Interested in Correlation Analysis: Rectify Individual Cortical Maps or Not

Discussions
1

Dear Brainstorm community,

I have a datatset in which I would like to correlate the source-localized ERP activity with behavioural data. Following the Workflow section (https://neuroimage.usc.edu/brainstorm/Tutorials/Workflows), under the "Constrained Cortical Sources" section. It mentions not to rectify the cortical maps for the subject average source file, but for the Group analysis, to rectify them.

Since my analysis is a correlation analysis, would the right procedure to be then to not rectify the cortical maps. The mention of cortical maps having ambiguous signs across subjects however makes me feel as though I should be rectifying the cortical maps before conducting any correlation analysis using all subjects.

Any insight would be greatly appreciated.

Best,
Paul

2

Rectified cortical source value can be interpreted as "there is that much activity in this region". Non-rectified values introduction a notion of direction of the dipole, which not be very relevant in comparison with behavioral data...

If you correlate discreet behavioral data with source results, aren't you interested in having only one single value per trial, maybe averaged over a short time window around your latency of interest? In that case, you should probably use the absolute value of the source values...
Another option could be to extract the power in a frequency band in a specific ROI. This would be a strictly positive value as well.