Comparison of ROC and AFC methods in a visual detection task.

Abstract

The visual observer performance for detecting low-contrast patterns by employing both the receiver operating characteristic (ROC) and the 9-alternative forced choice (9-AFC) method is investigated. For each observer and each experimental technique, the results were reproducible with respect to a repeated observation. When the results of different observers for each method are compared, a strong variation of values is found. From the ROC data obtained, the expected percentages of correct detection in a 9-AFC experiment is calculated and compared with the corresponding values of the actual 9-AFC experiment. For some observers, the latter values were significantly higher than the computed ones, whereas for others the results showed good agreement with predictions from theory.