TRENDS in Cognitive Sciences Vol.5 No.7 July 2001 http://tics.trends.com 1364-6613/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S1364-6613(00)01667-3 296 Opinion Opinion It is widely acknowledged that language plays a fundamental role in the development of human reasoning 1 . One key proposal is that certain forms of reasoning can only take place in explicit sentences of a natural language 2,3 . Grammatical capacity is seen to be necessary to reason out solutions to particular problems, such as ones involving theory of mind (ToM) understanding in which the task is to determine the relationship between others’ mental states (e.g. beliefs and desires) and behaviour. For example, in a ‘changed contents’ToM task 4 , the subject is shown a familiar container (e.g. a Smarties tube) and is asked to indicate what it holds. Then the unusual contents are revealed (e.g. pencils) and the subject is asked what a person who has not seen the contents would say is inside the container. Some features of a ToM involving eye gaze and emotion interpretation have been viewed as involving a ‘social-perceptual’ component that is language- independent and has its own developmental trajectory 5 . By contrast, it has been claimed that ToM reasoning such as reasoning on a changed contents task involves a ‘social-cognitive component’ that depends upon language, specifically the grammatical capacity to embed false propositions within true statements (‘Mary knows that John [falsely] thinks Smarties are in the tube’) 6 . Astington and Jenkins carried out a longitudinal study of the relationship between language and ToM reasoning in children, most of whom were 3 years old at the first time of testing 7 . The children were tested twice more at 3.5 month intervals. Language ability in the form of syntax was moderately correlated with later performance on changed contents and other ToM reasoning tasks involving the ability to predict the behavior of a person with a false belief. On this basis, it has been concluded that grammar is instrumental in ToM development. In this article, we examine the extent to which ToM and other forms of reasoning are supported by grammar. We consider an alternative account that language serves to support reasoning in terms of conversational awareness and experience of others’ intentions in communication. Reasoning in agrammatic aphasia One profitable avenue for investigation into the relationship between grammar and reasoning concerns the study of people who have severe forms of aphasia - an acquired disorder of language, resulting from damage to language-mediating regions of the cortex and associated sub-cortical structures 8 . In aphasia, there can sometimes be dramatic excisions of components of the language system, as in the case of severe agrammatic aphasia where the patient shows little or no ability to understand or construct sentences in any modality of language use (spoken or written). Even more extensive impairment is found in global aphasia where in addition to loss of grammatical ability there can also be substantial impairment of lexical knowledge. It is widely acknowledged that cognition is retained in many cases of aphasia 9 . However, previous research has not investigated sufficiently precise questions about the forms of thinking that are mediated by grammar; the nature and degree of language impairment in the patients studied was equally non-specific. If the claim is that language propositions supported by grammar are necessary for ToM and causal cognition, a test then requires that aphasic patients have no access to propositional language in any modality of language use. Similarly, questions of disruption of cognition in aphasia have been addressed through administering tests from non-verbal intelligence scales with no clear rationale as to why language might be implicated in, for example, visuo-spatial problem solving. It is only now that claims of specific forms of language mediation (e.g. language propositions) in specific types of reasoning (e.g. ToM) have been tested. Recent studies have adopted this more focused approach to the relationship of grammar to cognition in aphasia. These have revealed that patients with severe agrammatic aphasia, who have minimal access to propositional language, are capable both of ToM understanding and simple causal reasoning (see Box 1). By contrast, patients without aphasia, but who have had lesions to the right- (non-language dominant) hemisphere display impaired ToM reasoning, as well as difficulties in understanding sarcasm, jokes, and the conversational implications of questions 10,11 . This double dissociation between grammar and cognition indicates that grammatical language is neither a necessary or sufficient condition for causal reasoning and ToM understanding. M ind over grammar: reasoning in aphasia and development Michael Siegal, Rosemary Varley and Stephen C. Want Research on propositional reasoning (involving ‘theory of mind’ understanding) in adult patients w ith aphasia reveals that reasoning can proceed in the absence of explicit grammatical knowledge. Conversely, evidence from studies with deaf children shows that the presence of such knowledge is not sufficient to account for reasoning. These findings are in keeping w ith recent research on the development of naming, categorization and imitation, indicating that children’s reasoning about objects and actions is guided by inferences about others’ communicative intentions. We discuss the extent to which reasoning is supported by, and tied to, language in the form of conversational awareness and experience rather than grammar. Michael Siegal* Stephen C. Want Dept of Psychology, University of Sheffield, Western Bank, Sheffield, UK S10 2TP. e-mail: M.Siegal@ Sheffield.ac.uk Rosemary Varley Dept of Human Communication Sciences, University of Sheffield, Sheffield, UK S10 2TA.