The brain doesn't read words. It reconstructs meaning.
TRIBE predicts where in the cortex a sentence lights up — not because the brain stores dictionaries, but because meaning is built across regions, each carrying some part of what a word is.
Brain Mirror
Type something. Anything.
The brain above shows what published research associates with text semantically near yours — predicted activation across twenty regions, not a scan of you.
BGE-small embedding baseline · TRIBE inference: deferred
0 / 5000
Embedding baseline · BGE-small
TRIBE engine online · TRIBE v2 (Meta FAIR, d'Ascoli et al., 2026) model card · paper · CC-BY-NC-4.0 · text-only · average-subject
What you just saw is a prediction — not your brain.
A model of the average brain, trained on participants who lay still in scanners while sentences played. The gap between that average and yours is part of the show.
Here are some texts that surprised us.
Hover one to feel the pattern. Click to load it into the mirror above.
Three texts that surprised us