R. D. Sorkin
Huanping Dai
Dept. of Psychol., Univ. of Florida, Gainesville, FL 32611
A number of auditory scientists have proposed models for the multi-channel signal detection task: an observer must detect a signal that has components in several partially correlated channels (spectral, spatial, or temporal). This task has been reframed as a group detection problem: an array of observers must detect a signal by combining the observations of the individual detectors. It is shown how the performance of the ideal group detection system depends on the number of detectors, the correlation among detector observations, and the distribution of individual sensitivities. Also considered are some cases where the group decision is based on the binary outputs of the individual detectors. The results are applicable to a variety of different types of group detection systems. [Work supported by AFOSR.]