<b>Inference Complexity and the Logic Bias Effect in Conditional Reasoning</b>

The hybrid dual processing model views performance on reasoning tasks as the result of competing intuitive processes – some logical and some non-logical. According to the model, humans possess an extensive intuitive logic featuring multiple deductive inference forms. An important question for the hybrid model is whether there are limits to intuitive logic. One basis for claims that logical processing is largely intuitive is the presence of a logic bias effect on the dual instructional set paradigm. Implicit logical processing interferes with efforts to respond on the basis of belief to a greater extent than belief-based processing interferes with efforts to respond on the basis of logical validity. We manipulated inference complexity (defined by inference direction and the presence or absence of negation) on a conditional reasoning task by crossing conditional inference type (MP, MT) and conclusion wording (normal, contrary). We found that the extent of the logic bias effect varied, inversely, with the complexity of processing required by the inference. We also provide evidence that analytical thinking (AOT) is positively related to the logic bias effect while working memory span is negatively related to the effect. These results suggest that there are limitations to intuitive logic even within everyday inferences.