Whole-brain statistical analysis was performed using a multi-stage approach to implement a mixed-effects model treating participants as a random-effects variable. Explanatory variables for the alcohol–taste cues task were created by convolving delta functions representing the onset of the taste period for each trial type (see Figure 2) with a double-gamma hemodynamic response function (HRF) in FEAT. The event period was chosen to stay consistent with previous applications of this task (Filbey et al., 2008a; Filbey et al., 2008b). Alcohol and Water cue exposure were modeled as separate event types. The onset for each event was set at the first instruction to swallow (10 s after the initial taste cue was presented) with duration of 20 s plus the response time for the urge-to-drink rating. The delay in onset was chosen to capture the period in which participants are actively "experiencing" the taste of alcohol or water. We excluded the first taste period of each trial because this includes the time for the gustometer to deliver the liquid, which could be slightly startling to the subjects, and is likely too soon
