MIT study shows logical reasoning can operate without language processing

In a recent PNAS paper, MIT neuroscientists investigated whether logical reasoning depends on language. They worked with two stroke patients who had lost most language abilities and gave them non-verbal logic tasks involving number transformations and geometric matrices; both participants performed on par with a control group and communicated solutions through gestures or sketches. Parallel fMRI experiments with neurotypical adults showed that language-processing areas remained quiet during both inductive and deductive reasoning, while the brain’s multiple-demand network lit up for rule discovery but not for applying known rules.

Lead investigators Evelina Fedorenko and postdoctoral researcher Hope Kean interpret the results as strong evidence for distinct neural systems for language and thought. The work also has implications for understanding aphasia, suggesting that severe language loss does not diminish abstract reasoning capacity. Finally, the authors note potential relevance for artificial-intelligence research, where language-based models differ from human cognition.