[pymvpa] Papers discussing relationship between scanning parameters and MVPA performance?

[Gilles de Hollander]

Hi there,

I have been quite reading a bit about MVPA the past year, but found little
papers that focus on the very practical sides of MVPA. More specifically,
I'm looking for any literature that discusses the influence of different
scanning parameters, number of trials, number of participants and
Signal-to-Noise ratios on experimental power/classifier performance. Does
anyone know such a paper?

I have a paper in review right now and my reviewer says that we should have
looked into this. I think this sounds nice, but a bit naive: not that* *much
studies have been done and researchers don't publish stuff that doesn't
show significant effects. Also, I guess the influence of all these
parameters covary heavily with the task and ROIs at hand. To say you need n
subjects with m trials, depending on the SNR by factor x is not really
possible would be my hunch. Or am I missing something here? I'm glad to
hear your opinion about this.

Thanks,
Gilles de Hollander
PhD candidate at the Cognitive Science Center Amsterdam
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20130327/ddafe158/attachment.html>