Title: Discrimination measurement and experimental design. I. Variable-standard designs and categorical perception
Abstract:A continuous Signal Detection Theory analysis is presented for several variable-standard discrimination procedures (e.g., the ABX design) commonly used in psychoacoustic and speech experimentation. Th...A continuous Signal Detection Theory analysis is presented for several variable-standard discrimination procedures (e.g., the ABX design) commonly used in psychoacoustic and speech experimentation. The analysis shows that obtained data, in terms of either percentage of correct responses or d′-like measures, underestimate the discriminability of pairs of signals, and that the amount of underestimation depends on both the particular procedure used and the performance level obtained. Performance on such discrimination tasks is often compared to identification performance, the data being used to evaluate the hypothesis that an auditory dimension is perceived “categorically.” These comparisons, however, have been based on a theoretical framework which assumes (low) thresholds; we offer a restatement of the categorical perception hypothesis in terms of the continuous model, and test the hypothesis with existing data. For cases in which perception is found not to be categorical, the model provides a means for distinguishing the hypothesis that discrimination results from dual processing of categorical and noncategorical information from other explanations.Read More