A little help with pyMVPA RSA using modelled inputs?

I think this may be a simple question, but I’m a little stuck on the “right” way to do this.

I’d like to do an RSA on subject-level modelled data using the betas as inputs, ala Kriegeskorte 2008. It seems like pyMVPA is designed to operate on run-level data (with each line in the attributes file for a dataset corresponding to a single TR).

Is there a simple way to ask pyMVPA to operate on processed data? Any gotchas to know about?

(Sorry if I missed something – this seems like it’d be a common request, but I don’t see a lot of people asking about it, so I’m worried I’ve missed something obvious.)


Nevermind! I’m pretty sure I can just use fmri_dataset and specify inputs and targets manually.