Fig. 6
From: Short-term reward experience biases inference despite dissociable neural correlates

Gradual RPE integration into the dorsal striatum reduces model-free biases. a Scatter plot of the overall correlation (r = −0.70, p = 0.00012) between the individual slopes of the striatal gradients of RPE representation (as in Fig. 3c; averaged over both hemispheres) and the individual behavioural influence of RPEs on belief updating (regression weights displayed in Fig. 2d middle). b Individual striatal gradient slopes (averaged over both hemispheres), colour-coded according to the individual rank order of behavioural bias by RPE (short-term outcomes) on belief updating (red = strong short-term bias, green = low short-term bias). c Same as b with the intercept of the individual regression added to the plotted regression lines. In sum, weaker representations of the RPEs in dorsal striatum were associated with stronger biases of a participant’s behaviour by model-free processing of short-term outcomes