|
Spike Count ReliabilitySee Neuron (2000) paper: Abstract |
LGN responses are highly reliable. One aspect of this is
the reproducible number of spikes fired in response to a given
stimulus. The figure shows an LGN cell's responses to 1024 repeats of the same
black-and-white flickering stimulus (click to enlarge figure).
The spike count in any given
epoch of the LGN response is reliable, with
a trial-to-trial variance much less than the mean (ratio 0.2, if you
count spikes in 200 msec windows).
For comparison, a random Poisson sequence of events, like radioactive decay,
always has a variance equal to the mean (ratio of 1).
In the LGN and other visual neurons, refractoriness can explain
spike-count reliability. The figure below illustrates this with another
LGN cell. When the firing rate reaches a peak (a), the variability or Fano
Factor is lowest (b). The same data are replotted in (c) as the variance vs.
the mean count for each 1msec time bin. As the count gets higher, the data get
farther from the Poisson prediction (diagonal). The cell can be modeled as
a Poisson process with a refractory period imposed. The model is fit to the
observed firing rate (a) and refractoriness, but it accounts for how
variability changes with firing rate as well (red lines in a and b).
