Benjamin wrote:
[...]
> Hi Matthias,
>> While the hippocampus does 'know'
> Direction, it's primary function is =not=
> as a sort of "spatial memory".
I agree, but modelling spatial coding is a challenge already.
My model reduces the complexity of the system and still explains a lot
of its behavior, only based on movement statistics. Whatever drives this
movement statistics I don't care for here.
> I've reiterated what the hippocampus
> does in the msg I just posted.
Well, that's quite a debate for quite a while ;)
But I'd predfer not to discuss this aspect now as it would probably fill
a few threads of its own. I'll look up your msg.
> You can generate data by simulating
> food placements, mating opportunities,
> etc., relative to your 'rat's position.
>> But, if you actually want to simulate what
> the hippocampus does, you've also got
> to include relative-"novelty" in the mix, in-
> cluding the gating of "inverting reward."
That's why I don't call my model "model of the hippocampus", but "model
of hippocampal place fields"
> Anyway, it's useful to use generated data
> in the early stages of simulation because
> doing so enables one to rigorously check
> one's code. It's necessary to verify such
> rigor =before= 'trusting' the simulation
> with real-world data, which is, itself, laden
> with unknown-unknowns.
Right, that is what I've done. Analytical results and simulation results
match for the random walks. So I'm looking for real trajectories.
One standard experiment involves randomly distributed food pellets,
where the animal presumably rotates often. Here we find mostly
non-directional place fields. Otoh, in linear tracks or in the open
field (but with directed search behavior), place fields tend to be
directional [e.g. 1].
I can reproduce these effects with my model, solely based on the
movement statistics change of the rat.
So now I'm looking for more exact data on real rat trajectories to
verify my assumptions on the movements stats.
Thanks,
Mathias
Ref [1]:
Article{Markus1995,
author = "E.J. Markus and Y.L. Qin and B. Leonard and W.E.
Skaggs and B.L. McNaughton
and C.A. Barnes",
title = "Interactions between location and task affect the
spatial and directional firing of hippocampal neurons",
journal = "J Neuroscience",
year = "1995",
volume = "15",
number = "11",
pages = "7079--7094",
}
