In favour of a unified model of deductive reasoning