Decoding tactile sensation: Multiple regression analysis of monkey fingertip afferent mechanoreceptor population responses

How complex tactile sensations are encoded by populations of afferent mechanoreceptors is currently not well understood. While much is known about how individual afferents respond to prescribed stimuli, their behavior as a population distributed across the fingertip has not been well described. In this study, tactile afferent mechanoreceptors in monkey fingertips were mechanically stimulated, using a flat disc shaped probe, with several magnitudes of normal force (1.8, 2.2 and 2.5 N) and torque (2.0 and 3.5 mNm), in clockwise and anticlockwise directions. Afferent nerve responses were acquired from 58 slowly-adapting (SA) type-I and 25 fast-adapting (FA) type-I isolated single cutaneous mechanoreceptive afferents, recorded from the median nerve. At 10 ms time intervals after the application of torque begins, a multiple regression model was trained and evaluated to estimate the magnitude of the applied normal force and torque. Averaged results over the 200 ms period after the torque reaches its maximum indicate that SA-I and FA-I afferents can both estimate the applied torque value. FA-I afferents gave the lowest estimation error mean and standard deviation of -0.051 ± 0.334 mNm for a target torque of 2.0 mNm, and 0.003 ± 0.414 mNm for a target torque of 3.5 mNm. However, while SA-I afferents could estimate normal force well, there was no significant difference (ANOVA, p=0.173) in the FA-I estimates of normal force, as this force had already been held constant for one second before the torque loading phase under analysis began.