Fig 7. Controllability alters sequential decisions at 1st and 2nd-stages.
A. First-stage free choice proportion after a free and forced trial as a function of incoherence. B. Second-stage stay probabilities for the different action-state-reward trial types. Each subpanel represents a putative strategy followed by the subject. C. Estimated 2nd-stage stay probabilities. Error bars represent 95% CI. P-values are displayed for significant pairwise comparisons and adjusted for multiple comparisons. Statistics for significant comparisons are: Unrewarded&Coherent vs. Rewarded&Coherent (log odd ratio (OR) = -1.20, z = -5.4, p < 0.001), Unrewarded&Coherent vs. Unrewarded&Incoherent (log OR = -1.11, z = -3.62, p = 0.002), Unrewarded&Coherent vs. Rewarded&Incoherent (log OR = -0.843, z = -3.12, p = 0.010). Statistics for non-significant comparisons are: Rewarded&Coherent vs. Unrewarded&Incoherent (log OR = 0.083, z = 0.260, p = 0.994), Rewarded&Coherent vs. Rewarded&Incoherent (log OR = 0.354, z = 1.70, p = 0.323), Unrewarded&Incoherent vs. Rewarded&Incoherent (log OR = 0.271, z = 0.760, p = 0.872).