A fifty-fifty chance of being shocked is, according to a 2016 study in Nature Communications, more stressful than being told a shock is definitely coming. Researchers at University College London measured this in 45 volunteers across self-reports, pupil dilation, and skin conductance, and the curve held: stress peaked at the midpoint of uncertainty and fell away at either end.
That finding sits awkwardly against the story people tell about themselves when they settle a question early, before the evidence is fully in, before the outcome is clear. The story is usually about decisiveness. About not dithering. About having the confidence to commit. The UCL data offers a different reading: the decision may not be an expression of strength so much as a relief valve for a brain under considerable strain.
The study was led by Archy de Berker, with Robb Rutledge and Sven Bestmann among the co-authors at UCL’s Institute of Neurology. It mapped the relationship between uncertainty and stress with unusual precision, and found that the most stressful condition the brain can be in is not knowing what is coming, even when what might be coming is painful. A guaranteed bad outcome, it turns out, is measurably easier to bear than a fifty-fifty one.
The experiment
The researchers recruited 45 volunteers and placed them in a computer game in which they turned over rocks that might conceal a snake. Finding a snake earned a mild electric shock to the hand. Over the course of the game, participants learned which rocks were more likely to have snakes beneath them. The probabilities shifted throughout, meaning the degree of uncertainty each person experienced varied continuously and could be tracked.
To measure stress, the researchers used both self-reported ratings and two physiological markers: pupil dilation and skin conductance, the degree to which participants were perspiring. Both measures moved in tandem with uncertainty, not with the shocks themselves. Participants sweated more and their pupils dilated more not when they were shocked, but when they did not know whether they would be. And the peak of that stress response occurred when the probability of a shock was exactly 50 per cent.
“It’s much worse not knowing you are going to get a shock than knowing you definitely will or won’t,” said lead author Archy de Berker. “We saw exactly the same effects in our physiological measures — people sweat more and their pupils get bigger when they are more uncertain.”
The inverted-U relationship between probability and stress, with high stress in the middle and lower stress at the extremes, held consistently across participants. Zero chance of a shock was calm. One hundred per cent chance was also, relatively, calm. Fifty per cent was the worst of all.
What this means for how we decide
The study was designed to measure stress, not decision-making, but the implications for how people resolve uncertainty are direct. If the brain treats not-knowing as genuinely aversive, more aversive, physiologically, than knowing something bad is certain, then collapsing that uncertainty by committing to an outcome becomes a way of reducing distress rather than a neutral act of information processing.
This reframes a common behaviour in professional settings. When someone calls a meeting to a conclusion before all the information is gathered, or commits to a strategy before the analysis is complete, the conventional explanation is that they are decisive. That they have made a judgment about when enough evidence is enough. The UCL research suggests an alternative: the discomfort of not knowing may be driving the decision as much as any reasoned assessment. It is, in that sense, a stress management strategy wearing the clothes of leadership.
The same logic applies in negotiations, hiring, and investment. Any domain where sitting with ambiguity is part of the process, but where the pressure to resolve uncertainty can shorten the time horizon for information gathering.
The upside, and the limit of the framing
The study is not an argument against responsiveness to uncertainty. The researchers also found that participants whose stress responses tracked uncertainty most closely were better at predicting which rocks concealed snakes. In other words, the people most sensitised to not-knowing were also the most attuned to the probabilistic signals in their environment. The stress was doing something useful, sharpening attention in proportion to actual risk.
Senior author Dr Sven Bestmann framed this in evolutionary terms: “Appropriate stress responses might be useful for learning about uncertain, dangerous things in the environment.” The problem is not that the brain responds to uncertainty. It is that in modern professional contexts, where decisions play out over weeks rather than seconds, the stress signal can fire before the information is available to use it well.
The distinction the research opens up is between stress that is alerting, prompting closer attention to real signals, and stress that is simply uncomfortable, resolving itself into premature closure. The former is adaptive. The latter tends to look like decisiveness until the consequences arrive.
The computational model underpinning the study offers a template that researchers have since applied to a range of questions about how the brain manages ambiguity, including in clinical contexts where pathological intolerance of uncertainty plays a role in anxiety disorders.
For organisations, the research raises a pointed question. If unnecessary uncertainty imposes a measurable stress cost, and that cost tends to resolve itself in premature closure, then the conditions under which people make decisions matter as much as the decisions themselves. That is a different kind of problem than asking people to sit with ambiguity more comfortably. It locates the intervention upstream of the individual.
