Discrimination and Identification of Acoustic Transient Patterns.

Abstract

The relation between the discrimination and the identification of acoustic patterns can be addressed psychophysically or cognitively. The psychophysical approach predicts a monotonic relationship between performance on the two tasks. Cognitively, the relationship may depend upon the type of structure encoded from the patterns. Structure based upon similarities in runs could enhance discrimination but degrade identification. Hierarchical structural encoding might enhance both tasks. The relationship was investigated in three dual task experiments. In one experiment, trial and error learning was used whereas in the other two, observation of positive examples was used. All three experiments indicated that discrimination was superior to identification, and that the concurrent identification task improved discrimination performance above what has been obtained in single task discrimination studies. The effects of structure in the two tasks were equivocal but implied that the type of structured encoding is important and may be influenced by the procedure used to acquire the patterns. (Author)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 31, 1981
Accession Number
ADA110421

Entities

People

  • James A. Ballas
  • James H. Howard Jr.

Organizations

  • The Catholic University of America

Tags

Communities of Interest

  • Biomedical
  • Ground and Sea Platforms
  • Human Systems

DTIC Thesaurus Topics

  • Air Force
  • Applied Psychology
  • Artificial Intelligence
  • Cognitive Science
  • Computer Programs
  • Control Systems
  • Detection
  • Engineering
  • Human Factors Engineering
  • Identification
  • Military Research
  • Navy
  • Pattern Recognition
  • Psychology
  • Students
  • Systems Engineering
  • Task Performance And Analysis

Fields of Study

  • Psychology

Readers

  • Speech Processing/Speech Recognition.
  • Statistical inference.
  • Team-Based Human-Centered Cognitive Task Decision Making and Information Performance.