Stimulus Presentation Formats and Measurement Techniques for the Quantificationof Target Detection Performance

摘要

The forced-choice (FC) format for stimulus presentation and performanceassessment has been gaining popularity over other formats in a variety of human performance tasks, such as target detection and acquisition. Laboratory experiments, comparison investigations, and system performance assessments that require statistical testing, for example sonar system MDL, have been reported in the literature with claims that use of the FC procedure leads-to simplification of data handling and increased cost-effectiveness. These and other claims are examined in the present report and a number of concerns are raised about the nature of the information acquired when the FC method is used for the quantification of performance in tasks that primarily involve vigilance, monitoring, and search behaviors. These concerns may be summarized as follows. First, the kinds of performance outcomes assessed using the FC format are often different from those assessed by continuous-search (CS) procedures. Second, the FC format holds some aspects of response bias constant, making it impossible to assess many variables that are of paramount importance in sonar system development, such as the effects on performance of signal probability, motivation, training, and experience. Third, the procedures have sometimes been misunderstood, and the term forced-choice has been misapplied to situations that technically are not FC. Finally, although there may be some valid practical reasons for using the FC format for stimulus presentation in final system performance evaluation, there is no compelling evidence that the method is superior to others in terms of data analysis, programming, or cost efficiency for the controlled laboratory experiment....