The ability to pass tests of stimulus equivalence seems to mark an important distinction between animals and humans that is tied to language. Most other animals are unable to reliably pass equivalence tests. Even linguistically trained chimps seem unable to pass them (Dugdale & Lowe, 1990). Moreover, pre-linguistic children show similar poor performance. Sidman (1990) argues that an innate logic (Fodor, 1975) would confuse explanandum with explanans, i.e., logical ability needs to be explained by more primitive behaviour. We argue that abductive reasoning ability is more primitive in that it is required to learn basic stimulus-stimulus relations. Moreover, we propose that extending this ability beyond the temporal limitations of our neural equipment requires the ability to sustain a representation in working memory and this may be facilitated by associating it with an endogenous rehearsable response, i.e., a speech sound. In sum, extending the temporal range over which associations can be learned using associated sound may tie together the origins of reasoning, language and working memory.