The easy solution of the selection tasks with social contract rules, compared to the poor results of the original formulation of the tasks with descriptive rules, has been interpreted, in the framework of massive modularity hypothesis, as the evidence that humans are adaptively skilled to reason about particular deontic domains. Nevertheless, the two versions of the tasks are incomparable because they differ not only for the content of the rule, but also in terms of structural features that make their solution based on different types of reasoning-about and from a rule. In this study we disentangled these two aspects by testing type of reasoning (about vs. from) and content of the rule (descriptive vs. social contract) separately in order to establish their relative importance in human reasoning. In addition to these factors, we examined the putative effect of pragmatic expectation (neutral vs. disconfirming the status of the rule) on the participants' performance. Four hundred undergraduates participated in the study, with a 2x2x2 between-subjects design. Results showed that "reasoning from" tasks were better performed than "reasoning about" tasks, regardless of the content of the rule and the type of expectation.
