I saw your post on the auditory list.
In particular chaptar 4.
In short: I presented two tokens of guassian noise to listeners in each trial. Their task was to decide whether these were the same or different. Many trials were presented. Listeners were quite good at this for certain conditions. In one such condition we augmented the noises with extra noise that did not contribute to the listening task (fringes). This appeared to have a detrimental effect on performance. The fringes were also altered in several ways to investigate the effect of fringe properties.
In the discussion we induce that the detrimental effect is due to the the listener not being able to segregate the fringes into seperate auditory objects. Once sufficient segregation cues are introduced, and the listener is able to segregate target from fringe, the detrimental effect is partly mitigated.
Hope this helps,