Two-tone suppression in inner hair cell responses

Hearing Elsevier

HRR

187

Research, 40 (1989) 187-196

01233

Two-tone suppression in inner hair cell responses M.A. Cheatham and P. Dallos Auditory Physiology Laboratory, Frances Searle Building, Northwestern (Received

4 November

1988; accepted

University, Evanston, Illinois, U.S.A.

4 March

1989)

In an attempt to characterize certain aspects of two-tone suppression (2TS). ac receptor potentials were recorded from mammalian inner hair cells (IHC) in the third turn of the guinea pig cochlea. By comparing magnitude and phase changes occurring during suppression with predictions made on the basis of level-dependent responses to single-tone inputs, it is possible to determine whether 2TS is mimicked by simply attenuating stimulus intensity. Results indicate that the effects of suppression are not simulated by simple input attenuation for low probe levels which produce responses below saturation. In these situations, the suppressor causes a decrease in the magnitude of the ac receptor potential with the largest deviations measured at the characteristic frequency (CF) of the cell. Thus, frequency response functions become broader. Response phase goes through a lag/lead transition at CF, also opposite to the results expected by simply decreasing input to the cell. At higher probe levels, within the saturation region, the magnitude reductions produced during 2TS are largest for stimulus frequencies well below and well above CF. This effect partially reverses the broadening of frequency response functions seen at moderate intensities with possible benefits for the processing of complex stimuli at conversational levels. Although the magnitude data obtained at high probe levels are consistent with the attenuation hypothesis, the companion phase measures did not show the expected lead/lag transition through CF since phase changes were generally lags. Consequently, the high-level suppression data suggest that 2TS may reduce input to the IHC but in a way which is not equivalent to the attenuation of a single-input stimulus.

Cochlea;

Hair cell; Receptor

potentials;

Two-tone

suppression

Introduction The phenomenon of two-tone suppression (2TS) has engaged investigators of the peripheral auditory system since its discovery by Black and Cove11 in the thirties. Their experiments, along with those of Wever et al. (1940), demonstrated that the cochlear microphonic (CM) corresponding to a single tone input could be diminished by the addition of a second tone to the stimulus. Two-tone suppression has also been studied postsynaptically in discharge patterns of primary (Rupert et al., 1963; also see Evans, 1975 for a review) and secondary (Galambos and Davis, 1944) auditory neurons as well as in the whole nerve action potential (Dallos and Cheatham, 1977). The phe-

Correspondence to: M.A. Cheatham, Auditory Laboratory, Frances Searle Building, Northwestern 2299 Sheridan Road, Evanston IL 60208, U.S.A. 0378-5955/89/$03.50

Physiology University,

0 1989 Elsevier Science Publishers

B.V.

nomenon has also been demonstrated in cochlear mechanics (Rhode and Robles, 1974). (Note that since the literature on this phenomenon is voluminous, we are referencing only the first observation made at each recording site in the mammalian auditory system). Most of these studies concentrated on describing the effects of a second tone on response magnitude and neglected the accompanying phase measurements, even though several investigators (Hall, 1977; Allen, 1983; Deng and Geisler, 1985) emphasized that companion phase changes would be helpful in distinguishing between possible mechanisms underlying this phenomenon. Since magnitude and phase are linked in physical systems (Bode, 1945), concurrent measures of amplitude and phase can aid in pinpointing the site and mechanism of 2TS. Of the few studies in which response phase was measured, Arthur (1976) stated that this feature was relatively stable, albeit over a 40 o range. This,

188

coupled with reports (Nomoto et al., 1964; Arthur et al., 1971; Msller, 1975) that inhibition and excitation occurred with the same latency, suggested that phase shifts during 2TS might be negligible. Although Cheatham and Dallos (1982) demonstrated large phase shifts accompanying the CM magnitude reductions induced by a suppressor, there were difficulties interpreting these data since the CM reflects the outputs of hundreds of contributing hair cell generators. Because of this spatial ambiguity, it was not possible to know how the accompanying phase changes related to the underlying contributions from individual hair cells. However, more recent measurements from single units in the cat auditory nerve by Deng and Geisler (1985) demonstrate significant changes in the response phase of excitatory tones during presentation of high-frequency, non-excitatory suppressors. We have investigated 2TS by recording from individual inner hair cells (IHC). This is advantageous since presynaptic measures are not complicated by nonhnearities and time delays in the synaptic and spike generating processes. Neither are these measures confounded by the spatial filtering effects which make it difficult to estimate individual hair cell contributions, as in the CM response. It is also staightforward to study the effects of suppressors which are excitatory when presented alone on intracellular ac responses. This, of course, is difficult at the single unit level because suppressors inside the response area cause spike rate to increase, thereby making rate changes hard to interpret. The basic question that we wished to address was whether the frequency-dependent amplitude and phase changes resulting from introduction of a suppressor were different from those to be expected by simply reducing the level of the stimulus. (It should be emphasized that we in no way imply by this terminology that suppression is due to distortion in the sound). Although an ‘attenuation’ model for 2TS has not been specifically formulated, Rose et al. (1974) introduced the concept of an attenuating or damping factor to describe their results, as have countless others. In fact, Geisler and Sinex (1980) described suppression as ‘primarily an effective attenuation of the stimulus intensity’ (p. 331). While 2TS has been

observed in ac and dc receptor potentials of IHCs with high characteristic frequencies (Sellick and Russell, 1979; Dolan and Nuttall, 1988) no phase data were obtained.

Methods Adolescent albino guinea pigs (150-300 g) were anesthetized and maintained with urethane (1600 mg/Kg intraperitoneally, IP) and medicated with atropine (0.05 mg/Kg subcutaneously). Lidocaine (1%) was also infused subcutaneously along all incision lines. In later experiments, anesthesia was achieved using a combination of sodium pentobarbital (15 mg/Kg IP) and Innovar-Vet (0.3 cc/Kg) intramuscularly (IM). Supplemental doses for both drugs of 30% of that necessary for induction were administered every 75 min to maintain an adequate level of anesthesia. During data collection the animals were paralyzed with Flaxedil (5 mg/Kg IM) and artificially respirated with room air. Body temperature was maintained at 38 o C. The headholder was heated to prevent cooling of the cochlea (Brown et al., 1983). Receptor potentials were recorded from IHCs in the third turn where characteristic frequencies (CF) vary between 800 and 1000 Hz. Resting potentials ranged from - 5 to - 42 mV while the endocochlear potentials varied between +75 and + 58 mV. Electrodes were pulled from 1.2 or 1.5 mm O.D. glass capillaries with extruded fiber (Clark Electromedical Instruments, Reading, U.K., or AM Systems, Everett, WA) on a two-stage Brown Flaming (Sutter Instruments, San Francisco, CA) or BB-CH (Mecanex, Geneva, Switzerland) puller. When filled with 2 M potassium acetate tip resistances varied between 80 and 150 MS2. Further experimental details can be obtained elsewhere (Dallos et al., 1982; Dallos, 1985). Responses picked up by the recording electrode were preamplified (Dagan Model 8700) and lowpass filtered at 3000 Hz by an anti-aliasing filter. An automatic gain control system was employed to optimize amplification so that signal size was maximal without saturating the analog-to-digital converter. Responses were averaged on line using a laboratory computer (Digital Equipment Corp., PDP-11/73). Magnitude and phase data were later

189

sired, the response to the suppressor alone could also be obtained.

Results

50 msec Fig. 1. Schematic of stimulus presentation. In the top portion of the figure two partially overlapping 50 ms tone bursts (rise/fall time 0.625 ms) are presented with the suppressor preceding the probe. The probe alone condition is shown below. Dashed lines defme time windows used in the Fourier transformations. Window A gives the response to the suppressor alone; Window B, the region of overlap, provides the response to the probe in the presence of the suppressor and vice versa. Window C gives the response to the probe alone in the two-tone situation.

taken from the Fourier transforms of windowed averaged hair cell responses. Stimuli were generated in a closed system using a dynamic earphone (Beyer DT-48). The sound delivery tube included a con~nt~~ly placed probe tube for monitoring sound pressure level at the tympanic membrane. Output from a subminiature microphone (BT-1751, Industrial Research Products, Knowles Co., Elk Grove Village, IL) was carefully measured to assure that two-tone interactions did not occur in the acoustical signal. In the suppression experiments, the stimulus consisted of two pure tones, partially overlapping in time (Fig. 1). While the suppressor was always 1500 Hz, the frequency of the probe was variable. A single high-frequency suppressor was chosen since some investigators reported that low-side suppression may be quite variable in single units with low CFs (Arthur et al., 1971; Harris, 1979; but see also Costalupes et al., 1987). In order to obtain accurate measures of both magnitude and phase changes during suppression, the probe was presented twice, once alone and then again with the suppressor. This allowed us to use the same time window (B) when transforming both the alone and suppressed conditions. If de-

To address the question of whether suppression at the inner hair cell level results in amplitude and phase changes different from those predicted by a simple attenuation of stimulus intensity, we measured both the magnitude and phase of ac receptor potentiah produced in response to single sinusoids at variable sound levels. Fig. 2 provides frequency response functions measured in an IHC at four different levels: 80, 60, 40 and 20 dB. The data demonstrate that CF, bandwidth and both low- and high-frequency slopes are level-dependent. As level decreases, the frequency producing the largest response increases and the response area narrows as both low and high frequency slopes become steeper. (Although it is more common to discuss level-dependent features in terms of the changes associated with increasing level, we emphasize the reverse since comparisons are to be made with magnitude reductions induced by adding a second tone to the stimulus). In order to ease comparison with our suppression data, it is helpful to note the relative magnitude and phase changes that result from these 20

a0

dfl

60 dB

40 dB

20 de

/ d

-501

I

0.1

Frequedy

(kHz)

1 10

Fig. 2. IHC frequency response functions obtained at 20 (O), 40 (A), 60 (m) and 80 (0) dB. All sound pressure levels are measured in decibels re: 20 CPa. In this and all figures the 0 dB level on the ordinate corresponds to 2 mV peak.

190

dB decreases in level. For example, in Fig. 3 we plot magnitude reductions across frequency as level decreases from 80 to 60 dB, 60 to 40 dB and 40 to 20 dB. If the data were produced by a linear system, all plots would lie on the horizontal line at -20 dB. The bottom curve with open circles provides magnitude changes associated with a decrease in level from 40 to 20 dB. It is only for stimulus frequencies well below or well above CF that a 20 dB change in level results in an approximately 20 dB change in receptor potential. At CF there is only an 11 dB change in magnitude. In other words, the functions show most linear behavior, i.e. the largest changes, above and below CF while changes near CF are small. This, of course, is due to the presence of a frequencydependent nonlinearity which is robust for stimulus frequencies near the CF of the cell (Russell and Sellick, 1978; Dallos, 1985). Consequently, the relative magnitude changes measured in response to a reduction in input sound level from 80 to 60 dB (0) are minimal since the ac receptor potential is saturated at these levels for most frequencies. Analogous phase changes are plotted in Fig. 4 with phase leads corresponding to positive numbers. When single-tone level is reduced from 80 to O-

G- -5 -u

v

.

80-60

d8

_

60-40

df3

40-20

dB

-

g-10-

6

5

r” -

-15-t

-25 ~

-30s 0.1

10

Frequent+

(kHz)

Fig. 3. Relative magnitude changes are plotted across frequency as level is reduced from 80 to 60 dB (0). 60 to 40 dB (A) and 40 to 20 dB (0). Data are taken from Fig. 2. (It should be noted that the data points at 140 and 200 Hz for the 40 to 20 dB condition may be influenced by low-frequency noise).

50

-5

3

d

-6Oj0.1

10

Frequedy

(kHz)

Fig. 4. Relative phase changes for the same IHC as in Fig. 3.

60 dB, a lead/lag transition is observed around CF. A similar function is measured for level decreases from 60 to 40 dB but the transition is shifted to the right, possibly as a result of the slight increase in the frequency which causes the largest response. The resulting phase curves approach zero degrees at low and high frequencies again reflecting the frequency-dependent nonlinearity which underlies the phase change around CF. The more linear amplitude behavior measured at 40 versus 20 dB is also reflected in the phase data where changes are smaller. If phase were level independent, all three curves would lie on the line at zero degrees. Although phase changes corresponding to reductions in level are complex, the lead/lag transition around CF is as previously reported for IHCs (Dallas, 1986) and single units (Anderson et al., 1971) and is thought to reflect preceding mechanical events (Rhode, 1971; Rhode and Robles, 1974; Sellick et al., 1982; Geisler and Rhode, 1982). Since the level dependence of ac receptor potentials produced by IHCs is well documented, one can make predictions about the influence of suppression on these response patterns assuming that suppression operates as an effective attenuation of stimulus level. For example, if a suppressor produces a reduction of input to the IHC, then one would expect minimal decreases in response magnitude around CF, presumably because hair

191

o-

-lO-

is -0

-

V

;-20.e c g I

_ -3o-

-4ol 0.1

10

Frequedy

-,44 0.1

Fig. 5. o = An IHC frequency response function measured at 40 dB SPL. A = The same function obtained during presentation of the 1500 Hz suppressor at 60 dB.

cell output is saturated in this region. Larger decreases in magnitude would be expected above and below best frequency where responses are more linear. This magnitude characterization should be associated with a lead/lag phase transition as probe frequency increases. To test these predictions, we measured IHC frequency response functions in the presence of a 1500 Hz suppressor presented at 60 dB. The probe tone, introduced at 40 dB, was varied in frequency to map out the response area. Fig. 5 shows two frequency response functions. The dashed line represents the receptor potential obtained in the presence of the high-side suppressor. When the suppression induced magnitude changes are plotted versus frequency in Fig. 6 (i.e., the difference between the two curves of Fig. 5), one notes that suppression is greatest around CF and smaller for stimulus frequencies both above or below CF. This frequency dependence contrasts with the attenuation data in Fig. 3, where level dependent influences were minimal near CF. Companion phase data are illustrated in Fig. 7 and demonstrate a lag/lead transition around CF such that the suppressor induces phase lags below CF; phase leads above. To emphasize the contrasts between the phase changes induced by suppression and those result-

10

Frequedy

(kHz)

(kHz)

Fig. 6. Relative magnitude changes (i.e., the difference between the two curves of Fig. 5) induced by a high-side suppressor are frequency dependent such that the largest reductions occur at CF.

ing from attenuating the input, we excited four inner hair cells at 50 dB SPL with probe tones at 280, 800 and 1400 Hz. In Fig. 8, the solid line connects mean phase shifts at each of the three stimulus frequencies. These shifts, caused by a 20 dB attenuation of the probe from 50 to 30 dB, exhibit the lead/lag behavior referred to earlier. The dashed line is associated with phase shifts due

-3::0

Frequency

(kHz)

Fig. 7. Relative phase changes from the same cell as in Fig. 5. A line at zero degrees is provided for reference.

192 40 ALONE

30 z? 2o z^i 10

go t -lO--

75 dB

0

80 dB

6 -2o- a,

L

I

ATTENUATION

-

0” -3o-

\

65 dB 90

db

g -4o-I -50

-1 +::0

-i:0

Frequency

(kHz)

Fig. 8. Phase changes due to inputreduction versus 2TS. Error bars at 280,800 and 1400 Hz represent standard errors of the means. At each stimulus frequency, the means (N = 4) are connected with straight lines when the phase change occurred as a result of a 20 dB decrease in stimulus level from 50 to 30 dB. If the phase change was induced by the 1500 Hz suppressor, then the means are linked with dashed lines. In the latter case the probe was presented at 50 dB, the suppressor at 7C!dB. It should be emphasized that the magnitude changes produced by the suppressor were always less than 20 dB. In other words, identical magnitude changes were not produced by the two manipulations.

to the introduction of a suppressor. For these data the probe tones were again presented at 50 dB but the phase changes were induced by the 1500 Hz suppressor tone introduced at 70 dB. Standard errors of the means observed in the four sets of inner hair cell data are shown with vertical bars. While phase shift data for single tones exhibit the expected lead/lag transition with increasing probe frequency, the phase shifts due to 2TS are in the opposite direction. Phase lags occur for probe tones below CF, phase leads at higher frequencies. Minimal phase shifts are seen at CF. It is emphasized that the comparisons made in Fig. 8 are not meant to be quantitative since the magnitude changes produced in the single- versus the two-tone condition are not identical. The phase measures, however, clearly indicate opposing trends. The influence of a suppressor on fully saturated IHC ac receptor potentials was also investigated. Fig. 9 illustrates the influence of a high-side suppressor (1500 Hz) on the response area measured

Frequency

(kHz)

Fig. 9. This figure demonstrates changes in an IHC frequency response function obtained at 70 dB SPL (0). The 1500 Hz suppressor was introduced at 75 (o), 80 (A), 85 (m) and 90 dB SPL (0).

0

1

10

0.1

Frequedy

(kHz)

Fig. 10. Relative magnitude changes produced by the suppressor. Data reflect differences in dB between the frequency response function obtained for the probe alone condition versus that when the suppressor was also present. The four curves reflect differing amounts of suppression produced at the four suppressor levels indicated in Fig. 9. Data for the 75 dB 85 dB== suppressor = 0 and -----; 80 dB = A and -; . (Note that the data in and ------; and 90 dB = 0 and Fig. 10 were not taken directly from those in Fig. 9. This is because the frequency response function for the probe alone was remeasured for each suppressor level so that the magnitude and phase changes could be determined accurately. only one of these functions for the alone condition is plotted in Fig. 9).

193

Discussion

201

o-

‘;;; u 73 V

-2o-

a, cn

-4o-

6

-6O-

0' :

-6O-

~--looa -120-

-140 0.1

Frequen&

I

1 10

(kHz)

Fig. 11. Relative phase changes induced by the suppressor for the same cell whose data are plotted in Fig. 10. A line at zero degrees is provided for reference.

at 70 dB SPL. In this example, the parameter is suppressor level which changes from 75 to 90 dB. In contrast to Fig. 6, the largest magnitude decreases are produced at stimulus frequencies well below and well above CF. Consequently, the frequency-dependent amplitude reductions measured during suppression result in a sharpening of response areas which have become broader at moderate levels. In Fig. 10 relative magnitude reductions induced by the suppressor are plotted across frequency. Note that the 75 dB suppressor caused a 9 dB decrease at 200 Hz, a 1 dB decrease at 1000 Hz and a 14 dB decrease at 1800 Hz, consistent with the attenuation data in Fig. 3. Since responses around CF are saturated, magnitude changes are relatively small in this region. At frequencies below and above CF, where responses are more linear, much larger changes in magnitude are achieved. Companion phase changes at high probe levels are plotted in Fig. 11. These are generally phase lags with the largest lags recorded well above and well below CF. If 2TS did result in a reduction of input to the IHC, as indicated above by the magnitude reductions in Fig. 10, then the phase would be expected to go through a lead/lag transition around CF as demonstrated in Fig. 4.

Frequency response functions recorded from IHCs illustrate that CF, bandwidth and both lowand high-frequency slopes are level dependent. As intensity decreases, CF moves to slightly higher frequencies and bandwidth narrows. This occurs because of a strong compressive nonlinearity at CF which results in the steepening of both low and high frequency slopes as stimulus level decreases. Consequently, reductions in ac receptor potentials in response to a decrease in input are greatest for stimuli below and above CF. Companion phase data demonstrate a lead/lag transition as probe frequency increases. It should be emphasized that these attenuation effects, while quantitatively different, are qualitatively the same no matter where on the level scale they are measured. In other words, a fixed decrement has the same general effect at both high and low intensities. This phenomenon has been explained by assuming that damping is level dependent (Anderson et al., 1971; Hubbard and Geisler, 1972; Kim et al., 1973; Hall, 1977). Since we wish to learn if the magnitude and phase changes produced during suppression result from attenuating stimulus intensity to the cell, the two-tone IHC data will be discussed within this framework. Comparisons with suppression data at low probe levels indicate that the magnitude reductions measured during 2TS were largest around CF. Companion phase measures demonstrated phase leads for excitatory frequencies above CF, consistent with reductions in high frequency slope. Shallower low frequency slopes, associated with phase lags, were also obtained in response to the introduction of a high-side suppressor. Changes in CF were minimal. While the above changes are clearly inconsistent with the attenuation hypothesis, the lag/lead transition exhibited in the low-level, two-tone phase measurements is not unexpected as a similar pattern was demonstrated in the CM (Cheatham and Dallos, 1982; Fig. 1). In addition, data obtained from single units (Kiang and Moxon, 1974; Abbas and Sachs, 1976; Abbas, 1978; Javel et al., 1978; Geisler and Sinex, 1980) indicate that suppression is maximal at the CF of the cell, consistent with present low-level data.

194

This result implies a degradation in frequency selectivity. In fact, Kiang and Moxon (1974) demonstrated that the sharpness of single unit tuning curves was reduced by adding bands of noise to the stimulus. Consequently, the low-level IHC phase behavior is consistent with the observation that frequency selectivity becomes poorer in the presence of a high-side suppressor. When evaluating data obtained at higher probe levels, which saturate IHC receptor potentials, the phase measures are especially important. While the magnitude reductions caused by the suppressor are similar to reductions produced by attenuating input to the cell, the phase leads expected below CF were not observed. Consequently, 2TS may reduce input to the IHC but not in a way analogous to simply decreasing the level of a single-tone stimulus. Certainly the idea that 2TS cannot be adequately described by a simple attenuation model is not new. In fact, several investigators (Geisler and Sinex, 1980; Costalupes et al., 1987) have commented that rate intensity functions measured in the presence of a suppressor may not simply represent horizontal shifts along the abscissa since rates of growth may decrease with increasing level. Unfortunately, slope changes in the rate-intensity functions were observed for suppressors which were themselves excitatory so that changes in spike rate represent some complex addition of responses to both stimuli, making inte~retation of single unit data difficult. Deng and Geisler, (1985), however, used non-excitatory suppressors and measured complicated phase-level curves which were not simply shifted horizontally toward higher intensities as predicted by the attenuation hypothesis. Although changes in response phase at the single unit level have not received much attention, there are some reports which are consistent with IHC results. Both Arthur et al., (1971) and Javel et al., (1978) found phase lags when responses at CF were measured in the presence of a high-side suppressor. Similar results were also published by Robles et al. (1989) in their measures of two-tone interactions in basilar membrane mechanics. Although these data are compatible with IHC measures, they are not very helpful since the phase is in transition at CF. A more extensive study by

Deng and Geisler (1985) reports phase lags in association with rate reductions when units were excited at, above and below CF. This predominance of phase lags is consistent with the high-level IHC data. Before finishing discussion of the amplitude and phase changes observed during suppression, it is worth considering the implications that IHC data have for the mechanisms which underlie this phenomenon. For example, at low probe levels below saturation, a bi-directional model incorporating OHC feedback (Weiss, 1982; Rim, 1986; Geisler, 1986) may be necessary to explain the degradation in tuning which occurs because the largest magnitude reductions are produced at CF. In this scheme, introduction of the suppressor pushes output of the OHC toward saturation, thereby decreasing feedback which is thought to be level dependent (Zwicker, 1979; Zweig, 1988; Patuzzi et al., 1989; Yates et al., 1989). Since feedback may enhance responses around CF, its removal by the suppressor results in frequency specific magnitude reductions. In addition, if one assumes that feedback provides a negative damping (Neely and Rim, 1983), then its removal will increase damping, resulting in broader response areas. At higher probe levels, where a cell’s output is fully saturated, nonlinearities inherent in the forward transduction process are now expressed. Consequently, addition of the suppressor will produce the greatest magnitude reductions at frequencies well below and well above CF where responses are more linear. Decreases in receptor potentials produced at CF will only be observed at high suppressor levels which will decrease the CF response enough to release it from saturation. Thus, 2TS appears to partially reverse the broadening of frequency response functions seen at moderate levels. Although Merller (1983) states that it is doubtful that suppression facilitates frequency resolution, the suppression-induced changes in ac receptor potentials recorded from IHCs at conversational levels suggest that this idea should be re-evaluated. Acknowledgment This work was supported by grant NS08635 from the NINCDS.

195

References Abbas, P. (1978) Effects of stimulus frequency on two-tone suppression. J. Acoust. Sot. Am. 63, 1878-1886. Abbas, P. and Sachs, M. (1976) Two-tone suppression in auditory nerve fibers: Extension of a stimulus-response relationship. J. Acoust. Sot. Am. 59, 112-122. Allen, J.B. (1983) Magnitude and phase-frequency response to single tones in the auditory nerve. J. Acoust. Sot. Am. 73, 2071-2092. Anderson, D.J., Rose, J.E., Hind, J.E. and Brugge, J.F. (1971) Temporal position of discharges in single auditory nerve fibers within the cycle of a sine-wave stimulus: Frequency and intensity effects. J. Acoust. Sot. Am. 49, 1131-1139. Arthur, R.M. (1976) Harmonic analysis of two-tone discharge patterns in cochlear nerve fibers. Biol. Cybernetics 22, 21-31. Arthur, R.M., Pfeiffer, R.R. and Suga, N. (1971) Properties of two-tone inhibition in primary auditory neurons. J. Physiol. (London) 212, 593-609. Bode, H.W. (1945) Network Analysis and Feedback Amplifier Design. D. van Nostrand, New York. Brown, M.C., Smith, D.I. and Nuttall, A.L. (1983) The temperature dependency of neural and hair cell responses evoked by high frequencies. J. Acoust. Sot. Am. 73, 1662-1670. Black, L.J. and Covell, W.P. (1936) A quantitative study of the cochlear response. Proc. Sot. Exp. Biol. Med. 33, 509-511. Cheatham, M.A. and Dallos, P. (1982) Two-tone interactions in the cochlear microphonic. Hear. Res. 8, 29-48. Costalupes, J.A., Rich, N.C. and Ruggero, M.A. (1987) Effects of excitatory and non-excitatory suppressor tones on twotone rate suppression in auditory nerve fibers. Hear. Res. 26, 155-164. Dallos, P. (1985) Response characteristics of mammalian cochlear hair cells. J. Neurosci. 5, 1591-1608. Dallas, P. (1986) Neurobiology of cochlear inner and outer hair cells: intracellular recordings. Hear. Res. 22, 185-198. Dallas, P. and Cheatham, M.A. (1977) Analog of two-tone suppression in whole nerve response. J. Acoust. Sot. Am. 62, 1048-1051. Dallas, P., Santos-Sac&i, J. and Flock, A. (1982) Intracellular recordings from cochlear outer hair cells. Science 218, 582-584. Deng, L. and Geisler, C.D. (1985) Changes in the phase of excitor-tone responses in cat auditory-nerve fibers by suppressor tones and fatigue. J. Acoust. Sot. Am. 78, 1633-1643. Dolan, D.F. and Nuttall, A.L. (1988) Inner hair cell responses to tone and noise combinations. Abst. Assoc. Res. Otolaryngol. p. 23. Evans, E.F. (1975) The co&ear nerve and cochlear nucleus. In: W.D. Keidel and W.D. Neff @is.), Handbook of Sensory Physiology, Vol. V. part, 2, Springer-Verlag, Heidelberg, F.R.G., pp. l-108. Galambos, R. and Davis, H. (1944) Inhibition of activity in single auditory nerve fibers by acoustic stimulation. J. Neurophysiol. 7, 287-303.

Geisler, C.D. (1986) A model of the effect of outer hair cell motility on cochlear vibration. Hear. Res. 24, 125-131. Geisler, C.D. and Sinex, D.G. (1980) Responses to primary auditory fibers to combined noise and tonal stimuli. Hear. Res. 3, 317-334. Geisler, C.D. and Rhode, W.S. (1982) The phases of basilar membrane vibrations. J. Acoust. Sot. Am. 71, 1201-1203. Hall, J.L. (1977) Two-tone suppression in a nonlinear model of the basilar membrane. J. Acoust. Sot. Am. 61, 802-810. Harris, D.M. (1979) Action potential suppression, tuning curves and thresholds: Comparison with single fiber data. Hear. Res. 1, 133-154. Hubbard, A.E. and Geisler, C.D. (1972) A hybrid computer model of the cochlear partition. J. Acoust. Sot. Am. 51, 1895-1903. Javel, E., Geisler, C.D. and Ravindran, A. (1978) Two-tone suppression in auditory nerve of the cat: Rate-intensity and temporal analysis. J. Acoust. Sot. Am. 63, 1093-1104. Kiang, N. Y.-S., and Moxon, E.C. (1974) Tails of tuning curves of auditory nerve fibers. J. Acoust. Sot. Am. 55, 620-630. Kim, D.O. (1986) A review of nonlinear and active cochlear models. In: J.B. Allen, J.L. Hall, A. Hubbard, S.T. Neely and A. Tubis (Eds.), Peripheral Auditory Mechanisms. Springer Verlag, New York, pp. 239-249. Kim, D.O., Molnar, C.E. and Pfeiffer, R.R. (1973) A system of nonlinear differential equations modelling basilar-membrane motion. J. Acoust. Sot. Am. 54, 1517-1529. Meller, A.R. (1975) Dynamic properties of excitation and inhibition in the cochlear nucleus. Acta Physiol. Scand. 93, 442-454. Moller, A.R. (1983) Auditory Physiology. Academic Press, New York. Neely, S.T. and Kim, D.O. (1983) An active cochlear model showing sharp tuning and high sensitivity. Hear. Res. 9, 123-130. Nomoto, M., Suga, N. and Katsuki, Y. (1964) Discharge pattern and inhibition of primary auditory nerve fibers in the monkey. J. Neurophysiol. 27, 768-787. Patuzzi. R.B, Yates, G.K. and Johnstone, B.M. (1989) Outer hair cell receptor current and its effect on mechanics. In: J.P. Wilson and D.T. Kemp (Eds.), Cochlear Mechanisms Structure, Function and Models, Plenum Press, (in press). Rhode, W.S. (1971) Observations of the vibration of the basilar membrane in squirrel monkey using the Mossbauer technique. J. Acoust. Sot. Am. 49, 1218-1231. Rhode, W.S. and Robles, L. (1974) Evidence from Mossbauer experiments for nonlinear vibrations in the cochlea. J. Acoust. Sot. Am. 55, 588-596. Robles, L., Ruggero, M.A. and Rich, N.C. (1989) Nonlinear interactions in the mechanical response of the cochlea to two-tone stimuli. In: J.P. Wilson and D.T. Kemp @Is.), Co&ear Mechanisms - Structure, Function and Models, Plenum Press, (in press). Rose, J.E., Kitzes, L.M., Gibson, M.M. and Hind, J.E. (1974). Observations on phase sensitive neurons of anteroventral cochlear nucleus of the cat: Nonlinearity of cochlear output. J. Neurophysiol. 37, 218-253.

196 Rupert, A., Moushegian, G. and Galambos, R. (1963) Unit responses to sound from auditory nerve of the cat. J. Neurophysiol. 26, 449-465. Russell, I.J. and Sellick, P.M. (1978) Intracellular studies of hair cells in the guinea pig cochlea. J. Physiol. 284,261-290. Sellick, P.M. and Russell, I.J. (1979) Two-tone suppression in co&ear hair celIs. Hear. Res. 1, 227-236. Sellick, P.M., Paturzi, R. and Johnstone, B.M. (1982) Measurements of basilar membrane motion in guinea pig using the Mossbauer technique. J. Acoust. Sot. Am. 72, 131-141. Weiss, T.F. (1982) Bidirectional transduction in vertebrate hair cells: A mechanism for coupling mechanical and electrical processes. Hear. Res. 7, 353-360. Wever, E.G., Bray, C.W. and Lawrence, M. (1940) The inter-

ference of tones in the cochlea. J. Acoust. Sot. Am. 12, 268-280. Yates, G.K., Geisler, C.D., Patuzzi, R.B. and Johnstone, B.M. (1989) Saturation of receptor currents accounts for two-tone suppression. In: J.P. Wilson and D.T. Kemp (Eds.), Cochlear Mechanisms - Structure, Function and Models, Plenum Press, (in press). Zweig, G. (1988) The mechanics of hearing. Current Concepts in Hair Cell Function, Kresge Hearing Research Institute, University of Michigan, Ann Arbor, MI. Zwicker, E. (1979) A model describing nonlinearities in hearing by active processes with saturation at 40 dB. Biol. Cybernetics 35, 243-250.