Fig. 5: Transfer is moderated by evidence for metacognitive learning in Experiment 2.
From: Learning from outcomes shapes reliance on moral rules versus cost–benefit reasoning
a, Responses to the Sacrificial Harm Subscale from the OUS. b, Responses to the Deontology Subscale from the DCS. c, How much money participants donated to the charity supporting human challenge trials. Each panel compares responses between the CBR Success and Rule Success conditions as a function of the amount of evidence the participants’ responses in the moral learning paradigm provided for metacognitive learning quantified using Bayes factors (BF). BF values of >1 indicate evidence for metacognitive learning. The red dotted lines at BF = 10 indicate strong evidence of metacognitive learning; this is where the main effect of the experimental condition was tested. For all panels, the confidence bands indicate the 95% confidence level (N = 380). See Section 3 for a smoothed conditional means version of the plots.
