Noise-improved signal detection in cat primary visual cortex via a well-balanced stochastic resonance-like procedure

Abstract

Adding noise to a weak signal can paradoxically improve signal detection, a process called 'stochastic resonance' (SR). In the visual system, noise might be introduced by the image jitter resulting from high-frequency eye movements, like eye microtremor and microsaccades. To test whether this kind of noise might be beneficial or detrimental for cortical signal detection, we performed single-unit recordings from area 17 of anaesthetized cats while jittering the visual stimulus in a frequency and amplitude range resembling the possible range of eye movements. We used weak, sub- and peri-threshold visual stimuli, on top of which we superimposed noise with variable jitter amplitude. In accordance with the typical SR effect, we found that small noise levels actually increased the signal-to-noise ratio (SNR) of previously weak cortical visual responses, while originally strong responses were little affected or even reduced. Above a certain noise level, the SNR dropped a little, but not as a result of increased background activity - as would be proposed by SR theory - but because of a lowered response to signal and noise. Therefore, it seems that the ascending visual pathway optimally utilizes signal detection improvement by a SR-like process, while at the same time preventing spurious noise-induced activity and keeping the SNR sufficiently high.