A Novel Perspective into the Neuronal Encoding along the Retinal Pathway Employing Time- Frequency Transformation: Part Ii — for Color

This paper analyses how color information gets encoded along the retinal pathway in the time-frequency domain (TFD). A multinomial (multivariate polynomial) encoding for chromatic information is established based on the spatiotemporal spike trains corresponding to input stimuli varying in hue & saturation. Simulation tools are used to generate the spike trains from the retinal layers. The Gabor time-frequency transformation presented in the companion paper part I [1] is employed here for analyzing the dominant harmonic variations corresponding to the color stimuli. For the set of color input stimuli considered, a bivariate polynomial encoding is observed with the order of the bivariate polynomial and its coefficients encoding the variations in amplitude of the dominant harmonics. Analysis is pursued along the similar lines presented in part I. The simulation results pertaining to encoding for color variations are presented. This paper suggests a new mathematical formulation called the generalized “stochastic multinomial” encoding, for the retinal pathway in general.