The neural geometry of surprise: Graded boundary breakthroughs in the human brain

Authors

Hao, Y., Jiang, C.

Abstract

How whole-brain neural states reorganize when events violate observers' predictions remains unclear. We propose that naturalistic surprise is expressed neurally as a graded boundary breakthrough. The more surprising the event, the farther the whole-brain state moves beyond its non-surprise reference range. Using an openly available fMRI dataset of 49 participants viewing 36 magic videos, we show that surprise redistributed whole-brain geometry in a network-specific rather than uniform manner. To quantify this transition, we used non-surprise brain states to define a reference range along the surprise-related direction and measured how deeply each state moved beyond it. Boundary exceedance (E_BB) expresses this reference-normalized depth in bits; each additional bit means that a reference state is half as likely to reach at least that far. Exceedance deepened with independently rated surprise intensity, and this scaling was state-specific. The breakthrough state moved farther beyond the boundary as surprise increased, whereas the paired reference state did not. Peri-event trajectories showed that the event deepened an excursion already under way. Exploratory analyses linked exceedance to one-week recognition memory in a surprise-dependent manner. Together, these findings support boundary breakthrough as a systems-level account of naturalistic surprise, in which more surprising events carry the whole-brain state farther beyond its non-surprise reference range.

Preprint server: bioRxiv
The authors list and abstract were imported from bioRxiv on 09 Sep 2026.

Stats

1-terrible, 9-excellent. How would you rate this preprint? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.