Two-tone suppressionin auditory-nerve fibers: Extension of

a stimulus-response relationship*
PaulJ.Abbas
t andMurray
B.Sachs
Departmentof Biomedical
Engineering,
JohnsHopkinsUniversity
Schoolof Medicine,
Baltimore, Maryland 21205
(Received 16 June 1975; revised16 September1975)

Average
discharge
rateofsingle
auditory-nerve
fibers
in catswasmeasured
in response
to one-andtwo-
tonestimuli.Onecomponent(the"suppressor
tone")of eachtwo-tone
stimuluswasat a frequency
(f2)
whichproduced two-tone
suppression
at some
stimulus
levels.Theothercomponent (excitor
tone)
producedanincrease
in rateabovethespontaneous
ratewhenpresented alone.Fractional
response
was
definedas the drivenrate to the two-tonestimulusdividedby the drivenrate to the excitorpresentedalone.
Fractional
response is thusa quantitative
measureof theamount of suppression
producedbya suppress6r
tone.A number of qualitative
differences
werefoundin thedependenceof fractional
response
forf2> CF
andf• < CF. Forsuppressor tonefrequencies
greater
thanCF,fractionalresponse depends
onlyontheratio
ofsuppressorto excitor levels
(P2/POfora rangeof excitor
levels
(P0. For P• largeenough
to drivea unit
intosaturation,
fractional
responseincreases
with P•. For f2< CF, however,
fractional
responseis a
monotonicdecreasingfunction
of P2and P2/P•;it is alsoa monotonic
decreasing
function
of P• for P2/P•
fixed.Consistentwith theseresultsif the tonelevelratio P2/P• is fixed,rateis a monotonic increasing
functionof the overalllevelof a two-tonestimulusfor f2> CF; for f• < CF, rateis typicallya nonmonotonic
functionof overalllevel.For f2> CF, slopesof (log)fractionalresponse versus(log) P2/P• curvesare a

Downloaded from http://pubs.aip.org/asa/jasa/article-pdf/59/1/112/12144774/112_1_online.pdf

monotonic
decreasing
function
of suppressor
frequency
f2. Forf2< CF, ontheotherhand,slopedoesnot
depend
onf•. If f• isfixed,andexcitor-tone
frequency
(J])varied,
fractional
response
decreases
withf• for
f2> CF; for f• < CF, fractional
response
is independent
of f•.

Subject Classification:[43] 65.42, [43] 65.59.

INTRODUCTION nerve on one side was exposed by retracting the cere-

bellum media!ly. Glass micropipettes filled with 3 M
The transformation from sound pressure at the ear-
NaC1 were placed in the nerve under microscopicob-
drum to discharge patterns in auditory nerve fibers is servation and advanced by a hydraulically coupled
nonlinear. Recent studies have explored several non- microdrive from outside the sound-proofed room in
linear properties of this transformation (e.g., Sachs which the cat was placed. Physiological criteria, sim-
and Kiang, 1968; Goldstein and Kiang, 1068; Goblick ilar to those of Kiang et al. (1965) were used to ensure
and Pfeiffer, 1060). For example, ithasbeen shown that the electrode was not inadvertently placed in the
that average discharge rate in response to a tone at a cochlear nucleus.
fiber's characteristic frequency (CF) can be reducedby
the addition of a second tone of appropriate frequency In experiments prior to 20 November 1973, a Brfiel
and sound-pressure level (Sachsand Kiang, 1068). and Kjaer 1-in. condensermicrophone(model4132)
This reduction in rate has been called two-tone inhibi- was used as the sound source; the microphone was
tion or two-tone suppression. Sachs (1069) has devel- coupled to the bony meatus through a closed coupler.
In order to avoid the harmonic distortion inherent with
oped a quantitative relationship between discharge rate
and parameters of a two-tone stimulus for situations in condensor microphones, a Koss model Pro 4AA dynam-
which two-tone suppression occurs. ic earphone (specially selected and modified) was used
as' the source in all experiments after 20 November
In an attempt to gain insight into the mechanisms under- 1973. For frequencies below a fiber's characteristic
lying two-tone suppresion, we haveexpandedSachs' ear- frequency(where harmonicsare a problem), toneswere
lier studies to a broader range of stimulus parameters: In generatedbya Brflel and Kjaer beat-frequency oscilla-
all the two-tone stimuli used in Sachs' earlier study, tor (model 1022). With this combination of earphone
one of the tones was at fiber CF. Furthermore, the and oscillator, all harmonics were at least 70 dB below
previous study was limited to cases where the suppres- the fundamental, for fundamentals greater than 1 kHz;
sion was produced by tones whose frequency was for lower frequencies, all harmonics were more than
greater than fiber CF. In this paper we consider cases 65 dB down.
where the suppressing tone frequency is less than fiber
A calibrated probe tube leading to a Brflel and Kjaer
CF as well as cases where it is greater than CF. We 1 .
•-tn. microphone was placed at the mouth of the acous-
also examine two-tone suppression in cases where
tic coupler to allow calibration of the frequency-depen-
neither tone is at fiber CF. Data are presented from
dent sound-pressure level near the tympanum. In most
365 auditory-nerve fibers in 40 cats.
cases, sound-pressurelevels in the rate-level plots
I. METHODS shown were calculated after the experiment from the
acoustic calibration and voltage levels applied to the
A. Preparation and stimulus generation earphone. In later experiments(Figs. 9-14) wherewe
Healthy .adult cats, free from external ear obstruc- wished to compare data for different frequencies with
tion and middle-ear infections, were anesthetized with identical stimulus levels, the sound-pressure level as.
sodium pentobarbital or Dial in urethane. The auditory a function of frequency was input to the computer be-

112 J. Acoust.SOc.Am.,Vol. 59, No. 1, January

1976 Copyright
¸ 1976by theAcoustical
Societyof America 112
113 AbbasandSachs:Two-tonesuppression
in auditory-nerve
fibers 113

fore the experiment. Voltage into the earphone could by adaptation effects.
then be set to provide a specified sound-pressure level,
Most of the rate-level plots were taken with level
independent of frequency.
steps of 2 dB; in some cases the steps were 5 dB. In
the following graphs of discharge rate versus level,
B. Experimental paradigm
each point has been replaced by an average of itself
For all results presented here, sound-pressure level and its two neighbors. The first and last points in the
at the eardrum can be expressed as graph are not averaged. Data for each graph were
saved on digital tape and plotted off-line on a Calcomp
p (t) =Pt sin27rftt
+P2sin2•rf•.t. (1) plotter. In plots made in this way, the individual data
points are sometimes connected with straight lines to
The frequency(f•) andamplitudet(P•)are always such form a continuous curve.
that this tone producesan increase in rate (abovethe
spontaneousrate) whenpresentedalone. This tone will It is convenient to consider the results as an exten-
be called the excitor tone. The secondtone (f•.), desig- sion of Sachs' (1969) earlier paper. He expressed dis-
nated the suppressor tone, is always chosen in a fre- charge rate as the sum of a spontaneous
rate (R,p)and
quency range that will produce two-tone supression; a stimulus-driven rate (Rd). The discharge rate to a
i.e., for some values of amplitude P•., the rate to the single-tonestimulus, r(f•, Pt), was expressedas:
two-tone stimulus is less than the rate to the excitor
tone presented alone. Supressor-tone frequencies were r(ft, Pt)=Rd(ft, Pt)+ R,p. . (2)
chosen both above and below CF in each experiment.
The discharge rate to a two-tone stimulus, r(ft, Pt, f•.,

Downloaded from http://pubs.aip.org/asa/jasa/article-pdf/59/1/112/12144774/112_1_online.pdf

Discharge rate was measured as a function of the pa-
P•.), was also expressedas a driven and spontaneous
rameters Pt, ft, P•., and f•.. component. He proposed a simple formulation in which
When an auditory-nerve fiber was isolated, we deter- the two-tone driven rate could be expressed as
mined its characteristic frequency (CF), usually using
audiovisual cues while manually varying frequency and Rz•(fl, Pt, f•., P•.)=tto(f •, P•) . g(f ,, Pt, f•., P•.)+ tto(f •.,P•.),
level. After the CF was determined, the frequency and
where g is a multiplicative factor dependent on the
stimulus-level parameters were chosen for the two-
stimulus par•eters.
tone experiment. In all two-tone experimental runs
either the amplitudePt or the ratio of amplitudesP•./Pt In this paper we shall consider primarily the case
was held constant. We were thus left with one ampli- where there is no response to •e suppressor tone pre-
tude parameter to manipulate. Rate was measured as sentedalone, i.e., Ro(fz, P•)=O. In all cases where
a function of this level parameter as follows: Stimuli Ro(fz,P•) is takento be zero, Ro(f•, P•) was actually
were presented in 400-msec bursts once every 1.5 sec; measured and was zero. We define the "fractio•l
rise- and fall-times were 10 msec. In each experimen- resonse" as •e driven rate to •e two-tone stimulus
tal run, rate-versus-level functions couldbe determined divided by the driven rate to the excitor tone alone'
for two different stimuli, typically a single-tone and a two-
tone stimulus. Various amplitudes of the two stimuli were Rv(f •, P•, f•, P•) ß
presentedin a randomsequenceby a Varian Data 620/I fractio•l
response
= Ro(f•,
P•)
computer. Average discharge rate was computed over
the 400-msecburst. Once all levels for both stimuli
=r(fl,v,, A, - (4)
r(f•, P•) - R,• '
were presented, the results were averaged with those of
Fractional response is thus a quantitative measure of
previous sequences, andanother sequence was begun.
the effectiveness of the second tone in suppressing the
This process could be repeated as many times as de-
sired. Running plots of rate versus level for both
responseto the excitor tone; in cases where Ro(fz, P•)
=0 it is equal to g in Eq. 3.
stimuli were displayed online (see Sachs and Abbas,
1974).
II. RESULTS
In most of the experiments described below it is
essentialto comparequantitativelythe rate for twodif- A. Dependence
ofrateonP2;œ1,
P1,f2constant
ferent stimuli (typically a single-tone and a two-tone The effects of changing stimulus levels were first ex-
stimulus where we wish to measure the suppression of
amined by holding the excitor-tone level constant, while
discharge rate by th• addition of the secondtone). We
discharge rate was measured as a function of the level
have observed that the discharge rate of a fiber to iden-
of the suppressortone (P•.). This procedurewas used
tical stimuli may change with time because of adapta-
by Sachs in his earlier study. Discharge rate-versus-
tion. Care was taken in choosing stimulus parameters
P•. functions were measured for several values of Pt-
to minimize this effect. With a presentation rate of one
In addition, rate to the excitor tone alone was measured
400 msec burst every 1.5 sec, consistent changes in
for each Pt. The excitor-tone frequency(f0 was fiber
discharge rate to one stimulus were seldom observed CF.
from one presentation to the next. Furthermore, when-
ever the discharge rates for two stimuli were being Results for f•. > CF are considered first. Fractional
compared, the data were taken simultaneously through responseis plottedversusP•./Pt for six valuesof Pt in
the procedure outlined above. In this way we hoped to Fig. 1 (a). The curves for Pt < 40 dB SPL superimpose
ensure that our measure of suppression was not biased on one another, indicating that fractional response is

J. Acoust. Soc. Am., Vol. 59, No. 1, January 1976

114 Abbas and Sachs:Two-tone suppressionin auditory-nervefibers 114

p(t)= P•sin2nf•t * P2sin2nf2t f• = 125 kHz =CF rate was in saturation. Figure 2(b) plots the rate to
f= = 15 5 kHz the CF tone for four different levels; there is little
o Pe=50 dE}SPL change in rate, indicating that the levels are in the sat-
i.o ß
a
45
40
uration region. Fractional response is plotted versus
ß 35 P2/Pt in Fig. 2(a) for four different levels of Pt. The
curves do not superimpose upon one another as for low
-% • 25
levels. For any value of P•./Pt, as Pt increasesthe
-

fractional response also increases. For excitor-tone

levels that saturate the discharge rate of the nerve
o
fiber, the fractional response is no longer a simple
functionof P•/Pt as it is for lower levels. Data simi-
lar to those in Figs. 1 and 2 were obtained from 50
._1 fibers.
z
o
Figures 3 and 4 illustrate data from similar experi-
ments. In these figures, however, suppressor-tone
.i i I i i frequency(f•.) is less than fiber C F. Fractional re-
-20 -io o io 2•0
sponseis againplottedas a functionof P•/Pt for sever-
P2/Pi (dB)
al values of Pt- Unlike the case of f•. > CF, there is no
range of Pt over which the curves superimpose, indicat-
200 -
ing that the fractional response is not a simple func-

Downloaded from http://pubs.aip.org/asa/jasa/article-pdf/59/1/112/12144774/112_1_online.pdf

ß o

tion of the two-tone level ratio. As Pt increases, the

I00 -
curves shift to the left; i.e., fractional response de-
creaseswithincreasing
P t, forfixedtonelevelratioP•./P•.
_•.) -
.

_ p(t)=P,sin2wf,t +P2sin2wf2t
4/5/73
601 3 82 kHz = CF
5.00 kHz
,o J.0 -

P, dB SPL
ß o ß ß "' "' P,=70dBSPL
FIG. 1. (a) Fractional response plotted (log scale) as a func- o

ß o
tion of P2/P1 (dB)with P1 as a parameter. Fractional response o

is defined as the driven discharge rate to the two-tone stimu-

ß o ß 65
lus divided by the driven rate to the excitor tone alone. Two-
tone discharge rate was measured versus P2 for each level of
P•. In addition, rate to the excitor tone alone was measured
for each P•. Fractional response was then calculated and 060

plotted, each symbol corresponding to a constant level of P•. • .2-

f• = CF •- 12.5 kHz; f2 = 15.5 kHz. (b) Plot of driven discharge ß

rate to the CF tone as a function of P1. Data points are those ß55

used to calculate the corresponding fractional response curves

in (a). Note that P• levels have been chosen to elicit rates
I I i i I I I
well below and also at the saturation discharge rate of the -15 -5 5 15 25 55
fiber.
Pz/Pi
(dB) (•
200
dependentonly on the ratio of tone amplitudes(P•/Pt)
for these small values of Pt- These data confirm
IOO
Sachs' earlier observations which were limited to low
o ß D
excitor-tone levels. However, at higher levels of Pt -

ß
50 .

(45, 50 dB SPL), fractional response increases with

increasing Pt- Figure 1(b)' plots the discharge rate to
the excitor tone presented alone for the Pt values used
3/29/73,
in Fig. l(a). Rate to the excitor tone alone increases 8.05
with levels for levels below 45 dB SPL, but reaches a
rate saturation at about 45 dB. Comparison of Figs.
P, dB SPL
1 (a) and 1 (b) indicates that fractional response depends
only on the ratio of amplitudesP•/Pt for Pt in the range FIG. 2. (a) Fractional response plotted (log scale) as a func-
where rate is increasing. For Pt large enoughto satu- tion of P2/P• (in dB), withP• as a parameter. (b) Plot of
rate the fiber, fractional response increases with in- driven discharge rate to the CF tone as a function of P•. Data
creasing Pt. points are those used to calculate the corresponding fractional
responsecurves in (a). Note that the P1 levels chosenin this
In the experiment illustrated in Fig. 2 the excitor- case drive the fiber at or near the saturation ,rate. fl = CF
tone levels were purposely chosen so that the discharge = 3.82 kHz. f2 = 5.00 kHz.

J. Acoust. Soc. Am., Vol. 59, No. 1, January 1976

115 Abbasand Sachs:Two-tonesuppression
in auditory-nerve
fibers 115

p(t)=P,sin2 Trf,t + Psin2 constant

(i.e., suppressor-tone
levelwasheldata
fixed increment (in dB) above the excitor-tone level).

.9•© © ©•©•}•'• • Discharge

ratewas
measured
asafunction
oft:',the
.8• • •"r• f'=12.9
kHz=CF"overall
sUmulus
amplinde."
Discharge
rate
functions
•
.••
.
5•
• • ••
• •
f•'=
,.oo
kHz werealso
12/2o/72 amp•tude
measured for
of•e excitor
excitor
tone
tone presented
inthetwo-tone
alone.
(Notethatin Eq.5 theoverall•pli•de P equals
the
stimulus.)
• '[ • • • 2.,5 •e excitor-tone
frequency
was
again
limited
tofiber
CF; suppressor-tone frequencies were chosen both
• .4 •ove andbelow
CF.
Fibre 5 illustrates three examples of f• > CF. Dis-
• '• charge
rate
tothe
excitor
tone
alone
and
tothe
two-tone
• stimulus
arebo•plotted.
Ineach
case
thefractio•l
• .2 . responseis calculatedandplottedas a •nction of over-
•1 amplitude • (given in dB SPL). In each c•e there
is a range of levels over which fraction• response is

Downloaded from http://pubs.aip.org/asa/jasa/article-pdf/59/1/112/12144774/112_1_online.pdf

/. i
,o i i
,o i i •' -
• - o 85

ß 95

rs,o.s
sce,
P2/P• (dB)with P• as a parameter. The suppressor-tonefre-
quency•2) was chosenbelow fiber CF. f• = CF = 12.9 kHz. •
=1.00. • .2-

Fig•e
4shows
•at
•ebeha•or
creases
of
the
fractional
re-
sponsecurvesin the rangeof P[, whereR• •[, P[) in-
is similar to •e beha•or in •e saturation re-
I I
•on. •call thatquitedifferent
behavior
wasobserved .• 5 15 25 55 45
for f2 > CF (Figs. 1 and2) in the dynamicand saturation
P•/P• (dB)
regions. Results similar to those shownin Figs. 3 and
4 were obtained from 40 fibers. 200 -

o•e
We can summarize results presented above as fol- IO0 -
lows: For f•. >CF and at low levels of P•, when Px is in-
creased the level of P2 must be increased proportional- 5O
ly in order to producethe same amountof suppression. 5/17/73
However, when Px reaches levels close to those which 4.02
cause l:td (f•, P•) to saturate, increases in P• require
more than proportional increases in P2 in order to pro-
i i i i
duce the same amountof suppression(i.e., the same IO
65 75 85 95
fractional response). For f2 < CF, when Px is increased P!(dB SPL)
the level P2 must be increasedless thanproportionally
in order to produce the same amount of suppression.
FIG. 4. (a) Fractional response plotted (log scale) as a func-
B. Dependenceof rate on overall stimulus level; tion of P2/P! (dB)with P! level as a parameter. (b) Plot of
driven discharge rate to the CF tone alone as a function of P1..
fl, f2, P2/P1constant Data points are those usedto calculate the correspondingfrac-
Responsesfor bothf•. > CF andf•. < CF were measured tional response curves in (a). P1 levels have been chosento
underthe conditionP2/Px= constant. The stimulusin elicit rates well below and slso at the saturation discharge
this case can be written as rate of the fiber. Note that the levels used in this figure repre-
sent dB attenuation, i.e., dB attenuation from the maximum
p(t)=P(sin2•rf•
t +D sin2•rf2t)
, (5) voltage into the earphone (0 dB attenuation represents approxi-
mately 100 dB SPL at the tympanic membrane). fl = CF = 6.20
whereD= P2/P•. Thetone-amplitude
ratio (D) washeld kHz. f2 = 0.50.

J. Acoust. Soc. Am., Vol. 59, No. 1, January 1976

116 Abbas and Sachs:Two-tone suppressionin auditow-nerve fibers 116

p(t):P-(sin2wf•t+D.sin2Trf•t ) functions approach the same saturation rate. Note

also that at very low stimulus levels, fractional re-
200_ f.r. 1.0r
=D
/ I
sponse tends to increase as level decreases. This in-
3/29/73 crease in fractional response with decreasing level in-
8.01 dicates that two-tone suppression diminishes at low
enough stimulus levels. The curves of Fig. 5 are typi-
cal--of results obtained from 62 fibers.
ioo
- //7 C)o.5
• f, = 2.17 =CF
Results for a similar experiment donewith f•. < CF
f, =:3.20
///,..,, D =2$dB are illustrated in Fig. 6.
rate are plottedin Fig. 6(a)for severalvaluesofD =P•./Pt.
Excitor-tone and two-tone

The corresponding fractional responses functions are

04050 60 70 80 P,=P
plotted in Fig. 6 (b). As can be predicted from the pre-
6:5 7:5 8:5 9:5 103 vious experiment, the fractional response in each case
is a decreasing function of overall level P (equals ex-
$00
F citor-tone level). The resulting two-tone rate function
3/29/73
is typically nonmonotonic; rate increases with overall
4.01 level for small values of that level, then decreases for
higher overall levels. Note also that the value of D,
i.e., the level of the suppressor relative to that of the
f, = 1.60=CF excitor tone, apparently sets the level at which both

Downloaded from http://pubs.aip.org/asa/jasa/article-pdf/59/1/112/12144774/112_1_online.pdf

fz = 2.80
D = 29 dB
fl

12/27/73
,oo
[/ J iD=25
0 [ •
d8, I
•
I
p(t)
=P.
(sin
2wflt
+D-sin
2rr
f2t) P2/P[=D

200,•• I.Or-
11.2 $0 40 50 60 70 PI=P
/ i
i
I
I
i f,=8.10=CF zoo f•
f,= I1.0
I00 0.511- D=I4dB fz
,, ioo
i
• • io
I • 0
i
i
e D=30d8
i

41 51 61 71 81 P,=P • •o•o •o Fo ..=. • o

o
i • i • • Pz • P2
55 65 75 85 95 • 2oo
• f•
d8 SPL • 30 40 50 60 70 •0 • =P
< dB SPL

FIG. 5. Plots of discharge rate (solid curves) versus overall • • f•

stimulus level (P) with tone-level ratio (D) constant for three
fibers. Soundpressure is of the form p(t) =P (sin27rflt+D. D=•Sd•
sin 27rf2t)
, P2/Pi =D. Plots of calculatedfractional response i i i i i

(dashed curves) versus P are superimposed, with their ordi- %• 40 • 6• • P•P

nate scale drawn at the right. The discharge rate curves have
been smoothed (see discussion in Sec. I) and the adjoining data
f• 4/29/74
points connected with a straight line to form a smooth curve. 1.04
In each of the three graphs, the solid curve labeled "fi" repre- f, = 14 O= CF
sents the rate to fi alone [r(fi,P1)], and the solid curve labeled f2 f•: 8.0
'•fl +f2" representsthe two-tonedischargerate. There is no
responseto f2 presented alone. Fractional responseis calcu- D=40d8
lated at each level of P and plotted by the dotted curve ("f. r.").
ø3o4•o5•o6•o;o 'P,:P
The levels of eachof the componenttones (P1=P andP2) are
indicated on the abscissa. The fiber number, CF, suppressor
70.;0 ;0 100IlO P2
dB SPL
frequency and tone-level ratio (D) are indicated to the right of
the graph for each fiber.
FIG. 6. (a) Plots of discharge rate versus overall level P
for four values of tone-level ratio (D). The CF-tone rate
("fi")andtwo-tonerate ("fi +f2") are plotted. (b) Plot of frac-
approximately constant. This result is consistent with tional response versus overall level P, with tone-level ratio
Fig. 1. When the rate function for the CF tone (D) as a parameter. Fractional response is calculated from
approaches saturation, fractional response increases the data presented in (a) for each value of D. fi = CF = 14.0
sharply with level and the correspondingtwo-tone rate kHz. f2 = 8 kHz.

J. Acoust. Soc. Am., Vol. 59, No. 1, January 1976

117 Abbasand Sachs:Two-tone suppression
in auditow-nervefibers 117

however, at higher levels, the two-tone rate increases

500
1 D=ISdB
500
I D= 20dS
once more as the rate to the suppressor tone alone in-
creases. At high levels the two-tone rate function is
asymptotic to,that for the suppressor alone. Figures
6 and 7 are typical of results from 30 fibers.
The results of this section can be summarized as fol-
lows: For f,. > CF, P,./P• =D constant,bothtwo-tone
IOO IOO
discharge rate and fractional response are monotonical-
ly increasing functions of Pfor large P. For 3%.
< CF, frac-
tional response is always observed to be monotonically de-

o
I I ,I I
:,, l: :,",
'P
I
o
I I
6'o70' P
I , i
creasing and the two-tone rate functions are typically
nonmonotonic.

55 65 75 85 P• 60 70 80 90 P• C. Dependence
of rate on f2
The effects of changing suppressor-tone frequency
300 300 -
have been examined using the stimulus paradigm

I D=25dB 30 dB
described in Eq. 1. The excitor-tone frequency (f•)
was the CF and its level was held constant.
rate for a two-tone stimulus was measured
Discharge
as a function
•00 fl 200

Downloaded from http://pubs.aip.org/asa/jasa/article-pdf/59/1/112/12144774/112_1_online.pdf

f• of suppressor tone level (P,.). The data are plotted in
Fig. 8. In Fig. 8(b), curves are generated for several
values of the suppressor-tone frequency in the range
I00 ioo
21.0-30.0 kHz (greater than the CF). Two observa-
tions can be made: First, as f,. increases (gets further
from the CF) the curves shift to the right, indicating
/ / f•
that a higher level of suppressor tone is needed in
i order to produce the same amount of suppression.
o6o 7'o i
40 i
50 i
60 iP Secondly, the slopes of the curves for large P,. show a
75 . 95 P• 7'0 9'0 systematic decrease as f,. is increased. Both of these
dB SPL dB SPL observations were made earlier by Sachs (1969) for
5/2/74 >CF.
p(t)=P.(sin2nf,t+D.sin
2Tr
f•t) 402
fi=I40=CF
Figure 8(a) illustrates data from the same fiber; in
P2/P,=D f•:o5o this case the suppressor frequencies were chosen less
FIG. 7. Plots of discharge rate versus overall level P for than the CF. The curves shifted to the right when/2
four values of tone-level ratio (D). The CF-tone rate (fi) and was chosen further from the CF, but the slopes of the
two-tone rate (fi +f2) are plotted. In this case the fiber ex- two curves were approximately equal, even though the
hibited a response to the suppressor tone presented alone frequencies differed by an octave. An additional note
(labeledf2 in eachplot). fl=CF=14.0 kHz; f2=0.5 kHz. can be made concerning the relative slopes in Figs.

fractional response and two-tone rate begins to de- p(t)=P•sin2nf, t + P•sin2TTf2t

crease. Highervaluesof P•./P1causethe two-tonerate
to decrease at lower overall levels. Also the fractional
response curves for all four va!ues of 19appear to be
similar in shape; as 19increases the curves simply
30.0
shift to the left.
.0

Thus far discussion has been limited to the situation

in which there was no response to the suppressor tone
presentedalone,i.e., Ro (f,., P..)=0. In manycases, 20.0•/•
however, two-tone suppressioncan occur (i.e., the rate 5.00
rate to the two-tonestimulusis less tl{anthe rate to
the excitor tone alone) when the suppressor tone itself /o
elicits a response from the fiber. Figure ? illustrates P, (dS SPL) P, ( dS SPL) 4/26/75
1.02
results for some interesting cases with f,. < CF and f, =CF=17.8kHz
P•/P• = constant;in thesecasesthere was significant P,=60 dB SPL

responseto f2 presented alone. The discharge rate to

FIG. 8. Plots of fractional response versus P2 (P! constant)
boththe excitor andsuppressortor/ealoneas well as with suppressor-tone frequency (f2) as the parameter. The
the two-tone rate are plotted for several values of 19. characteristic frequency of the fiber is 17.8 kHz. Each sym-
As in the casewhere Ro (f,., P•.)=0, the dischargerate bol represents data for a different f2 (indicatedin kHz). In (a),
increases with overall level at low levels, then begins suppressorfrequencies were less than the CF; in (b)f2 was
to decrease at a level dependent on 19. In this case, greater than CF.

J. Acoust.Soc. Am., Vol. 59, No. 1, January 1976

118 Abbas and Sachs: Two-tone suppressionin auditory-nerve fibers 118

trates both two-tone and excitor-tone rate functions for

several values of f•. As f• is increased above the CF,
the two-tone rate is depressed further from the f• rate
f, p(t): P.(sin2Trf•t+ D-sin2•rf2t) than it is for f• = CF. The corresponding fractional re-
P2/P•=D sponse in Fig. 9(b) appears to be monotonically de-
I00 , =2 O0kHz
creasing with increasing f•. For f• > CF andf•. > CF it
has been observed in 46 fibers that fractional response
decreases with increasing f•.
or
200- For f• < CF (andf•. > CF), however, results are less
straightforward. In Fig. 9 one frequency was chosen
•j 1.0 2.0 2.40:CF below the CF. Whenf• = 2.00 kHz the two-tone rate
/ /
/ /
/ / f•+f2
r, function is closer to the excitor-tone rate than when f•
I00- / / f•=2.40=CF o i ::
ii
• V.-.•.•.....
' / 3,.00 =CF (2.40 kHz). The fractional response is conse-
/I
I/ n,. \ / quently slightly greater at 2.00 kHz than at the CF.
• .5 '•.._ ./
.. z Figure 10 illustrates data from another fiber, where f•
ß
o
was chosen at values less than and equal to the CF. As
. .
.
in Fig. 9, as f• is decreased the two-tone rate functions
I.• f• :...'...' u..oJ I I I I I get progressively closer to the rate function for ft
49 59 69 79 89 P•:P
<:[ ::.:'
f•+f2 I I I I I alone; i.e., there is less suppression. When the frac-
62 72 82 92 102 P= tional response is plotted versus overall level P•, how-
03 iiii f•=2.70

Downloaded from http://pubs.aip.org/asa/jasa/article-pdf/59/1/112/12144774/112_1_online.pdf

dBSPL (• ever, an apparent discrepancy appears. For a partic-
ular value of overall level P, fractional response de-
0j""( creases with decreasingf•; i.e., our measure of sup-
6o• f, // / pression increases with decreasing f•. The apparent
40 "/.//
'/f•+
fz discrepancy can be resolved by considering the effect
•
20k
//
/ /'/ f•=3.00
12/27/73
132

f==3.80 kHz
• '/ •.""'
ol,;.';'"
, , , D=I3dB
200

49

62 72 82 92 102
59 69
I I

d9 SPL
79 89
I I I
P•'P
ß

IOO
f'•/'•'
p(t):
P.(sin
2•rf,
t+D.
sin2•rf2
P2/R=D .

f, = 4.50 kHz
FIG. 9. (a) Plots of discharge rate versus overall level P D = 25dB

(D=P•./P1 constant);eachgraphrepresentsresponsesfor a
different value of excitor-tone frequency (fl). The abscissae I I , I I I

are the same for all four graphs; the levels of each component
are indi.cated. Rates to the excitor tone alone ("fl") and to the ..

two-tonestimulus ("f! +f•.") are plottedfor valuesoff! chosen --.. 200 [u 1.0- /
below (2.00 kHz), at (2.40 kHz), and above (2.70, 3.00 kHz) o 1•.20
the CF. (b) Plot of fractional response versus overall level
(P) with excitor-tone frequency as a parameter. Data plotted • I ,'?
ß
'J .5 - 2.60I
in (a) were used to calculate the fractional
bols used correspondto the same values of fl as in (a).
response; the sym-
'.',oo
t !; z
_o

•• J,
01 ,[/ f,=2.60
'-'T'"'",, i i i

8(a) and8(b). The slopeof the curvesfor f•.< CF are z

uJ 200 -
generally steeper than for f•.> CF. The data in Fig. 8
seem to indicate that for f•.> CF, as frequency de- dB
SPL
creases, the slopes of the curves approach the slopes
for f•.< CF• This trendhasbeenobserved
in seven ,oo- J //
other fibers for which we have data for both f•. < CF and i /.Jr,
+f, ;5/20/74
i..:/
1.03
f•. > CF. The results shownin Fig. 8 (a) are similar to
those from 16 fibers; those in Fig. 8(b) are typical of
results from 19 fibers. % ;o e'o

D. Dependence
of rateon fl dB
SPL
The effects of changing fz were explored with the FIG. 10. (a) Plots of discharge rate versus overall level P
stimulus in the form p(t) = P (sin2•ftt + D sin2•f•.t). Dis- (D=P2/.P{constant);eachgraph representsa different value
charge rate was measured as a function of overall level of excitor-tone frequency (fl). Excitor-tone rate (fl) and two-
P while P•./Pz= D was held constant. Rate-versus-level tonerate (fl +f2) are plottedfor values of j• chosenat (3.20
data were.generated for several different excitor-tone kHz) and below (2.60, 2.40 kHz) the CF. (b) Fractional re-
frequencie s . sponse versus overall level (P) with excitor-tone frequency
(fl) as a parameter. Data in (a) were used to calculate the
We considerfirst the casef•.> CF. Figure 9(a) illus- fractional response. CF = 3.20 kHz. f2 =4. 50.

J. Acoust. Soc. Am., Vol. 59, No. 1, January 1976

119 Abbas and Sachs:Two-tone suppressionin auditow-nerve fibers 119

p(t)= P-( sin2wf,t +D.sin2wfzt ) ing frequency. Results shown in Figs. 11 and 12 are
Pz/P, =D typical of those for 32 fibers studied for this stimulus
condition.
CF = ;5.20 kHz
D=25dB The effects of changingf• for f• < CF have been ex-
,,, I.O- 2.60 amined with the same paradigm. A typical examole is
z shown in Fig. 13. The two-tone and excitor-tone
o stimulus curves are generated for four values off•
n / I (includingthe CF); the sound-levelratio t>•/t>•is fixed
t¾ ..:' at 20 dB in all cases. The values off• have been chosen
JO.õ- X ........ so that there is a significant difference in threshold for
\ ,'....... ...oo kHz:CF responses to the excitor tones presented alone. As in
Figs. 6 and 7, the two-tone rate functions are nonmono-
i--
tonic. Note, however, that for each value of f• the two-
iz ß
tone rate function begins to decrease at approximately
U= I I
the same sound level, P = 50 dB SPL, even though there
o moo 200
are large differences in thresholds to the excitor-tone
DISCHARGE RATE
alone. Indeed, when the fractional response is plotted
{spikes/sec) $/20/74
1.0• as a functionof level I'Fig. 13(b)], the four curves
superimpose. At low levels the fractional response is
FIG. 11. Plot of fractional response versus discharge rate approximately one, indicating that there is little sup-

Downloaded from http://pubs.aip.org/asa/jasa/article-pdf/59/1/112/12144774/112_1_online.pdf

tofl alone with excitor-tone frequency as a parameter. Data pression. At higher levels the fractional response
plotted in Fig. 10 were used to calculate fractional response at
for all four curves decreases sharply to zero, following
several values of discharge rate tof t alone [r(fl,P1)]. The
data points were connected by a straight line to form a smooth the same functional relationship. Figure 14 shows ex-
curve. Calculations were made for the three values of fl and amples from four other fibers. The results shown in
plotted using solid, dashed, and dotted curves corresponding Figs. 13 and 14 are typical of those from 70 units
to the same values of fl as in Fig. 10 (a). studied in this condition.

Thus, for f• < CF, the amount of suppresion is inde-

of saturation of the fiber discharge rate. We saw in pendent of f•. Given Pt, f•, and P• the fractional re-
Figs. 1, 2, and 5 that as the excitor-tone rate reaches sponse is completely defined. On the other hand, for
saturation, fractional response begins to increase with
level and the two-tone rate approaches the same satura-
tion value. Now the rate to a CF tone reaches a satura- ' p(t) = P-(sin 2wf, t+D.sin2wf2t)
P2/R:D
tion at a sound level below that for the other excitor
tones. The corresponding increase in fractional re- 20 30 47:CF
z
sponse thus occurs at a lower level for f• = CF than for o

f• < CF. (The vertical dashedlines in Figs. 10(a) and (/3

10(b) indicate this level at which the CF rate function 0

begins to saturate. ) The observed decrease in frac-

tional response with decreasing f• is thus related to the
Z
0
/I.20 .5-
effects of saturation at different sound levels for dif-
0 I •-• I I
ferent f•. In discussion we suggest that rate saturation 3/28/74
IOO 2•o oI/3/74 IOO 2oo

occurs at a later stage in cochlear processing than does 1.05

D:ISdB
9.03
f:, = 2.40 D =20 dB f2:1.50
two-tone suppression. As we have defined it, fractional
response reflects both the suppression and the rate
saturation. For overall levels which drive a unit into
saturation, fractional response may thus not be a good Z50 I.O '

measure of suppression.
8.10=CF
•./5.50
5.20// _
ß . =
In order to eliminate this effect of saturation at dif-

05
CF
_.1 ,5- .5-
<1: '
ferent levels, we can compare fractional response at z

the same discharge rate for each excitor tone. In Fig.

11, the dataare replottedas fractional'responseversus
o

••.,•8.70
discharge rate to f, alone, r (f•, P,). Except for very o o
low rates where the small number of spikes recorded DISCHARGE RATE DISCHARGE RATE
($pikes/sec): (spikes/sec)
gives rise to large statistical fluctuation, the fractional 12/•7/7• 3/28/74
response shows a clear tendency to increase with de- ii.2 I10
D=15dB f2= I1.0 D=20dB f•,= 750
creasing f• below the CF. Figure 12 illustrates exam-
ples from'several other fibers, for f• values both above FIG. 12. Plots of fractional response versus discharge rate
and below the CF. The fractional response is again to f! alone with exeitor-tone frequency as a parameter. Each
plotted as a function of excitor-tone discharge rate. of the four graphs plots data from a different fiber. The value
The trend is clear for values of f• both above and below off! (in kHz) is indicated adjacent to each curve; values were
the CF; the fractional response increases with decrems- in general chosen both above and below fiber CF.

J. Acoust.Soc.Am., Vol. 59, No. 1, January1976

120 Abbas and Sachs: Two-tone suppressionin auditow-nerve fibers 120

•oor / (Fig. 1). However,at higherexcitor-tonelevelswhere

the rate of ft alone approaches saturation, the fractional

,oo
r
/'f•=145 p(t)=P'(sin2Trf,t +D-sin2•rf2t)
responseincreaseswith Pt for fixed P•./Pt (Fig. 2).
For f•. < CF, fractional response is not a function of
Pz/P, =D:20dB P•./Pt over any rangeof levels; for fixed P•./Pt, frac-
tional response decreases as Pt increases (Figs. 3 and
4).
:,oo f• = 14kHz=CF
(2) Whencurvesof fractionalresponseversus
are plotted for several values of suppressor-tone fre-
qunecy, f•., the slope of the curves in log-log coordi-
nates is a decreasing function of f•. for f•.> CF (Fig. 8).
• • | ........14.5
For f•. < CF, no change in slope is observed with changes

"'i 14.O
=CF
z 13.0
• o
o
a. 12.0 in f•. (Fig. 8). Slopesfor f•. < CF are greater than those
UJ :•00
//
for f•.> CF; i.e., slope is a decreasing function of
.J ß
/ f• =15.0 for all f•..
I 0

(3) In later experiments,the tone-level ratio

• •oo •,/ • was fixed and discharge rate and fractional response
• f,'fz u. 0•'0:5040 50 60 PIP
i , , i i i
40 50 60 70 80 P2 were measured as functions of overall level P (see Eq.
dB
SPL (• 5). As was to be expected from result 1, for f•. > CF,

Downloaded from http://pubs.aip.org/asa/jasa/article-pdf/59/1/112/12144774/112_1_online.pdf

the two-tone rate was always a monotonically increas-
ing function of overall level P (Fig. 5) and fractional

zoo
[ / 1/24/74
3.01
CF = 14.0kHz
response increased at high overall level P.
%.< CF and RD (%.,P•.)= 0, fractional responsewas
alwaysa monotonically
decreasing
functionof P (Fig.
For

,oo
I /f':
la.o f2 = 1.00kHz
6). The two-tone rate functionsfor %.< CF were typical-
ly nonmonotonic, showing a decrease at high levels.
0l/*"i.........
"•.........
• , ,
26 50 40 50 60 P•=P
i i i l (4) The effects of changing excitor-tone frequency
4O5O6O;o ;o P2
were examinedwith P•./Pt constant. For f•.> CF, frac-
dE)
SPL (• tional'response
decreased
(indicating
moresuppression)
FIG. 13. (a) Plots of discharge rate versus overall level P as ft increased with f•. constant (Fig. 11). In constrast,
(D=P2/P 1 constant)with eachgraph representingresponses fractional response was independent of ft for f•. < CF
to a different excitor-tone frequency (fl). The abscissaeare (Fig. 13).
the same in all four graphs, the levels of each of the component
tones are indicated. Excitor-tone alone (fl) and two-tone rate
("fl +f2") are plottedat values of fl chosenbelow (12.0, 13.0 p(t) =P.(sin2wf,t +D.sin2wf=t)
kHz), at (14.0 kHz), and above (14.5 kHz) the CF. (b) Plot of P/P :D
fractional response versus overall level (P) with excitor-tone 1/24/74 2/19/74
frequency as a parameter. Data plotted in (a) were used to 1.04 1.07

s { CF=14.4 CF= ;5.80

calculate the fractional response; the same symbols as in (a)
I.o - \ •. f,:•,.oo • f;=1.50
correspond to the different values offl. CF=14.0 kHz. f2
=1. oo.
}• -- '4.4=CF=f,
1.0q:,,,g'•
:•:•
D-'ZOdB
-- 3.80
=CF
=f,
•i --,34 : '•i..•:
"f' t --- 4 .oo
...... 14.9

f•. > CF, the fractional response is highly dependent on

0.5 -

0.5i'•"x•,
.....
3.40
ft (Figs. 10-12); as excitor-tone frequncy increases,
the fractional response decreases. ;o 30 40 50 60 p,=P
4;::)
õ• 6•0/0 8•0
III. DISCUSSION

A. Summary of results 2/22/74 2/22/74

3.06
3.10
•/,-• CF=8.30 CF: 0.60
A primary new result presented here is that the prop- Y.. tt ,,-5.5o 1.0 - f==0.20
• O=20d8
eries of two-tone suppression for the suppressing tone •,/•.......'..,.•
• -- 8.90=CF=f, •.(.,'•:
^ -- 0.60
--CF
=
frequency, f•., above CF are very different from those •,\ ---9zo :,,"• \%. .---o.5o
• • '•,, --- 0.40
for the suppressing tone below CF. For convenience
we shall review several of these differences:
0.5
0.5 :.•% ......
o.70
(1) We defined fractional response as the driven rate
to a two-tone stimulus divided by the driven rate to the
excitor tone alone. In an initial series of experiments,
% 30 40 50 60
ß,7,70 P, = P '20 3,0 40 50 60
,,'o ,'o ½o
P,=P

fractional response was measured as a function of sup- dB SPL dB $PL

pressor-tone level P•., with excitor-tone level Pt as
a parameter. For f•. > CF, fractional response depends FIG. 14. Fractional response versus overall level (P) with
onlyonthetonelevel ratio P•./Pt at moderatePt levels excitor-tone frequency as a parameter. Data from four fibers.

J. Acoust.Soc.Am., Vol. 59, No. 1, January1976

121 Abbas and Sachs:Two-tone suppressionin auditory-nervefibers 121

B. Comparison with previousstudies of dischargerate of a tone with increasing level corresponds to the de-
to two-tone stimuli crease in fractional response (increase in suppression)
with suppressor level (our Fig. 1). The attenuating
The observations made here are in general agree-
power of a tone is greater when its frequency is chosen
ment with those reported in previous studies using two-
closer to the CF; in our data, "threshold" for suppres-
tone stimuli. Two-tone suppression was observed in all
sion increases as the suppressor frequency is taken
fibers that were isolated long enough to permit a thor-
farther fromCF (Fig. 8). Rose et al. also observed
oughsearch. Suppressionwas observedfor suppressor that the increase in attenuating power for a given incre-
frequenciesaboveCF for fibers with CF's in the range ment in P•. dependsuponthe frequencyf•., at least for
0.10-29.5 kHz; suppression was seen for suppressor
f•. > CF; this observation probably corresponds to that
frequenciesless than CF for fibers with CF's in the made earlier (Sachs, 1969) and repeated here (Fig. 8)
range 0.42-28.7 kHz.
that the slope of fractional response versus P•. (in log-
Several portions of the study both repeated and ex- log coordinates) is a decreasing function of suppressor
tended earlier observations by Sachs (1969). His ob- frequency f•. for f•. > CF. Finally, Rose et al. consider
servation that the amount of suppression for f•. > CF is two-tone stimuli of the form of Eq. 5, where the tone-
dependent
onthe tone-levelratio (P•./P1)wasconfirmed level ratio is constant and the overall level is increased.
here for excitor-tone levels below saturation; Sachs They state that in such a situation the component which
did not present data for higher excitor-tone levels. Our is less effective at low overall levels dominates the re-

plots of fractional responseversus P•. for different sup- sponse at high levels. Forf•.< CF, our results would
pressor frequenciesalso agreedwith Sachs' earlier seem to be consistent with this statement; in our Fig.

Downloaded from http://pubs.aip.org/asa/jasa/article-pdf/59/1/112/12144774/112_1_online.pdf

results: The slope of the fractional response versus P•. 7 rate increases at low overall levels (where rate to
function decreased with increasing f•. (Fig. 8). Sachs the CF alone is increasing rapidly); it then decreases
did not investigate two-tone suppression in the condition (at levels where the rate to the CF has saturated but
f•.< CF, where quite different results have been ob- fractional response is decreasing toward zero); and
served here. finally, rate increases again at levels where the re-
sponseto f•. (the "less effective component")begins to
Arthur et al. (1971) have presented results in which
be significant. For f•. > CF, however, rate is a mono-
(P•./Pi) is constantanddischargerate is measuredas tonic function of overall level from threshold to satura-
a function of overall stimulus level. They found that
tion. Our methods do not allow us to separate our re-
the two-tone discharge rate could be nonmonotonic, de-
sponses to the two components in this case and no com-
creasing with overall level and then increasing at higher
parison with the Rose et al. result is possible.
stimulus levels [their Fig. 3(b)]. The example illus-
trated in Arthur et al. is for f•. less than the CF and is The origin of the differences between two-tone
consistent with our observations for f•. < CF (Figs. 6, 7, suppression for f•. > CF and f•. < CF is a matter of
and 12). speculation at this time. Indeed, the origin of two-
tone suppression must be a matter of speculation.
Roseet al. (1974)
ha•emademeasurements
ofre-
While Rhode and Robles (1974) have seen two-tone
sponseof cells in the anteroventral cochlear nucleus in
response to two-tone stimuli. A comparison with their suppression in basilar membrane displacement for
data is difficult, primarily because of the different re- f•. > CF, they are not clear about the casef•. < CF.
sponsemeasures which they employed. While re-
Furthermore, Kim's basilar membranemodel (Kim et
sponseshere were characterized by average discharge al., 1973)showsreduction of the rms value of the total
rate, Rose et al. considered details in the fine temporal output for f,. > CF but not for F,. < CF. Dallos e! al.
structure of responses to two-tone stimuli. They tried (1974) observed that the cochlear microphonic produced
to assess the relative contribution of each component by one tone can be reduced by the addition of a second
tone by fitting period histograms of the response by a tone of greater frequency but not by one of lower fre-
weighted sum of sinusolds at the componentfrequencies. quency. Legouix e! •f. (1973), on the other hand, found
this "two-tone interference" in cochlear microphonic
Furthermore, their stimulus always consisted of two
harmonically related tones, phase locked to one another, for interfering-tone frequencies both above and below
while here the frequencies were not harmonically re- the exciting frequency. Definition of the precise mech-
lated. Finally, their data were generally taken at stim- anism (or mechanisms) for two-tone suppression must
await further direct measurements of cochlear function.
ulus levels at which the discharge rate was in satura-
tion for most tone combinations. Nevertheless, sev- The results we have presented here, however, place
eral comparisons can be made. Rose et al. designate strong constraints on acceptable models of that function.
the measured response components to f• and f•. in the
joint responseto be A• and A•.•, respectively. The
attenuatingpower of A•.• is defined as the difference *This research was supported by the National Institutes of
Health (Grant Nos, 2-RO1-NS-05143 and 1-RO1-NS-12112)
(in dB) betweenA• and the responseamplitudewhen and the Air Force Office of Scientific Research (Contract No.
f• is presented alone. In order to. make comparisons F 44620-71-C-0024). M. B, Sachs is an NIH Research Car-
with their data we shall compare.the behavior of the eer Development awardee, P. J, Abbas was an NIH predoc-
attenuating power with that of the fractional response toral trainee,
which is our measure of two-tone suppression. Sever- tPresentaddress: Departmentof SpeechPathologyandAudiol-
al of their observations correspond well with those ogy, University of Iowa, Iowa City, Iowa 52242,
presented here: The increase of the attenuating power lIn our equationsthe amplitudesP1 andP2 andP will have

J. Acoust. Soc. Am., Vol. 59, No. 1, January 1976

122 Abbas and Sachs: Two-tone suppressionin auditory-nerve fibers 122

dimensions
dyn/cm•'. In the figures, levels will be usedin- Membrane Motion," J. Acoust. Soc. Am. 54, 1517-1529.
stead of amplitudes. Levels are given in dB SPL. Legouix, J.P., Remond, M. Co, and Greenbaum, H. B.
Arthur, R. M., Pfeiffer, R. R., and Suga, N. (1971). "Prop- 0.973). "Interference and Two-Tone Inhibition," J. Acoust.
erties of Two-Tone Inhibition in Primary Auditory Neurons," Soc. Am. 53, 409-419.
J. Physiol. (London)212, 593-609. Rhode, W. S., and Robles, L. (1974). "Evidence from
Dallos,-P., and Cheatham, M. A. (1974). "Cochlear Micro- M•ssbauer Experiments for Nonlinear Vibration in the
phonic Correlates of Cubic Difference Tones," in Facts and Cochlea," J. Acoust. Soc. Am. 55, 588-597.
Models in Hearing, E. Zwicker and E. Terhardt, Eds. Rose, J. E., Kitzes, L. M., Gibson, M. M., and Hind, J. E.
(Springer-Verlag, New-York). (1974). "Observations on Phase-Sensitive Neurons of Antero-
Goblick, T. J., and Pfeiffer, R. R. (1969). "Time-Domain ventral Cochlear Nucleus of the Cat: Nonlinearity of Coch-
Measurements of Cochlear Nonlinearities Using Combination lear Output," J. Neurophysiol. 37, 218-253.
Click Stimuli," J. Acoust. Soc. Am. 46, 924-938. Sachs, M. B. (1969). "Stimulus-Response Relation for Audi-
Goldstein, J. L., and Kiang, N. Y-S. (1968). "Neural Cor- tory-Nerve Fibers: Two-Tone stimuli," J. Acoust. Soc.
relates of Aural CombinationTone 2f1-f2," Proc. IEEE 56, Am. 45, 1025-1036.
981-992. Sachs, M. B., and Abbas, P. J. (1974). "Rate Versus level
Kiang, N. Y-S., Watanabe, T., Thomas, E. C., and Clark, Functions for Auditory-Nerve Fibers in Cats: Tone Burst
L. F. (1965). "Discharge Patterns of Single Fibers in the Stimuli,',' J. Acoust. Soc. Am. 56, 1835-1847.
Cat's Auditory Nerve," MIT Monogr. No. 35. Sachs, M. B., and Kiang, N. Y-S. (1968). "Two-Tone Inhibi-
Kim, D. O., Molnar, C. E., and Pfeiffer, R. R. (1973). "A tion in Auditory-Nerve Fibers," J. Acoust. Soc. Am. 43,
System of Nonlinear Differential Equations ModelUng Basilar- 1120-1128.

Downloaded from http://pubs.aip.org/asa/jasa/article-pdf/59/1/112/12144774/112_1_online.pdf

J. Acoust.Soc.Am., Vol. 59, No. 1, January1976