Hearing Research, 2 (1980) 135 - 149 © Elsevier/North-Hollar~d Biom.rdical Press

• .

SOME DUALISTIC PROPERTIES OF THE COCHLEAR MICROPHONIC *

MARTHA PIERSON ** and AAGE MOLLER ***

Division o f Physiological Acoustics, ENd=C/mrc, University of Gothenbut,~,, Sweden

(Received 1 May 1979; accepted 7 December 1979)

Within a small frequency range just above the characteristic frequer, cy of a differential electrode pair, cochlear microphonic input -output functions axe bimodal, exhibiting two maxima. The mor:~ sensitive, low'intensity response has a more limited linear operating range, and it is nmre labile due t) acoustic fatigue or hypoxia. After fatigue or nypoxia, the high-intensity response is revealed. TI',.~ latter operates 180 ° out-of-phase with the former, presumably due to its being generated more basal- ward in the cochlea. The difference in the lability of the two components suggests that the two sources are different types of hair eel/s: i.e., oute: and inner hair cells.

Key words: cochlea; cochlear microphonic; outer haL" cell: inner imir cell; acoustic fatigue; hypoxia.

INTRODUCTION

In this paper we describe cochlear microphonic (CM) input--output functions that exhibit two maxima separated by a 180 ° cancellation notch at 65 -85 dB SPL. These functions are conventionally described as having only a single maximum [8,12,!8]. The bimodal functions reported here are obtained when the CM is measured in a small frequency range, 1/5 to 1/2 octave above the characteristic frequency of a differential intraeoehlear pair of electrodes (where cl,araeteristic frequency is defined in accordance with Dallos [3] or Dallos et al. [6]. While the interpretatio, of CM measurement made at such relatively high frequencies is difficult due to the considerable phase gradient near the maximum of a traveling wave envelope [2,10,14,19], it is shown here that such inter- pretation is possible. The present study considers both the origin and the nature of the CM response in the two segments of these input-output functions (i.e., low-intensity and high-intensity segments).

Regarding the origin of the two segments of the response, we demonstrate that both are biological in origin since both disappear with anoxia. Secondbr, oo the basis of e×peri-

* Research was conducted according to the principles enunciated ir ~.he Guide for the Care and Uve of Laboratory Animals" prepared by the Institute of Labo~'atory A:nmal Resources. National Research Council. ** Present address: Neurobiology Department, A~med Forces Radiobiology Re,arch Institute, Bethesda, MD 20014, U.S.A. *** Present address: University of Pittsburgh School of Medicine, Eye and Ear Hospital, Department of Otolaxyngology, Eivision of Physiological Acoustics, Pittsburgh, PA 15213, (JS;.A.

136

ments using band-pass, f'fltered noise to suppress CM within restricted areas of the cochlea, it is likely that the low-intensity response arises more apically in the cochlea than does the high.intensity response. RegardL-.g the nature of the CM in the two segments, a dualistic response pattern was found during tonal stimulation or experimental manipula- tion (consisting of exposure to fatiguing stimuli or hypoxia). However, the nature of the response differences, both quantitative and qualitative, precludes the conclusion that the longitudinal location of the hair cells is accountable for their r~ponse properties. Instead, we infer that the low-intensity and the Mgh-intensity responses are generated by two different populations of hair cells that a.~e likely to be outer hair cells and tuner hair cells. The separation of the outer hair cell CM from the inner hai: cell CM in an extracellular recording has long been considered infeasible [3,8,9]. "l~is is because, of the summated nature of the potential, as well as the f'mding that the im.er hair cell CM is some 30-40 dB smaller that the outer hair cell CM under equivalent stimulus conditions [3,6,8]. However, the stimulus conditions of previous studies did not include the fzequency range used here.

It is noted that biimod~l CM input-output functions have been reported in two earlier studies [13,15], but the use of round window electrodes and vet./intense signals (120- 150 dB SPL) in those studies makes interpretation of the results difficult.

METHODS

More than 50 guinea pigs, weighing 250-300 g and anaesthezized vAth an intra- peritoneal injection of 1.5 g/kg urethane, were used in the presen!: study. The animals were maintained at a constant temperature of 38°C. A tracheostomy was performed and artificial respiration was used. The cochela was exposed by the ventrolatera! approach described elsewhere [ 1 ,I 7]. Sound was delivered in a closed system at the entrance to the bony meatus. Sound pressure levels were measured near the eardrum at the end of the experimentg. All cochlear microphonic recordings were differentially obtained using 2 tungsten electrodes which consisted of 24u wires that had been flame-sealed into pulled capillary tubes. The tips of the electrodes were placed in the scala vestibuli (SV) or the scala tympani (ST) through small, hand-drilled holes in the second turn of the cochlea. Prior to measurement, microphonic potentials were band-passed by a 1/3 octave filter (Brtiel and Kjaer Band Pass Filter; type 1618) which was tuned according to the stimul- ating frequencyo CM amplitude measurements were made with a Briiel and Kjaer mea- suring amplifier (t).pe 2608) either by reading the meter directly or by operating on the output.

Cochlear microphonic input-output functions have been measured and reported ~a two ways in the present study. First, an analogue of the amplitude of the recorded CM potential was o[ ~ained by full-wave rectification followed by low-pa~ filtering and sub- sequent averagirg (lntertechnique Physioscope DIDAC 800) of the response to a pure tone pulse with a linear rate of rise and fall, obtained using analog multipliers and trape- zoidal control si,mals. The low-pass filter had a cutoff frequency of 470 Hz, and the tone puls.~ generally i~ad rise and fall times of 40 ms and a total on time of 120 ms in a 1 s period. Because af the linear rate of rise and fall of the stimuli, both the CM amplitude and the stimulu; strength were displayed on linear coordinates. In order to present the

137

more familiar log-dog input-output curves and to attach numecica] values to the ampli- tude, the CM was also measured using continuous pure tones. This turns out to be a risky procedure because one component of the CM response, as measured in this study, is so labile that even a brief (about 1 s) exposure to an 80 dB SPL, continuous tone will deteriorate the response to some degree. Needless to say, therefore~ when continuous tor.es were used, they were switched off between successive measurements and were on about 500 ms total hn accordance with the time it took to read ,:he amplitude of the response.

The electrode configuration used in the present study requires further elaboration. During initial experiments, conventional differential electrodes we,re used: one in scala vestibuli and one ha scala tympani of turns 1, 2, and 3 of the guinea pig cochlea. After the study had progressed, howover, and in all results presented here,, a different type of differential recording was used. Two electrodes were placed within 1 mm of each othe~ in scala w.,stibuli of the second cochlear turn. The results from the two types of differ- ential recording are qualitatively similar although the, amplitude of the potentials recorded by the former type is larger than obtained by the latter. The reason for the change in electrode placement is to ensure that recording is made from a restricted area of the basflar membrane. In the c~nventional ST-SV electrode configure tion there is a possibility (see [2]) that a second-turn scala tympani electrode may pick up first-turn- generated CM to an excessive degree, whereas scala vestibuli electrodes do not seem to do so. The SV-SV electrode configuration used in the present study, where the distance between the electrodes is 1 mm (or less), is to be regarded as a bipolar electrode that records from a restricted area of the basflar membrane. It is commented fllat the SV-SV differential configuration seems to reject remote responses to a greater ~egree than the conventional SV-ST arrangement.

All results presented in this paper derive from individual but representative experi- ments.

RESULTS

Cochlear microphonic input-output functions are compared in Fig. 1 at different frequencies. Fig. la serves as a ref~,rence for the time course of pulsed stimuli at the four frequencies examined. Fig. lb co,,'lpares CM pulse responses at four intensity levels, and Fig. l e is a conventional presen:ation of CM input-output functions at these four frequencies. It can b~ concluded oa sight that the frequency at which the CM is measured determines the nature of the input-output function. It is noted that 2500 Hz is the characteristic [3,6] frequency of this electrode pair, while 4100 Hz is well above the characteristic frequency. At both 2500 and 4100 Hz, the curves display the regions most commonly seen in CM L,put-ou put functions: a linear region and a saturating region. However, in a comparison of the iaput-.output functions at 2500 Hz and at 4100 Hz, the one at 4100 Hz is linear to higher sound levels. At 2500 Hz the functions have a slope of 1 .(3. only below input levels of 40 dB; at 4100 Hz the functions are linear to 80 dB. These data are in opposition to the generalization [2,19] that deviation from lineariW occurs a't lower input levels as frequency ir~crease relative to the best frequency of an electrode pair. However, the generalization that the amplitude of the mi,-rophonic potential for

138

b.

SPLmIx,70 / ~ - ' ~

S p L m a x,.8 0 / - ~ " - ~ " ' ~

SPLmax,90 - ~ - ' - - " ~

f" L

... 100

g I11

1I[ ~ 0

i - o 1 -

o

I E

|

| " ~ - " i - i . . . . . . . I ~1 - 20 t,0 60 80 100 20 t,O 60 80 I00 20 /,0 610 80 100 2C

f / / /

~0 60 80 100

dB SPL dB SPL dE SPL dB SPL

Fig. 1. Differential (SV-SV) CH input--output functions at and above the characteristic frequency. (a) Pure tone pulses with 40-ms rise-fall times; 120 ms total duration; 1 s periods. (lo) CM amplitude as a function of pulse on time; data w~,e averaged, stored on punch tape, and traced as shown from a~ analog printout. (N.B. Amp!itude scales are not identical; both coordinates are line.~r.) (e) CM input-output curves measured ~': ith continuous pure tenes; dotted segments are extrapolated.

frequencies above the best frequency is smaller at any given input level [2,19] is borne out in Fig. It. In lhe linear regions o~" the 2500 Hz and ~'.he 4100 Hz functions, the CM magnitudes are quhe disparate, being approximately 45 dB sartaller at 4100 Hz.

By comparison to the 2500 Hz and the 4100 Hz curve~, the 3000 Hz and the 3600 Hz

139

cur~es show characteristics that have not been previously described for CM input-output functions at the intensity levds used here. The most notable feature of these latter func- tk~ns is the appearance of a second ma~'dmum at moderately high intensities. The 3000 Hz input--output function has two maxima separated by a notch at 85 dB. The 3600 Hz function has two maxima separated by a notch at 65 dB. As is apparent h~ the 3600 Hz pulse responses to a tone pulse lFig. 2b)at 75 dB and 80dB, the second maximum grows lineady to higher input levels than does the first m~ximum *. Thus the 3tX~0 Hz and the 3600 Hz responses exhibit chazacteristies that are a composite of the t~:o 'types' of response found at 2500 and 4 I(K~ Hz.

Using a conventional SV-ST electrode configuration the frequency ranges in which bimodal CM input--output functions have been obtained in turns 1,2, and 3 in the guinea pig cochlea are, respectively, 10-14, 2.5-4.0 and 1.25-1.75 kHz. In a given experiment the range is usually less than a quarter of an octave.

Comparisons between types of differential recordings

The present results are based on recordings made by differential electrodes placed in scala vestibuli (SV) within 1 mm of each oth~.r in the second turn of the cochlea. Such recordings are designated by the label SV-SV. While most intracochlear, differential measurements use the difference between scala vestibuli and scala tympani electrodes. SV-ST, we chose the SV-SV configuration simply to avoid any inadvertent pickup of remote, turn-one, scala tympani mierophonies. A discussion of the latter possibility can be found in [2].

We have found that the potentials recorded u~ing the SV--SV configuration are smaller in amplitude than those obtained using the conventione.l SV-ST configuration but the SV-SV electrode configuration is spatially selective inasmuch as measurements reveal a steep high-frequency cutoff. The difference in the results obtahled using SV-SV elec- trode configuration and the conventional SV-ST dectrode configuration are demon- strated in Figs. 2 and 3. This graph shows results from an experinlent in which three elec- trodes were placed intraeoehlearly in the second turn. Two of the electrodes were in scala vestiouli (with a distance of 1 mm), and a scala tympani electrode was paired with the more apical of the SV electrodes. Thus, SV-SV and SV-ST differential reeoT~ings of CM input---output functions can be directly compared in the same animal. It is seen from Fig. 2 that the SV-SV recordings are about 20 dB less sensitive than the conventiom~! SV-ST measurements. Second!y, both electrode configurations demonstrate notched input-output functions although the SV-ST pair demonstrates 'notching' at a lower frequency (2500) than does the SV-SV pair, presumably due to its mor~ apical location. It should be noted that notching functions can be recorded in all preparations, usi,g either type of differential electrode, as long as the animal is in healthy cor~dition and the cochlea is unfatigued.

* It is to be noted that while the high-intensity response appears to undergo an ove.-loading type of saturation, this is probably not the same phenomenon as seen in classic input -output functions. The decrease in CM after the second maximum has been observed in this study to be associated with tile appearance of subharmonics in the CM and in the sound at the eardrum. Such distortion components have been shown to arise at the eardrum and not in the cochlea [7].

140

E

>~;oo O

t~

"~ 10 J : .

G. o

=E

4t m

u

2500 Hz 3300 Hz 3950 Hz 4500 Hz

/ /

/ /

/

'1 40 60 80 I00 40 60 80 IO0 40

/

l// /

J o

t'

/ /

/ /

/ " ",

/ /

/ i ' x .

dB SPL

Fig. 2. Comparison of SV-ST and SV-ST differential recordings of CM input-output functions in one guinea pig. Stimufi arc 500-ms, continuous tones at the frequencies shown. Electrodes were placed as follows: two SV electrodes, 1 mm apart, in SV of the second turn, and one additional electrode in the .~cala tympani across from the more apical SV electrode. The less sensitive curve in each plot is the SV-SV response.

In Fig. 3 a frequency response function of 60 dB $PL is presented. Differential (SV- SV) second-turn electrodes were used. Measurements were made at close intervals in the frequency range above 2500 Hz. The curve shows a maximum a: 1000 Hz and demon- strates a cutoff of 40 dB/octave between 2500 and 3700 Hz. A secondary response area, albeit one of small magnitude, occurs above 3700 Hz. Above 4200 Hz, this secondary response also drops off at a rate just exceeding 40 dB/octave. Other than the occurrence of the secondary response area, this description is comparabie to the analysis made of functions mc'asured with SV-ST differential electrodes by DaUos et al. [6]. Their descPp- tion of frequency-response curves measured at low input levels in the guinea pig includes functions wf~ich pass through a relatively sharp maximum, after which they exhibit a steep (20-40 dB/octave) cutoff rate. The frequency of the 3 dB down point on the shoulder of ~these functions was considered to be the best frequency. Thus, 2200 Hz was the usual se,:ond-turn characteristic frequency for SV-ST electrodes in Dallos' experi- ments [6]. Due to the :,nsensitivity of the SV-SV electrodes, measurements ~.t levels com- parably low to those used by Dallos are not available. Accordingly a sharp maximum is not found in our frequency response curves, as illustrated in Fig. 3. Nonetheless, by choosing a frequency, just above which the CM exhibits a rapid cutoff, our estimation of characteristic frequency is in good agreement with that of DaUos et al. [6]. In Fig. 3, 2500 Hz approximates the characteristic frequency of this SV-SV electrode pair.

Using this procedure we found that the notching frequency is generally 1/5 to 1/2 octave abo~e the best frequency of a differential electrode pair, whether it is an SV-ST or an SV-SV electrode pair. The frequency chosen for exploring :aotching CM input- output functions for the experiment shown in Fig. 3 was 3244 ltz. Empirically, this frequency was found to be the one at which a notch occurs at the lowest sound pressure level when the animal is 9resented with 120-ms, shaped tone bursts. The notching

141

1(~0-

10-

J

1.0 0.5

\b

I 1.0 2.0 I 4.0 10.0

':mquency kHz

Fig. 3. CM isobar measured at 60 dB SPL with SV-SV differential electlodes. Stimuli consist of 5 00-ms continuous tones.

frequency (3244 Hz) in Fig. 3 is 0.43 octaves above the characteristic frquency, deter- mined as described above.

Phase relationship ~

Phase angle between sound and CM was measured at different sound intensities at a frequency where ~he input--output functi,:ns were bimGdal. These results were obtaine~ in experiments (not reported here) in wi~ch b o a the CM amplitude and phase were recorded continuously with the signal s~.ren~h was increased using a step attenuator. These data can be described quite simply since at the sound pressure level where the high- intensity portion of the response begins, the phase angle of the CM undergoes a rapid

142

180 ° shift. This phase relationship is then raaintained at all higher input levels. Although not explicity presented here, these phase relationships can be inferred from Figs. 3, 4 and 6, as discussed in the texls accompanying these figures.

It had seemed to us that a plausible explanation for the intensity-related 180 ° phase shift was that the reco;:ded potentials are generated by two separate populations of generators (hair cells), speci',dly separated from one another along the longitudinal axis of the cochlea, a distance within which the hydrodynamical events differ by 180". The strategy for evaluating tllis hypothesis was to use a 1000-Hz-wide, band-passed noise of successively lower center frequency to see, whether one portion of the input--output function could be selectively suppressed. A more basally located population would have its response diminished t)y a suppressing noise with a higher center frequency than erie that would degrade the :response of a more apical l~)pulation provided that appropriate intensity of the suppressing noise was chosen. The results of such experiments, an example of which is shown in Fig. 4, support the hypothesis that the high-intensity portion of the input--output functions arises from a more N ~ I location that does the low-intensity portion. The high-intensity portion of the CM response to a 3650 Hz pulse is clearly diminished by a suppressing nois,," with a center fr,.'quency of 5600 Hz. The range of center frequencies of the noise in which the responses could selectively be suppressed by noise was tound to be between 6200 and 5600 Hz. When the high.intensity response is selectively SUl:pressed, the low-irttensil:y portion of the response has an ampli- tude overshoot in the region of the notch. This enhancement of the low-intensity response is interpreted ag reflecting the (now) uncan~eHed CM from the apical source when the basal source, which operates with a phase dlfference of 180 °, is suppressed by the noise. That is, the enhancement is only apparenl, not physiological~ As the center frequency of the suppressor noise is gradually lowerec~., both populations become simul- taneously suppressed (as shown in Fig. 4 fo~: the 365(I ~ response in the presence of a 4900 ttz centered masker). These data sho~ that the minimum frequency separation of the center frequency of the noise, which selectively cause first the high-intensity portion and then both portions of the CM response to become smaller than they were in the absence of noise, is only 1/5 octave, which t~anslates into a distance of 0.45 mm in the guinea pig cochlea using data of Greenwood [11 ]. This separation between the two populations may be overestimated inasmuch as its ~:a'culatinn requires that the low- intensity response be smaller than its unsupFressed size. It might have been more accurate to require that the low.intensity response be smalh;r than its maximum size in the presence of the 5600 Hz ,noise. The important conclusio-~s, however, are that the two CM sources are spatially separate (although probably contiguous)and that the low-intensity response arises from a more apical location than does the high-intensity response.

Effects of fatigue

The low-intensity portion of the bimodal input-o :tput function is extremely labile due to acoustic fatigue. Exposures to 80 or 90 fiB SPL, continuous pure tones at the notchh~g frequency resulted in partial elimlination of the low-level response after less than 1 s of exposure. In Fig. 5, three responses to tone pulses are shown at the start of the experiment (a), after exposure to 500 ms tone bursts (b), and after a 15-min exposureto

. ~ 3 6 5 0 Hz Stimulus shape

143

Cochlear Microphonic response in abs÷nce of masking

Co~.hlear Microphonic response in presence of band pass noise with center frequency at 5600Hz

Cochlear Microphonic response in presence of band pass noise with center frequer~cy at /,900 Hz

Fig. 4. Sek:ctive masking of two CM components. Stimuli are tone pt, lses shaped as in Fig. la; data are reproduced as in Fig. lb except that amplitude scales are identical. Mask~r is a continuous bana- passed noise (bandwidth = 1000 Hz), approx. 70 dB SPL.

3600 Hz at 80 dB SPL (c). While the rapid decrement of the low.intensity portion of the input--cutput function in Fig. 5b is surprising due to the minimal exposure (incurred merely by measuring ;nput-output functions with 500-ms tones), it is even more dramati,: that the third CM response shown in Fig. 5c is no longer a two.population response since only the high-intensity response remains. I~ is noted that after fatigue an overshoot of the CM is often evidenced around the notch (see Fig. 5c). This is probably due to the elimination of the source which is responding 180 ° out.of-phase with the high- intensity source. If higher or lower frequency tones or band-p~ss noises were used as fatiguers, they also selectively fatigued the low-intensity re:;ponse. However, the most effective fatiguing signals had acoustic energy centered l/2 or~tave b~'~w the ~otching frequency, whether they were pure tones or band-pass noises. In th~ above exposure paradigms, the high-intensity portion of the response was ahnost never diminished,

144

b

Fig. 5. Difference in fatigability of two CM components. Stimuli are tone pulses shaped as in Fig. 1~; data a~-e reproduced as in Fig. lb. (a) CM response at 3600 Hz at the start of the experiment. Co) Partial elimination of low leve! response after several 500-ms presentations of 3600 Hz tones (used in bbtai;~ing an input-output function). (c) Total elimination of low level response after 15 rain exposure to 3600 Hz tone at 80 dB SPL.

aldtot~gh with very intense: (~>100 dB SPL) fatiguing signals, it also showed some decre- lne~lt.

Effects of hypoxia

hi a4dition 1~o the unequal effect of previous sound exposure on the two populations, there ~emed to be an untoward reduction of the low.intensity portion of ti:~e response during any metabolic impairment of the animal. Therefore, several experiraents were carried out in which hypoxia was produced by :educing the rate of artifical respiration.

145

It was .fiound that the amount of diminution of the low-intensity respoaw at the notching frequency depnded on both the percent reduction of the ventilation rate and the dura- tion of hypoxia. An extreme case is shown ti Fig. 6. To generate this figure, the respirator was adjusted to pump at 65% of normal. At 12.5 coin after the initiation of hno@j the, low-intensity type of response was elimiMed, whsreas the;: amplitude Gf the high-intensity response was unchanged. At the input ieve wllere the notch originally occurred, there w+as now an amplitude overshoot (due to the removal of the 180’ out-of-

During Hypoxia

_r-/7-2 After Hypoxia

Fig. 6. Differences in hypoxic lability of two CM components. Stimuli are tone pulses with 4O-ms rise-fall times; 90 ms total duration; 1 s period. Data are reproduced as in Fig. lb. The hypoxia response waa obtained 12.5 min after a 35% reduction in ventilation rate. The post-hypoxic response was obtahed 10 min after the respirator was turned back to normal (85 strckesjmin, 2.5 cm room aixlstroke).

146

phase, low-intensity CM). When the original ventilation rate was reestablished, the low- intensity response rapidly recovered, although as shown by the persistence of the over- shoot around the notch in Fig. 6c, the recovery was not entirely complete at the time of the recc~rding (often the original function did not entirely recover following severe hypoxia). It is notable that both portions of the response disappear within 1 rain after anoxia or death, which indicates both maxima are indeed biological in origin. These results suggest that the CM from the two populations are unequally subject to metabolic insult, the low-intensity response b,'.ing more susceptible to hypoxia than the high- intensity response, which is relatively resistant.

DISCUSSION

The impetus for this study was ovx observation that certain CM input-output func- tions exhibit two ma::irna seplrated by a notch at about 65 dB SPL. There was ;t 180 ° difference in phase o, the CM below ~he notch and above it. The shape of these input- output f'uncl:~ons strongly suggested tha~ they were generated by two populations of hair cells. Tl'~:t is, the low level po:tion of t i : response demonstrates 1 limited range of linear operation (by departing from lineanty at less than 40 dB SP~'.) whi? the high level response, allhough inferred to be less sensitive, is capable of oil'.rating linearly up to a level of abotJt 80 dB SPL. The 180 ° phase difference between t~:e responses below and above the notch further suggested that two populations contribute to the bimodal input- output functions. The thrust of the research was to identify the sources of fie two response types.

Pursuant to identifying these populations, the study in which a second sound w;Ls used to suppress the CM was conducted. An analysis of the relationship between the center frequencies of selective and norLselective band-pass suppressins nois,.'s (relative to the two components) indicates that the hi O-intensity source may be presumed to be located more basalward in the cochlea, but is probably locatea quite close, probably within OA5 mm of the low-intensity source. This spatial relationship can account for the phase differ- ence between the two CM components since hydromechanical phase can change by 180 ° within this distanc,-::. That is, if on,. • considers the instantaneous pattern of basilar mem- brane motion during pure tone stimulation, the above ~ata are consistent with a hypo- thesis that the low- and high-intensity CM components derive as responses to different half waves of the mechanical disLurbance (traveling wax'e). We feel the apical responm reflects events circumscribed by the leading half wave, while the basal response reflects events of the second half wave - the one at the so-called 'maximum' of the traveling wave (which is 180 ° out-of-phase wi~JL tlte leading half wave). !.t seems unlikely that the origin of either CM coraponent is mare proximal than this since the differential electrodes were located apical relative to loc'ation of the maximum of the.; traveling wave envelope at the frequencies being consider~:d here. Alternatively, one m,ght consider whether the two populations of generators might have an opposite polari~ tion. While the latter possibility might support a hypoth,~'sLs that two types of hair cells a~'e responsible for phase differ- ences, such a possibility is inconsistent with the data of D-,ulos et al. [4].

Because the two CM sources at the frequencies being s -~udied here are separated along the longitudinal axis of the cochlea, it can be inferred that a given signal operates differ-

147

ently on them. The question becomes one of whether hydromechaniical differenc ~ at two points along the basilar membrane are sufficient to account for the observed differences in the CM response at low and high intemities. The first indication that this is not the case follows quite simply from a consideration of which response ohginates at each loca- tion. Since the more basally located population rests on the basilar membrane near the '~naxi_mum of the traveling~ wave, one expects the low-intensity response to originate there, since hair cells so located would experience a wider excursion with a given input. In fact, however, the low-intensity response is presumed to arise on the leading edge of the trave ~ling wave whereas the high-intensity response is presumed to arise from a locati.~n compatible with the maximum basilar membrane excursion. Thus, the disparities in sen- sit~vity and linear operating ranges are not only inexplicable in terms of differences in basilar membrane amplitude at the two locations but also compounded in magnitude by those differences.

A second indication that the two populations differ by more than their longitudinal positions follows from the ~tudies of the extreme lability of one population following acoustic fatigue. Exposures to 80 or 90 dB SPL, continuous pure tones or band-pass noise,s, whether at the netching frequency or at lower or higher frequencies, eliminate ~:he low-intensity portior, of the bimodal CM input-output functions while only margin- ~Rly affecting the high-it~.tensity response. Since changing the frequency of the fatiguer does not increase the probability of diminution of the high-intensity re,~ponse, it is unlikely that the relative location of the two populations is related in any way to their fati~ability.

The pre~nt study demonstrably deals with two sources located at different points on the basitar membrane, which makes it inevitable that the two populations are operated on differently by a given signal whether it is a fatiguing, stimulating, or suppressing signal. Therefore quantitative differences such as fatigability, sensitivity, or linear operating ranges are useful merely as indications that the two population~ may differ by mo~e than longitudina~ location. In order to show more conclusively that the two CM r~:;ponses originated from two functionally different types of hair cells, it was necessary tcg emon- strate at least one un.equivocal qualitative difference between them. We have fo0nd two such differences, but for reasons of lirnited space, we report only of them here (see [16]).

l-lypoxia is a condition by which the traveling wave presumably is unchanged in its relative dimensions. Therefore, any diffe.ential effect on the two CM components arising during hypoxia can be assumed to originate because of the nature of the ger,,erating hair cells and not hydromeehanieal changes. Fig. 6 demonstrates that the low-intensity response vanishes within minutes after the ventilatiort rate is reduced, while 1:he high- intensity response is unaffected. The hypoxie CM function resembles the po,,;t-fatigue CM function as well as what had been inferred as the response character of the basally located group of ha~ cells: namely, an insensitive response with an extensive linear operating range. Therefore, not only is metabolic lability a dualistic property of ~he two populations, but the quantitative differences already described (fatigability, sensitivity, and dynamic range) may also arise in the nature of the two populations.

Having thus discounted the importance of longitudinal difference.,; in hydromechanics as a source of the two CM components, we have considered several alternative e×plana- tions for the dual properties of the CM. First, we condisered whether the number of hair

cells and their inte.rnal phase distribution within the two populations might account for the observed differences. Differences in sensitivity and linear operating ranges conceivably could be explained in this manner, but the magnitude of these differences make it theor&aUy unlikely that these factors would h&e a sufficient influence (see cakukdions in Es]). Moreover, these factors cannot account for the differences in lability, either acoustic or metabolic. Secondly, we entertained the possibility that the high-intensity response was of nonbiological origin since it resembles stray cagacitative pickup in its extensive linear operating range and its relative immunity to fatigue and hypoxia. This was discounted, however, when it was found that the high4ntensity response was abolished by anoxia or could be suppressed by band-pass noises. A third possibility was that some aberrant electtoanatomical property of the cochlea caused such functions to be recorded. However, this explanation is also hard to accept in view of the fatigue and hypoxia experiments. A fourth and the most interesting hypothesis is that the bimodal input--output functions are generated by two functionally different populations of hair cells which might be outer and inner hair cells, Of the hypotheses considered, only this one can be used to account for all of our findings. In this regard we propose that the population that is more sensitive, more limited dynamically, more fatigable, and more labile during hypoxia consists of outer hair cells, while the second population consists of inner hair cells. She designation that these two populations are inner and outer hair cells is consistent with the 30 to 40 dB decrease in CM sensitivity when cochleae suffer outer hair cell lesions [3,8,9]. Moreover, it seems appropriate that the more fatigable popula- tion be identified with the population of hair cells shown to be mare likely to suffer damage during acoustic trauma.

In speculating why outer and inner hair cells would exhibit these particular character- istics, it is not possible tc judge whether intrinsic differences in the hair cells or extrinsic differences in patterns of energy distribution would be accountable. Because inner hair cells rest on a relatively it?lflcxible portion of the basilar membrane, it is conceivable chat they ‘receive’ a signal that is lesser than or different from that received by outer hair cells during a given stimulus. That is, the separation of the cell types on the latero-modiolar aAs of the cochlea may account for differences in sensitivity, sijnamic range or fatig;lbil- ity. On the other hand, the differences found in metabolic labilty can be due only to intrinsic differences in the two types of hair cells, although it would be premature to speculate as to what divergent feature of the cells &ght account for the seemirjgly aerobic behavior of the fcbrmer and relatively anaerobic behavior of the latter. Th_% is not to say that intrinsic differences cannot account for the dualism found in the dynajnic properties or fatigability t)f the two cell types as well. In point of fact, evidence exists that the appropriate stim ~1~s is different from outer hair cells and inner hair cells inas- much as the outer hair cell CM is proportional to basilar membrane displacement while the inner hair cell CM is proportional to basilar membrane velocity [4].

In conclusion, we believe that within a narrow range of frequencies 1 /S to l/2 oct,ave above the characteristic frequency, the CM originating from one population of hair c:lls (that is most likely the outer hIair cells) can be distinguished from that of another popula- tion of hair cells (most likely being the inner hair cells), by using intensity as a parameter. At every electrode location there is a narrow window defined by frequency and intensity in which the CM from o,*le of the populations becomes small by comparison to that of the

149

other population. We believe this fortuitous recording situation has enabled us to begin cataloguing differences in receptor potentials from these two populations of cells. Tke factors influencing these differences and their physiological meaning will be questions for future study.

ACKNOWLEDGMENT

This work was supported by the Swedish Work Environmental Fund and the Swedish Medical Research Council (Project 04X-90 and P-3251 ).

REFERENCES

Ill

[2I [3l

[4l

[5l

[6l

[7l

[8l

[9I

1101

1111

[12]

113]

[14l

[15]

[~6]

1171

[181

[19]

Dallos, P.J. (1969): Comment on the differential-electrode technique. J. Acotst. Soc. Am. 45, 999-1007. Dallos, P.J. (1973): The Auditory Periphery, pp. 218 --317. Academic Press, bev, York. Dallos, P.J. (1973): Cochle~ potentials and cochlear mechanics. In: Basic ~iechanisms of Hear- ing, pp. 335-376. Editor: A M~ller. Academic Press, New York. Dallos, P.J., Billone, M.C., Durrant, J.D., Wang, C.Y. and Raynor, S. (1972): Cochlear inner ard ot:ter hair cells: Functional differences. Science 177,356-358. Dallo~, P.J. and Cheatham, M.A. (1976): The production of cochlear potentials by ironer and outer hair cells. J. Acoust. Soc. Am. 60,510-512. Dallos, P.J., Cheatham, M.A. and Ferraro, J. (1974): Cochlear mechanics, nonlinearities, a'~d cochlear potentials. J. Acoust. Soc. Am. 55,597-605. Dallos, P.J. and Linnell, C. (1967): Even-order subharmonics in the peripheral auditory system. J. Acoust. Soc. Am. 40, 1381-139 I. Dallos, P.J. and Wang, C.-Y. (1974): Bioelectric correlates of kanamycin intox:ication. Audiology, 13,277-289. Davis, H., Deatherage, B.H., Rosenblut, B., Fernandez, C., Kimura, R. and Smith, C.A. (1958): Modification of cochlear potentials produced by streptomycin poisoning and by extensive venous obstruction. Laryngoscope 68,596-627. Engebretson, A.M. and Eldredge, D.H. (1968): Model for the nonlinear characteristics of coch- lear potentials. J. Acoust. Soc. Am. 44, 548-554. Greenwood, D.P. (1961): Critical bandwidths and the f,:equency coordinates of the b Jsilar mem- brane. J. Acoust. Soc. Am. 33, 1344-i356. Honrubia, V. and Ward, P.H. (1968): Longitudinal distribution of the cochlear m~crophonics inside the cochlear duct (guinea pig)~ J. Aeoust. Soc. Am. 44, 951-958. Karlan, M.S., Tonndorf, J., Khanna, S.M. (1972): Dual origins of the cochlear microphonic.~ - inner and outer hair cells. Ann. Oto-Laryngol. 81,696-704. Kohlloffel, L.U.E. (1971): Studies of the distribution of cochlear potentials along the basilar membrane. Acta Oto-Laryngol. Suppl. 288, 1-66. Lawrence, M., Wolsk D. and Schmidt, P. (1962}: Inner ear response to bigh-leve! sounds. J. Acoust. Soc. Am. 34, 102-108. Pierson, M. and M~ller, A. (1970): The effect of modulation of basilar membrane po., ~tion on the cochlear microphonic. Hearing Res. 2,151-162. Sweetman, R.H. and Dallos, P.J. (1969): Distributior~ pattern of cochlear combination tones. J. Acoust. Soc. Am. 45, 58-71. Tasaki, I., Davis, H. and Legouix, J.P. (1952): The space-time pattern of the cociflear micro- phonics. J. Aeoust. Soc. Am. 24,502-518. Whitfield, I.C. and Ross, H.F. (1965): Cochlear microphonic and summating poten:ials and the outputs of individual hair cell generators. J. Acoust. $oc. Am. 38, 126 - 131.