[pymvpa] normalization: zscore by example?

[Yaroslav Halchenko]

agree and good idea IMHO (if of any help of cause).  Since for
searchlights just a single value recorded in the searchlight's center
and thus it is already notspecific  to that voxel (or specific up
to the size of ROI), demeaning pattern within should be as legit
as searchlight itself ;)

On Tue, 08 Nov 2011, J.A. Etzel wrote:
> Perhaps, if uniform differences are a concern, we should
> mean-subtract across the voxels within each separate searchlight.
> This could still introduce a bit of blurring (information leakage),
> but at a smaller scale ...

-- 
=------------------------------------------------------------------=
Keep in touch                                     www.onerussian.com
Yaroslav Halchenko                 www.ohloh.net/accounts/yarikoptic