Professional Documents
Culture Documents
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
suchthat with two-.tonestimulation,usinglow-to-moderate A. Measuring acoustic distortion
soundlevels,intermodulationdistortionisgeneratedasa by-
productof the amplificationprocess(Davis, 1983;Strube, The primary tones0 e1 lower,f2 higherin frequency)
1985;Zwicker, 1986). This impliesthat distortionmeasure- weresynthesized by programmable,PhilipsPM 5193 func-
mentscould be usedto study activeprocesses in cochlear tiongenerators.The two continuoustoneswerefedto the ear
mechanicsand may bear somerelation to auditory sensitiv- canalvia the two probeloudspeakers. To measurethe sound
ity. Someanimal studiesdo, indeed,suggestthat a relation- field in the ear canal during the two-tone stimulation,the
shipexistsbetweenacousticdistortionandauditorysensitiv- probemicrophoneresponse wasamplifiedandfedto a Hew-
ity. Zurek et al. (1982) measuredADP levelsandbehavioral lett-Packard spectrum analyzer (3561A). Here, the re-
thresholds in the chinchilla, both before and after noise ex- sponsewas sampled,digitized, and fast Fourier trans-
posure.A significantcorrelationwasfoundbetweenthresh- formed.The spectrumof the responsewasdisplayedon the
old shifts and reduction in ADP level. Brown (1987) and analyzer.
Lonsbury-Martinet al. (1987) alsoshowedthat the levelof The peakenergyof the response wasmeasuredwithin a
the ADP bearssomerelationto subjectsensitivityin normal selectedbandwidth. For experimentsin which a fixed fre-
rodentand rabbit ears,respectively. quencyseparationratio (f2/f 1) wasmaintained,thisanaly-
Resultsof theseandotherstudieshaveledto thepropos- sisbandwidthwas calculatedas a proportionof thef 1 fre-
althattheacoustic distortion product 2f 1-- f,2couldbeused quency(f 1/20), resultingin a narroweranalysisbandwidth
to assesscochlearfunction (Zurek et al., 1982; Brown and at the lower frequenciesto improvenoiserejection.For ex-
Kemp, 1984;Horner et al., 1985;Brown, 1987; Lonsbury- perimentswhich involvedvaryingthe frequencyseparation
Martin et al., 1987) and thusasan objective,indirecttestof ratio, a fixednarrow bandwidthwasusedto ensurethat only-
auditorysensitivity.The clinicalpotentialof thesemeasure- one of the componentsfell within the analysisbandwidth.
ments makes it worthwhile to explore the relationship The frequencyresolutionof each analysisbandwidth(the
betweenacousticdistortion and auditory sensitivityin hu- "effectivebandwidth") wascalculatedby dividingthe width
man subjects. of the analysisbandwidthby the numberof analyzerfilters
The followingexperiments(i) make a detailedinvesti- (400); e.g.,an analysisbandwidthof 100Hz gavean "effec-
gationof the behaviorof humanADP levelasstimuluspa- tive bandwidth" of 0.25 Hz. The frequencyof the distortion
rametersare manipulated,(ii) explorethe "fine structure" productwascalculatedfrom the frequencies of the primar-
of this responseacrossfrequencyand its relationto the abi- ies. Sincethe primarieswere synthesized,this calculation
lity of the stimulusand distortionfrequencyregionsto re- gavea precisevaluefor the distortionfrequency.If the fre-
emit sound,and (iii) explorethe relationshipbetweenthe quencyof the measuredpeak deviatedfrom the calculated
ADP level and subjectiveauditory thresholdsacrossfre- distortion frequency by more than the "effective band-
quency. width," then it was assumedto be noise rather than a true
distortion reading. Where distortion level fell below the
I. METHOD AND EQUIPMENT
noisefloor, the data points are identifiedby the symbol^
A total of 34 human subjectswere usedin this study. beneaththe datum point. Sincethis wasa "peak" noiseread-
They were of both sexes( 19 female, 15 male) and rangedin ing,it wasalwayssubstantiallyhigherthanthe averagenoise
agefrom 15-50 yearsold. At the time of testing,noneof the floor. The level of backgroundnoisevariedwith frequency,
subjectsconsideredthat they had any hearing problems, analysisbandwidth, and subject.
nonehad a historyof repeatednoiseexposure,but two had a Following each distortion reading,the magnitudesof
historyof childhoodmiddle ear infections.Pure-toneaudio- the stimuli were alsomeasured.All magnitudereadingsare
metry wascarriedout on 30 subjectsand showedthat 9 sub- in dB SPL (dB re:20/zPa). Stimulussetting,datacollection,
jectshad thresholdsmore than 20 dB abovenormal at oneor and storage were controlled by a BBC microcomputer
more frequencies(3 females, AFB, TP, and MD, and 5 throughan IEEE interface.The systemwastestedfor arte-
males, RBG, PD, JC, CG, MJG, and LA). Becauseof the factsacrossthe frequencyrangef 1 = 500-8000 Hz. Distor-
possiblerisksto the subjects,tympanometricmeasurements tion was undetectablein a 1-cc cavity with both tones at
were not made,exceptin the earsof the two authors.Both 73dB SPL. Subjectswere thereforenot testedwith stimuli
provedto havenormal middle ear function. above70 dB SPL. Sincelow-frequencyreadingswere par-
Each subjectwas seatedin a sound-treatedbooth and a ticularlytime consuming(due to the highnoiselevelandthe
speciallydesigned,Perspex,meatal probe was fitted into narrowbandwidthsemployedfor measurements),mostsub-
theirearcanal.Rubbereartipswereusedto improvetheseal jectswere testedacrossthe frequencyrangef 1 = 1-8 kHz
in the ear. The probecontainedthreeKnowlestransducers: (seeTable I for detailsof the frequencyrangeinvestigated).
two 1712loudspeakers andan 1843microphone.The micro- Usingthisprocedure,detailedobservations weremadeof the
phoneresponse wasfully calibratedin 25-Hz stepsacrossthe level of human acousticdistortionas the frequencyand/or
frequencyrange200 Hz-10 kHz. The stimuluslevelswere level of one or both of the stimuli were varied.
measuredby the probe microphoneand correctedduring All 34 subjectswere screenedfor acousticdistortion
each experimentalsessionto compensatefor variations acrossthef 1frequencyrangeof 500 Hz (or 1kHz) to 8 kHz
acrossfrequencyin the outputof the Knowlesloudspeakers at 1/3 octave intervals to give a distortion "audiogram"
andvariationsacrossfrequencyof theacousticpropertiesof (Schmiedt,1986), usinga stimulusfrequencyseparationra-
individual ears. tio (f2/fl) of between1.1 and 1.3, and with stimuluslevels
822 J. Acoust.Soc. Am., Vol. 88, No. 2, August1990 S.A. Gaskilland A.M. Brown:Acousticdistortionproducts 822
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
TABLE I. Thefl,f2, and2fl --f2 frequenciesin Hz, at 1/3-octintervals Irregularitieswereclearlyvisiblein the lowerlevelresponse
overthefrequencyrangeinvestigated;f2/f
1wascommonly 1.225.Frequen-
ciescorrespondingto frequency
ratiosof 1.075and 1.375indicatetheex-
curve.The response curveswerenormalizedby multiplying
tremesof therangeinvestigated.
fl frequenciesmarkedwith an asterisk the latter curve by 10. The spectrumof this responsewas
wereusedonly occasionally. dividedby the spectrumof the responseto the higher level
stimulus.This procedurewas repeatedfour times to enable
f2/fl 1.075 1.225 1.375 differentiationof the repeatablecochlear-generated irregu-
fl f2 DP f2 DP , f2 DP
laritiesfrom the random backgroundnoisewhich had been
*500 538 462 612 388 688 312 exaggeratedin the low-level responseby normalization
*630 677 583 772 488 866 394 (after Furst et al., 1988).
*794 853 734 972 615 1091 496
1000 1075 925 1225 775 1375 625
1260 1354 1165 1543 976 1732 787
D. The acoustic properties of the ear probes
1587 1706 1468 1944 1230 2183 992
2000 2150 1850 2450 1550 2750 1250 The followingprocedurewasusedto investigate the in-
2520. 2709 2331 3087 1953 3465 1575 fluenceof the acousticpropertiesof the ear probe on the
3175 3413 2937 3889 2460 4365 1984
4000 4300 3700 4900 3100 5500 2500
recordedresponse. A Knowlesloudspeaker andmicrophone
5040 5418 4662 6174 3906 6930 3150 were fixed at one end of a 1-cccavity with the probe to be
6350 6826 5873 7778 4921 8731 3968 tested fitted into the other end. A small amount of wire wool
8000 8600 7400 9800 6200 11000 5000
was placedinsidethe cavity to minimize standingwaves.
The four probesusedfor distortionmeasurements variedin
the numberof transducersthey housedandthe impedanceof
the sounddelivery and receptiontubes.
Eachprobewasinsertedin turn into the samepositionin
of between 40 and 65 dB SPL. In addition to this, the effect of the cavity and the sound-pressurelevel in the cavity was
stimulusleveland frequencyseparationwasstudiedin detail recordedby the cavity microphone.The cavity loudspeaker
in 24 of the 34 subjects.The stimulusfrequenciesfor these delivered a tone swept acrossthe frequency range 100-
investigationswere thoseat which high levelsof distortion 10 000 Hz. This was repeatedwith a probewith blanked-off
had previouslybeendetectedin the distortion"audiogram." inlet and outlet tubes (the referenceprobe) usedin placeof
The resultsshownare from the subject'sright ear unlessthe an experimentalprobe.
letter L follows their initials. The sound-pressurelevel acrossfrequency measured
with eachtest probe in placewas comparedwith that mea-
B. The measurement of human auditory sensitivity sured with the referenceprobe in place. The difference
With essentiallythe sameequipment,subjectivethresh- betweenthe test responseand the referenceresponsegavea
oldswere measuredin 30 subjectsusingan automaticcom- relativemeasureof the acousticimpedanceof the probe as-
puter-controlled adaptive procedure (based on Levitt, semblyand thus indicatedthe influencethat eachtest probe
1970). The stimuli were tone pulses(500 ms in duration had on the recordedresponse.For frequencies below4 kHz,
with 5-msriseand fall time), which weredeliveredthrough the responses measuredwith each of the test probesin the
oneof the probeloudspeakers. The lengthof eachpresenta- cavityshowedno significantdifferencefrom thosemeasured
tion was dependenton subjectresponsetime with a maxi- with the referenceprobe in the cavity. Above 4 kHz the
mum of 5 s. The subjectwas required to pressa response acousticimpedancedifferedslightlybetweenall four probes.
buttonwhenthe tonepipswereaudible.The soundlevelwas
II. RESULTS
decreasedwhen the subjectmade two consecutivecorrect
responses,but increasedwhen a singleincorrect response A. Frequency separation
wasmade.The level wasalteredin stepsof 5 to 2 to 1 dB as The levelof2f 1 -- f2 wasmeasuredasa functionof stim-
thresholdwasapproached.
Therewereeight"turnarounds" ulusfrequencyseparation(ratio = f2/f 1) in 11subjects( 17
before the threshold sound level was measured in dB SPL
ears). In eachexperimentalsequence, with a constantd
e1 fre-
usingthe probe microphone.This procedurewas repeated quencyand a variablef2, the ratio wasincreasedfrom 1.075
for each 1/3-oct interval from 500 Hz (or 1 kHz) to 8 kHz. to 1.325 or 1.375 in increments of 0.025 or 0.05. This was
repeatedat differentd
e1frequenciesand at differentequaland
C. The measurement of stimulus-frequency-evoked relativestimuluslevels.In five of the subjects(ten ears), the
emissions
effectof frequencyseparationwas measuredat eight differ-
A 100-scontinuousfrequencysweepfrom 500 to 5500 entfl frequencies,1/3 oct apart from 1 kHz, at eachof four
Hz wasgeneratedby a Hewlett-Packard functiongenerator different stimulus levels (L 1 = L 2 at 50, 55, 60, and 65 dB
(3314A) and deliveredto the ear usingthe probedescribed SPL). In the remainingsix subjects(sevenears), the effect
above.Stimulusfrequencyemissionswere measuredin one of frequencyseparationwas measuredat a selectionof the
subjectusingan averagingfunction on the signalanalyzer eightfrequencieswith eitherequalor unequalstimuluslevels
which measuredthe peak energyfor eachtime sampleand (seelegend).
displayedthisasa spectrum.The acousticresponseof the ear Figure 1 showsexamplesof how the pattern of the re-
was measuredto two successivesweeps,the first at 40 dB sponsewith increasingfrequencyseparationis affectedbyf 1
SPL ( q- or -- 6 dB), the second 20 dB lower in level. frequency (for a constantlevel of stimulation). Figure 2
823 J. Acoust.Soc. Am., Vol. 88, No. 2, August1990 S.A. Gaskilland A.M. Brown:Acousticdistortionproducts 823
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
20 20
J J
rolo rolo
•-0 •-o
j -10 j -10
-2O
,0 •.• f.2 f.• f.4 1"1 1"2 1"3 1"4 1"5
?2/?1 ?requency ratio 72/71 ?requenc y rotio
showsexamplesof how the pattern of the responsewith in- smoothgrowth to a maximum followedby decline,while in
creasingfrequencyseparationis affectedby stimuluslevel others the curve coveringmidrange ratios showeda fine
(for a constantf l frequency).Each graph in Figs. 1 and 2 structureof peaksand troughs.These featureswere more
showsthe resultsfrom an individualsubject.The two figures commonwhenf 1 frequencies werebelow4 kHz. To increase
illustrate that while the absolute ADP level varies with stim- the detail of the responsecurve, the frequencyseparation
ulus frequency,stimuluslevel, and subject,the pattern of ratio was incrementedin stepsof 0.01 (1% off l) in six
ADP levelincreaseto a maximum followedby decline,seen subjects.This was repeatedat severaldifferentf l frequen-
with increasingstimulusfrequencyseparation,is preserved. cies,and in each casestimulus levels ofL 1 = 55 and L 2 = 40
In three of the subjectsshownin Fig. 2 [ (a), (c), and (d) ] dB SPL were used (the choice of stimulus level was based on
maximumlevelsof distortionweregeneratedby the highest resultsthat will be describedin Sec. IIB). Figure 3 shows
stimuluslevelsused.However,therewereexceptions to this examplesof the fine structureseenwith small changesin
asshownin Fig. 2(b) wherethe reverseis true. This reflects frequencyseparationratio, from a relativelysmoothcurveas
marked saturationof the responseat low stimuluslevelsin in (a) to a highly structuredcurve as in (d). The actual
this subject. ADP level and the pattern of ADP level with increasing
The "optimal ratio" generatingthe maximallevelof dis- frequencyseparationremainsremarkably stablewith time.
tortion varies only slightly with frequency,stimuluslevel Individualcurvesin (c) and ( d ) showthesequence repeated
'(up to 65 dB SPL), and subject.A reliable estimateof the after varioustime periodsof up to 15 monthshad elapsed
optimal ratio couldbe made in 11 ears,givinga meanopti- (seecaptionfor details).
mal ratio of 1.225 (65 trials, s.d. = 0.03). In somesubjects,wherea minimumof threedatapoints
While the increase and decline were common to all sub- fell on a straight-linesegment,it was possibleto calculate
jects, the shapeof the curve at the interveningfrequency individualratesof declinein ADP levelwith increasingfre-
separations varied.Somecurvesdescribeda comparatively quencyseparation.Becauseof the intersubjectvariation in
824 J. Acoust.Soc. Am., Vol. 88, No. 2, August1990 S.A. Gaskilland A.M. Brown:Acousticdistortionproducts 824
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
2O 2O
rolo B o
•-0
_o _o -10
ADP level acrossfrequency,it was not possibleto make B. Growth of ADP as a function of stimulus level
slopemeasurements acrossthe samefrequencyrangein ev-
1. Incrementing the levels of both stimuli
ery subject.
There was considerableintersubjectvariability in de- In eightsubjects(nine ears), the ADP levelwasmea-
cline rates. Table II shows how the rate of decline with in- suredasequallevelstimuliwerecovaried.The two stimuli
creasingfrequencyratio is affectedby stimulusfrequency. wereincrementedfrom 30 to 70 dB SPL in stepsof 3 or 5 dB.
Despitethe gapsin the data (due to unmeasureable slopes), This wasrepeatedat severaldifferentfrequencycombina-
thereisevidenceof a steeperdeclinewith increasingfrequen- tions,with a constantfrequencyseparationratio of 1.225.
cy. Of the 11 examplesin Table II, 8 showthistrend. Com- Resultsfrom two subjectsare shownin Fig. 4. The ADP
paringparts (a) and (b), it is clearthat stimuluslevelsof growswith a slopejust below1 (the meanof 11 slopeswas
L 1 = 55 and L 2- 40 dB SPL elicit steeperdecline rates 0.91, s.d.- 0.09). In approximately
half of the trials,when
than L 1 = L 2 -- 55 dB SPL, even if the variation with fre- the stimuluslevelsreach 60-65 dB SPL, the ADP level satuo
rates.
quencyis bornein mind. Table III showshow the rate of
declinewith increasingfrequencyratio is affectedby stimu-
luslevel.To excludeanyeffectof frequency,Table III shows 2. Incrementing the level of one stimulus relative to the
other
slopesmeasuredwithf 1 = 4 kHz only.While no cleartrend
wasapparentwith increasingstimuluslevel,in four of the Usingfivesubjects,thelevelof f 1 (L 1) wasincrement-
five subjectsshownin Table III the most gradualrate of ed in 5-dBstepsfrom 40 to 65 dB SPL, relativeto a station-
declinewasseenat the highestlevel of stimulation(65 dB ary levelœ2(L 2 ). Thiswasrepeatedusingdifferentconstant
SPL). levelsoff2 from 25 to 65 dB SPL. The choiceof 40 dB SPL
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
2O 2O
-10
-2O
-2010 i.i f.2 f.3 f.4 i'-5 FIG.3.Thefinestructure
ofthepattern
of
72/71 ?rmqumncy ratio 72/?1 ?r•guency rotio ADP level with increasingstimulusfre-
quencyseparation. ADP levelshownasa
functionof frequencyseparation. f l was
20' 2O
constant;f2wasvariedto increasethefre-
quencyratio from 1.09to 1.31or 1.35in
stepsof 0.01. L 1 = 55 and L 2 = 40 dB
SPL. Results from four ears are shown. In
.o -lO _• -10
-2o -20.
o i.i i'.• f.m f.4 f.5 D i'.l f.2 ½3 ½4 ½5
72/71 ?nequency riorio 72/71 ?requency rotio
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
3O
TABLE II. The effectof stimulusfrequencyon the rate of ADP levelde-
clinewith increasingfrequencyseparation.To excludeany effectof stimu- (a)
luslevel,parts(a) and (b) showresultsobtainedwith L 1 = L 2 = 55 dB
SPL and L 1 = 55 and L 2 = 40 dB SPL, respectively. 2O
(a) L 1 = L2 = 55 dB SPL
Rate of decline dB/oct c•10
fl (Hz)
Four ears KT AC AC(L) AMB
793 70 ........ •o
1000 '" 74 -" 112
1587 '" 100 81 '"
2000 "- 118 86 '" -10
3174 116 136 ......
4000 ...... 222 128
5039 '-' 156 ......
-2020 go
(b) L 1 = 55 and L2 = 40 dB SPL
Level o? each stimulus dB SPL
fl (Hz) Rate of declinedB/oct
3O
Seven
ears SAG SAG(L) JST TJ CAW TP FJN (b)
-10
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
(a)
52
3O 30
m 2O
27••
-10.
, ) •o •o go •o ')o •o
L•vel o½ lel Stimulus dB SPL Lev•) o½ ½2 Stimulus dB SPL
FIG.5.GrowthofADPasonestimulusisincremented(in5-dBsteps)
relative
totheother,
fromsubject
CMP.fl = 1kHz;f2?fl = 1.225. (a) shows
individual
growth
curves
withincreasing
L 1;(b)showsindividual
growth
curveswithincreasing
L 2.Numbersmarked
byeachcurveindicate
thelevelofthe
stationary
stimulus
indBSPL.Individual
curvesareplotted
witha 10-dB
offset:
Toindicateabsolute
levels,
horizontal
barstotherightindicate
0dBSPLfor
each curve.
828 J. Acoust.
Soc.Am.,Vol.88, No.2, August1990 S.A. GaskillandA.M. Brown:
Acoustic
distortion
products 828
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
(• (a)
2O
• 20
•10
N10
I
o0
-10
-•0
-20•!o Jo •o •o •o Vo
-20. Lmvml o• •! Stimulus dB SPL
• •o •o •o •o •o õo
L•v•l o• F1 Stimulus dB SPL
3O
3O (b)
(b)
_j 2•
• 20
........... ...................
ß.... ---
•_ i o T •-.'•f " '..............
N•0
I
o0
010
-•0
_2oj
o Wo
do Level oF •'2 Stimulus dB SPL
•0
-20,
•' •o •o •o do Vo •o
L•vel oF F2 Stimulus dB SPL
FIG. 7.Variations
inADP levelgrowth
patternsacrossfrequency. Growth
ofADP asonestimulusisincremented
in steps
of 3 dB,relativetotheother
FIG. 6. Variationsin ADP levelgrowthpatternsacrosssubjects.
Growthof stationarystimulus,
fromsubject
SAG.Measurementsmadeateightdiffer-
ADP asonestimulusis incremented (in 5-dBsteps),relativeto the other entf l frequencies.
F 1frequency
code:
.... 2160Hz; .... 2333Hz;-
stationarystimulus,from four subjects(SAG, AMB, CMP, and MU). --=2519 Hz; ---=2721 Hz; --=2939 Hz; --3174 Hz;
fl = 4 kHz,f2/fl = 1.225.(a) increasing L 1,L 2 = 40 dB SPL;(b) in- ....... 4666HZ;.... 5039Hz.f2/fl = 1.225.
(a) Increasing
L 1,
creasingL 2, L 1 = 55 dB SPL. L 2- 40dB SPL; (b) increasing
L 2, L 1 -- 55 dB SPL.
829 J.Acoust.
Soc.Am.,Vol.88,No.2,August
1990 S.A.Gaskill
andA.M.Brown:
Acoustic
distortion
products 829
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
2O 2O
(a)
ß o
c-
-lid
-10 ' : ' ' ' ' ' '
1000 2000 3000 4000 5000
Stimulus Frequency kHz (F1 Frequency For ^DP)
?1 Frequency Hz
2O
(b)
FIG. 8. Average ADP levels across frequency range fl = 1-5 kHz;
L 1 = 55, and L 2 = 40 dB SPL;f2/fl = 1.225.Data from 12 subjects(6
males,averageage= 29 years;6 females,averageage= 26.7 years);audi-
tory thresholdswithin 20 dB of normal. Vertical lines indicate standard • •o
deviation.
• 0
frequencyrange,this differencewasonly significantat 13%
level [F(1,10) = 2.646,p = 0.013]. The great intersubject
variability in distortion"audiograms"mentionedaboveis
reflectedby the large standarddeviationsmarked on the
graph. -lO
0.5
2. Stability o! tY•eADP "audiogram" St imulus Frequency kHz (?1 Frequency For ADP)
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
15
gram is broadlyfeatured,while 1/9- and 1/18-oct stepsre-
veal the fine structureof this pattern.
lO In eight subjects(ten ears), fine detaileddistortionau-
diogramswere measuredand the stabilityof the fine struc-
ture tested.In eachsubjectthree 1/18-oct distortionaudio-
gramswereconsecutively measuredoverthesamefrequency
range.1/18-oct intervalscouldplaceup to threeconsecutive
f l frequencieswithin the sameauditory filter (filter band-
widths estimated from psychoacousticexperiments;see
-5 I I I I I
Moore and Glasberg, 1983). Examplesof the resultsare
20
shownin Fig. 11 (in eachexamplestimuluslevelsofL 1 -- 55
andL 2 = 40dBSPLandf2/fl ratio of 1.225wereused). In
15 (a), the probewasremovedand replacedbetweentrials.
Each of the ten ears tested showed a fine-structured dis-
10
tortion audiogram.This could be reproducedon successive
measurementwith varyingsuccess, dependingon the prox-
imity of the responseto the noisefloor. In five earsthe re-
sponsewaswell abovethe noise,asin Fig. 11,andwashighly
reproducible,thus excluding probe instability and back-
groundnoiseascausativefactorsof thisfinestructure.With-
in the fine structure,the ADP levelchangesat ratesof up to
-5 I I I I I 250 dB/oct, with an averageacross10ears(23 measures)of
2O 141.4 dB/oct (s.d. -- 39.4).
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
2O 2O
J
n
u•
1: lO
_J J
-lO -lO
Stimulus Frequency kHz (F1 Frequency For ^DR) Stimulus Frequency kHz (F1 Frequency For ADP)
20 20
J(b) I (d)
ß
?•
• :....**'"'"
"
-o lO 1: lo /'"; ........... 'b;....
>
J J
-10 , -10 ,
2 4 ]. 2
Stimulus Freouencv kHz (F1 Frequency For ADP) Stimulus Frequency kHz (F1 Frequency For ADP)
FIG. 11.Thestability
ofthefinestructure
ofdistortion
audiograms.
Eachgraph
shows
three
distortion
audiograms
taken
fromanindividual.
ear(two
subjects;
fourears).
Readings
taken
at1/18-oct
intervals.
Withtheexception
of(a),probe
fitwas
notadjusted
betweenconsecutive
distortion
audiograms.
In
each
case
L 1= 40andL 2= 55dBSPL;f2/f
1= 1.225.
Results
shown
arefromsubject
SAG, (a)right
ear,(b)leftearand
from
subject
CMP,(c)right
ear,
and (d) left ear.
fect the fine structure of the acoustic distortion without sub- To facilitate the comparisonbetweenan individual's
stantiallyalteringthe major features. acousticdistortionlevelsand their auditorysensitivity
acrossfrequency,thesetwo differentresponses
are plotted
E. ComparingADP levels with subjectivethresholds
on the samegraph, (seeFig. 14). They sharea common
across frequency
abscissa,but have individual ordinates: left-hand ordinate
In thefollowingexperiments
distortion
audiogramsand for distortionreadingsandright-handordinateforthreshold
subjective
threshold
audiogramsarecompared.In eachof26 readings.Thealignment of thetwoaxesisarbitrary.An au-
ears (23 subjects:11 female, 12 male), a distortion audio- ditorythreshold curvebasedonthatderivedbyLarsonetal.
gram(f2/fl = 1.225;L2 = 40or45dBSPL,L 1> L 2by15 (1988) for the ER-3A insertearphoneis alsomarkedon
dB) and a subjectiveaudiogramweremeasuredat 1/3-oct thesefiguresto indicatenormalhearinglevels.
intervalsacross
thefrequencyrangef1 = 500Hz (or 1kHz) Previousstudiessuggestthat acousticdistortioncomes
to 8 kHz (seeTableI for corresponding
f2 and distortion predominantly from the primaryfrequency region.There-
frequencies).
In 8 of these26 ears,troughs
in sensitivity
at fore,if ADP levelandauditorysensitivity areto beplotted
oneor morefrequencies
weremorethan 20 dB belownormal on theordinate,it isappropriatefor theabscissa
to beoneof
levelsof hearing. the stimulusfrequencies.Sincethe ADP level mustbe de-
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
FIG. 12. Two overlaid derived stimulus
frequencyemissionspectrafrom subject
SAG. See method section for experi-
mental details.
I I i I I I I t I i
pendentuponboth of the primaryfrequencies, shoulddis- dentof L 2). The linear portionof the L 1 growthcurvesug-
tortion levelsbe comparedto auditorysensitivityat the cor- geststhat it may be possibleto extrapolateADP level down
respondingfl or f2 or the arithmeticmeanoffl andf27 to threshold,i.e., directly relating ADP levelsgeneratedby
The arithmeticmean is inappropriatesincethis study has stimuli ofL 1 = 55 andL 2 = 40 dB SPL to activity at thresh-
shownmarked variation in distortiongeneratingability of old. F 1 was therefore chosen for the abscissa and threshold
regions1/18-octapart,and the frequencyof the arithmetic measurements were made at thefl frequency.
mean is not a stimulusfrequency.The resultsreportedin Figure 14 showsexamplesfrom male and female sub-
Sec.IIB. showthat the ADP levelis more dependenton L 1 jects of the correspondence betweenthe featuresof the dis-
than L 2 (in somecasesADP levelappearsalmostindepen- .
tortion "audiogram"and the subjectiveaudiogramthat can
be obtainedwith appropriatestimulusparameters.Individ-
ual graphsshowpeaksand troughsin distortionlevel coin-
cidingwith peaksand troughsin sensitivity.
i i I I
Of the 26 ears tested, 25 yielded suificientdata for a
comparisonto be made of the distortionand thresholdau-
diograms. By subjectiveassessment,all but 5 of the ears
clearly showeda correspondence at somefrequencyrange.
L.
A quantitativemeasureof the correspondence for eachear is
FIG. 13. The relation of distortion fea-
m (a) givenin Table VI. The tableshowsthat 21 of the 25 earsgave
turesto stimulusfrequencyemissionsin
subjectSAG. (a) Detail of derivedstim- a negativecorrelationand that a statisticallysignificantcor-
ulus frequencyemissionspectrumfrom relationbetweenthe two responses
wasseenin half of the
Fig. 12 acrossthe frequencyrange2.6- ears tested. (A significantpositivecorrelation was seenin ß
' 833 J. Acoust.Soc. Am.,Vol. 88, No. 2, August1990 S.A. Gaskilland A.M. Brown:Acousticdistortionproducts 833
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
4O -20 4O -2O
Subject SAG Subject AMB
30 -10 30 -10
2O
10 lO 10
' ! \
2O o 2O
30 -lO • 3O
-2O 40 4O
(a) (d)
-•0 50 -30 5O
0.5 0.5
Stimulus Frequency kHz (F! Frequency For ADP) Stimulus Frequency kHz (F1 Frequency For ADP)
4O -20 40 -20
Subject SAG(L) Subject JST
-10 3O -lO
J
m 2O
• lO 10 lO
•
o
0 / 2O q-
o
0 2o
> -10
-20 4o
(e)
50 -:30 5o
0.5 1 2 4 B 16 0.5
Stimulus Frequency kHz (F1 Frequency For ADP) Stimulus Frequency kHz (F1 Frequency For ADP)
40 -20 40 -20
Subject RBG(L) Subject PD(L)
30 -lO 30 -10
•m 10 lO 20
............................
lO ..................
ß
"•"'0 ,•,
lO
• 0 20 o 2o
o
-2O 4o -20 40
(c)
-30 -90 50
0.5 oso o.s
Stimulds Frequency NHz (F1 Frequency For ADP) Stimulus Frequency kHz (71 Frequency ?or ADP)
834 J. Acoust.
Soc.Am.,Vol.88,No.2, August
1990 S.A.Gaskill
andA.M. Brown:
Acoustic
distortion
products 834
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
TABLE VI. Correlationbetweensubjectiveand distortionaudiograms.
(a) 10 fi•malesubjects
Number of Stimulus level Correlation Level of
Subject data points Age (dB SPL L 1/L 2) coefficient significance
835 J. Acoust.Soc. Am., Vol. 88, No. 2, August1990 S.A. Gaskilland A.M. Brown'Acousticdistortionproducts 835
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
40 -2O
tion areused(not exceeding60 dB SPL), andL 1 isset 15dB
Subject GS(L)
higherthanL 2, 2f 1 --f2 distortionis commonly35 dB be-
-10 low L 2, with highestlevelsof 20 dB belowL 2. Severalpre-
J
viousstudieshavereportedvery low levelsof humanacous-
tic distortion,for example,65 dB belowstimulirecordedby
Kemp and Brown (1983) and 55-65 dB below stimuli re-
lo cordedby Harris et al. (1989). Our resultsindicatethat the
comparativelylow ADP levelsrecordedin the abovestudies
• 0 may havebeendueto theuseof highlevelsoff2 stimulation.
o
The frequencyseparationof the two stimuli can be in-
> -10 creasedby changingthe frequencyof oneor bothof the stim-
uli. Increasingthe frequencyseparationresultsin an in-
creasein ADP level to a maximum followedby a decline,
-20: 40
whetherr 1 (Wier etal., 1988;Furstetal., 1988),f2 (Brown,
1987;Lonsbury-Martinetal., 1987), orfDP (Wilson, 1980;
-90 50 Harris et al., 1989) remainsfixed. However, it has alsobeen
0.5 1 2 4 8 18
shownthat thelevelof anADP isdependent upontheprox-
Stimulus ?requency kHz (71 ?requency?or ADP)
imity of the distortionfrequencyto a spontaneous emission
4O -2O (Furst et al., 1988;Weir, 1988). If the frequencyseparation
SubJect GS (L) is alteredby adjustingthe frequencyof both of the stimuli
3O -10
such that the distortion frequency remains constant,it
would be possiblefor the distortionfrequencyto fall con-
tinuouslyat a narrowregionof abnormallyhighexcitability,
oo
e.g., at a spontaneous emissionor siteof very high re-emis-
N
sion.This would givean uni'epresentative pictureof the ef-
,• lO lO
fectof increasing stimulusfrequencyseparation. By varying
N
the frequencyof oneof the stimuli (thusvaryingthe distor-
• 0 2o •
o tion frequency),this situationwas avoided.
Our re•ult• •how that the averageontirnal frem•ene.
................. r v• d
30 •
>
separationratio is 1.225. This is in closeagreementwith
Harris et al. who report an optimal ratio of 1.22. However,
4O the higheststimuluslevelsusedin our study(L 1 -- L 2 -- 65
dB SPL) were the lowest levels used by Harris et al.
-90 (L 1 -- L 2 -- 65-85 dB SPL), and therefore the data from
0.5
the two studiesbeardirectcomparisononlyat thislevel.Our
Stimulus ?requency kHz (?1 ?requency For ADP) resultsshowedlittle variationin the optimalratio acrossfre-
quencyanddid not showthesystematic dependence onstim-
ulus level reportedby Harris et al. (1989), who saw some
evidenceof a decreasein the optimal ratio with increasing
FIG. 16. As Fig. 14 from subjectGS: (a) measurements taken at onsetof
transientcochlearimpairment of unknown aetiology;(b) measurements
frequencyand with decreasingstimuluslevel. Harris et al.
taken2 weekslater,whenhearinghadalmostreturnedto normal.Readings reportedthesetrendsfor stimuli above65 dB SPL. Our re-
takenat 1/3-oct intervalsfromfl = 500 to 8000 Hz. L 1 - 55 andL 2 = 40 sults indicate that such trends are not shown with lower
dBSPL;f2/fl = 1.225.Thresholdreadings+ -- + ßADPreadings*- - -*; stimulus levels.
dottedline = normalhearinglevel.
As the frequencyseparationis increasedbeyondthe op-
timal ratio, a rapid declinein ADP level is seen.At eachf 1
frequency,the rate of declineshowedconsiderablevariation
of measurements,distortioncouldonly be measuredacrossa between subjects.In agreementwith the observationsof
narrow frequencyrange;in the secondset,ADP levelshad Harris et al., our results showed that the decline with in-
increasedby up to 7 dB and could be recordedacrossthe creasingfrequencyratio becomessteeperat higherfrequen-
whole frequencyrange investigated.These resultsclearly cies.However,the rangeof slopesandthe maximumgradi-
showthat acoustic2f 1 - f2 canfollowtemporarychangesin ent (30-420 dB/oct) describedhereweregreaterthanthose
subjectivesensitivityin human subjects. reportedby Harris et al. ( 102-134dB/oct). In our studythe
mostgradualratesof declinewith increasingfrequencyratio
III. DISCUSSION
tendedto be seenat the highestlevel of stimulation.This
With detailedknowledgeof ADP levelbehavior,stimu- supportsthe observationby Harris et al. of a progressive
lusparameterscanbechosento generaterecordableacoustic decreasein declinerate with increasingstimuluslevel.Re-
distortionin all our subjects.In somesubjectsADP can be sultsfrombothstudiesthereforesuggest that the moregrad-
recordedacrossa four-octavefrequencyrange.The results ual slopesobservedby Harris et al. may be dueto the useof
reported here show that, when moderatelevelsof stimula- comparativelyhigh levelsoff2 in the latter study.
,
836 d. Acoust.Soc. Am., Vol. 88, No. 2, August1990 S. A. Gaskilland A.M. Brown:Acousticdistortionproducts 836
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
The width of the responsecurveobtainedwith increas- bury-Martin et al., 1987; Martin et al., 1988) and human
ing frequencyseparationmay reflectmechanicaltuning of studies(Pasanenet al., 1987;Wier et al., 1988), and sup-
the cochlea.Broadly tuned basilarmembrane (BM) travel- ports the idea that distortiongenerationis governedby a
ing-waveenvelopes wouldcontinueto interactandthusgen- normalized power seriesnonlinearity (Goldstein, 1967).
eratedistortionat wider frequencyseparationsthan sharply However, studiesof rodent distortionhave shownthat by
tuned envelopes.Experimentallybroadeningthe envelopes incrementingthe level of only one stimulus, more rapid
by increasingthe level of stimulationdoesindeedresult in growth is seen (Brown, 1987). To our knowledge,ADP
wider ADP responsecurves. growthwith one stimulusincrementedrelativeto the other
Psychoacoustic and acousticmeasurementsagreethat has not previouslybeenreportedfor human subjects.Our
distortion levels decreaseat wide frequency separations findingsshowevidenceof compression,particularly for low
wherethereis insufficientinteractionbetweenBM traveling stimuluslevels.This compressive growthandsaturationisin
wave envelopesto generatedistortion. However, psycho- accordancewith modelsof the cochlearamplifier (CA) and
acousticmeasurements showno declineat closefrequency supportsthe notionthat ADP is the by-productof this am-
ratios;instead,perceiveddistortioncontinuesto growasthe plification process;i.e., compressiveADP growth reflects
frequencyseparationis decreased(Goldstein, 1967;Wilson, the CA adjustingfor stimuluslevel, amplifying relatively
1980). more at low levelsthan at high. ADP level saturationmay
It is likely that the basisof this differencelies in the reflect the "upper limit" of the amplificationmechanism.
differing origins of acoustic and perceived distortion. With higherlevelsof stationaryL 2, theslopeasa functionof
2f 1 -f2 is generatedat a frequencyregion(the primaryre- L 1,approachedthat predictedby classicalpowerseriesnon-
gion) whichis not tunedto thedistortionfrequencyandthus linearity,but neveractuallyreachedthe predictedslopeof 2.
wherevibrationwill be positivelydamped.From the siteof This suggests that at higher levelsoff2, the nonlinearityis
generation,distortionpropagatesboth basally,to be record- lesscompressive. LouderstationarylevelsofL 1did not seem
ed directly in the ear canaland, apically,to peak at its own to affectthe slope.This isprobablydueto thefactthat stimu-
characteristicplaceon the BM wherethe vibrationis-more lus overlapoccursat thef2 site,and thusthe normalization
sustaineddue to negativedamping,givingriseto perceived processis not as dependenton L 1 as it is on L 2.
distortion.If distortionis generatedat the beatrate (Kemp Predictedpatternsof the BM traveling-waveenvelopes
and Brown, 1986), then,whenthe stimuliare closetogether may explainwhy, with a ratio of 1.225and L 2 at 40-45 dB
(and the beat rate is slow), distortionwill be generatedin SPL, the maximum level of distortion is seenwhen the stim-
discreteburstsat the primaryfrequencyregion.This region uli are 15 dB apart. At closeratios,equallevel stimuli in the
is not "tuned" to the ADP frequency,the vibrationwill be ear canalare nearlyequalin levelat the overlapregionon the
positivelydamped,and the re-emitteddistortionwill, after BM where distortion is thought to be generated (Hall,
some frequencydispersiondue to different propagation 1974). For close ratios, ADP level maxima are seen when
times,emergeassuccessive burstsof energy.Apicallytravel- thestimuliareof near-equal
level(KempandBrown,1986).
ing distortionburststhat arrive at the distortionfrequency At wider ratios,equallevelsin the ear canalwill not be equal
placemay lead to sustainedvibration,due to local negative at the point of overlap:There will be a larger contribution
damping,whichmay thenprovidealmostcontinuousstimu- fromf2 thanfl. Thus,to equalizecontributionsat the over-
lation to the afferentnervefibersfrom this region.Decreas- lap (and maximizeADP level), L 1 mustbe sethigherthan
ing the frequencyseparationresultsin a decreasein the beat L 2 in the ear canal (Wiederhold et al., 1986). Thus it ap-
rate. For closelyspacedstimuli, the root-mean-square level pearsthat the optimal level separationis dependentupon
of distortionovera periodof timemaydiffergreatlyfromthe frequencyseparation.Further studies(seeBrown and Gas-
peak level emitted at eachbeat cycle.The ear canal micro- kill, 1990) indicate that the optimal level separationdoes
phone collectsdata which are translatedinto an rms level. vary with ratio and that at wideratiosthe optimallevelsepa-
The distortionfrequencyplace,ontheotherhand,maymea- rationisgreaterthan at closeratios.With increasinglevelsof
surenearpeakenergyof a slowlydecayingoscillationwhich stimulationthe travelingwaveenvelopebecomesbroaderat
is reinforcedat eachbeat cycle.Decreasingthe frequency its peak.Therefore,for a particularratio, higherlevelstimuli
ratio •would, therefore, result in a decreasein the level of will resultin greaterbasalwardextensionofthell vibration
acousticallyrecorded,but not necessarilyof perceiveddis- and thus the difference between the levels of the two stimuli
tortion. at the generationsitewill diminish.This couldexplainwhy
Another feature of the responsewith decreasingfre- the optimallevelseparationwasseento decreasewith higher
quency separationis that, while the level of 2fl-f2 de- levels of the constant stimulus.
creases,higher-ordercomponentstend to increasein level Measuringacousticdistortionacrossa wide frequency
and number (Brown, 1987). It is possiblethat in psychoa- range,but at small frequencyintervals,showsthat human
cousticstudies,thetotal (n •- 1)f 1 - nf2 distortionpackage distortion audiogramsmeasuredwith moderate stimulus
may contributeto the perceptionof 2f 1 -f2 at closeratios, levelsnot only exhibitbroadfeaturesacrossfrequency,they
while the comparableacousticstudieshave measuredthe alsoshowa fine structureof abrupt changesin ADP level
level of 2fl -f2 alone. superimposed uponthesefeatures.Within the finestructure,
With covariedstimuluslevels,the growthrateof acous- the ADP levelcan changeat ratesof up to 250 dB/oct. An
tic distortionhasa siopeof approximately1. This confirms individual'spatternof featuresandfinestructureremainsta-
the resultsof previousanimalstudies(Brown, 1987;Lons- ble with time. The fine structureof distortion audiograms
837 J. Acoust.Soc. Am., Vol. 88, No. 2, August1990 S.A. Gaskilland A.M. Brown:Acousticdistortionproducts 837
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
resemblesthreshold microstructure (Elliott, 1958; van den audiogramsacrossat leastpart of the frequencyrangeinves-
Brink, 1970;Thomas, 1975;Kemp, 1979b;Long, 1984). tigated. A statisticallysignificantcorrelationbetweenthe
Using moderate (and unequal) stimulus levels, the two responses acrossthe wholefrequencyrangewasseenin
comparisonof ADP fine structureand stimulusfrequency 50% of the ears.There are severalpossiblereasonsfor the
emissionstrengthin one subjectdemonstratedthat ADP mismatches seenin whichtwo responses do notcorrespond
levelsareonly slightlyaffectedby the maximaandminimain acrossthe whole frequencyrange.First, the ADP level is
the SFE responsewhen thesefall at the f2 and distortion dependentuponboth of the stimuli,but the comparisonis
frequencies. The characteristic broadfeatureof ADP level with sensitivityto only one of the stimulusfrequencies. A
seen with increasingfrequencyseparation (increaseto a frequencyseparationratio of 1.225setsf 1 andf2 0.294of an
maximum followed by decline) was preserveddespite a octave apart. Detailed distortion audiogramshav.e shown
highlystructuredSFE spectrum.However,it ispossible that that therecanbesignificantchangesin the levelof distortion
in subjectswith moreextremevariationsin localexcitability as the stimulimoveonly an •8 of an octave(0.055of an
(shownby extremevariationin level in their SFE spectra), octave).Thusa ratioof 1.225couldplacer1andf2 at regions
this variation may dominatepatternsof ADP level across of very disparatesensitivity,If the mechanicalresponseat
frequencyand obscurethe broadfeatures.The subjectsre- thef2 regionisimpaired,thendistortionmaynotbegenera-
portedby Wier et al. (1988) and Furst et al. (1988) may ted, despitea sensitive f 1 region.This may explainwhy in
representsuch a condition,where the ADP responseis somecasessensitivityto f 1 was goodwhile distortionwas
'dominated by a particularly active stimulusfrequencyor absentor reduced,In such casesa closercorrespondence
spontaneous emissionregion. may have beenachievedif the distortionaudiogramhad
In any interpretationof the variationsacrossfrequency, beencomparedto subjective thresholdsat thef2 frequency.
the transmission propertiesof the "wholeear" mustbe con- Second,it is possiblethat in subjectsat somef 1 frequencies
sidered.Distortion generatedby moderatelevelsof stimula- f2/fl = 1.225wasnot the idealfrequencyseparation.For
tion is a cochlearphenomenon, andthereforethe presence of example,Fig. l(b) showsthat whenfl = 2 kHz, subject
distortionat any one frequencycombinationsindicates that FJN has a "notch" in the response curve when
the cochleais activeat the primary frequencyregion.How- f2/f 1 = 1.225.Thismayhavecontributedto thefactthat the
ever, the absolute level of acoustic distortion must reflect distortionand subjectiveaudiogramsdid not significantly
both the middle ear and cochlearresponse.The impedance correlatein this subject.Third, the distortionaudiogram
of the whole ear, includingany contributionmade by the couldbe influencedby strongSFEs,thusobscuringthe cor-
micromechanics of the organof Corti, determinesthe abso- respondencewith auditory sensitivity. Last, some of the
lute levelsof acousticdistortion. This includes ( 1) the trans- "mismatches" werealmostcertainlydueto subjectinexperi-
missionof the stimuli to the point of distortiongeneration, ence of the threshold task. Some of the best matches were
(2) distortiongeneratingcapabilityof the particularstimu- obtainedfrom subjectswho had participatedin other audi-
lus region of the cochlea, (3) the strengthof re-emission tory experiments.
from the distortionfrequencysite,and (4) the transmission This study appearsto be the first demonstrationof a
of the distortion out into the ear canal. frequency by frequency corespondencebetween human
However, the passivepropertiesof the middle ear can- acousticdistortionand auditorysensitivity.The resultssug-
not.explain thesharpdeclineseenwithincreasing stimulus gest that, if appropriatestimulusparametersare chosen,
separationor the compressive growthcurves.Variationsin acousticdistortioncouldbe usedto make an objectiveinves-
fine structureof ADP level acrossfrequencyare numerous tigation of auditory function, where middle ear function is
and rapid (ratesof up to 250 dB/octave). Middle ear prop- normal.The resultsfrom subjectGS suggestthat distortion
ertiesalonecouldnot be responsible for this fine structure, "audiograms,"taken periodically,could be usedto monitor
thoughthey may influenceits frequencycomposition(see thresholdshiftsof cochlearorigin. This may be especially
discussionvan den Brink, 1965). beneficialto peoplewho risk noisedamagein their working
Stimuluslevelsfor the distortionaudiogramswerecho- environment.Sincethe presenceof acousticdistortionun-
sensuchthat distortionreadingscamefrom the linear por- questionablyreflectscochlearactivity, distortionmeasure-
tion of the ADP growth curve. Assumingthat this linear mentscouldform thebasisof a valuablescreeningtechnique
growth extendsdown near to thresholdand that the CA to detect early signsof cochlearimpairment. The results
governsboth thresholdand distortionoutput providesan from GS, PD, and RBG suggestthat a test basedon ADP
explanationfor the relationshipbetweenADP generatedby measurements would be most usefulfor indicatingthe fre-
moderate level stimuli and auditory sensitivity reported quencyregionsover which the cochleais functioningnor-
here.The meansby whichthe two responses areobtainedare mally, ratherthanfor evaluatingthe degreeof impairmentat
very different.The subjectivemeasureof eachthresholdre- a specificfrequency,sincethe ADP tendsto be lost in the
sponseis made to a quiet, singlefrequency;the objective noisewhen thresholdsare elevatedbeyond20-30 dB.
measureof acousticdistortionis made to a pair of stimuli An audiometric test based upon acoustic distortion
well abovethresholdlevel. In addition, the distortionprod- measurements would not only eliminatethe problemsasso-
uct frequencydiffersfrom the frequencyat whichthreshold ciatedwith subjectivemeasurements(e.g., subjectcoopera-
is measured. tion, anticipationof stimulus,experimenterbias), but would
By subjectiveassessment, 80% of the earstestedshowed alsotake a fractionof the time requiredto measureequiva-
a correspondencebetweenthe distortion and the threshold lent subjectivethresholds.
838 J. Acoust.Soc. Am.,Vol. 88, No. 2, August1990 S.A. Gaskilland A.M. Brown:Acousticdistortionproducts 838
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03
ACKNOWLEDGMENTS len, J. L. Hall, A. Hubbard, S. T. Neely, and A. Tubis, (Springer,New
York), pp. 306-313.
This work was supportedby the Medical Research Khanna, S. M., and Leonard,D. G. B. (1986). "Relationshipbetweenbasi-
Council(grantto A.M.B. ) andtheScienceandEngineering lar membranetuning and hair cell condition,"Hear. Res. 23, 55-70.
Kim, D. O., (1980). "Cochlearmechanics: implicationsof electrophysiolo-
ResearchCouncil (grant to S.A.G.). Someequipmentwas
gical and acousticalobservations,"Hear. Res.2, 297-317.
providedfromgrantsfromtheWellcomeTrustandSmithk- Kim, D. O., Molnar, C. E., and Matthews, J. W. (1980). "Cochlear me-
line Foundation. Transducers were kindly supplied by chanics:Nonlinearbehaviourin two-toneresponses as reflectedin coch-
Knowles Electronics. Thanks are due to T. Summers and D. lear-nerve-fiberresponses
and ear-canalsoundpressure,"J. Acoust.Soc.
Am. 67, 1704-1721.
Moreman for technical assistance,and to I. J.Russell and
Larson,V. D., Cooper,W. A., Talbott, R. E., Schwartz,D. M., Ahlstrom,
M.Kosslfor helpfulcommentand discussion.
The authors C., andDe Chicchis,A. R. (1988). "Referencethresholdsound-pressure
acknowledgewith thanksthe valuablecomments
of the re- levelsfor the TDH-50 and ER-3a earphones,"J. Acoust. Soc. Am. 84,
46-51.
viewers.
Levitt, H. (1970). "Transformedup-downmethodsin psychoacoustics," J.
Acoust.Soc.Am. 49, part 2, 467-477.
Liberman, M. C., and Dodds, L. W. (1984). "Single-neuronlabelingand
chronic cochlearpathology.III. Stereociliadamageand alterationsof
thresholdtuning curves,"Hear. Res. 16, 55-74.
Long, G. R. (1984). "The microstructureof quietand maskedthresholds,"
Hear. Res. 15, 73-87.
Lonsbury-Martin, B. L., Martin, G. K., Probst, R., and Coats, A. C.
(1987). "Acousticdistortionproductsin rabbit ear canal. I. Basicfea-
Anderson,S.D., andKemp,D. T. (1979). "The evokedcochlearmechani- turesand physiologicalvulnerability,"Hear. Res. 28, 173-189.
cal response
in laboratoryprimates,"Arch. Otorhinolaryngol.
224, 47- Martin, G. K., Lonsbury-Martin, B. L., Probst, R., and Coats, A. C.
54. (1988). "Spontaneous otoacousticemissions in a nonhumanprimate.I.
Brink, G. van den (1970). "Experimentson binauraldiplacusisand tone Basic features and relations to other emissions," Hear. Res. 33, 49-68.
perception," in FrequencyAnalysis
andPeriodicityDetectionin Hearing, Martin, G. K., Lonsbury-Martin, B. L., Probst, R., Scheinin,S. A., and
editedby R. PlompandG. F. Smoorenburg (Sijthoff,Lieden,The Neth- Coats,A. C. (1987). "Acousticdistortionproductsin rabbitearcanal.II.
erlands), pp. 362-374. Sites of origin revealedby suppressioncontoursand pure-tone expo-
Brown, A.M. (1987). "Acousticdistortionfrom rodentears:A compari- sures," Hear. Res. 28, 191-208.
sonof responses fromrats,guineapigs,andgerbils,"Hear. Res.31, 25- Moore, B.C. J., andGlasberg,B. R. (1983). "Suggested
formulaefor calcu-
38. lating auditory-filter bandwidthsand excitation patterns," J. Acoust.
Brown, A.M. and Gaskill, S. A. (1990). "Measurementof acousticdistor- Soc. Am. 74, 750-753.
tion revealsunderlyingsimilaritiesbetweenhumanandrodentmechani- Mountain, D.C. (1986). "Electromechanicalpropertiesof hair cells," in:
cal responses,"J. Acoust.Soc.Am. 88, 840-849. Neurobiologyof Hearing: The Cochlea,editedby R. A. Altschuler,D. W.
Brown,A.M., andKemp,D. T. (1984). "Suppressibility ofthe2fl -f2 stim- Hoffman, and R. P. Bobbin (Raven, New York), pp. 77-90.
ulatedacousticemissionin gerbiland man," Hear. Res.13, 29-37. Neely, S. T., and Kim, D. O. (1983). "An activecochlearmodel showing
Brown A.M., McDowell, B., and Forge,A. (1989). "Effectsof chronic sharptuningand high sensitivity,"Hear. Res.9, 123-130.
gentamicin treatmenton hair cellscanbe monitoredusingacousticdis- Pasanen,E.G., Wier, C. C., and McFadden, D. (1987). "Reciprocalrela-
tortion products,"Hear. Res.42, 143-156. tion betweenthe growth of an emittedcubicdistortionproductand the
Davis,H. (1983). "An activeprocess in cochlearmechanics,"
Hear.Res.9, suppression of a spontaneousotoacousticemission,"J. Acoust.Soc.Am.
79-90. Suppl. 1 81, S8.
Elliott, E. (1958). "A rippleeffectin the audiogram,"Nature181, 1076. Rosowski, J. J., Peake, W. T., and White, J. R. (1984). "Cochlear non-
Evans,E. F. (1975). "The cochlearnerveand cochlearnucleus,"in Hand- linearitiesinferred from two-tonedistortionproductsin the ear canal of
bookof SensoryPhysiology, editedby W. D. Keidel and W. D. Neff the alligator lizard," Hear. Res. 13, 141-158.
(Springer,New York), Vol. 2, V/2, pp. 1-109. Schmiedt, R. A. (1986). "Acoustic distortion in the ear canal. I. Cubic dif-
Furst, M., Rabinowitz, W. M., and Zurek, P.M. (1988). "Ear canal acous- ference tones: Effectsof acute noise injury," J. Acoust. Soc. Am. 79,
tic distortionat 2f l-f2 fromhumanears:Relationto otheremissionsand 1481-1490.
perceived combinationtones,"J. Acoust.Soc.Am. 84, 215-221. Strube,H. W. (1985). "The shapeof the nonlinearitygeneratingthe combi-
Gold, T. (1948). "The physicalbasisof theactionof thecochlea,"Proc.R. nation tone 2fl-f2," J. Acoust. Soc.Am. 79, 1511-1518.
Soc. London Ser. B 135, 492-498. Thomas, I. B. (1975). "Microstructureof the pure tone threshold,"J.
Goldstein,J. L. (1967). "Auditory nonlinearity,"J. Acoust. Soc.Am. 41, Acoust.Soc.Am. Suppl. 1 57, S26-S27.
676-689. Wiederhold,M. L., Mahoney,J. W., and Kellogg,D. L. (1986). "Acoustic
Hal, J. L. (1974). "Two-tone distortionproductsin a nonlinearmodelof overstimulationreduces2fl-f2 cochlearemissionsat all levelsin the
the basilar membrane," J. Acoust. Soc. Am. 56, 1818-1828. cat," in: PeripheralAuditory
Mechanisms, editedby J. B. Allen, J. L. Hall,
Harris, F. P., Lonsbury-Martin, B. L., Stagner,B. B., Coats, A. C., and A. Hubbard,S. T. Neely, and A. Tubis. (Springer,New York), pp. 322-
Martin, G. K. (1989). "Acousticdistortionproductsin humans:System- 329.
atic changesin amplitudeasa functionoff2/fl ratio," J. Acoust.Soc. Wier, C. C., Pasanen,E.G., and McFadden, D. (1988). "Partial dissocia-
Am. 85, 220-229. tionof spontaneous otoacoustic emissionsanddistortionproductsduring
Horner, K. C., Lenoir, M., and Bock, G. R. (1985). "Distortion product aspirinusein humans,"J. Acoust.Soc.Am. 84, 230-237.
otoacousticemissions in hearing-impairedmutantmice,"J. Acoust.Soc. Wilson,J.P. ( 1980). "The combinationtone,2f l-f2, in psychophysics and
Am. 78, 1603-1611. ear-canalrecording,"in: Psychophysical, Physiological, and Behavioural
Kemp, D. T. (1979a). "Evidenceof mechanicalnonlinearityandfrequency Studiesin Hearing, editedby G. Van den Brink and F. A. Bilsen (Delft
selectivewave amplificationin the cochlea,"Arch. Otorhinolaryngol. U.P., Delft, The Netherlands), pp. 43-50.
224, 37-45. Wit, H. P., Langevoort,J. C., and Ritsma,R. J. (1981). "Frequencyspectra
Kemp, D. T. (1979b). "The evokedcochleamechanicalresponseand the of cochlearacousticemissioris("Kemp-echoes"),"J. Acoust.Soc.Am.
auditory microstructure--Evidence for a new element in cochlear me- 70, 437-445
chanics,"Scand.Audiol. Suppl.9, 35-47. Zurek, P.M., Clark, W. W., and Kim, D. O. (1982). "The behavior of
Kemp, D. T., and Brown, A.M. (1983). "A comparisonof mechanical acousticdistortionproductsin the ear canalsof chinchillaswith normal
nonlinearitiesin the cochleaeof man and gerbil from ear canal measure- or damagedears,"J. Acoust.Soc.Am. 72, 774-780.
ments,"in Hearing:Physiological Basisand Psychophysics, editedby R. Zwicker,E. (1979). "A modeldescribing nonlinearities
in hearingby active
Klinke and R. Hartmann (Springer,Berlin), pp. 82-88. processeswith saturationat 40 dB," Biol. Cybernet.35, 243-250.
Kemp, D. T., and Brown,A.M. (1986). "Widebandanalysisof otoacoustic Zwicker, E. (1986)." 'Otoacoustic'emissionsin a nonlinear cochlearhard-
intermodulation,"in PeripheralAuditory Mechanisms, editedbyJ. B. AI- ware model with feedback," J. Acoust. Soc. Am. 80, 154-162.
Redistribution subject to ASA license or copyright; see http://acousticalsociety.org/content/terms. Download to IP: 131.156.157.31 On: Wed, 26 Nov 2014 02:37:03