In everyday life, humans often make decisions without full knowledge of the consequences. To improve their decisions, people often try to obtain more information prior to making a choice. We’ve developed new ways to examine how people sample information before committing to a decision. Our findings show how individuals vary evaluate the cost of effort against potential rewards, and why time pressure can paradoxically lead to more efficient sampling of information.

Computational modelling revealed that cognitive effort itself carries a cost, improving explanations of both healthy and impaired behaviour. People with hippocampal lesions, for example, sample rapidly but inefficiently. By contrast, individuals with subjective cognitive decline (memory symptoms often due to depression and anxiety) show excessive sampling, reflecting hypersensitivy to uncertainty. Patients with Parkinson’s disease and small vessel cerebrovascular disease, on the other hand, commit to decisions too quickly, tolerating high levels of uncertainty and making poorer decisions. Together, these results illuminate the neural computations that underpin optimal decision-making and explain impulsive or indecisive behaviour in clinical populations.