Abstract
In daily social activities, we often touch others in natural and intuitive ways to share thoughts and emotions—such as tapping to get one’s attention, shaking hands to express gratitude, or caressing to soothe one’s anxiety. Indeed, it has been widely reported that social affective touch is of critical importance for both human development and social connections. Nowadays, many efforts are attempting to replicate social touch communication using actuated haptic displays. However, it remains challenging to render the rich and subtle emotional expressions delivered by human touch. To achieve this goal, more needs to be understood regarding how touchers exactly adapt their contact strategies to various emotional meanings and social dynamics, and how such different physical skin contact elicits distinct sensory and perceptional responses. Specifically in this work, we addressed this gap by precisely quantifying the social touch delivery using 3D computer vision, analyzing emotional perception using psychophysics, and modeling the underlying peripheral neural coding using machine learning. First, we developed a high-resolution 3D tracking system utilizing a depth camera to capture physical skin contact delivered by bare hands. Time-series contact attributes were calculated including contact area, indentation depth, contact velocities, etc. The tracking system was systematically validated for measurement accuracy and its applicability in social touch scenarios. With this quantification capability, we then conducted psychophysical user studies and correlated perceptions of emotional valence and arousal with contact attributes. Social factors such as relationship status were also analyzed, where we found that while emotion recognition accuracies are similar across couples and strangers, valence and arousal ratings are significantly higher for couples. This discrepancy could be related to couples’ fine-tuning of their contact delivery. Next, to further investigate the neurophysiological mechanism of social touch, we conducted microneurography experiments to record single peripheral tactile neurons in response to human touch stimuli. Classification analyses revealed that SA-II and HFA afferent subtypes can effectively decode social touch expressions, with their most informative firing patterns identified accordingly. Our results also show that peripheral neurons decode social touch expressions at time scales 1-2 orders of magnitude longer than those needed for non-social touch. Overall, this work uncovers the details of social touch communication from the perspective of contact delivery, neural pathways, and resulting emotional perception. These insights contribute to future design of haptics that achieve high effectiveness and realism in mediated social touch communication.
