Classifying Neural Signals Related to Action Perception

Activity from individual neurons in primary motor cortex (M1) and ventral premotor cortex (PMv) can be modeled as a linear function of the direction of arm movement. If this signal is generated, it represents an extrinsic construct built upon various parallel coordinate transformations from the areas of the brain associated with vision. Previous work suggests that the PMv is a unique area in which to investigate how these visual signals are transformed into motor commands. The purpose of this study was to evaluate evidence of this visuospatial transformation in PMv and M1. In the first set of experiments, we recorded neuronal activity in M1 and PMv while each monkey reached to 14 targets represented in a 3D virtual environment. The monkeys hand was hidden from view and was represented on the display by a cursor. Goal position was represented with a spherical target. Across experiments, identical hand movements were performed with varying views of the task. Each view dissociated one of three putative coordinate frames: hand-centered velocity, cursor- centered target location, and displayed motion. In the second set of experiments, each monkey passively observed visually congruent replays of the trials from the first set of experiments. This paradigm was used to evaluate neuronal response in an action-related context without the monkey having the intention to move. In the third set of experiments, each monkey passively observed object motion on the display. This paradigm was used to evaluate neuronal response to a dynamic visual stimulus without action context. Results were as follows: 1) During active reaching, neural activity corresponded to the three coordinate systems in distinct anatomical locations, each with different latencies. This suggests a systematic substrate for visuospatial transformation. 2) Activity from a subpopulation of units located in M1 and PMv corresponded to hand velocity during active movement only. This implies the presence of a motor command-related signal. 3) Activity from a subpopulation of units located in anterior M1 and PMv corresponded to cursor-centered target location during action context, with or without the intention to move. This suggests the presence of a signal related to the visualized goal of a learned action. 4) Activity from a subpopulation of units located in anterior PMv corresponded to displayed motion of a stimulus in any context. This suggests a signal broadly related to sensory feedback