Slope

For each subject we estimated separate beta values for the two picture types (pain, neutral) in each of the 4 runs using a standard GLM approach. To control for habituation effects that are not specific for the pain stimuli, the run-wise difference between the beta value for pain and no-pain condition was calculated yielding 4 contrast maps per subject, one for each run. We assumed that habituation could be demonstrated as a linear decrease in this beta difference over the runs. At each voxel and for each subject, the slope of the best-fitting regression line was estimated.