Development of Target Reaching Gesture Map in the Cortex and Its Relation to the Motor Map: A Simulation Study

摘要

The motor maps in the cortex are topologically organized, just like the sensory cortical maps, where nearby locations in the map represent behaviors of similar kinds. However, there is not much research on how such motor maps are formed. In this paper, as a first step in this direction, we developed a target reaching gesture map using a selforganizing map model of cortical development (the GCAL model, a simplified yet enhanced version of the LISSOM model). The inputs were target reaching behavior of a two-joint arm on a 2D plane (2 DOF), encoded as a time-lapse image where time was encoded as the pixel intensity. For training, 20,000 random arm movements were generated where each arm movement started at a random initial location and moved toward one of 24 predefined target locations. The resulting gesture map showed global topological order where nearby neurons represented gestures toward nearby target locations, comparable to the motor map reported in the experimental literature. Although our simulations were based on a sensory cortical map development framework, the results suggest that it could be easily adapted to transition into motor map development. Our work is an important first step toward a fully motor-based motor map development (e.g. using proprioceptive input), and we expect the findings reported in this paper to help us better understand the general nature of cortical map development, not just for the sensory but also for the motor modality.