How children and adults keep track of real information when thinking counterfactually.

Thinking about counterfactual conditionals such as "if she had not painted the sheet of paper, it would have been blank" requires us to consider what is conjectured (She did not paint and the sheet was blank) and what actually happened (She painted and the sheet was not blank). In two experiments with adults (Study 1) and schoolchildren from 7 to 13 years (Study 2), we tested three potential sources of difficulty with counterfactuals: inferring, distinguishing what is real vs conjectured (epistemic status) and comprehending linguistic conditional expressions ("if" vs "even if"). The results showed that neither adults nor schoolchildren had difficulty in the comprehension of counterfactual expressions such as "even if" with respect to "if then". The ability to infer with both of these develops during school years, with adults showing great ability. However, the third source factor is critical: we found that the key to young children's difficulty with counterfactual thinking was their inability to differentiate real and conjectured information, while adults showed little difficulty with this.

Concessive and semifactual interpretations during reasoning with multiple conditionals.

The present research evaluates how people integrate factual 'if then' and semifactual 'even if' conditional premises in an inference task. The theory of mental models establishes that semifactual statements are represented by two mental models with different epistemic status: 'A & B' is conjectured and 'not-A & B' is presupposed. However, following the principle of cognitive economy in tasks with a high working memory load such as reasoning with multiple conditionals, people could simplify the deduction process in two ways, by discarding: (a) the presupposed case and/or (b) the epistemic status information. In Experiment 1 and Experiment 2, we evaluated each of these hypotheses. In Experiment 1, participants make inferences from two conditionals: two factual conditionals or one factual and one semifactual, with different representations. In Experiment 2, participants make inferences with a factual conditional followed by two different semifactual conditionals that share the same representations but differ in their epistemic status. Accuracy and latency data suggest that people think of both the conjectured and the presupposed situations, but do not codify the epistemic status of either when the task does not require it. The results are discussed through theoretical predictions about how people make inferences from different connected conditionals.