Reduction of unique noise in the psychophysics of hearing by Group Operating Characteristic analysis

Group Operating Characteristic (GOC) analysis can be used to reduce the effects of unique noise on data from psychophysical experiments on hearing. Unique noise can come from physiological sources such as heartbeat and breathing and from other internal sources such as criterion variability, inattention, and faulty memory. The theory of GOC analysis is described and applied to the results of frequency discrimination experiments with human, pigeon, and simulated observers. The method yields relatively unattenuated measures of sensitivity, recovers Receiver Operating Characteristic curves from noisy choice data, and estimates the relative magnitudes of unique and common noise.