A study published on July 22, 2026, in PNAS by cognitive neuroscientists at MIT's McGovern Institute for Brain Research, led by associate professor Evelina Fedorenko and postdoc Hope Kean, reveals that logical reasoning does not rely on the brain regions responsible for language processing. The research tested stroke patients with severe aphasia and healthy participants undergoing functional MRI scans, demonstrating that individuals with profound language impairments can still successfully solve complex logic puzzles and abstract reasoning tasks. The authors challenge the long-standing philosophical and cognitive assumption that language serves as the necessary foundation or machinery of thought. By utilizing non-verbal puzzles involving number sequences and geometric designs, the study highlights that abstract thinking and rule induction possess independent cognitive mechanisms that operate entirely separately from linguistic systems in the human brain. The core findings indicate that while language is frequently used to communicate instructions or articulate logical conclusions, the actual mental execution of formal reasoning occurs outside the brain's language network. Furthermore, functional brain imaging confirmed that language regions remain inactive during inductive reasoning tasks, shifting our understanding of how complex problem-solving and cognitive architecture are organized. These insights carry significant practical implications for clinical evaluations, suggesting that communication difficulties in conditions like aphasia, stuttering, or foreign language barriers do not equate to a lack of intelligence or reasoning capability. Additionally, comparing human cognitive separation of language and logic with artificial intelligence models like ChatGPT provides a new benchmark for designing future AI systems that attempt to mimic human reasoning.
