Effects of Context on the Classification of Everyday Sounds
Abstract
The effects of context on the classification of everyday sounds was examined in five experiments. Context was produced by meaningful sounds and by phrases describing an environmental scene. All experiments presented listeners with pairs of test sounds that are confused in identification, but which are discriminable. These test sounds were presented for classification in isolation, and embedded in sequences of other everyday sounds. Three types of embedding sequences were used: 1) sequences consistent with the correct response; 2) sequences biased toward an incorrect choice; and 3) neutral sequences composed of randomly arranged sounds. Two paradigms, binary-choice and free classification were used. The results indicated that context could bias the response against the correct response, but did not raise performance above isolated classification performance. Performance was consistently poorest in biased context and best in both isolated and consistent context. Performance in random context depended upon the paradigm and the performance measure. In the free response paradigm, biased sequences produced responses that were appropriate for the context but incorrect as classifications of the sound. A signal detection analysis indicated that sensitivity in detecting a sound that is out-of-context remains constant for different paradigms, and that response bias is conservative, especially with a free response paradigm. Labels added to enhance context generally did not change the effects of context, suggesting that sounds alone are usually sufficient to generate these contextual effects. Keywords: Audition; Hearing; Classification; Pattern perception; Complex sound; Context. (JHD)
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1989
- Accession Number
- ADA214943
Entities
People
- James A. Ballas
- R. T. Mullins
Organizations
- George Mason University