Berlin 2015 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

BP: Fachverband Biologische Physik

BP 2: Neurophysics I

BP 2.3: Talk

Monday, March 16, 2015, 10:00–10:15, H 1058

Up and down states in top-down input may be advantageous to the transmission of information about weak signals — •Felix Droste and Benjamin Lindner — HU Berlin & Bernstein Center for Computational Neuroscience Berlin, Berlin, Germany

How a single neuron transmits information about a sensory stimulus is heavily dependent on the state of the embedding network. This allows higher cortical areas, providing so-called top-down input, to influence the processing of a bottom-up signal. Cortical networks may be in asynchronous-irregular states, where the population firing rate is roughly constant in time. In this case, analytical approaches to signal transmission are well established. However, the firing rate may also be time dependent. It may, for instance, jump between so-called up and down states (observed under anesthesia, during sleep, or quiet wakefulness). While the effect of such input on information transmission has been studied in experiments, theoretical approaches have so far been lacking.

Here, we model up/down input to a leaky integrate-and-fire neuron by dichotomous noise, a simple stochastic two-state process. We derive exact results for the spike train power spectrum and the susceptibility for pure two-state input and expand this to incorporate additional fluctuations around the states, which take into account the shot-noise nature of the input. This allows us to compare signal transmission with an asynchronous-irregular background to that in the presence of up/down state input. We find that if the overall input firing rate is low, up/down states may be beneficial for information transmission.