well as the caudate regions of interest (ROIs) were anatomically defined using the high-resolution MPRAGE anatomical images, segmented on a subject-specific basis in native space using FMRIB's Integrated Registration and Segmentation Tool (FIRST) in FSL. The average time course of the right ventral striatum and the caudate were extracted from motion-corrected, high-pass filtered image data (same pre-processing steps as outlined above). PPI analyses were conducted using components of SPM and FSL’s FEAT. The models were identical to the first-level model described above with the inclusion of three additional regressors: ‘psychological’, ‘physiological’, and ‘psychophysiological interaction’. These regressors were generated separately by computing three vectors using the PPI algorithms implemented in SPM5 (http://www.fil.ion.ucl.ac.uk/spm/software/spm5). The psychological vector was specified by a delta function with Alcohol-cue events represented by 1 and Water-cue events represented by −1 (zero centered), the physiological vector estimated ‘neural’ activation of the right ventral striatum via hemodynamic deconvolution of the average pre-processed time course, and the psychophysiological interaction vector was the product of the two. The three vectors were convolved with a hemodynamic response function (using SPM functions), prior to inclusion in FSL’s FEAT model. A whole-brain contrast image for the PPI was computed from this model and submitted for
