New study published in Nature Communications (open access) by Archy de Berker on the causes of stress. We show that uncertainty plays a major role in determining both subjective and physiological stress responses. Subjects played a computer game in which they turned over rocks and had to guess whether there were snakes under them. If there was a snake, they received a mild electrical shock. Subjects learned where the snakes were, making good guesses even when probabilities changed over time.
We used computational models to estimate the uncertainty that subjects experienced during the task. Pupil dilation and perspiration were both correlated with a subject's uncertainty. In a previous study published in PNAS (also open access), I showed that self-reported happiness reflected the cumulative impact of past rewards and expectations. Here, we used the same computational approach to show that self-reported stress reflected the cumulative impact of past shocks and uncertainty about those shocks. It is worse not knowing whether you will get a shock than knowing you definitely will or won't. We also found that subjects for whom acute stress responses closely tracked uncertainty were better at guessing where the snakes were, suggesting that getting stressed at the right times might be beneficial.