Psychologists are beginning to uncover the rational basis for many of the biases revealed over the last 50 years in deductive and causal reasoning, judgement and decision-making. In this paper, it is argued that a manipulation, experiential learning, shown to be effective in judgement and decision-making may elucidate the rational underpinning of the implicit negation effect in conditional inference. In three experiments, this effect was created and removed by using probabilistically structured contrast sets acquired during a brief learning phase. No other theory of the implicit negations effect predicts these results, which can be modelled using Bayes nets as in causal approaches to category structure. It is also shown how these results relate to a recent development in the psychology of reasoning called "inferentialism." It is concluded that many of the same cognitive mechanisms that underpin causal reasoning, judgement and decision-making may be common to logical reasoning, which may require no special purpose machinery or module.