Fig. 5.
Prefrontal activations involved in computing choices. a Topography of frontal activations negatively correlated with relative chosen normalised utilities (vchosen–vunchosen) and with relative chosen state beliefs (Bchosen–Bunchosen) with no interactions between these factors (p > 0.05, uncorrected), from model MIX. Activations are thresholded at p < 0.05 (T-tests, voxel-wise corrected for family-wise errors over the frontal lobes: MNI coordinate Y > 0; cluster-wise threshold p < 0.05) and superimposed on sagittal and coronal anatomical slices (MNI template, X = −6 and Y = 32). Frontal activations are located in the dmPFC. See Supplementary Table 4 for coordinates of activation peaks. b, left: sizes of linear effects (betas) for relative chosen beliefs and normalised utilities (green and orange, resp.) averaged over the dmPFC activation cluster shown in a (leave-one-out procedure removing selection biases). b, right: activations averaged over the dmPFC activation cluster shown in a and plotted against relative chosen beliefs (no quadratic component factored out) and normalised utilities (the quadratic component being factored out is shown in Fig. 7), z-scored within each participant. All error bars are s.e.m. across participants (N = 21)