Category Archives: stochastic resonance

On the Positive Effects of Variability

from reference 1

An emerging idea in neuroscience is that noise is a good thing, in moderation. Neural activity is very noisy, there is a large degree of variability in the temporal and frequency domains of the spiking of brain cells. It is thought that this variability actually contributes to the robustness of the system. One concrete example is stochastic resonance. In that phenomenon, randomly perturbing neural activity can push it over some threshold such that a sensory event is detected, or an ambiguous decision is made, one way or the other. It may be difficult to see this as beneficial, but especially as we are fantastic learning machines, simply making a decision, or having a perceptual event occur (even when there has been none) contributes to the system’s learning abilities far more than indecision or non-detection.

In a more macroscopic example, a recent paper analyzing variability in brain activity across several age groups has found it to be quite positive. “Behaviorally, children showed slower, more variable response times (RT), and less accurate recognition than adults. However, brain signal variability increased with age, and showed strong negative correlations with intrasubject RT variability and positive correlations with accuracy. Thus, maturation appears to lead to a brain with greater functional variability, which is indicative of enhanced neural complexity. This variability may reflect a broader repertoire of metastable brain states and more fluid transitions among them that enable optimum responses. Our results suggest that the moment-to-moment variability in brain activity may be a critical index of the cognitive capacity of the brain.1

1. McIntosh AR, Kovacevic N, Itier RJ. (2008) Increased brain signal variability accompanies lower behavioral variability in development. PLoS Comput Biol. 4(7):e1000106.