 |
 |
Dear FSL experts I wish to get some clarification on the concept behind paired t-test using higher-level FEAT package. My goal is to identify changes in functional connectivity (FC) associated with learning. I performed resting-state scans before and after the learning paradigm. I also measure their performance before and after learning. I then set the design matrix as attached (2 different designs), where change in behavioral score was assigned as EV1. The rest of the EVs take care the within-subject means as mentioned in the FEAT guide. Once I ran the analysis, I obtained the statistical map showing areas that changed significantly associated with the score improvement. I wanted to plot the correlation between change in FC and change in score. Supposed there is this SMA cluster in my higher-level output. Taking this as mask, I extracted subject-level average COPE inside the mask and manually subtracted their respective post and pre value. What I have is the change in FC. I'm very puzzled because this FC-change is not correlated with the change in score although the cluster appeared significant from the higher-level FEAT. In other words, should I trust my output without doing secondary cluster test by manually correlating FC-change and score-change? Am I missing any link? I appreciate some inputs. Thanks! ~Amanda