Whom to trust? Inferred source credibility and response borrowing in a memory task

We investigated whether people can discriminate between sources of information that are either credible or respond at random, based only on their own knowledge and the responses provided by these sources. In three experiments, participants were asked to judge the validity of trivia statements. Some statements were accompanied by true/false responses provided by either a credible source or a source whose responses were random. In Experiment 1, participants first saw a set of easy questions, which provided the basis for assessing the relative credibility of the sources, before responding to a set of difficult questions, where response borrowing was assessed. In Experiments 2 and 3, participants solved a test composed of difficult questions only, but only after studying the correct responses to all these questions. In Experiment 2, there was no delay between the study and test phases, whereas in Experiment 3, the delay was 24 hours. In all experiments, more participants explicitly identified the more credible source in the postexperimental questionnaire than misidentified the noninformative source as credible. However, differentiated response borrowing—borrowing more responses from the credible than the noninformative source—emerged only in Experiment 2. Therefore, people can often explicitly infer source credibility from the responses the sources provide. However, using these inferences to regulate response borrowing is relatively less likely and happens only under specific, favorable circumstances.