Perceptual selection based on alphanumeric class: Evidence from partial reports

In two experiments, subjects were shown brief displays of mixed digits and letters. In partial report, the task was to report as many digits as possible while ignoring letters, or vice versa. In whole report, both digits and letters were to be reported. There were two main findings. The probability of reporting digits (letters) was greater when letters (digits) could be ignored (partial report) than when they could not (whole report). When letters (digits) could be ignored, the probability of reporting a given digit (letter) increased as the number of digits (letters) in the display decreased. Both findings held even at exposure durations so brief that only two or three characters were ever reported, minimizing the chance of output interference. While the tendency was for hit rates and false alarm rates to covary, the major findings held even in the ROC curves of individual subjects, and so were not due to simple criterion shifts. The results suggest that digits and letters can be separated “preattentively,” that is, prior to a limited-capacity system (LCS) in visual perception. Thus, in partial report, the main competition for LCS comes from characters in the target class, nontargets being rejected at a prior level.