AUDITORY PROCESSING OF UNCERTAIN STIMULI

The long-term goal of this project is a better understanding of the
interaction of peripheral and central auditory processes that determine
performance under conditions of stimulus uncertainty, that is, when one or
more of the physical properties of the sounds cannot be predicted. Unlike
many psychophysical studies which seek to minimize the influence of central
processes, these experiments examine conditions in which central processes
significantly influence or dominate performance. Our previous work has
focused on the large effects of masker frequency uncertainty in the
detection of a known signal. In the proposed work, we undertake a more
general examination of the effects of stimulus uncertainty, adopting a
common approach to the measurement of both masker and signal uncertainty.
The absence of such an approach has hindered the development of more
general models of the effects of uncertainty. We test the hypothesis that
uncertainty effects may best be understood in terms of the ratio of
relevant (to be processed) variability to irrelevant (to be ignored)
variability, in an information-based analog to the ratio of
signal-to-noise-powers that govern energy-based masking. We first develop
stimuli and procedures which allow us to quantify the amount of masker
uncertainty, and then use these to explore the effects of irrelevant
uncertainty both alone and in combination with effects of relevant
variation or effects of energy-based masking. The experiments proposed are
grouped in three series which examine: 1) how best to measure uncertainty;
2) the sources of the uncertainty effects; and 3) the combination rules for
the effects of irrelevant uncertainty, relevant uncertainty, and
energy-based masking. Because stimulus uncertainty is inherent in everyday
listening situations, these experiments will contribute to more realistic
models of auditory processing that include both peripheral and central
processes.