This note discusses how information contained in a neural message is transmitted depending on two schemes of encoding: stochastic or deterministic. For the first case, it is shown that the rate of information loss is minimized for a range of signal to noise ratios entering the channel with noise and signal amplitude of the same order of magnitude. In contrast, at the deterministic limit, (i.e., signal amplitude very large compared with the noise) the rate of information loss increases; approximately by a power law of the distance traveled by the message. The exponent depends linearly on the time constant of the function relating speed of propagation vs period of excitation of the axon.
Dante R. Chialvo,
A. Vania Apkarian,
"One more reason why neurons need to be noisy", Proc. SPIE 2036, Chaos in Biology and Medicine, (5 November 1993); doi: 10.1117/12.162729; https://doi.org/10.1117/12.162729