Hi AFNI Experts!

I have a question concerning 3dANOVA3 and do not really know, how to set up all questions I am interested in and/or if this program or another one is capable to do this job.

I am interested in studying 2 factors across subjects.
fixed Factor 1 (Emotion) has 3 levels (pleasant, unpleasant, neutral)
fixed Factor 2 (Cognition) has 2 levels (working memory task, labeling task)
random Factor 3 (Subjects)
These factors were constructed based on an rapid event related paradigm, where subjects had to perform a working memory and labeling task using stimuli of three valences (pleasant, unpleasant, neutral). As input for 3dANOVA3 six conditions have been used:
P1, U1, N1, P2, U2, N2
where P, U, N means the valence and '1' means during the working memory and '2' during the labeling task.
The main goal of the study is to investigate, how different valences affect brain areas involved in working memory performance.
Other goals are, to investigate, how these valences affect encoding, where I want to take the labeling task as reference.
Furthermore, I want to study, how the differences in valence found during the working memory task differ from activations during the labeling task.

Some questions were quite easy to answer in a mixed effects model using 3dANOVA3 (type 4) as setting up contrasts using (-adiff/bdiff or -a/bcontr options). However, I would need something like -abdiff/abcontr) to set up contrast between valences dependent on their factor 2 level as I could do that in 3dDeconvolve.

Please, give me any advice, how I could extract these information using 3dANOVA3 or if I have to go different ways to answer these questions?

Thanks,

Lukas