NATURAL
MOVIES EVOKE TEMPORALLY PRECISE RESPONSES IN CAT VISUAL CORTEX THAT ARE
NOT
PREDICTED BY NON-UNIFORM POISSON PROCESSES WITH RELATIVE REFRACTORY
PERIODS
Shih-Cheng
Yen1, Jonathan L. Baker1,
Jean-Philippe Lachaux2, Charles
M. Gray1
1Center
for Computational Biology, Montana State University, Bozeman, MT, USA
2INSERM
U280, Lyon, France
The
responses of striate cortical neurons are often modeled using
non-uniform
Poisson processes. This model has proved to be a good model of
responses to
simple, artificial stimuli, but has yet to be tested on naturalistic
stimuli.
We recorded the responses of 46 isolated neurons in striate cortex of
anesthetized cats while presenting repeated naturalistic movie
sequences with
frame durations of 35 ms (29 Hz). We compared the spike-count and
spike-time
variability of the neuronal response across repetitions in each 35 ms
window
with 1000 sets of surrogate responses generated using a non-uniform
Poisson
process with a relative refractory period. We found spike-count
variability
(Fano Factor) to be largely comparable to the surrogate responses, with
4.7%
and 3.4% of the windows (with a mean spike count of at least 1) showing
lower
variability than 95% and 99% of the surrogates respectively. These
windows were
found in 10 and 8 cells respectively. On the other hand, we found
spike-time
variability to be much lower than the surrogate responses, with 34.5%
and 21.6%
of the windows showing lower variability than 95% and 99% of the
surrogates
respectively. These windows were found in 37 and 28 cells respectively.
In
addition, we found a highly significant correlation (Pearson's r =
0.21,
Spearman's r = 0.22, p<0.001 for both) between the spike-count and
spike-time variability, indicating that windows with low spike-time
variability
tended to have low spike-count variability. Our results suggest that
when
stimulated by natural movies, neurons in the striate cortex of the cat
exhibited spike times that were not well modeled by non-uniform Poisson
processes with relative refractory periods, and that low spike-time
variability
was highly correlated with low spike-count variability.
Support
Contributed By: NEI and the Kopriva Fellowship