A ventral focus was evident in the explore participants and also across the entire group but did not differ reliably between learn more groups. The more ventral

focus is closer in proximity to both the region of RLPFC associated with exploration by Daw et al. (2006) and the region associated with tracking reward value of the unchosen option by Boorman et al., (2009; though see Supplemental Information for an analysis of branching and the expected reward of the unchosen option in the current task). We did not obtain region by effect interactions and so are not proposing that a functional distinction exists between these dorsal and ventral subdivisions. Nevertheless, activation clusters in these two subregions were clearly spatially noncontiguous and were reliable under partially overlapping contrast conditions. Thus, future work should be careful regarding the precise locus of effects in RLPFC and their consistency across conditions. Beyond RLPFC, we also consistently located activation in SPL in association with relative uncertainty in the explore BVD523 group. Although this region was not reliably different between explorers and nonexplorers, the relative uncertainty effect was found to be reliable in SPL in explorers across the alternate models tested

here. Previous studies have reported activation parietal cortex along with RLPFC during tasks requiring exploration (e.g., Daw et al., 2006). However, the locus of these effects has been in the intraparietal sulcus

(IPS) rather than in SPL. Effects in IPS were less consistently observed in the current study, and ROI analysis of IPS defined from previous studies failed to locate reliable relative uncertainty effects in this region (see Supplemental Information). This comes in contrast to the effects in RLPFC, which are highly convergent in terms of neural locus. The reason for the variability in parietal cortex cannot be inferred from the present data set. However, one hypothesis is that it derives from differences in attentional demands between the different tasks. For example, SPL has been almost previously associated with endogenous, transient shifts of spatial and object-oriented attention (Yantis et al., 2002 and Yantis and Serences, 2003), perhaps as encouraged by the clock face design, and thus, the direct relationship between exploration and identification/attention to new target locations on the clock. However, such hypotheses would need to be tested directly in subsequent experiments. Previous studies have not found an effect of uncertainty on exploration (Daw et al., 2006 and Payzan-LeNestour and Bossaerts, 2011).