AI and the World's Hard Problems · Life sciences · arXiv · 2026-10-20
Spontaneous brain-wide rotational dynamics constrain cue-evoked behavior
The authors develop Rotational Component Analysis (RCA), a dimensionality-reduction method for detecting low-variance rotational modes in brain-wide neural data. Applied to monkey, human, and mouse recordings, it shows that resting rotational phase in mice predicts whether an animal pulls a lever after a cue. Rotational irreversibility also captured early psychosis in one human dataset where principal component analysis found no group difference.