Dear all,

I would like your expert advice on the following issue, brought up by one of our reviewers.

We have used a within-subject between-sessions design. In session 1, subjects performed conditions A and B. In session 2, subjects performed conditions C and D. We wanted to compare the overlap between the contrasts A > B and C > D.

We have used a full factorial model with one factor (session) and with the contrasts A>B and C>D (defined at the first level) as inputs. We then looked at the conjunction (and disjunction) between the two sessions.

Although the analysis is maybe not the most elegant one, the question is whether it is wrong. Some people have argued that the use of a full factorial model instead of a flexible factorial model is allowed if the data (contrast images) are already a subtraction of conditions for each subject (as is the case here), since then between-subject differences will be removed to some extent. A flexible factorial model indeed showed very similar results for our data.

Furthermore, is it allowed to use a conjunction analysis (as implemented in SPM) for our within-subject design or is this only valid for between-subjects designs? 

Best regards,

Wouter De Baene

​