Interpreting Within Group Source Differences and Relating to Sensor ERP Differences

Discussions
,
1

Dear Brainstorm community,

in a within subject design, I have 11 participants who are tested twice, once at time A, and then again at time B.

I have a significant effect at the sensor level (corrected for multiple comparisons), and looking at the ERP (30 - 50 ms), the component I am interested in decreases in negativity (smaller negative deflection) from time A to B (please ignore the Controls).

06%20PM
06%20PM2478×1646 397 KB

I then wanted to do source localization of this effect.

After following the tutorials on 'Differences' as well as the 'Statistics' and 'Workflows' in the context of paired samples, I got to my source localization (in t-statistic values):

59%20PM
59%20PM1232×446 248 KB

Now, having looked at both the statistics and differences, I'm arriving to the interpretation that Time B > Time A in regards to activity in the source level based on looking at |A|-|B|, due to me being interested in simply which condition is greater/lesser than the other with regards to activity.

My confusion then comes in the context of applying the absolute value to my source files, and based off the negative deflection of time A being greater than time B for the ERP, is it wrong to have assumed that for my source localization, time A should have exhibited greater activity than time B?

Any thoughts on this would be greatly appreciated.

Thank you,
Paul

2

Hi Paul,

In general, yes, when you observe higher EEG magnitudes, you can expect higher source magnitudes as well. But that would be true when looking at all the EEG sensors together. If in your ERP with all the sensors superimposed on the same graph you observe a peak (some positive values + some negative values), you would observe a similar peak at the same latency in the source maps. Higher peak-to-peak values in the EEG would probably correlate well with higher source magnitudes (= absolute values of the min norm maps).
However, looking at one single sensor, it is difficult to guess whether its increase leads to higher or lower magnitude of the source maps.

If you are not sure to understand the signs of the maps you get, you could try to create a small scout in your region of interest, and plot the corresponding signals for all the files you have (the two conditions, the difference, the difference in absolute value) - make sure you select "relative" in the scout tab, so that you don't get an additional confusing absolute value applied to the source signals.