Wordless Logic

MIT neuroscientists showed that logical reasoning and language run on entirely separate neural systems. Even people with severe language damage from stroke solve complex syllogisms as well as healthy controls — while brain scans show language regions go silent during reasoning tasks. Toggle modes to see which network fires.

Hover over a brain region to read what the MIT study found about it

Study published
July 2026
PNAS — Fedorenko lab, MIT
Participants
2 groups
healthy adults + stroke/aphasia patients
Key finding
Fully separate
logic runs without language regions
Aphasia impact
Zero
logic scores unchanged after stroke
Syllogism puzzles — like the MIT tasks0/0 correct
Premises
All A are B
All B are C
Conclusion — valid?
All A are C

MIT's researchers used exactly this type of formal syllogism. Stroke patients with almost no ability to produce or understand spoken language scored identically to healthy controls — demonstrating reasoning proceeds without language.

MIT research pipeline

Why this matters:For decades, cognitive scientists assumed that abstract thought — especially formal logic — requires internal language: a “language of thought” voiced silently to ourselves.

The Fedorenko lab’s PNAS study punctures that assumption. fMRI showed language-network regions (Broca’s and Wernicke’s areas) firing during sentence tasks but staying nearly silent during syllogism and rule-induction tasks — while the multiple-demand network (bilateral parietal and lateral prefrontal cortex) did the opposite. Crucially, stroke patients with severe aphasia — unable to form sentences or understand speech — solved the same logic puzzles just as well as people with intact language, ruling out any residual “inner voice” shortcut. Reason, it turns out, is not narrated.