Frontiers in Systems Neuroscience Systems Neuroscience

Abstract

Our current study exploits the fact that in the thalamus it is possible to record, with one extracellular electrode, the afferent input together with the thalamic output, and that each relay cell is driven primarily by a single retinal ganglion cell For a series of stimulus sizes we compared the single spike information content (Brenner et al., 2000) in individual LGN cells with that of the retinal ganglion cells that drove the LGN. Although there are some variations among cells, we generally fi nd that, to compensate for its lower fi ring rate, the LGN packs more information than the retina into each spike, but as stimulus size increases this trend becomes weaker. Since single spike information is only an approximation to the total information a channel can carry, we also computed the information contained in pairs of spikes. We fi nd that a second spike in a pair contributes information synergistically for short intervals in the retina, but is mostly redundant for the LGN over most of the spike time interval scale, almost independently of stimulus size. We also ask: how does the LGN cell 'decide' which retinal spikes to ignore? Does it perform optimally, from the perspective of transmission of information about the stimulus, or is its editorial policy a compromise that includes other considerations, such as the overall state of arousal, contextual information and so on? We therefore edited the retinal spike train in several ways to produce artifi cial LGN spike trains of the same number of spikes as the one recorded in the laboratory. The comparison of information content of real