THE DETECTION OF MARKOVIAN SEQUENCES OF SIGNALS.
Abstract
The influence of constrained stimulus sequences on detection was studied in a two-alternative temporal forced choice task with feedback. Three observers listened to a weak pure tone embedded in noise whose probabilities of occurrence and repetition in an interval were governed by a first-order Markov process. Each observer listened to examples of each of nine different Markov chains. Results were : (a) A single function relating detections to false alarms fitted individual sets of data well, in agreement with the theory of signal detectability, except (b) that detection was higher for more extreme repetition probabilities; (c) Responses depended strongly on the previous stimulus with (d) the dependence being peculiar to a given chain; (e) Detection probabilities increased during runs of signals in the same interval, yet (f) probability of detection on the first trial of a run in a given interval did not depend on the length of the preceding run in the other interval. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 15, 1964
- Accession Number
- AD0609457
Entities
People
- Edward C. Carterette
- Morton P. Friedman
Organizations
- University of California, Los Angeles