Formal model comparison and result consistency across models and replicability across subject groups.

<p><b>(a)</b> The model fit of the Hybrid model was compared with other nested models. The Hybrid model (see <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1012602#sec006" target="_blank">Materials and Methods</a> for model structure) explained participants’ behavioral choices the best (smaller integrated Bayesian information criteria (iBIC) indicates better fit). Note that we performed a parameter recovery analysis (S2) and showed that we could separately identify each parameter. <b>(b)</b> In an independent sample of healthy controls (N = 57, [<a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1012602#pcbi.1012602.ref008" target="_blank">8</a>]), we show the same relationship (as in <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1012602#pcbi.1012602.g002" target="_blank">Fig 2d</a>) between non-social model fit (–LL) and individuals’ social conformity heuristic (Pearson’s r = 0.51, <i>P</i> = 5.78e–05, robust correlation) as was found in the subjects in the current study. Weight parameters for social conformity heuristic were estimated using the Hybrid model, while model fit was calculated from the Solo risk preference model only using Solo trials. Models for healthy controls were estimated in a similar fashion as the original study participants, with the exception that inverse temperature was constrained between 0 and 50, and risk preference was constrained between 0 and 2. Each point represents an individual participant, and lines are the regressions between the indicated parameters; Error bars represent s.e.m.</p> <p>(PDF)</p>