Dear SPM experts, My colleagues and I are attempting to do the MVPA for our event-related fMRI data analysis. We are doing the preprocessing with SPM12. Our preprocessing steps go as following. 1. time series realignement [Realign: Estimate & Reslice] 2. coregistration between fMRI and sMRI [Coregister: Estimate] 3. Segment 4. normalisation to a standard space [Normalise: Estimate & Write] 5. fMRI model specification 6. Model estimation Now we are supposed to do the step "Contrast manager" step(in fact, t-test) to begin the final MVPA work. To illustrate, say we have two contrasts: "*larger reward vs. small reward*"; "*larger loss vs. small loss*" for the MVPA. Our aim is to see whether the MVPA can discriminate the comparison between them. To reiterate, the comparison means: "larger reward vs. small reward" COMPARED TO "larger loss vs. small loss". So I was wondering whether I can directly do this t-test: *(larger reward- larger loss)* vs.* (small reward - small loss)*. From the mathematics perspective, this t-test is equal to the comparison as I described above. This opens a further question, can we do the MVPA with only one t-test, say "larger reward vs. small reward" in my case? Any comments on our analysis are much appreciated. Thank you very much, Simi Luck