Spike Timing Precision in the Visual Front - End

To assess whether the output of the cat retina and LGN is sufficiently precise to support a temporal coding scheme, we determined whether spike timing in recorded spike trains is more precise than would be expected on the basis of the firing rate alone. We compared spike time deviations (temporal offsets between nearest spikes in subsequent responses to repeated presentations of the same stimulus) in recorded and Poisson simulated spike trains. We found that spike time deviations for recorded responses were significantly smaller, suggesting that there is information in the timing of spikes. The effect is robust and can not be explained by e.g. refractoriness of the spike generator. The increased spike timing precision in front end visual responses varied with the momentaneous spike rate. Spike timing precision in recorded and simulated spike trains was comparable at low firing rates (0 20 spikes s), but progressively diverged with increasing spike rates. Above about 80 spikes s, spike timing in retinal ganglion cell responses was a factor two more precise than would be expected on the basis of the firing rates alone. We conclude that spike timing in the output of the cat retina and LGN is sufficiently precise to enable temporal decoding of visual information at subsequent stages of visual processing.