DECISION PROCESSES IN DETECTION AND DISCRIMINATION

DESCRIPTION (provided by applicant): The goal of this research program is to develop a better understanding of fundamental aspects of human hearing by characterizing differences in decision processes for three tasks that use the same basic stimuli in different temporal configurations: intensity discrimination, increment detection, and forward masking. The classic and ubiquitous energy detector model of the decision process provides a good account of data for intensity discrimination. An alternative to energy detection, however, is required to account for results obtained in increment detection and forward masking, as well as the majority of other detection and discrimination tasks. Alternative models of the decision process, in particular "template matching", are currently poorly defined and make vague predictions. We will characterize properties of the decision process in these tasks by determining the effects of variability in overall stimulus level and the effects of introducing background noise. Performance measures include adaptive thresholds, psychometric functions, direct measures of internal noise, and the correlation of specific features of stimuli with each subject's response on individual trials. The specific aims are: 1) to determine the relation between peripheral nonlinearity and the slopes of psychometric functions; 2) to compare decision processes in intensity discrimination and increment detection; 3) to determine the effects of noise on intensity discrimination, increment detection and forward masking; and 4) to develop and compare direct measures of internal noise. This research effort will result in a better characterization of alternative decision processes, a detailed test of specific features of the decision process associated with each of the three basic psychophysical tasks considered here, measures of internal noise appropriate to each task, and a better understanding of the interaction of peripheral compression and internal noise in multi-stage models of masking.