The emperor is still under-dressed

Reply to Cosmides et al. (CTFB), by Buller and Fodor CTFB [1] miss the main point. The ‘Buller–Fodor hypothesis’ concerns only the logical form of mental representations of obligation rules [2]. A subject’s performance on reasoning tasks is determined by his/her mental representation of the logical form of the stimulus material (for Wason tasks, a ‘conditional rule’), not by its surface grammar. The mental representation of logical form is, in turn, a function of contextual variables, including background information ([3], p. 279). Predicting performance on a reasoning task thus requires information about both the stimulus material and the factors that influence how subjects interpret it. Nor is it denied that there can be residual effects of content on subjects’ performance when the logical form of the experimental stimuli is property controlled. If, however, logical form is not properly controlled, then it’s fallacious to attribute observed response asymmetries to content variables. CTFB are repeatedly guilty of this fallacy. Correspondingly, the predictions that CTFB claim follow from the ‘Buller–Fodor hypothesis’ do not, because CTFB derive them from the surface grammar of the conditional rules alone. Consider ‘If you clean up spilled blood, then you must wear rubber gloves’ [4]. This formula can express either an obligation (an employer, for fear of liability, requires it of employees) or a prudential recommendation (issued as a public health advisory). As the wording is the same in either case, whether a subject represents it as an obligation or as a recommendation will, of necessity, be determined by contextual variables. Strictly speaking, only rules expressing obligations (and their converse, prohibitions) are deontics, although we have no objection to calling all ‘must’ statements deontics. But there will then be crucial logical differences between deontics that express obligations and deontics that express recommendations (what CTFB call ‘precautions’). For ‘It’s obligatory that P’ entails ‘It’s prohibited that not-P’, whereas ‘It’s recommended that P’ does not. Given this logical difference, subjects will apply one set of inferential principles to rule formulations construed as obligations and another set to formulations construed as recommendations. So, of course CTFB have been able to manipulate performance with ‘deontic conditionals’ to produce ‘double dissociations’ of obligations and recommendations. The question is: what accounts for this dissociation? We suggest that CTFB have merely manipulated contextual variables that determine whether