The difference between generating counter examples and using them during reasoning.

Abstract

The aim of this article is to provide insight into the types of long-term knowledge that are used for solving causal conditional inferences. Two taxonomies were constructed to map the types of counter example. The available counter examples are traditionally probed via a counter example generation task. We observed that there are some significant differences in the types of counter example retrieved in the reasoning task versus the generation task. The generation task can be used for predicting answers that sprout from a reasoning process that takes counter example into account, but some participants use a different reasoning process in which the available semantic information is not used as contrasting evidence. Nonetheless, we found that the results of the generation task validly predicted the proportion of inferences accepted as well as the number of counter examples used during reasoning.