The impact of effort on the search for information

Abstract

In everyday environments, people encounter vast amounts of information but are limited in what they can realistically use in making decisions. Because acquiring information often requires effort, individuals may favor less demanding sources even at the expense of their accuracy. However, it remains unclear how effort costs and the predictive validity of information jointly shape information search in decision-making. This study investigates how these factors interact in probabilistic environments. Across two experiments, participants explored a two-dimensional digital environment, using their mouse cursor to uncover cue information used to predict answers in a two-alternative forced choice task. Cue predictive validities ranged from 50% to 100%, and effort was manipulated through cursor speed and cue distance. Experiment 1 tested how effort costs and predictive validity influenced sampling when validities were explicitly presented on screen, whereas Experiment 2 required participants to learn predictive validities through trial and error. In both experiments, higher effort reduced sampling of less valid cues, while perfectly predictive (100%) cues were consistently prioritized even under greater effort, demonstrating a certainty effect in information search. When predictive validities were hidden in Experiment 2, effort exerted a stronger deterrent effect than it did when predictive validities were known (Experiment 1), and individual variability in exploration strategies increased. Analyses further indicated that participants treated perfectly predictive cues as categorically distinct, and the functional form of effort’s influence was similar to that of a delay discounting curve. Together, these findings support a cost-benefit view of information search, in which people weigh expected information gain against the cost of effort. Effort constrains information sampling, but perfectly predictive information resists the cost of effort much more than less predictive information.