Evaluation of Neural Encoding Schemes in Multifiber Activity Using Information Theory: An Application to the Vibrissal System

FARFÁN FD; ALBARRACÍN AL; FELICE CJ

Congreso; II Reunion Conjunta de Neurociencias; 2010

Sociedad Argentina de Investigación en Neurociencias

Currently, the vibrissal texture encoding is studied through the neural activity of one neuron(or a few neurons) of trigeminal ganglion and/or cortex. Here we analyzed various neural encoding schemes which could be related to texture sensing in anesthetized rats during active touch. The multifiber discharge from a deep vibrissal nerve during the sweeps different materials with similar degrees of roughness and sandpapers of different grain size were analyzed. Three neural encoding schemes were proposed: RMS scheme, cumulative event count (CEC) and inter-event time (IET). These events were detected by using an event detection algorithm based on multiscale decomposition of the signal. All encoding schemes were evaluated quantitatively through the Shannon s mutual information formula. The CEC scheme provides more information than RMS scheme. By using the IET scheme, the information values were higher than those obtained with RMS and CEC schemes. In conclusion, information of sandpaper roughness was found to be better encoded by IET scheme, while for surfaces of different materials, CEC scheme was optimal.Finally,in this study we have demonstrated (and quantified) the existence of two biologically plausible neural coding schemes based on temporal patterns of events.