[Mne_analysis] Repost: alternative to decoding across time

Dirk van Moorselaar dirkvanmoorselaar at gmail.com
Tue Jun 16 14:06:47 EDT 2020
Search archives:

        External Email - Use Caution        

Dear mne users,

I am sorry if this is a repost. I sent this mail earlier in June but did not find it back in the mailing list so I am sending it again. 

We’ve conducted a study where we want to (using mne) differentiate between three mental states (thinking, feeling, and resting), entered in 12 second blocks (with 12 trials per condition). Ultimately we'd like to see if training in meditation makes these states more different (as would be indexed by higher classification acc.). 

Thus, for our purposes decoding along the time-course of the 12 seconds is largely irrelevant. We simply want the best way possible to differentiate between the three states, which might be better achieved by collapsing across time (the states are likely to be highly variable over time, i.e., at any specific time points, making classification difficult). 

So, we’re seeking advice on ways to decode these conditions that might lend better classification than time-course decoding. Put differently, we would be interested in a way that enables a clumping together of the time information. 

We were thinking of splitting the 12s trial data into 1s epochs, as this would increase our power. Does anyone have experience with a similar method or can suggest a better option?


Dirk van Moorselaar

More information about the Mne_analysis mailing list