Adaptation in spiking neurons based on the noise shaping neural coding hypothesis

Abstract

Shin, Koch and Douglas [Shin, J., Koch, C., & Douglas, R. (1999). Adaptive neural coding dependent on the time-varying statistics of the somatic input current. Neural Computation, 11, 1983-2003] proposed an adaptive neural coding model that makes spiking neurons adapt its input/output relation to the stimulus statistics. In a surprisingly precise manner, the adaptive neural coding model has been supported by recent experiments. However, the previous report has two problems: (a) although the adaptive neural coding model was developed based on the noise shaping neural coding hypothesis, their connection was not explained clearly in the previous report; and (b) the previous model did not suggest a biologically plausible method to estimate the stimulus mean and variance from spike-evoked intracellular calcium concentration. In this paper, I present how the noise shaping neural coding hypothesis produced such a precise model without any available experimental data at that time. Moreover, I propose a computational model for a biologically plausible signal statistics extraction from spike-evoked intracellular calcium concentration. An asymmetry in contrast adaptation time between increasing and decreasing variance, observed in biological experiments, is explained using the signal statistics extraction method. In addition, a new perspective on the relationship between the spike train of spiking neurons and EEG (or local field potential (LFP)) is suggested based on the noise shaping neural coding hypothesis.