No stone unturned: Prevalence effects in interactive search are different than those in visual search

When carrying out a search for a target object, manipulation with the environment may be required to successfully detect the target. These searches are known as interactive searches. Many real-life examples of interactive search involve searching for targets that are unlikely to be present or are said to have low target prevalence. To date, the effects of low target prevalence upon interactive search behaviors remain unclear. We conducted two experiments to examine search exhaustiveness in interactive search, focusing on whether searchers were less exhaustive when prevalence was low, primarily in terms of behavior during target-absent trials. For both experiments, we found a standard effect of low prevalence on response accuracy, such that low-prevalence targets were more likely to be missed than high-prevalence targets. However, through the utilization of Bayesian analyses, we found strong evidence against the influence of prevalence upon response times and all other search exhaustiveness measures during target-absent trials. In other words, contrary to traditional visual search findings, changes in response accuracy were not a result of reductions in search exhaustiveness. We conclude that, during interactive search, even when prevalence is low, searchers operate under a no-stone-unturned approach. Under this approach, searchers are unwilling to provide an "absent" response without checking most-if not all-possible places, regions or areas in a display that could contain a target.