Seminar Abstract

6 Dec 2006:
Speaker:  Simon Durrant
Venue:     Devonport Lecture Theatre, Portland Square
Time:       14:00

Turn up the noise: suprathreshold stochastic resonance in neural systems

Neural firing is classically modelled as a stochastic point process, reflecting the variability of firing times within a spike train. From the widely-held rate-coding perspective, where signal is first order (mean value), and the variance is treated as noise, this poses an obvious question about the possible functional role of this noise; in short, why are neurons noisy? Traditionally, noise has been mostly seen as an undesirable by-product of network effects, to be minimised by the system where possible. However, recent research on suprathreshold stochastic resonance (SSR) has shown possible benefits of noise to populations of simple threshold units. My research outlines the first application of SSR to small networks of the most widely used neuron model (the classical integrate-and-fire model), and demonstrates how noise can be beneficial in neural processing.