Mechanisms and mechanics of auditory masking

Abstract

It is rare to hear just one sound at a time. More often than not, our experience of listening to someone speak, or hearing a piece of music, is made more challenging, and possibly more annoying, by the presence of interfering sound, be it traffic noise, air-conditioning rumble, or other people talking at the same time. In extreme situations, such as in a noisy restaurant or bar, the background noise may prevent us from even hearing the conversation we are trying our best to follow. This phenomenon is known as masking, and it occurs in every sensory system. In this issue of The Journal of Physiology, a study by Recio-Spinoso & Cooper (2013) traces the origins of masking down to the first stages of electro-mechanical processing in the cochlea. When multiple sound sources are present at the same time, the variations in air pressure produced by each source combine to form a single pattern of rapidly varying air pressure at each of the listener's eardrums, causing them to vibrate. The auditory system must then decompose the patterns of vibrations at the eardrums to provide us with information about the individual sound sources, such as their identity and location. Given this daunting challenge, the question might be how we are ever able to hear out sounds from a mixture, rather than why our ability to do so sometimes fails. A crucial first step in this decomposition process is accomplished within the cochlea, where different frequencies within a complex sound maximally stimulate different places along the length of the basilar membrane – a structure that runs the length of the cochlea. This frequency-to-place mapping, or tonotopic organization, is a fundamental organizational principle of auditory coding and is maintained throughout lower structures of the auditory system, up to and including primary auditory cortex (Pickles, 2012). Masking tends to occur when the tonotopic representation of one sound interferes with that of another. The ‘classical’ masking studies in both physiological and perceptual (behavioural) experiments have involved embedding a tonal signal in a noise masker. Physiological correlates of masking have been identified at various levels of the auditory system beginning in the auditory nerve (Delgutte, 1990), but the current study by Recio-Spinoso & Cooper (2013) is the first to report the responses of the mammalian basilar membrane to this classical stimulus combination. The study shows that many of the non-linear neural phenomena produced by the interaction of tones and noise, such as frequency-specific suppression of tones by noise (and vice versa), have their origins in the mechanical response of the basilar membrane. In addition, some of the more linear aspects of auditory processing, such as the constant signal-to-noise ratio at threshold over a large dynamic range, provide an early physiological correlate of the perceptual invariance we experience when listening to the same sound over a wide range of stimulus levels. The new study demonstrates that many attributes of auditory masking, previously observed in both neural and psychophysical experiments, are established at the level of mechanical transduction in the cochlea. The data also provide a valuable resource with which to test and validate non-linear computational models of the peripheral auditory system, which in turn can be used as front ends for devices as diverse as low bit-rate audio codecs (such as those used for MP3 audio compression) and automatic speech recognizers. Important remaining questions include the extent to which cochlear responses to complex sounds are altered by efferent control (activated by prior sounds, or ‘top-down’ effects such as attention) in awake, behaving animals, and the still-controversial issue of how similar human cochlear tuning is to that of typical laboratory animals (Shera et al. 2010; Joris et al. 2011), such as the chinchilla and gerbil studied by Recio-Spinoso & Cooper (2013).