At least three types of baselines were used in the studies described here. Bjork et al. (2004) defined the baseline as the mean signal value averaged across the entire time series. In the Ernst et al. (2005) paper, 18 (of 129) trials were fixation trials that served as baseline. That is, all contrasts of interest were compared to trials in which the participant was presumed to be doing nothing but staring at a fixation cross (refer to Thomason et al., 2005 above to note how this may be problematic). Similarly, Galván et al. (2006) used the intertrial interval as the relative baseline, during which the participant was presented with a fixation cross. Finally, van Leijenhorst et al. (2009) and Geier et al. (2009) did not define an implicit baseline and instead generated contrast images between different trial types (e.g., certain versus uncertain reward trial types). All authors presumably had good reason to choose the baseline they did and there is no standard baseline in the field but, clearly, small differences in baseline can have dramatic effects on final results. For
