Cochlear Structure and Function

Graeme K. Yates

I FUNCTION AND STRUCTURE OF THE COCHLEA

A Transduction of Acoustic Stimuli

The cochlea is required to transduce minute, rapid fluctuations in the atmospheric baseline pressure into a neural code on the auditory nerve. In doing so it must make available to the brain as much as possible of the information contained in those fluctuations. Sound is a mode of energy transfer by longitudinal motion (i.e., in the direction of propagation) of air molecules, and typical fluctuations occur on a time scale from tens of milliseconds down to microseconds. The amplitudes are extremely small fractions of the baseline pressure; the conventional physical reference level of 2 × 10− 5 Pa is equivalent to the pressure change caused by driving a standard 10 ml syringe into its barrel by a mere 10− 8 mm while a sound pressure level (SPL) of 120 dB is equivalent to pushing such a barrel in by 0.1 mm. The lower level is close to the noise level expected as a consequence of simple shot noise across a receiving window the size of the tympanic membrane. The larger value, although still extremely small, is some 106 times larger in pressure and highlights the extreme range that is apparently of interest to hearing animals. This combination of rapid change, the very small magnitude, and the wide range of pressures places special demands on the cochlea that have been solved in an elegant and efficient way.

B Hair Cells and Mechanical-to-Electrical Transduction

The basic mechanical sensory unit is the hair cell (Russell 1981). Hair cells take a wide variety of forms in the acoustico–lateralis system but share a common configuration in having two types of ciliary processes protruding from the apical ends: a kinocilium and several stereocilia. The former is a true cilium with a characteristic structure, while the latter are actually microvilli with rootlets projecting into their supporting base. Many hair cells lack the kinocilium, showing only the vestigial basal body of the kinocilium, and it is not required for mechanical transduction. Apparently, the stereocilia are responsible for the mechanical sensitivity in the cochlear hair cells. Mechanically sensitive ion channels exist near the tips of the stereocilia (Jaramillo & Hudspeth, 1991) and deflection of the stereocilia toward or away from the basal body modulates the standing current through them, resulting in a receptor current (Corey & Hudspeth, 1979). This current then develops a receptor potential across the basal membrane of the hair cell, which in turn modulates transmitter release from the afferent synapse. The basal end of the cell receives afferent innervation with typically many afferent terminals to each hair cell.

1 Speed Limitations

If mechanically induced opening and closing of the ion channels of the stereocilia is to modulate the transmembrane potential by changing the resting current through the hair cells, then the channels of any one hair cell must collectively have an electrical impedance approximately equal to that of the base of the cell. This expectation is confirmed by the measurements of Sellick and Russell (1978), who showed that the resistance of guinea pig inner hair cells (IHCs) was reduced by at most 50% when driven at very high SPLs by low-frequency stimuli. Thus, the receptor current is determined by the state of the ion channels and by the basal properties of the cell, and the receptor potential is determined by the electrical impedance of the cell membrane. Typically, cell membranes have large shunt electrical capacitances and so the receptor potentials are low-pass filtered representations of the receptor current, restricted to rise times on the order of a millisecond or so. Because the transmembrane potential determines the afferent synaptic response, this means that a simple hair cell is incapable of encoding sounds that vary on a time scale significantly faster than a millisecond.

Similarly, the afferent nerve cannot drive action potentials at a rate faster than, at most, a thousand per second and in particular cannot rapidly modulate its rate of action potential production. This is because it too is limited by the electrical properties of its membranes.

2 Dynamic Range Limitations

A second problem for hearing imposed by the limitations of hair cells is dynamic range, the range of sound intensities that is sufficient to stimulate the receptor and yet not overload it. Typically, an afferent synapse of the acoustico–lateralis system requires as much as 1 mV of receptor potential to stimulate an increase in transmitter release above the spontaneous rate (Sand, Ozawa, & Hagiwara, 1975), yet the receptor potential saturates at a level of a few tens of millivolts, implying a dynamic range of at most 30:1, or about 30 dB. The lower limit is determined by the threshold properties of the synapse while the upper limit is determined in part by saturation of the mechano–electrical transduction process itself, partly by the reduced electro–chemical potential across the transduction channels caused by the change in the internal potential of the cell and partly by saturation of the transmitter-releasing mechanism of the synapse.

The afferent nerve similarly has a restriction on its dynamic range. The maximum sustainable rate of action potentials is on the order of a few hundred to a thousand per second while a reasonable minimum would be on the order of ten or so per second. Rates slower then this could not pass information quickly enough. Hence, the dynamic range of an afferent nerve fiber would be around 10–100:1, or between 20 and 40 dB.

Both of these limitations, speed and dynamic range, would be a major problem for hearing. Without some mechanism to overcome them a large amount of acoustic information, valuable for survival, would be lost; and so most animals have evolved specialized preprocessing mechanisms to defeat these limitations.

C Acoustic Preprocessing

1 Time-Domain Filtering of the Stimulus

The problem of coding a wideband acoustic signal when only very low-pass channels (the hair-cell/afferent nerve channels) are available may be solved by breaking the wideband signal up into many narrowband signals and transmitting each signal separately on an independent narrowband channel. Thus, if a suitable preprocessor could analyze the wideband acoustic signal through a series of parallel, overlapping, narrowband filters it could then pass on all the information in the original signal by transmitting the amplitude and phase of each of its constituent filters. Because filters can change their amplitudes and phases only slowly, at a rate inversely proportional to their bandwidths, the information rate on each channel would now be quite low, easily handled by the narrowband channels. In effect, the preprocessor would convert a single, wideband signal into a number of narrowband signals. All information contained in the original signal could be preserved to be reconstructed at the receiving end of the channels.

This is precisely what the cochlea does. The acoustical signal received by the middle ear is passed through to a partial Fourier analyzer, a parallel series of narrowband filters. This converts the information contained in the rapid temporal variations of the stimulus into a parallel set of information channels with a slow temporal variation. Apparently the cochlea preserves only the amplitude information, at least for the higher frequencies, and discards the phase of the stimulus.

2 Dynamic Range Compression

The range of sound intensities that the cochlea can handle could be increased by some form of mechanical amplitude–compression before the stimulus is applied to the stereocilia of the IHCs. That is, if appropriate mechanical preprocessing could reduce the change in vibration of the basilar membrane (BM) produced by a given change in the sound pressure, then the dynamic range of the IHCs would be increased.

Again, this is precisely what is accomplished by the cochlea. The nonlinear transduction properties of the outer hair cells (OHCs), presumably similar to those of the IHCs, are used as a template to compress the BM motion over the useful range of hair cell transduction.

D Structure of the Cochlea

Functionally, the mammalian cochlea consists of a transduction organ, the organ of Corti, stimulated by a hydrodynamic surface wave propagating on the BM. In most mammals it exists as a cavity in the petrous temporal bone of the skull, a section through which is represented diagrammatically in Fig. 1A. It is a long, tapered tubular structure, divided into three chambers. The uppermost chamber of Fig. 1A is the scala vestibuli, which is in direct mechanical communication with the stimulus-induced displacements of the middle ear. It is filled with perilymph, similar to other extracellular fluids and high in sodium, low in potassium. The middle chamber, scala media, is mechanically probably a part of scala vestibuli but is chemically and electrically quite distinct. It is filled with endolymph, similar to intracellular fluids, being high in potassium and low in sodium. The membrane separating the two upper chambers, Reissner’s membrane, appears to be very compliant mechanically but provides chemical isolation between the compartments and some electrical isolation for low frequencies. The lower chamber, scala tympani, is terminated at...