Abstract
Low frequency of target appearance in visual search increases the chance that an observer will respond ‘absent’, resulting in an increase in correct rejection rates, but also a decrease in hit rates (the ‘prevalence effect’). Additionally, searching for more than one target at a time reduces the chance that a target will be detected (the ‘dual-target cost’). Both the prevalence effect (Wolfe et al., 2007) and the dual-target cost (Menneer et al., 2007) have recently been cited as a cause for concern for those working in airport X-ray screening, where screeners search for multiple threat items that appear infrequently. Here, we present two experiments in which target prevalence was varied across a full range of presentation rates (across two experiments, prevalence levels of 2%, 20%, 24%, 50%, 76%, 80% and 98% were examined), finding that, although the dual-target cost did not interact with the prevalence effect, there was a surprising imbalance between the low and high prevalence conditions. In very high prevalence conditions (80%, 98% prevalence), participants were exceedingly biased towards responding ‘present’. Although the opposite effect was observed in the very low prevalence conditions (2%, 20%), with a bias towards responding ‘absent’, the effect was considerably weaker. Thus, there appears to be some intrinsic imbalance in the manner in which ‘present’ and ‘absent’ responses are made as prevalence varies.
Supported by a grant from the UK Department for Transport and the Engineering and Physical Sciences Research Council