[Marsbar-users] %signal change

[Jozina de G. <Joz...@in...>]

Hi marsbar-users and Matthew,

I'm puzzled about a result concerning the %signal change in comparison with 
an classical analysis with betas...

I have two "events" A and B with a duration of 2 seconds that I modeled 
with a single HRF regressor. I contrasted A-B using betas and found a 
delta-beta of 0,03990071 (for one subject and a given ROI). Then I decided 
to rather calculate the %signal change of Marsbar to illustrate my results 
than these tiny little delta-betas. So, I determined the mean %signal 
change over all sessions for the events A and B and determined A-B %signal 
change and here I found a delta-%signal change of -0,02275, so a negative 
value !!! How is it possible that in one case the difference between A and 
B is positive (so, A evokes a higher BOLD signal than B) and in the other 
one it is negative (so, %signal change in B is higher than in A)? It is 
noteworthy to mention that I do not find this opposition for every ROI, 
otherwise I could think I made a mistake in defining the contrast...

Is there anybody who can help me with this puzzle??

Thanks in advance,

Jozina.

-----------------------------------------------------------------
Jozina de Graaf
Mediterranean Institute of Cognitive Neuroscience
Marseille