It’s difficult for us to imagine what our mental lives would be like without language. Some theorists have even gone so far as to argue that language and logical thought are one and the same thing. A new brain imaging study challenges this notion by showing that logical inferences based on simple “not”, “or”, “if”, “then” terms activate a separate, though overlapping, network of brain regions compared with logical inferences based on grammatical judgements.
Martin Monti and colleagues scanned the brains of fifteen participants while they judged the accuracy of conclusions flowing from two kinds of logical argument. One kind was a more pure form of logic, such as “If both X and Z then not Y”, whilst the other kind was based on grammatical rules, such as “It was X that Y saw Z take”. The two types of inference were intended to be of comparable difficulty and to be equally valid (or invalid) but crucially only the grammatical version involved the interpretation of language-related roles such as “object” and “subject”.
As expected, inferences drawn from the grammar-based logic activated a swathe of brain regions usually associated with language functioning, including the Wernicke-Broca circuit, as well as other regions associated with working memory and executive functioning. Judgements about the purer logical arguments also activated regions associated with memory and mental effort, but did not activate the core language areas of the brain. Instead, the grammar-free logical problems triggered activity in prefrontal regions previously associated with logical reasoning.
Monti’s team said their findings were hard to reconcile “with the claim that language and logic are a unitary phenomenon”. Rather, they argued their results are consistent with language and logic being separate processes. The grammar-based statements appeared to be solvable using language networks of the brain, whilst purer logic was dealt with by a distinct neural network not dependent on language. The researchers concluded that their work supports earlier findings by others. For example, it’s been shown that it is possible to have numerical concepts without the words for those concepts . The new and old findings together show that “much of thought is not embedded in language”, they said.
Monti MM, Parsons LM, & Osherson DN (2009). The boundaries of language and thought in deductive inference. Proceedings of the National Academy of Sciences of the United States of America PMID: 19617569