Why do Hearing-impaired Listeners Not (always) Enjoy Cocktail Parties?

Christophe Micheyl - cmicheyl@umn.edu
Andrew J. Oxenham - oxenham@umn.edu

Department of Psychology
University of Minnesota
Minneapolis, MN 55, USA.

 

Popular version of paper 4aPP2
Presented Thursday morning, April 22, 2010

159th ASA Meeting, Baltimore, Maryland

 

Imagine yourself at a cocktail party. You can hear chatter, glasses clinking, and soft ambient music. But try as you might to focus on what the person next to you is saying, their voice is extremely difficult to separate from other voices and sounds. Even piecing together a few snippets of coherent speech out of this jumble of sounds requires sustained attention, and tremendous effort. This is what individuals who suffer from hearing loss often experience in situations where multiple sound sources compete for attention. Understanding the origin of these selective-listening difficulties, so that they can be addressed more effectively, is an important goal of current research in psychoacoustics (the scientific study of auditory perception) and auditory neuroscience (which is concerned with the biological basis of hearing). Here, we introduce recent findings from our laboratory, which shed some light on this issue. These findings indicate that listeners ability to extract information from concurrent sounds can be predicted, to a large extent, by the frequency resolution of the auditory sensory organthe cochlea.

To understand these findings, it is useful to think of the cochlea as a spectrum analyzer: it breaks down complex sounds (such as musical notes or vowels) into their elementary components, mapping each frequency to a different place along a vibrating membranethe basilar membrane. On this membrane sit two types of cells that both play a crucial role in hearing. The inner hair cells transform mechanical vibrations of the basilar membrane into electrochemical signals, which are transmitted to the brain via the auditory nerve. The outer hair cells amplify the vibrations of the basilar membrane selectively, thereby boosting soft sounds, while at the same time making the ear more fine-tuned in frequency. Damage to the outer hair cells (which can result from exposure to loud sounds, ageing, viruses, and other causes) usually leads to reduced auditory sensitivity (soft sounds are no longer detected) and to reduced frequency selectivity (schematically, each place on the basilar membrane now responds to a broader range of frequencies).

Figure 1 provides some insight into the effect of reduced cochlear frequency selectivity on the internal representation of sounds in the auditory system. In this example, the stimulus is a complex tone, that is, a tone that contains multiple frequencies in it. It could be a musical sound. The frequency components of the tone are shown as solid black lines; each line represents one component. The blue and red curves show simulated patterns of auditory excitation in response to the tone. These patterns were computed using a mathematical model of the ear (Glasberg and Moore, 1990). They indicate how the spectral components of the complex tone (solid black lines) are represented inside the auditory system of a normal-hearing listener (blue curve), and the auditory system of a listener with a 40-dB hearing loss, and frequency selectivity reduced by a factor of 2 (red curve). Note how the spectral components of the tone are represented as salient peaks in the normal excitation pattern, but lost in the impaired case, due to insufficient frequency resolution.

 

Figure 1. Patterns of excitation produced by a harmonic complex tone in a normal auditory system (blue curve), and in an auditory system with reduced frequency selectivity (red curve).

If you observe the figure carefully, you will notice that the frequency components of the tone (solid black lines) are regularly spaced, with frequencies equal to all integer multiples of 400 Hz. This reveals that this complex tone is harmonic, with a fundamental frequency (F0) of 400 Hz. Harmonic complex tones are a very important class of sounds, which encompasses most musical sounds and many biological communication signalsincluding vowels. They usually have a distinct pitch, which is determined by the F0the higher the F0, the higher the pitch. Pitch is a very important auditory attribute. It is used to convey melody in music, and prosody in speech. In addition, pitch can be used to distinguish sounds, such as a male voice and a female voice, because male voices usually have a lower pitch than female voices. Thus, voices and other sounds can be tracked selectively over time based on their pitch.

In order to determine whether the presence of salient peaks in auditory excitation patterns is a good predictor of listeners ability to accurately perceive the pitches of concurrent harmonic complex tones, we measured pitch-discrimination thresholds for a harmonic complex tone (target) presented simultaneously with another harmonic complex (masker) under a wide variety of stimulus conditions. We then computed auditory excitation patterns for the mixture of target and masker, and looked for covariations between the magnitude of the peaks in these patterns, and the thresholds obtained by the listeners. We found that, in all conditions in which the excitation patterns evoked by the mixture of target and masker did not contain salient peaks, the listeners thresholds were high, indicating that the listeners were unable to accurately hear out the pitch of the target. Conversely, in all conditions in which the excitation patterns evoked by the mixture of target and masker did contain salient peaks, the thresholds achieved low discrimination thresholds. Thus, it appears that the presence of salient peaks in simulated auditory excitation patterns is a good predictor of listeners ability to accurately hear out the pitches of concurrent sounds.

These findings lend support to the hypothesis that the fine-grain frequency analysis performed by the cochlea determines in large part how successful listeners are at analyzing concurrent sounds perceptually, and that the loss of this fine-grain frequency analysis following cochlear damages can go a long way toward explaining why hearing-impaired individuals often have considerable listening difficulties in settings where multiple sound sources are simultaneously activesuch as cocktail parties.

[Work supported by NIH grant R01 DC05216.]

Glasberg, B. R., and Moore, B. C. J. (1990). "Derivation of auditory filter shapes from notched-noise data," Hearing Res. 47, 103-138.

Micheyl C., Keebler, M.V., Oxenham, A.J. (2010) Pitch perception for mixtures of spectrally overlapping harmonic complex tones, J. Acoust. Soc. Am. (in press)

Micheyl, C., and Oxenham, A. J. (2010) "Pitch, harmonicity, and concurrent sound segregation: Psychoacoustical and neurophysiological findings," Hearing Res. (in press)