Analysis of the rat nerve activity with an autoregressive parametric model

Abstract

Studies of vibrissal tactile discrimination show that rats can use their vibrissae to distinguish objects differing only in surface texture. Here we propose to use an autoregressive (AR) parametric model to analyze the electrical nerve discharge when vibrissae are sweeping surfaces differing only in roughness. The results were compared with the control situation: vibrissae sweeping the air. The AR model order was evaluated in both theoretical and practical ways using the Akaike information criterion (AIC) and calculating the delta value from the experimental data. The value obtained was five and the parameters a1 and a2 show more clearly the differences between control and surface measurements. The results show that the AR model is a good tool to use in this procedure.