This paper studies two types of cognitive factors which have been assumed to underpin people’s interpretation of conditional promises and threats: logic and socio-cognitive assumptions about what conditional promisors and threateners are obliged and permitted to do. We consider whether the logic of conditionals is compatible with the socio-cognitive assumptions underlying their interpretation or whether the two come apart. From the classical logical accounts of conditionals, almost all modern theories have inherited a constraint which specifies that a conditional cannot be true if its antecedent is true and consequent false. This logical constraint is widely assumed to constitute, at least partially, a conditional’s semantics, or ‘core meaning’. A replication of Beller et al.’s (2005) study, reported in this paper, calls for revisiting this long-standing, cross-theoretically assumed constraint. As predicted, we have found that, in English, conditional promises are generally consistent with this logical constraint, but threats are not. Our findings provide evidence for the existence of a new usage-based category of conditional threats, and support the claim that the observed logical asymmetry in the interpretation of conditional promises versus threats is just an epiphenomenon of a socio-cognitive symmetry which pertains to people’s assumptions about the deontic commitments of both conditional promisors and threateners. Based on (i) the observed lack of uniform application of the logical constraint and (ii) a consideration of individual variation in the interpretation of conditional promises and threats, we argue that an exemplar approach to conditionals is a plausible option.