback processing. Here, we present a hierarchical neural network model control to simulate these capabilities, based SURE REACH model. is able integrate visual proprioceptive representations, but, by now, relies only ongoing movements. extend that processes both, feedback. In simulated reaching experiments demonstrate considerably enhances accuracy original controller. Moreover, ability combine or adjust task-specific constraints not affected. Finally, discuss results, propose further enhancements, outline model's relevance for other domains cognition. I. INTRODUCTION Human sensorimotor truely amazing. An asto- nishing spectrum can guide different types be processed. Besides proprioception vision, various sensations organized skills, both as modality. Furthermore, adaptive flexible degree unrivaled any artificial system. Even clumsiest among us readily novel situations, example, when moving with heavy winter clothes opening door while balancing stack folders holding cup coffee. How are achieved brain yet well understood. may account vision representations processing task-dependent control.