MIT Study Indicates Logical Reasoning Can Occur Without Language

Cognitive neuroscientists at MIT’s McGovern Institute for Brain Research have found that logical reasoning is possible even when individuals experience severe language impairments. The research, led by MIT associate professor Evelina Fedorenko, highlights that the brain regions responsible for language processing are not activated during reasoning tasks. These findings were published in the journal PNAS.

The debate over the relationship between language and thought has persisted for centuries. While many scholars believe language is vital for cognitive processes, Fedorenko and postdoctoral researcher Hope Kean have proposed that abstract reasoning may rely on distinct brain mechanisms separate from the linguistic faculties.

Kean explains that “abstract thinking has properties that look a lot like language,” but highlights potential limitations, noting that ordinary language may lack the precision required for logical reasoning. The sequenced nature of language can further complicate how multiple pieces of information are processed in reasoning tasks.

Investigating Reasoning Without Language

To explore how the brain handles logic independent of language, Kean collaborated with Rosemary Varley from University College London, who specializes in acquired language disorders. They studied two participants who suffered strokes, resulting in significant language processing difficulties.

The research included creating logic tasks based on numerical relationships and visual patterns. In one task, participants were asked to identify hidden rules that transformed lists of numbers. Another involved choosing geometric patterns to complete a series. Results indicated that the participants with language impairments performed comparably to a control group and could convey their findings through gestures.

“It really upends a theory that says that symbolic rule induction is not possible without linguistic capacities,” said Kean.

Brain Imaging Reveals Distinct Pathways

Further analysis was conducted using MRI scans on healthy adults to observe brain activity during logical problem-solving. Some tasks were aimed at identifying language-processing regions, while others examined the brain’s “multiple demand network,” which supports complex problem-solving abilities.

The study verified that the brain’s language system did not activate during either inductive or deductive reasoning tasks. Surprisingly, while the multiple demand network was active during inductive reasoning, it showed no involvement with deductive reasoning.

This evidence supports the notion that language and logical reasoning are governed by separate systems within the brain. Earlier studies from Fedorenko’s lab have also demonstrated that various forms of thought, like object categorization and social reasoning, do not rely on language.

Implications for Understanding Aphasia

The findings hold significant implications for understanding aphasia, a condition characterized by impaired language ability. Professionals recognize that a loss of linguistic skills does not equate to a decline in intelligence. Individuals with aphasia may still engage in complex activities such as chess or puzzles, despite challenges in communication.

Fedorenko remarked, “This research adds to a growing body of work establishing that even severely aphasic individuals can preserve their ability for abstract logical thought, a defining feature of our species.” She emphasized the need to educate the public that linguistic difficulties do not reflect overall intelligence or capability.

Insights for Artificial Intelligence

The study’s implications extend to the realm of artificial intelligence. Current large language models, such as ChatGPT and Claude, are purely text-based yet can simulate various forms of reasoning. However, the research indicates that human logical thought and language function distinctly.

Kean noted that exploring the differences between human cognitive processes and AI reasoning could inform future developments in artificial intelligence. As research continues, understanding how the human brain accomplishes reasoning will remain a pivotal area of inquiry.