Sam Nastase

Just to start a preliminary map: PyMVPA wraps nibabel for loading GIFTI surfaces (https://github.com/PyMVPA/PyMVPA/blob/master/mvpa2/datasets/gifti.py) and also loads NIML datasets (https://github.com/PyMVPA/PyMVPA/blob/master/mvpa2/datasets/niml.py), which is a surface-data format used by AFNI/SUMA. Then you're...

No, I don't think this is worth delaying the release. More of just a stylistic issue—shouldn't negatively affect actual analysis results.

@cbaldassano thanks for the comment! I'm trying to understand your question here... In the one-sample permutation test, the sign-flipping is applied to the correlation values resulting from the ISC computation...

Hmm, I'm still unsure what to do about this. As far as I know, nobody has done the work to see how well various randomization tests will control false positive...

@cbaldassano @tristansyates if you're interested (or know anyone else who might be), we could put our heads together and systematically evaluate how well these models control FPRs with simulated data...

Starting to look into this simulation—will share soon.

Okay here's a first pass at getting FPRs for different tests using simulated data: https://github.com/snastase/isc-tests. The notebook should render interactively via Binder ([link](https://mybinder.org/v2/gh/snastase/isc-tests/master); takes awhile to build). Feel free to...

@tristansyates—interesting! I'm curious how `phaseshift_isc` and `timeshift_isc` would measure up here as well (you think more similar to Chris's blue distribution?).

Sorry—lost track of this for a few weeks. I agree with @cbaldassano that this will take a bit more work to fully resolve, so I'm not sure we should hold...

Hi @Yibeichan, thanks for pointing this out! Like @CameronTEllis said, you're correct that it breaks when receiving nans specifically when `pairwise=True` because of the `assert np.allclose(voxel_iscs, voxel_iscs.T)` line. This is...