Abstract
Previous models of touch have linked skin mechanics to neural firing rate, neural dynamics to action potential elicitation, and mechanoreceptor populations to psychophysical discrimination. However, no one model spans all levels. The objective of work herein is to build a multi-level, computational model of tactile neurons embedded in cutaneous skin, and then validate its predictions of skin surface deflection, single-afferent firing to indenter shift, and population response for sphere discrimination. The model includes a 3D finite element representation of the distal phalange with hyper- and visco-elastic mechanics. Distributed over its surface, a population of receptor models is comprised of bi-phasic functions to represent Merkel cells' transformation of stress/strain to membrane current and a leaky integrate-and-fire neuronal models to generate the timing of action potentials. After including neuronal noise, the predictions of two population encoding strategies (gradient sum and euclidean distance) are compared to psychophysical discrimination of spheres. Results indicate that predicted skin surface deflection matches Srinivasan's observations for 50 micron and 3.17 mm diameter cylinders and single-afferent responses achieve R(2) = 0.81 when compared to Johnson's recordings. Discrimination results correlate with Goodwin's experiments, whereby 287 and 365 m(-1) spheres are more discriminable than 287 and 296 m(-1).