Home > Research > Publications & Outputs > On the generation and evaluation of inferences ...

Associated organisational unit

View graph of relations

On the generation and evaluation of inferences from single premises.

Research output: Contribution to journalJournal article


<mark>Journal publication date</mark>04/2003
<mark>Journal</mark>Memory and Cognition
Issue number3
Number of pages12
Pages (from-to)467-478
<mark>Original language</mark>English


A theory of how individuals construct mental models to draw inferences from single premises was tested in three experiments. Experiment 1 confirmed a counterintuitive prediction that it is easier to generate inferences between conditionals and disjunctions than it is to evaluate them. Experiment 2 replicated this finding, but an advantage found in the first experiment for conditional-to-disjunction over disjunction-to-conditional inferences was removed with different sentence contents. Experiment 3 showed that disjunction-to-conditional inferences were facilitated when premises expressed familiar indicative relations, whereas conditional-to-disjunction inferences were facilitated when premises expressed causal relations. The results indicate that small changes in task format can have large effects on the strategies that people use to represent and reason about different sentential connectives. We discuss the potential for theories other than mental models to account for these results. We argue that, despite the important role played by single-premise inferences in paraphrasing logical forms during inference, mental logic theories cannot account for the results reported here.

Bibliographic note

Ormerod was lead author, holding a Leverhulme Trust Study award grant to develop the work at Princeton. First experiment included in Richardson's PhD, second