Cocktail Parties for Cats and Humans: 
Spatial Hearing in Noisy Environments

Norbert Kopco
Hearing Research Center
Dept of Cognitive and Neural Systems
Boston University

In 1953, Cherry used the term "cocktail party phenomenon" to describe
the remarkable human ability to "recognize what one person is saying 
when others are speaking at the same time." Cherry suggested that 
spatial hearing significantly contributes to this ability, allowing 
humans to selectively attend to sounds coming from one direction while 
suppressing other sounds. Here we present results of a behavioral and 
computational study that investigated the effect of spatial separation 
on audibility of simple non-speech sounds masked by broadband noise. 
We examined how different processing stages in the auditory pathway 
contribute to the benefit of spatial separation and we studied how the 
spectral content of the stimuli influenced performance. The results 
showed that monaural factors (i.e., listening with the ear that is 
spatially closer to the stimulus source vs. the noise source) 
determine performance for broadband stimuli, while both monaural and 
binaural information (i.e., processing of stimuli from both ears) is 
used when listening to low-frequency stimuli. To analyze the results, 
we proposed a simple model based on neurophysiological data from the 
cat Inferior Colliculus. The model succeeded only partially in 
describing the results, suggesting that more central, cortical 
processing must be considered even for simple non-speech stimuli.