Discoveries about attitude aggregation have prompted the re-emergence of non-reductionism, the theory that group agency is irreducible to individual agency. This paper rejects the revival of non-reductionism and, in so doing, challenges the preference for a unified account, according to which, agency, in all its manifestations, is rational. First, I offer a clarifying reconstruction of the new argument against reductionism. Second, I show that a hitherto silent premise, namely, that an identified group intention need not be determined by member attitudes (...) according to a rule, e.g., majority, is false. Third, I show that, on rejecting this premise, the aggregation results lead instead to the conclusion that, in contrast to individual agency, group agency is non-rational. (shrink)