Figure 3.

Analysis based on the full computational model

(A) MBCA parameters for each format (S, PI, and RI) and outcome serial position (first/second). The asterisks correspond to the significance of the format x position interaction (see the X) and to the standard-retrospective inference contrast for the second outcome (red line).

(B) Common-outcome reward effects (MFCA), as in Figure 2C, but based on simulations of the full model. The asterisks correspond to the significance of the main common reward effect.

(C) Common-outcome reward effects (MBCA), as in Figure 2E, but based on simulations of the full model. The asterisks correspond to the significance of a triple common reward x format x position (see the X) interaction and to a common reward x standard versus retrospective inference format interaction for the second outcome (red line).

(D) MBCA impairment in the retrospective inference format (i.e., the difference between MBCA parameters for the standard and retrospective inference format) for the first (x axis) and second (y axis) outcomes. Each point corresponds to an individual participant. Regression (thick) and identity (thin) lines are imposed.

Error bars correspond to SEM across participants calculated separately in each condition. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001. See also Figure S4 for model comparisons, Figure S5 for parameter recovery and trade-offs in the full model, and Table S1 for best-fitting model parameters.