VoL 58, No. 4 Gober 1987. Printd in U.S.A
The PenonalizedAuditory Cortex of the
MustachedBat Adaptationfor Echolocation
NOBUO SUGA, HIDETO NIWA, IKUO TANIGUCHI,
a,No DANIEL MARGOLIASH
Departmentof Biology, WashingtonUniversity,St. Louis, Missouri 63130
ily sharplytuned in frequency,the personalization
ofthe auditory cortex or systemis not
1. In the mustachedbat, Pteronotuspar"resting"
for the detection of wing beats of
only
suited
nellii, the
frequency of the constant-frequency component of the second insects, but also for the reduction of the
harmonic (CFz) of the orientation sound masking effecton echolocationof conspeci(biosonar signal) is different among individ- fic's biosonarsignals.
5. Becausethe orientation sound is sexualswithin a rangefrom 59.69to 63.33kHz.
ually
dimorphic and the auditory cortex is
The standard deviation of CF2 resting frepersonalized,
the tonotopic representation of
quency is 0.091 kHz on the averagefor indicortex is also sexually dimorthe
auditory
vidual bats. The male's CF2 resting frequency (61.250 + 0.534 kHz, n : 58) is phic.
1.040kHz lowerthan the female's62.290 +
0.539 kHz, n : 58) on the average.Females' I N T R O D U C T I O N
resting frequenciesmeasuredin December
are not different from those measured in
The shapeand size of the sensorymaps in
April when almost all of them are pregnant. the cerebral cortex show differences among
Therefore,the orientation sound is sexually individuals as well as common featuresfor a
dimorphic.
species(e.9.,12, 17, 21, 24 for the auditory
2. In the DSCF (Doppler-shiftedCF pro- cortex;7 for the somatosensory
cortex;28 for
cessing)area of the auditory cortex, tono- the visual cortex). Such differenceshave
topic representationdiffers among individ- been consideredas random variation and
ual bats. The higher the CF2 resting fre- rarely have been correlatedwith behavioral
quencyof the bat's own sound,the higherthe differences.Since large amounts of data acfrequenciesrepresentedin the DSCF areaof cumulated by severalneurobiologistsindithat bat. There is a unique match between cate that the functional organizationof senthe tonotopic representationand the CFz sory systemsis altered by postnatal experiresting frequency.This match indicatesthat ence (e.g.,7, 8, 27), variations in the shape
the auditory cortex is "personalized" for and sizeof cortical sensorymapsare likely to
echolocationand that the CFz resting fre- havebehavioralcorrelates.In the mustached
quency is like a signatureofthe orientation bat, Pteronotusparnellii, certain areasof the
sound.
auditory cortex differ among individuals, not
3. If a bat's resting frequency is normal- only in shapeand size(12, 24), blt also in
ized to 6 I .00 kHz, the DSCF areaoverrepre- tonotopic representation(21). The tonotopic
sents60.6-62.3 kHz. The central region of representation in the auditory cortex is
this overrepresented
band is 6l.l-61.2 kHz. unique, and its unique aspectsare directly
This focal band matchesthe "reference" fre- related to the amplitude spectrum of the
quency to which the CF2 frequency of a bat's orientation sound. which is also called a
Doppler-shifted echo is stabilized by biosonar signal (18). Therefore, this species
Doppler-shift compensation.
ofbat offers us an excellent opportunity for
4. Since DSCF neurons are extraordinar- the detailed examination of whether a small
SUMMARY AND CONCLUSIONS
0022-3077/87$1.50Copyright @ 1987The American PhysiologicalSociety
SUGA, NIWA, TANIGUCHI, AND MARGOLIASH
differencein the functional organization of it reflects the small differences in CF frean auditory cortex or systemamong bats is quency among orientation soundsproduced
correlatedwith a small differencein biosonar by different individuals (23, 26). However,
the data supportingthis last statementis limsignalamong them.
"velocity-sensi- ited.
The mustachedbat has a
The aim ofthe presentpaper is to present
tive" echolocationsystem.It emits orientaindicating that the distribution of the
major
two
data
consisting
of
tion sounds each
portions: a long constant frequency (CF) best frequenciesof neurons in the DSCF area
portion and a short frequency-modulated varies according to the CF2resting frequency
(FM) portion. Each portion consists of four of a bat's own orientation sound and that
harmonics, so that there are eight compo- females' resting frequencies are significantly
nents (CF1-+,FMr-r) in each orientation higher than the males'. We conclude that
sound.(The third portion, a short initial FM, tonotopic representation ofthe auditory cor"personalized"for echolocationand is
has been ignored, becausethe frequency tex is
sweepin it is very small.) The CF compo- sexuallydimorphic.
nents are suited to obtaining velocity information. To optimize the analysisof velocity M E T H O D S
information, the bat performs an auditory
The methods of the present researchwere
behaviorcalled Doppler-shift compensation,
nearly
the sameasthosepreviouslydescribed(2 1),
by which the CFz component of Doppler- so that only the main portions of theseare sum"reference"
shiftedechoesis stabilizedat the
marizedbelow.
higher
kHz
than
frequency.This is 0.1-0.2
"resting"
frequency, which is Materials, surgery, and recording
the bat's CFz
the frequency of the CF2 component of the of neural activity
Six mustached bats, P. panellii rubiginosus,
orientation sound emitted by the bat not in
Panama were used for electrophysiological
from
(2,
15).
flight
3,
was sexed. In
In the orientation soundsemitted by mus- experiments. Only one of these injected with
preparation for surgery,eachbat was
tached bats from Jamaica and Panama, the
a l:50 mixture of Fentanyl-droperidol(InnovarCF2 component is always predominant. Its Vet 4.08 mglkg body wt). The dorsal part of the
frequency diffen among individuals within a skull was exposed,and the flat head of a 1.5-cmrange between 59 and 64 kHz, averaging long nail was attachedwith glue and cement.The
-61 k}Jz (24). ThLeauditory system of the bat waskept separatelyfor 2-3 daysin the animal
mustachedbat is highly specializedfor fine room to allow for recoveryfrom surgery.
At the beginning of eachdayJong experiment,
frequencyanalysisof the CF2 component of
biosonar signals: /) the auditory periphery the resting frequency of the CFz component of
has an array of neurons extremely sharply orientation sounds emitted by each animal was
voltage signal with a frequency-totuned to 60-62 kIIz (19, 25); 2) the central expressedby a
The voltage signals for -20
voltage
converter.
auditory systemproduces6A-62 kHz tuned orientation soundswere storedon the screenof a
"level-tolerant" freneurons, which show
storage oscilloscope, and the center voltage of
quency tuning becauseof lateral inhibition these,i.e., approximate averageresting frequency,
(10, ll, 21,25,26\; and -l) among several was measured. During an experiment, an unanareas of the auditory cortex, the DSCF esthetizedbat was put in a Lucite restraint, which
(Doppler-shifted CF processing)and CF/CF was suspendedwith an elastic band in the center
(CF/CF combination sensitive) areas are of an echo-attenuatedsoundproof room. To imhrghly specialized for processing the infor- mobilize the bat's head,the nail was locked into a
given
mation carried by the CF2 component of a metal rod with setscrews.The animal was
mealworms during
crushed
and
sometimes
water
Doppler-shiftedecho. The DSCF area overthe experiments, and its surgical wound was
representssounds between 6l and 62 kHz treatedwith local anesthetic(Xylocaine)and anti(18, 2l). The CF/CF area extractsvelocity biotic nitrofurazone ointment (Furacin)' The
information by comparing the CF compo- temperatureof the soundproof room was mainnents of the emitted orientation sound with tained at 30-32"C. After the experimentsthe anithose of Doppler-shifted echo (20, 22, 24). mal was returned to the animal room. The same
The tonotopic representationin the DSCF bat could be used for up to 2 wk (4 8-h sessions)
and CF/CF areasis sufficiently detailedthat for studiesofthe responsepropertiesofneurons in
645
PERSONALIZED AUDITORY CORTEX
the DSCF area. Throughout this period, special
care was given to the animal by application of
antibiotic nitrofurazone ointment to the surgical
wound as well as maintenanceon antibiotic (tetracycline)-lacedwater.
For electrode penetration into the DSCF area,
holes of -50-pm diam were made in the skull
with a sharpened needle. During this procedure,
animals showedno sign of distress,yet responded
vigorously to accidental touching ofthe face.This
indicated their awarenessand ability to express
pain if so needed. A vinyl-coated tungsten wire
electrode with a tip diameter of 5-7 pm was insertedobliquely or orthogonally through the holes
into the DSCF areafor recording action potentials
from singleneuronsor clusterr ofa few neurons.
An indifferent tungsten wire electrode was placed
on the dura mater near the recording electrode.
Action potentials were fed into a level detector to
isolate action potentials of one or two neurons.
The instruments usedfor recording were the same
as previouslydescribed(21).
Acoustic stimulation
A CF tone wasdelivered at a rate of 4.5/s from a
condenserloudspeakerplacd72.5 cm in front of
the bat. The duration and risedecay time of the
CF tone were 30 and 0.5 ms, respectively. The
frequency and amplitude of the tone bunt were
manually varied to measure best frequencies
(BFs) for excitation of neurons. The instruments
used for delivering acoustic stimuli were the same
as previouslydescribed(21).
Data acquisition and processing
Action potentials of a single neuron or a cluster
of a few neurons were amplified and displayed on
the oscilloscopescreen.Action potentials exceed-
ing the threshold of a level detector triggered uniform electric pulses that were fed into an audiomonitor and also to the beam intensification of
the oscilloscopefor monitoring the action potentials exceedingthe trigger level. The BF ofan isolated neuron was measured by observing its responseon the oscilloscopescreenand by listening
to the responseon the audiomonitor.
M easurement of resting frequency
To study the distribution of the CF2 resting frequency within a bat population, the resting freqr+enciesof I l6 Jamaican mustachedbats, P. parnellii parnellii, were measured by a computer
(PDP I l/23+). Each bat was isolated from the
colony and was placed in a small cage,in which it
produced trains of orientation sounds spontaneously. The signal from a microphone placed
l5-30 cm from the bat was amplifred and filtered
(50- to 65-kHz band pass,24 dB/octave slope).
The voltage reference level of the filtered signal
was set by inspection with an oscilloscopeto detect only the vocalized signal. Every l0 positivegoing crossingsproduced by a comparator gateda
TTL signal frrst high then low. The times of transition of the TTL pulse were measured with a
l-MHz real-time clock (Data Translation 2769).
By measuringgroups of l0 rather than individual
transitions, the internrpt rate was reduced from
-60 to 6/ms, and the precision of measurement
was increased.Each orientation sound consistsof
a short initial FM component, a long CF component, and a short FM component. The resting frequency of each orientation sound was estimated
from the average of the TTL pulse transition
times occurring during a l0-ms window, which
openedwith a l0-ms delay fron the onset of the
sound to exclude the initial FM component from
'e "e:+
62.O
N
r
J
.E
o
c
o
=
E
o
ll.
61.0
60.o
.-
ill
80.59'
f-l
B
ffE'l't
-1
l
82'14
I l
er'eel
_1
r'!1111
,.r.ee i ,
i ,
, ' :! 1 ' ' \
i I I I
H50
20 msec
Ir l
msec
nc. 1. Orientation sounds(biosonarsignals)emitted by mustachedbats at rest.,4: soundsemitted by 4 bats
(a-d). B: 3 trains ofsounds (a-c) emitted by one individual bat. The upper andlower traces,respectively,represent
the oscillogramsofthe soundsand the frequenciesofthe secondharmonicsofthese sounds,i.e., the outputs ofthe
frequency-to-voltage
converter.The frequencyofthe CF2componentofeach soundis representedby the flat portion
of the lower trace.
SUGA. NIWA. TANIGUCHI.
646
the measurement.The CF component of the orientation sound typically lasts for 15 ms, but to
guaranteethat only the CF component was measured, trials in which any single point dropped
below 55 kHz (i,e,,trials with a contribution from
the terminal FM component of the orientation
sound) were automatically rejected.The final
resting frequencyassignedto the animal was the
averageof 200 soundsthat were usually emitted
by a bat within 5 min. During the courseof handling the animals, the sex was ascertainedby inspectingthe genitalia.
AND MARGOLIASH
RESULTS
CF2restingfrequencY
Orientation soundsemitted by mustached
bats eachconsistof a short upward sweeping
initial FM component, a long CF component, and a short downward sweepingterminal FM component. Figure 1,4showsexamples of orientation sounds emitted by four
different bats. The CF2 resting frequencies
and
61.69,61.70,
were60.59,61.28,61.60,
A
62
6't
C F t r e s t i n gf r e q u e n c yi n k H z
N
Ioz
;
o
Eor
c'
o
.g 61
at,
o
ot
II
o
20
Days
r.rc. 2. Differencesin CF2restingfrequencyamongmustachedbats.l: distribution of CFzrestingfrequenciesof
orientation soundsemitted by 58 males(fitled circtes)and 58 females(opencircles).The mean + SD of CF2resting
3
frequencywere61,250-r534Hzfor 58malesand62,290+ 588Hzfor53fernales.B: CF2restingfrequenciesof
females(a-c) and 2 males (d and e) measuredover 32-38 days. Each data point indicates an averageof CF2 resting
frequenciesof 200 orientation soundsemitted by a bat within 5 min. The means t SD of CFz resting frequenciesof
thesebats are listed in Table 1.
647
PERSONALIZED AUDITORY CORTEX
62.14 kHz for the six Panamanianmustachedbats,which were usedfor the present
electrophysiologicalexperiments.The standard deviation in CF2 resting frequency
rangedbetween0.08 and 0.l4kHz, i.e.,0.13
and 0.23Voof the CF2 resting frequency,
when 200 soundswere sampled(4). This
within-animal variation in restingfrequency
wasso small that the differencesamong these
six mean restingfrequencieswere all significant exceptfor the differencebetween61.69
and 61.70kHz.
The bats, when excited, often emitted
trains of orientation sounds.The CF2resting
frequency of the first sound of each train
tended to be slightly lower than that of the
rest (Fig. l,B). The standarddeviation of CF2
restingfrequencywould be 50-100 Hz, if the
first sound of each train of orientation
soundswasexcludedfrom the measurement.
To examinethe distribution of CF2resting
frequencieswithin the population, the CF2
resting frequency was measuredin I 16 Jamaican mustached bats (Fig. 2). In these
A
B : P 7 - 1 5 - 8 1L 9
RF=62.14kHz
C:P4-23-81L8?
R F = 6 0 . 5 9k H z
UA
A: DSCF
b: FM-FM
c: cF/cF
C
61.0
6?'4kH'
60.6
r
l
\- -l-
\
ji;\Wit'dj,:i,
62.2
r62.4
/no
\r
251
)..^
2.51
---358,
ry7!----as.o
I
I
\
,'xrf"
^,,
)r1
*qt*ii\\
-o"B\i
':.'.\:1iil'llTl;
o.e,
\;
r.'-:--'Ki;v1wI
*n.ooo-- \QqI
0.92
2143.-- _-?
2.44
1.92
,
0.5 mm
Add 6O.00 kHz
,
O.Smm
- -''-."-"i"
1.20
1 . 11
1.27
144
Add 60.00 kHz
nc. 3. Distributionsof bestfrequencies(BFs)in and around the Doppler-shiftedCF processing(DSCF) areaof 2
mustachedbats: P7-15-81 female (B) and P4-23-81,presumably amale (C). Ar dorsolateralview of the left
cerebrum.The primary auditory corlex(dashedlina) containsthe DSCF (a), FM-FM area(b), and CF/CF area(c).
dataobtainedfromneuronslocated
Thebranchesofthemediancerebralarteryareshownbythebranchinglines.B:
at 400- to 500-pm depth by 25 orthogonal electrodepenetrations.C: data obtained from neurons located at
different depthsby 9 oblique electrodepenetrations.In .B and C, the lowest BF is circled.Each BF is expressedby
subtracting60.00 kHz from a measuredvalue, becauseof a spacelimitation in C. Thus, for example, 1.20 means
61.20.The decimal point indicatesthe recordingsite of neural activity. lso-BF contour lines labeled,e.9.,62.0 kHz
are obtained by estimating the relative location between higher and lower measuredBFs (solld lines). Certain
portions ofthe contour linesaredrawn to enclosethe regioncontaininglower measuredBFs.The largedashedcircle
representsthe estimatedouter edgeof the DSCF area.The CF2restingfrequency(RF) was 62,14kHzfor P7-15-81
(.8)and 60.59 kHz tor P4-23-81(C\.
648
SUGA. NIWA. TANIGUCHI. AND MARGOLIASH
bats, the CFz resting frequenciesranged between59.692and63.334kHz. The distributions of the male's and female'sresting frequenciesoverlappedeachotherby 33%o.The
mean and standarddeviation of CF2 resting
frequencywere 61.250+ 0.534 kHz for the
males(n : 58) and 62.290+ 0.588 kHz for
the females(n : 58). The 1.040kHz differencein resting frequency was significant, P <
0.001 (x2 test).The high restingfrequencyof
the females was not due to pregnancy, because there was no significant difference in
resting frequency between pregnant bats collectedand testedin early April and nonpregnant bats collectedand testedin the middle
of December(Table 1,4).Therefore,the orientation sound was sexuallydimorphic.
In two males and three females,the CF2
resting frequency was measuredevery late
afternoonover 3 1-38 daysbetweenlate February and early April (Fig. 2.8). The average
resting frequenciesand standard deviations
of thesebatsare listed in Table lB. The standard deviations ranged between 67 and lI7
H4 averaging9l Hz. The resting frequency
tended to drift lower or higher over 10-20
days. An unusually large.drift was observed
TABLEl. Restingfrequenciesof CF2
componentsof orientation soundsof
Pteronotusparnellii parnellii from Jamaica
A.
Sexes
Males
Males
Females
Females
B.
Bat ID
No.
Fig.2Ba
Fis.2Bb
Fig. 2Bc
Fis.2Bd
Fig.2Be
Resting
Frequencies,
Hz
61,388+ 543
6l,lll + 525
62.278 + 646
62,302 + 431
(Mean SD:536)
Resting
Frequencies,
Hz
62,067 + lO9
61,556+ 83
'77
61,418+
+
60,885 67
60,691+ ll'7
(Mean SD:91)
No. of
Bats
Month
Measured
29
29
30
28
April
December
April
December
No. of
Sounds
No. of
Days
200x37
200x 38
200x37
2 0 0x 3 1
200x37
37
38
37
31
3"1
Values are meanst SD. CFr, constant frequencyof
the secondharmonic.
in one male within the initial I I days of the
measuremen
ts (Fig. 2B e).
CFz restingfrequencies and BFs
of neurons in the DSCF area
The sizeand shapeof the DSCF:rea were
slightly different among individual bats.
Tonotopic representation in this area was
higly systematic in some bats, but less so in
others.Iso-BF contour lines were commonly
eccentric,aspreviouslydescribed(18, 21). In
addition to thesevariations, the range of frequencies representedin the DSCF area was
also different among different bats. For instance, BFs were predominantly between
60.5 and 61.5 kHz in bat P4-23-81(Fig. 3C)
and between 62.0 and 63.0 kHz in bat
P7-15-81(Fig. 3,8).P7-15-81was a female
and P4-23-81 was presumably a male becauseof its very low CFz resting frequency.
The CF2restingfrequenciesof P4-23-81and
P7-15-81were60.59 and 62.14kHz, respectively. The difference in resting frequency is
1.55 kHz, which is sirnilar to the difference
in frequency representation between the
DSCF areasof thesetwo animals.
Across individual bats, a consistent relationship between CF2 resting frequency and
BFs of DSCF nourons was observed.This is
clearly shown by BF histograms (Fig. a).
Since the standard deviation in CF2 resting
frequencywas about +0. I kHz on the average(4;0.091 kHz in Table 1,8),the histograms were plotted with a 0.20-kHz classwidth. The center frequency of one of the
classesfor each bat was at the CF2 resting
frequencyof that bat. [n Fig. 4A, the modes
of histograms a, b, and e are 0.2-0.4 k}Iz
higher than the resting frequency. This was
also true in the data obtained from bat
PI 2-4-78,which are not shown in Fig. 4A.ln
histograms c and d, the mode is at the same
frequency asthe bat's own resting frequency.
In addition to the difference in frequency between the resting frequency and the mode of
BF histograms, the shape of BF histogram
was also different among individual bats.
Among the data obtained from different
bats. there were some variations in relationship betweena resting frequency and a histogram. However, there were two common
features shared by all the data: /) the higher
the resting frequency, the higher were the
PERSONALIZED AUDITORY
c
20
o
c
b or.og
61.28 I
o
B
l s
c
649
CORTEX
U
A
P4-23-41
P 6 - 11 - 8 1
P6-15-81
P7-15-81
P12-1-80
N
40
55
44
20
23
o
b 10
ll
E
3
z
.5
6 bats
o
c
N: 269
940
=
o)
c
o
o
2E 2 0
f
z
61
62
63
Best freq. of DSCF neuronsin kHz
dai
Frc.4. Distributions of best frequencies (BFs) of DSCF neurons recorded from mustached bats' l:
obtained from 5 different bats (a-e) were separately plotted. The numbers and arrows indicate the CF2 resting
frequenciesof thesebats and their BF histograms.8: a composite BF histogram. The BF histograms for 6 different
bats were normalized by shifting thesealong the frequency axis by a difference between6 1.0OkHz and the bat's own
resting frequency. Then thesewere added to produce the composite BF histogram'
BFs of DSCF neurons; and 2) the population
of neurons with BFs higher than the resting
frequency was much larger than that of
neurons with BFs lower than the resting frequency. These data indicate that the functional organization of the DSCF area
matches the properties of the bat's own bio-
sonar signal to pro@ssthe information carried by the CF2 component of positively
Doppler-shifted echoes.
To obtain a composite BF histogram for
six bats, individual BF histograms were normalized to 61.00 kHz, which is closeto the
mean resting frequency of the population of
650
SUGA. NIWA. TANIGUCHI.
AND MARGOLIASH
modes of the distributions were 61.7 kHz,
and both the DSCF areas predominantly
contained neuronstuned to soundsbetween
61.5and 62.0kHz
N
s>
c
right DSCF 54
9
= z o
- l e ft D S C F 4 4
c)
c
o
15
o
E
2 1 0
61
62
Best freq. of DSCF neuronsin kHz
t tc. 5. Distributions of best frequenciesin the left
(rtiled circles) and right (open circles) DSCF areasof a
mustachedbat. The CF2resting frequency ofthis animal
was61.69kHz.
bats. That is, each histogram was shifted
along the frequency axis by the difference
between the CFz resting frequency of each
bat and 61.00 kHz. Then the histograms
were addedtogether(Fig. aB). The composite BF histogramis peakedat a 6l.l- to 61.3kHz band. It showsthat the DSCF arearepresentssoundsfrom 60.6 to 62.3 kHz, applying a criterion of more than four neuronsper
0.2-kHz band and that the distribution of
DSCF neurons tuned to different frequency
bandsis highly skewed:50,23, and 87oof all
neurons are tuned to the first, second,and
third 0.5-kHz bands above the resting frequency,respectively,and only 18 and l%oare
tuned to the first and second0.5-kHz bands
below the restingfrequency.The DSCF area
representsthe soundbetweenthe CF2resting
frequency and 0.5 kHz above it with a disproportionately large region.
In one animal, suftcient data were obtained for a comparison in BF betweenthe
DSCF areasin the left and right hemispheres.
The CF2 resting frequency of this bat was
61.69 kHz on the average.The distributions
of BFs measuredin the left and right DSCF
areas were nearly the same (Fig. 5). Both
DISCUSSION
The personalizedauditory system
During flight, the mustachedbat performs
Doppler-shift compensation:the bat reduces
the frequency of orientation sounds to stabilize a positively Doppler-shifted echo at a
preferred frequency (reference frequency),
which is 0.1-0.2 kHz higher than the bat's
own CF2restingfrequency(2, 3, l5). Because
of this compensation,a bat with a CF2 resting frequencyat 61.0 kHz, for example,predominantly receivesDoppler-shiftedechoes
with a CF2componentat 6 1 1-6 | .2 kHz and
a CF3componentat 91.7-91.8kHz.
The mustachedbat has a "velocity-sensitive" echolocation system.Its auditory system is specializedfor extracting velocity information from both the CF2 and CF3 components of Doppler-shifted echoes(22, 24,
26). The peripheral auditory systemcontains
two groups of neurons extraordinarily
sharplytuned to either 60-62 kHz or 90-93
kHz (19, 25). The primary auditory cortex,
which is tonotopically organized,overrepresentsthesetwo frequencybands(1, 18, 2l).
In particular, the over-representationof
6l-62 kHz by the DSCF areais phenomenal
(18,2l).
The histogram displaying the distribution
of BFs in the DSCF area, normalized to a
61.0-kHz resting frequency,indicatesthat
the DSCF area overrepresentsa sound between 60.6 and 62.3 kHz. The histogram is
peakedat a 6 l. l- to 61.2-kIlz band (Fig. 4^B).
This "focal" band matchesthe band for the
referencefrequency.Eighteen, 50, and 23Vo
of DSCF neuronsrepresent60.5- to 61.0-,
61.0-to 61.5-,and 61.5-to 62.0-kHzbands,
respectively. Therefore, 9l%o of DSCF
neuronsrepresentsoundsat around the bat's
own referencefrequency and resting frequency.The DSCF areais undoubtedly specialized to representbiosonar information
carried by the CF2 component of "stabilized" Doppler-shifted echoes.[The echoes
from flying insectsalso fall near the reference
frequency, but not at the referencefrequency
becauseof their own flight speed(16).] Since
PERSONALIZED AUDITORY CORTEX
the focal band variestogetherwith CF2 resting frequency,the functional organizationof
the DSCF area matched the bat's own biosonar signal.
In the auditory cortex of the mustached
bat, there is the CF/CF area,where neurons
are tuned to combinations of the CFr of the
orientation sound and the CF2 or CF3 of a
Doppler-shifted echo to extract target velocity information (22,24,26). The functional
organization of this area matchesthe CF2
and CF3frequenciesof the bat's own orientation sound (26). Located anterior to the
DSCF area, the anterior primary auditory
cortex overrepresents92-94 kHz ( 1). The
tonotopic representationof this area is expectedto match the bat's own CF3reference
frequency.
The match between the properties of a
bat's own orientation sound and the functional organization of its auditory cortex indicates that each bat has a private line for
echolocation, which is "communication
with the environment." In other words, both
the orientation sound and the auditory cortex of the mustachedbats are personalized.
This match also indicatesthat the frequency
of the CF components of the orientation
sound is like a signature.It has not yet been
examinedwhetherthis personalizationof the
auditory system(settingof the bat's tuner) is
becauseof a genetic code and/or auditory
experience.
The personalizationof the auditory cortex
apparently originates from the periphery.
The best frequencyof cochlearmicrophonic
response(CM) is sharply tuned at -61k}Iz
and slightly differs from bat to bat (14). In
two bats studied, it has been demonstrated
that the CM best frequencywas found to be
0.2 kHz higher than the CF2 resting frequency(2, 3). Neuronssharplytuned at -61
kHz are disproportionatelylarge in population at the periphery(19,25), in the inferior
colliculus (13), and in the auditory cortex
( I 8, 2 I ). In two batsstudied,it hasbeendemonstratedthat the population of inferior collicular neuronsis disproportionatelylargeat
the CM best frequency(13). Theseobservations indicate that the peripheral and central
auditory systemsare specializedfor the detection and fine frequencyanalysisofsounds
at -61 kHz, i.e., the CFz frequency of biosonar signals.However, none of thesepapers
651
showsthe data demonstratingthat the CM
best frequencyor focal band or tonotopic
representationat -61 kHzvary significantly
from bat to bat, according to the bat's own
CF2 resting frequency.No data demonstrating the personalizedauditory system has
previouslybeen published,although this has
beenstatedin the past (2, 16,23). The papers
cited in this paragraph,however,are very
consistentwith the view that the auditory periphery and subcortical auditory nuclei are
personalized.
The horseshoebat, Rhinolophusferrumequinum, emits an orientation sound consisting of a long CF component and a short FM
component. As reviewed by Schnitzler and
Ostald (16), its auditory systemis apparently
personalized according to the CF2 resting
frequency in the bat's own orientation
sound. Flowever, it is not yet demonstrated
that the CM best frequency or focal band or
tonotopic representation varies significarrtly
from bat to ba! according to the bat's own
CF2 resting frequency.
Variation in CF2 restingfrequency
and the range oftonotopic representation
in the DSCF area
The CFz resting frequency is quite constant in some animals, but it varies significantly in others, as shown in Fig. 28 and
Table l,B. Over the 40 dayswhen CF2resting
frequencieswere measured,the bat's health
appearedto be good,and they were alert and
emitted a lot of orientation sounds when
they were placedin a small cage.The health
and alertnesswere unlikely sourcesto the
observeddrifts/variations in CF2 resting frequency. The room temperature was nearly
the sameover those days. Further, sincethe
directions ofthe drifts in frequencydiffered
among the bats studied over the samedays,
the room temperaturecould not be the cause
for the drifts.
We don't know the causeof the observed
drifts in CF2 resting frequency. However,
these present intriguing problems: whether
there is a positive correlation between the
range of tonotopic representation in the
DSCF area and the magnitude of drift/variation in CF2 resting frequency over a prolonged time period, and whether tonotopic
representation changestogether with a drift
in CF2 resting frequency. One may speculate
652
SUGA, NIWA, TANIGUCHI, AND MARC'OLIASH
that the difference in frequency tuning ofthe
DSCF area among bats (e.g.,a differencein
shapebetween BF histograms a and c of Fig.
4A) is significant and that the amount of
drift/variation in CF2 resting frequency is
positively correlated with the frequency tuning of the DSCF area.
The personalized echolocationsystem
and reductionof masking
In radio communication, a specificcarrier
frequency within a particular band is assignedto eachstation,and a tuner is adjusted
to receive signalsfrom a particular station.
The assignmentof a different carrier frequencyto eachstation and an array ofsharp
filters in eachtuner are essentialfor communication and reduction of masking. In contrast, in the mustachedbat, the tuner is fixed
and the emitter is adjusted so that the returning signal (Doppler-shifted echo) is
maintained at the referencefrequency to
which the tuner is set. The central auditory
system contains groups of neurons with
sharp level-tolerantfrequency-tuningcurves
at and around the bat's own restingand ref(10, 11, 13,21,24,25,
erencefrequencies
26). Therefore, the personalization of the
biosonar signal and auditory system is not
only suited for the detection of wing beatsof
insects,but also for the reduction of the
masking effect on echolocation of sounds
produced by conspecifics,as discussedpreviously (20,22) and below in detail.
The CF2restingfrequenciesofthe I l6 bats
studied ranged from 59.69 to 63.34 kHz.
That is, they spreadover 3.64 kHz. The
standarddeviation in CFz resting frequency
is 0.091 kHz on the averagefor individual
bats. This implies that each bat has not a
particular single frequency,but a particular
narrow frequency band that overlaps with
other's. How many different bats can theoretically be discriminated in terms of CF2
restingfrequency?A l.16 SD on either side
ofthe meancorrespondsto a75Voconfidence
level. Assumingthat the distribution of bat's
CF2 resting frequenciesis uniform over the
3.64-kHzband,3.64kHz is dividedby 0.21I
kHz. The result. 17.3.is the number of bats
to be discriminated'l57oof tlaetime usingthe
CF2 resting frequency as the only cue. If
there are additional acoustic cuessuch asthe
harmonics and envelope of orientation
sounds,a much greaternumber of batscould
be discriminated. However, what is primarily
important to each bat is not the discrimination of these 17 bats but to discriminate its
own biosonar signalsfrom the sounds produced by conspecifics.The distribution of
bat's CF2 resting frequency is not uniform
overthe 3.64-kHzband (Fig.2A).Therefore,
a bat emitting orientation sounds at either
the loweror higherend of the 3.64-kHzband
could easily discriminateits own biosonar
signalsfrom thoseproducedby conspecifics.
On the other hand, a bat emitting orientation
soundsat the centerofthe band would facea
harder discrimination. The personalization
of the echolocationsystemwould reducethe
masking effect of conspecific'ssounds on
echolocation,but it alone is apparently not
sufficient to make the echolocation system
tolerant to masking, and other mechanisms
should be consideredtogether with this, as
previouslydescribed(24, 22).
The personalized echolocationsystem
and prey detection
Johnsonet al. (6) first pointed out the following: the "flight-speed" Doppler-shift
compensation adjusts the carrier frequency
ofan insectechoto the centerofthe particular bat's sharp sensitivity peak. Then the periodic frequency modulation (FM) in the
echo due to the wing beats of the insect
occurs as the frequency sweepsback and
forth through the sensitivitypeak.Becauseof
the sharp tuning of the cochlea,the FM is
converted into large-amplitudemodulation
to stimulate periodically auditory neurons.
Johnson et al. clearly pointed out that the
auditory systemwith sharptuning at the CF2
frequencyis a specializationfor the detection
of wing beatsof insects.This theory hasbeen
well accepted.The mustachedbat is extremely sensitiveto the wing beat (5). Its peripheral neurons with a BF at -61 kHz are
extraordinarily sharply tuned and are sensitive to very small periodic FMs (19).
Neurons in the DSCF and CF/CF areasof
the auditory cortex are also sharply tuned at
-61 kHz, and many of them are also sensitive to small periodic FMs (22).
The horseshoebat, R. ferrumequinum,
emits orientation sounds.eachofwhich consists of a long CF component followed by a
short FM component. As reviewed by
PERSONALIZED AUDITORY
Schnitzler and Ostwald (16), behavior and
coding related to wing-beat detection have
been more extensively studied in this species
than in the mustachedbat.
Sexual discrimination with CF zfrequency
Becausethe CF2restingfrequencyis significantly different betweenmalesand females,
there is a possibility that CF2 resting frequency servesas an acousticcue to identify
sex. The distribution of male's resting frequencyoverlapswith that of female'sby 33Vo
(Fie. 2A). If mustachedbats usethe CF2resting frequencyfor sexualdiscrimination, the
percentcorrectlevel of discrimination would
be 6'7Vo.
Sexual dimorphism of the auditory system
In prior studies,the CF2 restingfrequency
was 60.87 + 0.48 kHz for 77 P. p. rubiginoszs from Panama (24) and 61.21 + 0.68
kHz for 34 P. p. parnellii from Jamaica(26).
The exact male-to-female ratio is not known
in these samples but it was probably about
4: l, becausethey were imported at this ratio.
There is no significant differencein resting
The
frequencybetweenthesetwo subspecies.
present study of CF2 resting frequency revealed sexual dimorphism of orientation
sound in the Jamaican mustachedbats. Of
the six Panamanianmustachedbats usedfor
the electrophysiologicalexperiments,we had
sexedonly one(P7-15-81),which wasfemale
and emitted orientation soundsat the highest
restingfrequencyamong the six. There is no
reason to doubt that orientation sounds of
Panamanianmustachedbats are also sexually dimorphic.Sincethe orientationsound
is sexuallydimorphic and the auditory cortex
is personalized,the auditory system of the
mustachedbat is sexuallydimorphic. The
mustachedbat is the first example of sexual
dimorphism of tonotopic representationin a
certain range of frequencies,in spite of the
fact that both malesand femaleshear sounds
CORTEX
653
in a wide rangeof frequenciesfrom <10 kHz
to > 100kHz. As in many different speciesof
mammals. the male mustachedbats tend to
be slightly larger than the femalesand their
soundsare slightly lower in frequency than
thoseof females.
Soundsprodued by males and femalesin
many speciesare commonly different in frequency and/or structure. But a differencein
the auditory systembetween males and females has been found only in the tree frog,
Elentherodactyluscoqui. The communication sound of the male tree frog consistsof
"Ko" and "Ki," which are respectwo notes:
tively emitted for territorial maintenance
and femaleattraction. The male's ear is sensitive to the low-frequencynote Ko, but not
to the high-frequencynote Ki. In contrast,
the female'sear is sensitiveto Ki, but not to
Ko (9). This is the best example of sexual
dimorphism of the auditory system. It
should also be noted that in the zebra finch,
the auditory responseof the vocal system
is quite different between males and females(29).
ACKNOWLEDGMENTS
We thank S. J. Gaioni for his helpful comments on
this article.
This work on mustached bats has been supported by
JavitsNeuroscienceInvestigatorAward RO l-NS I 7333.
D. Margoliashwas supportedby a training grant from
the National Institutes of Health to the Department of
Anatomy and Neurobiology, Washington University
Medical School.
H. Niwa, Dept. of Otolaryngology,
Presentaddresses:
Medical School,Nagoya l-hriversity,65 Tsurumai-cho,
Showa-Ku, Nagoya 466, Japan;I. Taniguchi, Medical
ResearchInstitute, Tokyo Medical and Dental llniversity, l-5-45 Yushima, Bunkyo'Ku, Tokyo 113,Japan;
D. Margoliash, Dept. of Anatomy, Ihiversity of Chicago,1025East 57th St., Chicago,lL 60637.
Received I December 1986; accepted in final form I 5
May 1987.
REFERENCES
l. ASANUMA,A., woNG, D., AND Suce, N. Frequency and amplitude representations in anterior
primary audilory cortex of the mustached bat. J.
Neurophysiol.50: I 182- l 196, 1983.
2. HENSoN,O. w., JR., HENsoN, M. M., KoBLER,
J. 8., AND PoLLACK,G. D. The constant fr€quency
component of the biosonar signals of the bat, Pteronotus parnellii parnellii. ln: Animal Sonar Sys-
tems, dited by R. G. Busnel and J. F. Fish. New
York: Plenum, 1980,p. 913-916.
3. HENSoN,O. W., Jn., Scuurren, G., AND VATER,
M. A comparative study of the physiological properties of the inner ear in Doppler shift compensating
lorrts(Rhinolophw rouxi and Pteronotusparnellii)- J.
Comp. Physiol. 157:587-597, 1985.
4. HoRrKAwA, J. AND Suce, N. Biosonar signalsand
654
SUGA, NIWA, TANIGUCHI, AND MARGOLIASH
cerebellarauditory neurons ofthe mustachedbat. ,/.
Neurophysiol. 55: 1247-1267, 1986.
5. GoLDMAN, L. J. aNo HENSoN,O. W., Jn. Prey
recognition and selection by the constant frequency
bat, Prcronotus P. parnellii. Behav. Ecol. Sociobiol.
2 : 4 1 1 4 1 9 ,1 9 7 7 .
6. JoHr.rsoN.R. A., HENSoN.O. W. JR.. AND GoLD
MAN,L. R. Detectionof insectwing beatsby the bat,
Pteronotusparnellii. J. Acoust. Soc. Am. 55: S53,
1974.
7. MERZENTCH,
M. M., JENKTNS,
W. M.. AND
MTDDLEBRooKS,
J. C. Observationsand hypotheses
on specialorganizationalfeaturesofthe central auditory nervoussystem.In: Dynamic Aspectsof Neocortical Function, edited by G. M. Edelman, W. E.
Gall, and W. M. Cowan.New York: Wiley, 1984,p.
397-424.
8. MovsHoN, J. N..lNo VAN SLUyTERS,
R. C. Visual
neural development.Annu. Rev. Psychol.32:
477-522,1981.
9. NARTNS,
P. M. AND CApRANTcA,
R. R. Sexualdifferencesin the auditory systemof the tree frog
Eleutherodactyluscoqui. Science Wash. DC 192:
378-380,r976.
10. OrsEN,J. F. Functional organizafionof the medial
geniculate body of the mustachedDar(Ph.D. thesis).
St. Louis, MO: WashingtonI-lniv., 1986.
I l. O'NEILL, W. E. Respons€s
to pure tones and linear
FM componentsofthe CF-FM biosonar signal by
signal units in the inferior colliculus of the mustachedbat. J. Comp. Physiol. 157:797-8 I 5, I 985.
12. O'Nprr-r-,W. E. eNn SucA, N. Encoding of targetrange information and its representation in the auditory cortex ofthe mustachedbat. J. Neurosci. 47;
22s-255. 1982.
13. PoLr-AK,G. ,c.NnBooeNneuEn, R. D. Specialized
characteristics
ofsingle units in inferior colliculusof
mustachebat: frequency representation,tuning, and
discharge patterns. J. Neurophysiol. 46:. 605-620,
1981.
14. PoLLAK,G., HENsoN,O. W., JR.,eNn Novrcr<, A.
Cohlear microphonic audiogramsin the "pure
tone" bat, Chilonycteris parnellii parnellii. Science
Wash.DC 176:66-68,1972.
15. ScnNnzr-en, H.-U. Echoortungbei der Fledermaus
Chilonycterisrubiginosa.Z. Vgl.Physiol.68: 25-38,
t9'70.
16. SCHNITZLER,
H.-U., ANDOSTwALD,J. Adaptations
lor the detection of fluttering insectsby echolocation in horseshoebat. ln: Advancesin Vertebrate
Neuroethology,editedby J. P. Ewert, R. R. Capranica, and D. J. Ingle. New York: Plenum, 1983,p.
801-827.
17. Suce, N. Amplitude-spectrum representationin
the Doppler-shifted-CFprocessingareaof the audi-
tory cortex of the mustachedbat. ScienceWash.DC
196: 64-67, 1977.
l8. Suce, N. ANDJEN,P. H.-S. Disproportionatetonotopic representation for processing species-specific
CF-FM sonar signals in the mustache bat auditory
cortex.ScienceWash.DC 194:542-544,1976.
19. Suce. N. exo JeN. P.H.-S. Further studieson the
peripheral auditory system of 'CF-FM" bats specialized for fine frequency analysisof Dopplershified echoes.J. Exp. Biol. 69:207-232, 1917.
20. Suce, N., KuJrRAr,K., AND O'Nru-r-, W. E. How
biosonarinformation is representedin the bat cerebral cortex. ln'. Neuronal Mechanisms of Hearing,
editedby J. Sykaand L. Aitkin. New York: Plenum,
1 9 8 1p, . t 9 7 - 2 t 9 .
2 l. SucA, N. ANDMANABE,T. Neuralbasisof ampli
tude-spectrumrepresentationin the auditory cortex
ofthe mustachedbat. "f.Neurophysiol.47: 225-255,
1982.
22. Suce, N., NIwA, H., ANDTANrcucHr,L Representation of biosonar information in the auditory cortex of the mustachedbat, with emphasison representationof target velocity information. ln; Advancesin VertebrateNeuroethology,edited by P.
Ewert, R. R. Capranica,and D. J. Ingle. New York:
Plenum,1983,p. 829-867.
23. SucA, N. ANDO'NEILL,W. E. Auditory processing
of echoes:representationof acoustic information
from the environment in the bat cerebralcortex.In:
Animal Sonar Systems,dited by R. G. Busnel and
J. F. Fish.New York: Plenum, 1980,p. 589-611.
24. SucA, N., O'NEILL,W. E., KUJIRAI,K., eNo MrN.lae, T. Specialization of "combination-sensitive"
neuronsfor processingofcomplex biosonarsignals
in auditory cortex of the mustached bat. J. Neurop hysiol. 49: | 573-r 626, 1983.
25. Suce, N., Srvrr,roNs,J. A., lNo JeN, P. H.-S. Peripheral specialization for fine analysis of Dopplershifted echoesin "CF-FM" bat Pteronotusparnellii.
J. Exp. Biol.63: 16l-192, 1975.
26. Suce, N. AND Tsuzuru, K. Inhibition and leveltolerant frequencytuning in the auditory cortex of
bat.J. Neurophysiol.53:1109-l 145,
the mustached
I985.
27. TELLER,D. Y. ANDMovsHoN, J. A- Visualdevelopment. VisionRes.26: 1483-1506,1986.
28. VAN EssEN,D. C., NEwsoME,w. T., ANDMAUNsnll, J. H. R. The visual field representationin
striatecortexof the macaquemonkey: asymmetries,
anisotropies,and individual variability. Vision Res.
24:429-448, 1984.
29. Wtlltnvs, H. Sexualdimorphism of auditory activity in the zebra finch song system.Behav. Neural
Biol. 44:470-484, I 985.