Characterizing the performance limits of high speed image triage using Bayesian search theory

Abstract

The rapid serial visual presentation (RSVP) modality has been used in conjunction with neurophysiological and behavioral responses to identify targets within large volumes of imagery efficiently. The research reported here uses optimal search theory to characterize the limits of this approach. Search theory is used to inform the estimation of detection functions. These functions provide a principled basis for selecting presentation parameters that balance search efficiency and accuracy. Detection functions are also used to characterize individual differences in search performance and to assess the extent to which the RSVP presentation modality generalizes across a class of complex targets. © 2011 Springer-Verlag.