The Bernstein Center Freiburg

Informal Seminar
Gabriela Mochol

The Nencki Institute of Experimental Biology
Department of Neurophysiology
Warsaw, Poland

Stochastic properties of Superior Colliculus neuronal responses to moving stimuli
Wednesday, July 14, 2010

11:00 h sharp
Bernstein Center Freiburg
Hansastr. 9a
Superficial layers of catís superior colliculus (SC) receive main retinal input from two classes of ganglion cells, Y and W belonging to different visual processing channels. It turns out that different sources of visual signal lead to differences in response variability of collicular neurons. To explore this issue four stochastic models of experimental data were studied: inhomogeneous Poisson process (IP), parametric and nonparametric versions of inhomogeneous Markov interval process (IMI) and inhomogeneous Gamma process (IG). The variability of surrogate spike trains derived from these models were then compared with the experimental data results.
The extracellular single unit activity was recorded from superficial, retinorecipient layers of the SC of anesthetized and paralyzed cats. As a visual stimuli the light spot moving with different velocities was used. On the basis of the velocity response profiles the recorded cells were classified into group receiving input from Y channel and group receiving input from W channel.
The time-dependent variability of responses of these neurons was quantified by Fano factor (FF) calculated in discrete time windows. The FF for cells responding to low-velocity stimuli (W input) increased with the increase in the firing rate. The dynamics of activity of the cells responding to fast moving stimuli (Y input) correlated negatively with changes of FF.
The nonparametric IMI models could account well for the properties of responses to fast stimuli and revealed drop of variability during increase of firing rate. Similarly, but with lower extend, did parametric gamma IMI model. For most of the neurons the best fit of the IG model resulted in low order of gamma which, in principle, made it possible to attain variability higher than one (IP level) observed in experimental data for low velocity responses. However, the study of IG surrogate data did not reveal changes of variability consistent with the experimental data at any velocity. The variability of IP surrogate data fluctuated around theoretical value equal one and was not correlated with changes of response level at any velocity.
No model of the data could fully recover dependencies between firing rate of SC cells and their response variability even though all models mimicked well the average evoked level of neuronal activity.
The talk is open to the public. Guests are cordially invited!