Back to PLS Helpbrain score vs. signal changes
Posted on 03/19/12 10:48:42
Number of posts: 60
Dear all,
I ran block-pls using HRF function for a paradigm which is consisted of three different conditions. The PLS patterns look very interesting. In particular, LV2 reflected condition A vs. condition B, whereas condition C is roughly around zero. However, when I correlate the signal changes of the most contributive region to LV2 for condition A and brain score of LV2 for condition A, the correlation is very low (r = 0.11), although the general pattern (condition A vs. condition B) holds in the most contributive region! Given that the brain score reflects how strong each subject contribute to each LV, I expect the correlation value (between brain score cond A and signal change cond A) to be pretty high. Isn't it strange or have I missed something?
/Alireza
/Alireza
Replies:
Untitled Post
I'm Online
Posted on 03/19/12 10:55:36
Number of posts: 291
Have you compared with the result from SPM?
brain score vs. signal changes
Posted on 03/19/12 11:01:06
Number of posts: 60
I did!
The correlation between BS cond1(from LV2) and Beta value cond1 (the most contributive region to LV2) is also pretty low!
The correlation between BS cond1(from LV2) and Beta value cond1 (the most contributive region to LV2) is also pretty low!
Untitled Post
I'm Online
Posted on 03/19/12 11:49:59
Number of posts: 229
Alireza...
keep in mind that you ran a taskPLS - which is assessing mean differences across conditions and/or groups...
there is no need for there to be a correlation between conditions in the presence of a mean difference...
The reverse also holds - you may find opposite correlations in two conditions with some external measure, but no mean difference.
cheers,
nancy
keep in mind that you ran a taskPLS - which is assessing mean differences across conditions and/or groups...
there is no need for there to be a correlation between conditions in the presence of a mean difference...
The reverse also holds - you may find opposite correlations in two conditions with some external measure, but no mean difference.
cheers,
nancy
Login to reply to this topic.
