The Brain’s Bad Bet That the World Won’t Stop Shifting

Somewhere in the machinery of a delusion is a wager about how much the world changes. A person leaves a card game convinced the rules keep flipping underneath them, so they hedge, second-guess, and read hidden intent into a run of ordinary bad luck. New research following adults recovering from an acute psychotic episode reports that this wager, an inflated expectation that the environment is volatile and unstable, loosens as delusions ease, moving in step with symptom severity over six months rather than sitting fixed in the mind. That it moves at all is the finding that matters, because a mental habit that shifts during recovery is one a treatment might eventually reach. The work appears in Biological Psychiatry: Cognitive Neuroscience and Neuroimaging.

The wager has a technical name, a volatility prior: your standing expectation, before any fresh evidence arrives, of how likely the rules around you are to change. Set it too high and the brain starts treating noise as signal.

In schizophrenia-spectrum disorders, that setting often runs abnormally high. The world reads as chaotic and unpredictable even when it is holding still, and the mind, straining to explain the churn, reaches for a story. Someone is following me. The looks mean something. Delusions are the most common symptom of psychosis, strongly held beliefs that survive any amount of contradicting evidence, and they respond poorly and unevenly to the antipsychotic drugs that remain the front-line treatment. Over 70 per cent of people in a first episode of psychosis experience them, and a large catchment study of first presentations put the figure higher still, with persecutory themes the most common.

“Patients deserve more effective treatments,” says Julia Sheffield, a psychologist at Vanderbilt University Medical Center who led the work. To watch the wager change, you have to catch people at the moment it is most likely to move. Sheffield and her colleagues recruited 75 adults with schizophrenia-spectrum disorders, all admitted to a psychiatric hospital with delusional thinking, and began tracking them within weeks of discharge, as recovery was underway.

Reading a Bet Off a Deck of Cards

You can’t ask someone for their volatility prior; they don’t know the number any more than you know yours. So the researchers inferred it from behaviour. Participants played a card game with three decks, hunting the one most likely to pay out, knowing the best deck could change without warning. Partway through, unannounced, the payout odds shifted, tilting the floor under anyone who had grown confident. How players chased or held through that churn, fed into a layered statistical model of belief called a Hierarchical Gaussian Filter, yields an estimate of how volatile each person expected the game to be. The 75 patients did this alongside 71 comparison participants, six times each across six months.

At the start, the patients’ volatility priors sat well above the comparison group’s, alongside more severe paranoia and delusional thinking. That much echoed earlier snapshots. The new part was the follow-through. Within the patient group, volatility priors fell over the six months, and crucially they tracked delusion severity as they went, the expectation of an unstable world easing as the delusions themselves loosened their hold. The link held specifically for delusions and paranoia, not for depression or worry, and it survived adjustment for antipsychotic dose and baseline cognitive ability. Something fairly particular is moving here, not a general lift in mood dragging everything up with it. That movement is the crux of what Sheffield set out to test: whether the expectation is “a state-marker (temporary condition) that could be modified with treatment,” or “a trait-marker (permanent vulnerability) that could be a target for prevention.” A thing that shifts as you recover is the first kind.

The specificity cuts finer still. The expectation of volatility tracked persecutory delusions, the kind about being watched or pursued, more tightly than grandiose ones, and it followed how strongly a belief was held and how much it preoccupied a person rather than how much distress it caused. A picture emerges of a particular cognitive stance, braced for instability, that belongs to a particular flavour of delusion.

The Trap That Practice Sets

Hand someone the same puzzle six times and they get better at the puzzle, a problem when what you are measuring is meant to be their state of mind, not their practice. The researchers built in defences, six versions of the card game, each introduced as new, and lean on a telling contrast: the comparison group’s priors dropped sharply between the first two sessions, the fingerprint of simple learning, while the patients’ decline unfolded gradually across the whole six months and never flattened into normal. Patients were still showing elevated priors at the sixth sitting, so the task had not merely been over-learned. Even so, the authors say that practice effects remain a confound they mitigated rather than banished.

The caution runs deeper than the arithmetic. Because the study only watched these measures move together and changed nothing, it can’no’t show that the skewed expectation drives the delusion rather than the reverse/ Or that some third thing nudges both. And a second hedge the researchers refuse to drop: delusions, paranoia and the priors themselves were all still elevated at six months, none fully normalised. Whether this expectation ever resets to typical, or is better read as a hybrid, part passing state and part standing trait, the data leave open. The team’s own earlier work points both ways, finding ordinary priors in patients without current paranoia but also a link between the expectation and a history of childhood adversity, the sort of thing that leaves a lasting mark.

What survives the caveats is the direction of travel. For decades delusions have been treated bluntly, with drugs that damp down the brain’s dopamine signalling and often disappoint. A measure that starts high and comes down as a person recovers, and comes down in step with the very symptom clinicians are trying to shift, is the kind of foothold the field has been short of. Cameron Carter, the psychiatrist who edits the journal, calls it “the first time that volatility priors have been assessed longitudinally in psychotic disorders during recovery,” which is partly why a modest single-site study is drawing notice. It hints that cognitive behavioral therapy for psychosis, already recommended for people recovering from an episode, may already do some of this work without naming it—and now has a specific target to aim at: recalibrating someone’s sense of how changeable the world really is.

What nobody yet has is a way to talk about volatility with a patient in the room, a shared vocabulary for a bet the mind is placing without announcing it. Build that, and the wager stops being something only a card game and a statistical model can read. It becomes something two people might change together.

FAQ