Brain Plots Emotions on a Hexagonal Grid, Scans Suggest

Researchers at Yumeng Ma's team ran 18 healthy adults — 13 of them female, average age around 40 — through four brain-imaging runs inside a 3T fMRI scanner. Two runs used faces: five-second videos digitally blending seven emotion categories, happy, surprised, angry, afraid, sad, disgusted and neutral, from one state into another. The other two used phrases pulled from eight everyday contexts, paired into five-second sequences that shifted emotional tone. After each clip or pair of phrases, participants judged whether pleasantness or arousal had risen, fallen or stayed put.

The scans delivered a striking shape. Activity in the ventromedial prefrontal cortex, the region handling emotion regulation, decision-making and self-control, showed sixfold hexadirectional symmetry as people tracked emotional change. Brain patterns looked more alike when emotional paths lined up with the same 60-degree grid angles, and less alike when they didn't.

The grid didn't care how the emotion arrived. Patterns estimated from written phrases predicted responses to morphing faces, and the reverse held too. That points to the brain encoding shared emotional meaning rather than surface visual features, according to the study published in PLOS Biology.

Years of behavioral work already showed people organize emotions, words and facial expressions along one shared map with two axes: valence, how positive or negative something feels, and arousal, how calm or energized. Spatial navigation research had found the brain lays out physical space with grid-like signals repeating every 60 degrees. Existing theories credited the vmPFC with merging context and emotional value, but never explained what format it used to store emotion knowledge.

A second signal sat alongside the grid. How strong the hexagonal pattern looked had nothing to do with whether participants answered consistently or contradicted themselves. Instead, the vmPFC carried a separate reading of the distance between two emotions on the valence-arousal map — and the crisper that distance signal, the steadier the judgments.

The researchers conclude the vmPFC stores emotional knowledge on an interconnected, coordinate-based grid, letting emotions be understood in relation to one another instead of as disconnected categories. That organization, they suggest, is what lets the brain use what it knows about emotions to predict what's coming and adapt its decisions.