J.P. Wilson, Department of Communication & Neuroscience, Keele University, Staffs ST5 5BG, UK
ABSTRACT
From an early time frequency analysis has been thought to be mechanical in origin. The first direct measurements, however, showed consistently poor tuning. This led to the concept of a second filter mechanism beyond the basilar membrane. Electrical tuning properties in some species appeared to support this view. With the discovery of otoacoustic emissions, an earlier hypothesis that mechanical amplification might enhance tuning received support. The most recent mechanical measurements demonstrate the extreme vulnerability of basilar membrane tuning and show that under good conditions it can indeed be adequately sharp. Although the outer hair cells are the prime candidates for the cochlear motor, their mode of action is not yet fully worked out. Furthermore there still appears to be room for other mechanisms in some non-mammalians. Some important aspects of psychophysical phenomena such as masking curves, equal loudness curves, recruitment, hearing loss, distortion products, and pich shifts now appear to be explicable at the mechanical level.
INTRODUCTION
In addition to its obvious function as that of sensory transducer converting the acoustical input into a code of nerve impulses, the cochlea also performs the first stage of sensory analysis separating out the different regions of the frequency spectrum into different channels for parallel transmission to the brain. This âplaceâ analysis is based on mechanical processes but in a way which turns out to be more complex than originally thought, and is still not completely understood in detail. For this reason it is probably helpful to approach the topic from an historical viewpoint explaining how our present understanding has unfolded. This may also be appropriate because in certain respects our present conceptions are closer to the starting point than to some of the points en route.
Inevitably the topic of cochlear mechanics focuses on âplaceâ aspects of hearing rather than temporal aspects. This does not imply that temporal processing is unimportant but rather that temporal features of the signal have to be âseenâ through the filters of the cochlea. In certain respects these filters appear to have evolved in such a way that loss of temporal information is minimised.
With parallel evolution of hearing organs it is inevitable that differing mechanisms receive different emphasis in different species and comparative studies may allow a mechanism that is embryonic or vestigial in one species to be understood more readily when observed in a more fully developed form. Where, however, completely different paths are taken, this approach may not be useful.
This review will start with mammalian hearing and consider other developments where these might be appropriate.
There are several areas of research which are closely linked with the present topic which will be discussed in passing, but which are now too large to be considered fully in a review of this length. These include models of the cochlea (see de Boer, 1980, 1984), otoacoustic emissions (see Kemp, 1978, Wilson, 1984), micromechanical processes, and cochlear potentials (see Dallos, this volume).
MAMMALIAN COCHLEAR MECHANICS
History
It is appropriate to start with Hermann von Helmholtz (1863, 1877) because he brought together anatomical findings, a physical model, and psychophysical considerations in a form which was to hold sway for many decades. So persuasive infact, that it probably stifled further research in this area! Helmholtz considered that the basilar membrane consisted of a series of simple second-order resonators each tuned to a slightly different frequency and, invoking MĂŒllerâs theory of specific nerve energies, each connected by a separate receptor and nerve fibre to the brain. In 1863 he postulated that the mechanical resonators were the rods of Corti but by 1877 it was realised that birds did not have these and that the structure showing most change in dimension from base to apex of the cochlea was the width of the basilar membrane. Furthermore, the basilar membrane (BM) was observed to split easily along the direction of its transverse fibres. From this, Helmholtz concluded that it could be stretched strongly across its width and very weakly along its length and thereby providing a graded series of independent vibrating strings in which the tuning frequencies were lowered to the appropriate range by the masses of the (mechanically independent) overlying structures. He estimated the amount of mechanical damping in the resonators from psychophysical observations. By considering the lowest tone for which the ear could follow a rapid alternation of notes, he concluded that its Q- factor must be about 25 (Q10dB = 8.4), a result remarkably close to modern estimates! He commented that the illustration (Fig. 1a) that he had derived for intensity as a function of frequency might, with suitable scaling, also represent intensity as a function of position. It is interesting to note that Helmholtz considered his theory partly proved by an observation of V. Hensen that different auditory hairs of the tail of the opossum shrimp were tuned to different frequencies. Hensen used a keyed bugle as acoustic stimulator and perhaps this rates as the first direct observation of mechanical auditory tuning!
Fig. 1 (a) Helmholtz (1877) resonance curve showing variation of intensity (amplitude)2 over a range (a-c) of four semitones or over a distance of 1.5mm on the BM. (b) Computed instantaneous âtravelling waveâ patterns (1-5) and envelope (solid) for a bank of independent resonators. The vertical arrows represent adjacent reeds observed by BĂ©kĂ©sy (1960, Fig. 12.11(a)) which misses the intermediate region of interest.
The principle of the telephone, which had been recently invented, appeared so relevant to mechanisms of hearing that Rutherford (1886) applied it in a theory in which it was assumed that the auditory nerve fibres carried the temporal information present in the acoustic waveform. Much temporal information is indeed carried by the pattern of nerve firings and aspects of this concept still receive support in modern measurements and theories, although each channel must be preceded by its own cochlear filter.
Georg von BĂ©kĂ©sy (1928 see 1960) was the first to observe the vibration pattern of the cochlear partition. This was mostly done by sprinkling fine silver crystals on Reissnerâs membrane and observing the vibration with a stereo microscope using stroboscopic illumination. He made most of these measurements in human cadavers and guinea-pigs, but also in the chicken, mouse, rat, cow and elephant. When plotted as a function of frequency, the results from different specie...