Humans recognize biological motion visually, even from such a minimalistic setup as point-light displays. The Mirror Neuron Systems Theory (MNST) is thought to account for this ability. However, the MNSTa??s conceptual framework examines this type of motion perception as a top-down (vision-to-action) process. This view cannot explain how humans are capable of integrating that visual information that they perceive in the movements of others with the re-afferent kinesthetic information emerging as a bottom-up percept from their own movements. We address this new question here. A new paradigm and new analytics are introduced to extract statistics from data generated by wearable sensors registering natural movements from unconstrained behaviors. We ask the extent to which people are capable of discriminating their own motions from those of others. We recorded 16 subjects performing sports routines and walking. For each subject their veridical movements were recorded at 15 joints across the body (240Hz Polhemus-Liberty). The angular velocity was analyzed to extract signatures of noise from the peak joint-velocity in each segment and from the time to reach that value. In two independent sessions subjects decided ME vs. not-ME as they watched the animation of an avatar endowed with the actual physical movements and with their noisy variants in rotational space and in time. We found that subjects were accurate well above chance for the temporal-noise cases. In the spatial-noise cases ?? of the subjects performed below chance and needed more trials to begin responding correctly. Power laws linking velocity-dependent parameters were found at three levels: abstract (decision), physical (body motions) and hybrid (hand pointing the decision). Hand speeds while deciding were the most predictive and reliable of all three, while decision response times were the most variable. We discuss our results in the context of a newly proposed framework to redefine the mirror-neuron-systems theory.
展开▼
