Endogenous neural noise and stochastic resonance

We discuss the relationship of endogenous neural noise (ENN) to performance of behavioral tasks and to information processing in the brain. Spontaneous neural activity is closely linked to development and perception, and is correlated with behavior Some of this activity is probably related to internal processing of task- and goal-relevant information, but some is simply noise. Two previous studies have reported correlations between performance on behavioral tasks and measures of neural noise and have characterized these relationships as intrinsic stochastic resonance (SR). We argue that neither of these studies demonstrated intrinsic SR, and discuss several alternative ways of measuring ENN in humans from EEG or MEG records. Using one of these, random-phase power in the 30-50 Hz range 1 sec before the onset of the signal we demonstrate a kind of intrinsic SR that optimizes detection of weak visual signals. Minimum response time was obtained when this EEG measure of ENN was in a middle decile. No other measure of ENN was related either to response time or to an unbiased measure of detection accuracy (e.g., ï). A discussion of the implications of these findings for the study of intrinsic SR concludes the paper.