Very basic questions

Hi Experts,

First of all, thank you for developing the toolbox, for me at least, I find it much easier to use compared to other ones out there.

As someone who is fairly new to programming and mvpa, I am having some trouble with understanding the results and displaying issues.
I ran a correlation classification because i only had 2 runs, using searchlight (12mm).Then using AUCminuschance (unbalanced data), I took it to twosample t-test in spm to compare 2 groups (Low vs. High group).
The result indicated 3 clusters significant for G1>G2 only, but no clusters fo G2>G1. Does this mean the activations in those clusters are increased for G1 than g2? I’m just a little confused because I understand that mvpa is about the pattern of activations within the space.
Additionally, is there anyway to get the activity maps or performance map?

I am sorry if these questions are very layperson.

Best,
Erin.

Hi Erin,

To make sure I understand, I assume you ran one classification analysis for each participant in Group 1 (classifying A vs. B) and one for each participant in Group 2. Is that correct? In that case, what the results tell you is that the difference (or to be more precise, the discriminability) in the classified patterns representing classes A and B is larger in Group 1 than in Group 2.

Btw: You only need AUC in case there is an imbalance in A vs. B, not if there is an imbalance in Group 1 vs. Group 2 (but using AUC shouldn’t harm you).

I’m not quite sure I understand. Did you run a searchlight analysis? This should give you a map per participant, which you would have used for your two-sample t-test.

Best,
Martin