I am interested in finding out about computational methods (using
Matlab) to model primate inferotemporal (IT) neuron response to color
pictures.  Specifically, I have a large amount of IT spike train data
from monkeys looking at color pictures and would like to develop a
probabilistic model of an IT neuron that can predict spikes (or firing
rate) given a color image.

I would be interested in seeing information/discussion on previous
models of IT neurons (even if flawed) or a list of possible approaches
to the problem.
My first approach to the problem will be using reverse correlation to
try to find a spike-triggered average image which causes the neuron to
fire.  I have been warned that this technique is apt to fail because
IT neurons are nonlinear when compared to neurons in V1 etc...but it
seems that it would be a good place to start.
Additionally, any sources on using Matlab for computational
neuroscience applications would also be appreciated.
Please do not include information on computation models of other types
of neurons, unless techniques/approaches are directly relevant to
models of IT neurons.  This question may be difficult to answer; I'm
not sure if this has been done successfully or not, so I am not
necessarily expectly a perfect solution (especially not for $50), just
information on what has been tried before and how one might use
established computational neuroscience techniques to start examining
the problem.  Thanks.