search

cvnlab / GSN

A toolbox for generative signal and noise modeling
BSD 3-Clause "New" or "Revised" License
7 stars 1 forks source link

rdmfun transpose redundancy in rsanoiseceiling.m/py? #3

Open jacob-prince opened 2 years ago

jacob-prince commented 2 years ago

the default rdmfun behavior is to transpose a (voxel x condition) data matrix and then apply the pdist function to achieve a (condition x condition) RDM. however, in each monte carlo simulation, the signal and measurement variables already come in (condition x voxel) format, requiring each matrix to be transposed when passed to rdmfun() only to be re-transposed immediately afterward via rdmfun().

is there a reason for this double-transpose? if not, for clarity, we should consider removing both by modifying the line which sets the default rdmfun() and the line where rdmfun() is applied to signal and measurement.

kendrickkay commented 2 years ago

No real reason for the double transpose;it's just necessary for everytihng to work out. It just seems more natural for people to think in terms of "voxels x conditions x trials" as the standard format. Do you agree? If not, what do you suggest?