Manuel Beiran, BCCN Berlin / TU Berlin

Effects of neural heterogeneity vs. dynamic noise on population coding of time-dependent stimuli

Processing information from the outer environment is essential for the interactions of all organisms. However, the response of sensory neurons to a repeated stimulus is usually highly unreliable, and furthermore, neurons of the same type commonly show a wide variety of behaviors. Therefore, two classes of variability are observed: (i) irregularity in the response at the single neuron level due to neural noise and (ii) heterogeneity at the population level. The goal of this Thesis is to understand and compare the effects of these two classes of variability on the information transmission properties.

Numerical simulations and analytical approximations have been developed to study the encoding properties of two equivalent ensembles of uncoupled leaky integrate-and-fire neurons: a homogeneous population of identical neurons receiving independent noise vs a deterministic population of neurons with heterogeneous firing rates. It has been found that both a finite noise intensity and a certain level of heterogeneity optimize the transferred information about the stimulus. Moreover, heterogeneity achieves a better performance than noise in processing of weak stimuli.

Additional Information

Thesis defence in the international master program Computational Neuroscience

Organized by

Benjamin Lindner / Robert Martin