Autocorrelation properties of single neurons

Abstract

Recent studies have shown that stochastic resonance (SR) can occur in biological neurons (Douglas et al, 1993). This behaviour is characterised by possible enhancement of signal detection in the presence of increasing noise levels, and by phase locking between neuron activation and weak periodic input signals. It has been suggested that this behaviour may occur in the human nervous system, and may play an important part in sensory perception. In this paper we show that a single neuron in the presence of noise is capable of performing an autocorrelation an an incoming periodic stimulus, and thereby can even detect periodicities which have no spectral energy (so-called "missing fundamentals"). There is substantial evidence for this mechanism in the current data on acoustic perception (Horst et al, 1986, Wightman, 1973). Autocorrelation has been proposed in the past as a mechanism in sensory systems; these results suggest that it can take place at the simplest level in the human sensory system.