Dear experts, We studied three groups of subjects using a conditioning paradigm . We expect that during extinction the amygdala habituates in the 3 groups differently. In the main analysis I compared the CS+ with CS- using appropriate t-contrasts (because of the small groups I used a fixed effect model). Now I want to compute a time by event interaction for the different events within groups. I tried to use the time x event interaction option and typed in a vector (length 8 for CS+ and CS- respectively) with a linear decrease (e.g. -4 -3 -2 -1 1 2 3 4, because the reversed order seems to show an increase in the design matrix). Now I don't no exactly how to model the best fitted different regressors for the interaction. Should they been all mean corrected ? Is it necessary to compare interaction terms between CS+ and CS-, when in my case the CS- is already habituated during acqusition or can I compare the main effect for CS- with the interaction regressor for CS+ ? Do I have to include these regressors already in the main analysis or is better to create a new design matrix ? Thanks for any comments Ralf Dr. Ralf Veit Institut fuer Medizinische Psychologie Universitaet Tuebingen Gartenstr. 29 Tel.: +49(0)07071 72074 Tuebingen e-mail: [log in to unmask] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%