‘‘Damped’’ wideband noises and ‘‘ramped’’ wideband noises were constructed; the former have a repeating, exponentially decaying amplitude‐modulation function, the latter a repeating, exponentially rising amplitude‐modulation function. They sound remarkably different; there is a continuous background hiss in the ramped percept that is absent in the damped percept. An experiment is reported showing that listeners could reliably discriminate the two over a wide range of modulation half‐lives and periods. The short half‐life at which discrimination failed was between 0.5 and 1 ms, similar to other measures of temporal resolution. The longest half‐life at which discrimination failed increased with increasing modulation period, from 6 (12.5‐ms period) to 160 ms (100‐ms period). The leaky integrator model of temporal resolution, together with either the standard deviation or max/min decision statistics, failed to predict the psychometric functions. The auditory‐image model, which preserves phase‐locking information through the integration process, can predict the psychometric functions. The decision statistic is the hiss correlate, the activity present within the auditory image averaged across frequency, integration interval, and time. The current simulations are the first to use the auditory‐image model to predict quantitatively psychophysical data.

This content is only available via PDF.
You do not currently have access to this content.