The neural basis of stereognosis and its application to neuroprosthetics

PROJECT SUMMARY When we interact with objects using our hands, we are able to easily distinguish between our keys and our phone, and can do so even without visual cues. This ability to sense the three-dimensional structure of an object through haptic exploration alone is termed stereognosis and relies on the integration of two distinct streams of sensory information: tactile signals from the fingertips contacting the object relay information about local features (e.g., edge location, curvature, texture), and proprioceptive information from the muscles relay information about the overall shape and size of the object. While the integration of tactile and proprioceptive signals has been observed at higher order stages of somatosensory processing (Brodmann’s area 2, secondary somatosensory cortex, parietal ventral area), the neural mechanisms underlying this integration remain largely unknown. Given that the hand is a highly deformable sensory sheet, there are likely unknown neural processing mechanisms unique to the somatosensory system that underlie this integration process and a new framework will be necessary to understand how stereognosis can arise. The goal of the present study is to better understand the principles of multimodal integration that give rise to stereognosis by characterizing the responses of multimodal neurons in area 2 during grasping (Aim 1) and by developing computational models of how tactile and proprioceptive signals are integrated to give rise to object representations that are independent of how objects are grasped (Aim 2). We anticipate that the computational models will inform the interpretation of our neurophysiological results and deep novel insights into the neural mechanisms of stereognosis. Not only will the results of the study contribute to basic science, but they will also have implications for translational research and clinical applications. Our study of neural coding along the primate neuraxis informs our work toward more dexterous brain-controlled prostheses, which involves inferring motor intent but also restoring sensory feedback. Indeed, our ability to dexterously interact with objects, even without vision, depends on neural representations of objects. We anticipate that a deeper understanding of object representations in higher order somatosensory cortices, including area 2, will allow us to leverage these representations to improve the informativeness of intracortical microstimulation-based somatosensory feedback, thereby conferring greater dexterity to the brain-controlled bionic hands.