Vibrations transmitted throughout the hand and arm during touch contact play a central role in haptic science and engineering but are challenging to model or experimentally characterize. Here, we present SkinSource, a data-driven toolbox for predicting skin vibrations across the upper limb in response to user-specified input forces. The toolbox leverages impulse response measurements that encode the physics of vibration transmission across the hands and arms of four participants and provides software tools for analyzing the predicted skin responses. We show that the SkinSource predictions closely match experimental measurements and confirm the underlying assumption of linear vibration transmission in the skin. We also demonstrate through several usage examples how SkinSource can act as a versatile computational platform for haptic research applications, such as characterizing vibrotactile transmission in the skin, engineering haptic interfaces, and investigating touch perception.

ABSTRACT Touching an object elicits skin oscillations that are biomechanically transmitted throughout the hand, driving responses in thousands of tactile receptors, including numerous exquisitely sensitive Pacinian corpuscles (PCs). Accepted descriptions of PC functionality characterize their response properties as highly stereotyped, based on experimental data gathered when stimuli are applied near the receptor. However, during natural touch, spiking activity in the majority of PCs is evoked by transmitted skin oscillations that are modified by biomechanical filtering. This filtering mechanism, stemming from dispersive wave dynamics in the skin, bears some similarity to the pre-neuronal filtering of auditory signals by the basilar membrane, a mechanical process that is instrumental to perception. Thus, we sought to clarify how skin biomechanics might influence tactile information encoding in the periphery. We used vibrometry imaging and computational neural experiments to examine the influence of biomechanical filtering on neural activity in whole-hand PC populations. We observed complex, location- and frequency-dependent patterns of filtering that were shaped by tissue mechanics and hand morphology. This source of biomechanical modulation diversified PC population spiking activity and enhanced tactile information encoding efficiency. These findings indicate that biomechanics furnishes a pre-neuronal mechanism that facilitates efficient tactile encoding and processing.

Manual touch interactions elicit widespread skin vibrations that excite spiking responses in tactile neurons distributed throughout the hand. The spatiotemporal structure of these population responses is not yet fully understood. Here, we evaluate how touch information is encoded in the spatiotemporal organization of simulated Pacinian corpuscle neuron (PC) population responses when driven by a vibrometry dataset of whole-hand skin motion during commonly performed gestures. We assess the amount of information preserved in these peripheral population responses at various spatiotemporal scales using several non-parametric classification methods. We find that retaining the spatial structure of the whole-hand population responses is important for encoding touch gestures while conserving the temporal structure becomes more consequential for gesture representation in the responses of PCs located in the palm. In addition, preserving spatial structure is more beneficial for capturing gestures involving single rather than multiple digits. This work contributes to further understanding the sense of touch by introducing novel measurement-driven computational methods for analyzing the population-level neural representations of natural touch gestures over multiple spatiotemporal scales.

The sense of touch can convey semantic and emotional information in social or computer-mediated interactions. Touch plays an essential role in communication with individuals affected by multiple sensory loss, many of whom use modes of touch communication that can be broadly described as tactile sign languages. Few technologies exist today to support such interactions. Here, we present a smart bracelet for facilitating tactile communication and interaction. The smart bracelet captures and analyzes vibrations that are elicited in the skin via touch gestures performed on the hand. We demonstrate the utility of this system for supporting communication via the Deafblind Manual alphabet, which is a tactile sign language. This smart bracelet can classify signed letters with greater than 90 % per-letter accuracy. These results show how existing modes of tactile communication can be integrated with information technologies. This work may furnish new paradigms for human-computer interaction via self- and interpersonal-touch contact.