Abstract:
Listeners can detect a weak tonal signal in noise more successfully when the signal is at an expected frequency than when the signal frequency is unexpected [e.g., Dai et al., J. Acoust. Soc. Am. 89, 2837--2842 (1991)], suggesting that listeners focus their attention on frequency regions where distinctive information is likely to be found. This ability may benefit listeners faced with the task of discriminating natural sounds in noisy environments. However, demonstrations of selective weighting of acoustic information in discrimination tasks are relatively few. Results will be reported of an experiment in which subjects were required to discriminate two complex tones consisting of seven equal-amplitude components. The tones were distinguished by an increment in the amplitude of a single spectral component. When the increment was added to an expected component, discrimination performance was better than when an unexpected component was incremented. A similar result was found in a second experiment in which the levels of all spectral components were randomized between each trial. The relatively better discrimination performance shown for expected discriminanda is interpreted as an effect of attention directed by expectation to the spectral region encompassing the incremented component. [Work supported by UK MRC.]