Fig. 8 | Nature Communications

Fig. 8

From: Prefrontal mechanisms combining rewards and beliefs in human decision-making

Fig. 8

Prefrontal activations associated with value components of normalised utilities. a, left: topography of fMRI activations positively correlated with relative chosen normalised utilities vchvun (red) from Model MIX (no other PFC activations exhibited this effect) and superimposed on sagittal anatomical slice X = +4 (MNI template, see Supplementary Table 4 for MNI-coordinates of activation peaks: yellow cross; results from the second regression analysis). a, right: MR signal changes averaged over the cluster shown on the left and plotted against relative chosen normalised utilities vchvun, z-scored within participants. Error bars are s.e.m. across participants (N = 21). b When normalised utilities were broken down into its value components including proposed rewards Vprop and reinforcement-Learning values VRL (third-regression analysis), the results showed that vmPFC activations actually correlated with relative chosen proposed rewards (left) and not RL-values (no other PFC activations exhibited this effect, except in a cluster located in the right dorsolateral cortex, BA 8). The only PFC activations associated with RL-values were located bilaterally in the orbitofrontal cortex (laOFC), which activations increased with the unsigned relative RL-values between bandits, irrespective of choices (positive quadratic effect). Activations were 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). Lighter colours indicate larger statistical T-values. Numbers in brackets indicate MNI-coordinates of activation peaks. See Supplementary Table 5 for MNI-coordinates of activation peaks

Back to article page