Dear Paul,
> My paradigm has three random stimulus conditions A, B and A+B. What I
> have done, which I feel is incorrect, is to set 3 different contrasts.
> One contrast for A, one for B and a final one for A+B. However, when
> running the contrast for A or B, off periods will include the time
> frames during which A+B is active. This does not seem to be correct
> because during the A+B epoch, both conditions A and B are active
> therefore this is not a true off time.
This is not really the case. The contrast is testing for the
parameters associated with the epochs designated as A and B relative to
rest. The effects of the A+B condition are modelled by the third trial
type and therefore do not effect the A or B contrasts. [I am assuming
here that you have modelled three trial types: A, B and A+B).
> Is it possible to analyze the data using the off periods prior to and
> following each A epoch and exclude the time periods when the other
> epochs (B and A+B) are on? If not why?
It is possible but not necessary.
> Do you have a better suggestion for contrasts that will allow me to see
> activations from A, B and A+B. Of particular interest are the brain
> regions active specifically during the A+B condition.
One might consider a factorial perspective. What you are interested
in is the extra activation when A and B are presented together. This is the
interaction AxB. To assess this interaction you could use a contrast
-1 -1 1 for A B and A+B.
or
1 -1 -1 1 for R A B and A+B if you have modelled rest explicitly.
I hope this helps - Karl
|