Specificity quantification of texture discrimination processes in vibrissal system

LUCIANNA, FACUNDO; FARFÁN, FERNANDO DANIEL; ALBARRACÍN, ANA LIA; RUIZ, GABRIEL; FELICE, CARMELO JOSÉ

Huerta Grande, Córdoba

Congreso; XXIX CONGRESO ANUAL DE LA SOCIEDAD ARGENTINA DE INVESTIGACION EN NEUROCIENCIAS; 2014

SOCIEDAD ARGENTINA DE INVESTIGACION EN NEUROCIENCIAS

Rodents can discriminate finely textured objects through active rhythmic movement of their whiskers. The tactile discrimination process originates in the vibrissa primary afferents, in that, if two stimulus evoke distinct responses, probably less sweeps are necessary to recognize their origin. In this work, such process is called ?specificity? and we have quantified it as the number of sweeps required to discriminate friction situations. For this, we analyzed the multifiber discharge from a deep vibrissal nerve when the vibrissa sweeps materials (wood, metal, acrylic, sandpaper) having different textures. We polished these surfaces with sandpaper (P1000) to obtain close degrees of roughness and we induced vibrissal movement with two-branch facial nerve stimulation. The afferent discharges were characterized according to their spectral content, amplitude and autoregressive modeling. These features were used to train two-level perceptrons with the back-propagation algorithm. The neural networks were subjected to processes of training, which consisted in presenting the features belonging to 25, 20, 15, 10 and 5 sweeps respectively. Subsequently the features of new sweeps were subjected to classification processes. Preliminary results have shown that neural networks were able to discriminate friction situations by simply observing the features of 5 sweeps. In this way, we quantified the specificity of peripheral vibrisal system through misclassification.