Dear Conn users

In a seed-to-voxel second-level analysis, I am planning a t-contrast between two different resting–state sessions: session A (1) & session B (-1). How can I know if the connectivity results of this t-contrast are based
on positive or  negative (Anti-)  correlations ?

One way to distinguish between the positive and negative correlations might be to first) define a mask from z-transformed r-maps from positive and negative correlations separately and second) to then use these masks separately to restrict the between-sessions analysis for connectivity maps on only positive or negative correlations respectively. This way it should facilitate the interpretations, because the results would be limited to regions showing for example anticorrelations only.

Do you consider such a procedure as being valid?

Thank you very much for your thoughts!

Best,
Diana

Dear Diana,

Yes, that is perfectly valid. The standard way to present this would be as a conjunction of two contrasts, one contrast (for example) could be defined as [1,1] (for selecting only those areas that show positive correlations on average across both conditions), and the second contrast could be defined as [1,-1] (for selecting those areas that show higher connectivity in the first condition compared to the second condition). The conjunction of the two (those areas that show significant one-sided effects in *both* of these individual contrasts) will inform you which areas show higher positive connectivity in the first condition compared to the second condition (similarly, for example, the conjunction of [-1,-1] and [-1,1] will tell you which areas show higher negative connectivity in the first condition compared to the second)

Best
Alfonso
Originally posted by Diana Wotruba: