Sharpening of Frequency Tuning by Inhibition in the Thalamic
Auditory Nucleus of the Mustached Bat
NOBUO SUGA, YUNFENG ZHANG, AND JUN YAN
Department of Biology, Washington University, St. Louis, Missouri 63130
INTRODUCTION
Because frequency analysis is the most fundamental function
of the auditory system, the physical and physiological bases of
frequency analysis and coding in the auditory periphery have
been well studied. To extend the studies on the periphery, much
research on the central auditory system has been performed
and frequency-tuning curves of many single neurons have been
measured. However, the question has only partially been answered as to where, how, and why the tuning curves change
in shape from peripheral ones. Therefore the processing of
auditory signals in the frequency domain by the central auditory
system remains to be studied further.
In the central auditory system of the cat, Katsuki et al. (1958,
1959a) found that the higher the level in the auditory system,
the sharper the frequency tuning of neurons, and that neurons
in the medial geniculate body (MGB) of the thalamus showed
the sharpest frequency tuning. However, Aitkin and Webster
(1972) reported that ‘‘in general, tuning curves [of MGB neurons of cats] are less sharp than those of cochlear nerve fibers’’
(page 376). Calford et al. (1983) reported that ‘‘no difference
in sharpness of tuning was found between samples of units
from nuclei in the lemniscal auditory pathway [of cats], although units from the anterior auditory field showed broader
tuning than those in the lemniscal pathway’’ (page 395). The
conclusion of Aitkin and Webster (1972) and Calford et al.
(1983), which is different from that of Katsuki et al., is probably due to differences in defining the sharpness of frequencytuning curves of neurons and also in sampling neurons (Suga
1995; Suga and Manabe 1982).
Unlike quasitriangular tuning curves of peripheral neurons, the tuning curves of central auditory neurons show a
large variation in shape. Among these, pencil- or spindleshaped tuning curves (e.g., Fig. 2) have a very narrow bandwidth regardless of sound pressure levels, so that they are
called ‘‘level-tolerant’’ sharp frequency-tuning curves,
level-tolerant tuning curves, or level-tolerant excitatory areas
(Suga 1977; Suga and Manabe 1982). For simplicity, neurons showing level-tolerant tuning are hereafter called leveltolerant neurons. (Here, level-tolerant means that the width
of a frequency-tuning curve stays narrow regardless of sound
pressure levels. This definition is not related to any physiological or psychological interpretation, but only to the shape
of an excitatory frequency-tuning curve. The definition does
not at all specify a rate-level function. If a level-tolerant
neuron shows no excitatory responses to sounds at high
sound pressure levels because of upper threshold, it simply
means that the bandwidth of its tuning curve is infinitely
narrow and that the neuron sends no frequency information
about these sounds to other neurons.)
Level-tolerant neurons have been found in the central auditory systems of many different species of animals, such
as little brown bats (Condon et al. 1994; Grinnell 1963;
Suga 1964a,b, 1965a,b, 1 1971; Suga and Schlegel 1973 1 ),
1
‘‘Inhibitory’’ tuning curves or areas are commonly found on both sides
of a level-tolerant tuning curve.
0022-3077/97 $5.00 Copyright q 1997 The American Physiological Society
2098
/ 9k0f$$ap31
J864-5
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
Suga, N., Y. Zhang, and J. Yan. Sharpening of frequency tuning
by inhibition in the thalamic auditory nucleus of the mustached bat.
J. Neurophysiol. 77: 2098–2114, 1997. Unlike the quasitriangular
frequency-tuning curves of peripheral neurons, pencil- or spindleshaped frequency-tuning curves (excitatory areas) have been found
in the central auditory systems of many species of animals belonging to different classes. Inhibitory tuning curves (areas) are commonly found on both sides of such ‘‘level-tolerant’’ sharp frequency-tuning curves. However, it has not yet been examined
whether sharpening of frequency tuning takes place in the medial
geniculate body (MGB). We injected an inhibitory transmitter
antagonist, bicuculline methiodide (BMI), into the MGB of the
mustached bat to examine whether frequency tuning is sharpened
by inhibition in the MGB and whether this sharpening, if any,
occurs in addition to that performed in prethalamic auditory nuclei.
Thirty-seven percent of thalamic Doppler-shifted constant frequency (DSCF) neurons mostly showing a level-tolerant frequency-tuning curve had an inhibitory area or areas. BMI changed
the inhibitory areas of these neurons into excitatory areas, so that
their excitatory frequency-tuning curves became broader. However, the BMI-broadened excitatory frequency-tuning curves were
still much narrower than those of peripheral neurons. Our results
indicate that level-tolerant frequency tuning of thalamic DSCF
neurons is mostly created by prethalamic auditory nuclei and that
it is further sharpened in 37% of thalamic DSCF neurons by lateral
inhibition occurring in the MGB. The comparisons in sharpness
(quality factors) of frequency-tuning curves between peripheral,
thalamic, and cortical DSCF neurons indicate that the skirt portion
of tuning curves is sharper in the above order, and that their tip
portion is not significantly different between the peripheral and
thalamic DSCF neurons, but significantly sharper in the cortical
DSCF neurons than in the thalamic DSCF neurons. Therefore the
central auditory system has inhibitory mechanisms for the progressive sharpening of frequency tuning. DSCF neurons in the primary
auditory cortex were recently found to show facilitative responses
to paired sounds. That is, they are combination sensitive. In the
present studies, we found that thalamic DSCF neurons also showed
facilitative responses to paired sounds. The responses of thalamic
DSCF neurons to acoustic stimuli consisted of a slow and a fast
component. BMI mainly increased the slow component and an
excitatory transmitter antagonist, D-2-amino-5-phosphonovalerate
mainly suppressed the slow component. Therefore the response
pattern of these thalamic neurons is shaped by both g-aminobutyric
acid-mediated inhibition and N-methyl-D-aspartate-mediated facilitation.
SHARPENING OF FREQUENCY TUNING IN AUDITORY THALAMUS
/ 9k0f$$ap31
J864-5
second harmonic CF component (CF2 ), which is Ç61 kHz
(Bodenhamer and Pollak 1983; Pollak et al. 1972; Suga and
Jen 1976, 1977; Suga and Manabe 1982; Suga and Tsuzuki
1985; Suga et al. 1975).
In the mustached bat, iontophoretic injections of an inhibitory transmitter antagonist (bicuculline methiodide, BMI)
broaden frequency-tuning curves of collicular (Yang et al.
1992) and cortical neurons tuned to Ç61 kHz (Sawaki et
al. 1992). The amount of broadening is very limited in the
cortex compared with that observed in the inferior colliculus.
In the MGB, neurons sharply tuned to Ç61 kHz are located
in its ventral division (MGBv) (Olsen 1986; Wenstrup et
al. 1994). Unlike that of the cat (Rinvik et al. 1987) and
the rat (Winer and Larue 1988), the MGBv of the mustached
bat contains an extremely small number of GABAergic neurons, but many GABAergic endings. These endings must
predominantly originate from the prethalamic auditory nuclei and the thalamic reticular nucleus (Winer et al. 1992).
In rats, glycinergic endings are practically nonexistent in the
MGB (Aoki et al. 1988; Wenthold and Hunter 1990). This
is presumably also the case in the mustached bat. In the
present research, therefore, we iontophoretically applied bicuculline to single thalamic neurons and examined whether
sharpening of frequency tuning by inhibition took place in
the MGB; whether this sharpening, if any, was to cause
further sharpening in the frequency tuning of prethalamic
neurons; and whether sharpening was ‘‘practically’’ completed in the MGB. We obtained data that clearly indicate that additional sharpening by inhibition takes place in
the MGB.
Neurons in the primary auditory cortex of the mustached
bat had been thought to respond primarily to single frequencies, like those in other mammals. However, Fitzpatrick et
al. (1993) found that cortical Doppler-shifted CF (DSCF)
neurons are sensitive to a combination of two sounds, although they are undoubtedly neurons in the primary auditory
cortex. These neurons are sharply tuned to a sound of
Ç61 kHz and their response to a Ç61-kHz sound is facilitated by a sound at Ç26 kHz. Because the frequency-tuning
curves of cortical DSCF neurons at Ç26 kHz are broad, they
show a facilitative response to a combination of pulse CF
(Ç30 kHz) or FM (sweeping Ç30 to 24 kHz) and echo CF2
( Ç61 kHz) when an acoustic stimulus simulates the pulseecho pair. The additional aim of the present paper is to
explore whether thalamic DSCF neurons show a facilitative
response to a pulse-echo pair, as cortical DSCF neurons do,
and whether the facilitation, if any, is mediated by N-methylD-aspartate (NMDA). We obtained data indicating that thalamic DSCF neurons show the facilitative response to the
pulse-echo pair and that NMDA is involved in evoking the
facilitative response.
METHODS
The materials and methods in the present experiments were basically the same as those described in Suga et al. (1983) and Saitoh
and Suga (1995). Therefore only the main parts of the methods
are described below. The protocol of our research was approved
by the animal study committee of Washington University (St.
Louis, MO; Approval No.: 890486).
Animals and surgery
Seven Jamaican mustached bats, Pteronotus parnellii parnellii,
weighing 11–14 g were used in this study. The CF2 resting fre-
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
Yuma bats (Suga 1969 1 ), mustached bats (Olsen and Suga
1991; O’Neill 1985; Suga and Manabe 19821; Suga and
Tsuzuki 19851; Suga et al. 1975, 1 1979; Yang et al. 1992 1 ),
horseshoe bats (Metzner and Radtke-Schuller 1987; Rübsamen and Schäfer 1990; Vater et al. 1992 1 ), big brown
bats (Casseday and Covey 1992 1 ), pallid bats (Fuzessery
1994 1 ), mice (Ehret and Moffat 1985), cats (Abeles and
Goldstein 1970; Evans and Nelson 19731; Godfrey et al.
1975; Greenwood and Maruyama 19651; Phillips et al. 1985;
Schreiner and Mendelson 1990; Schreiner and Sutter 1992;
Spirou and Young 19911; Sutter and Schreiner 19911; Young
and Brownell 1976 1 ), starlings (Lepplesack 1974), and
leopard frogs (Fuzessery and Feng 1982 1 ). Iontophoretic
injections of inhibitory transmitter antagonists broaden leveltolerant frequency-tuning curves (Sawaki et al. 1992; Vater
et al. 1992; Yang et al. 1992). Therefore the central auditory
system has inhibitory mechanisms for sharpening frequency
tuning.
Level-tolerant neurons already exist in the cochlear nucleus of the mustached bat (Suga et al. 1975) and the cat
(Young and Brownell 1976). In the little brown bat, neurons
with level-tolerant tuning are rare in the cochlear nucleus,
but are common in the inferior colliculus and the auditory
cortex (Suga 1964a,b, 1965a,b). This is also true in the
mustached bat (Olsen and Suga 1991; O’Neill 1985; Suga
and Manabe 1982; Suga et al. 1975; Yang et al. 1992).
Therefore it is almost certain that sharpening of frequency
tuning takes place at different levels of the central auditory
system, but may not always be progressive.
Katsuki et al. (1959a) speculated that sharpening of frequency tuning may be completed in the MGB. However, the
research has not been performed to examine whether the
sharpening by inhibition takes place in the MGB and whether
this sharpening, if any, takes place in addition to what occurred in the prethalamic auditory nuclei. The aim of our
present research is to obtain the answers to these two questions with the mustached bat. The reason we used the mustached bat instead of the cat in our research is described
below.
If a particular species of animal frequently uses constant
frequency (CF) and/or quasi-CF sounds for its very important acoustic behavior, it is expected that the neural mechanisms for fine frequency analysis of these sounds are enhanced in this species and can be more easily explored with
this species than with a species that does not use or rarely
uses these sounds. [a quasi-CF call is a slightly frequencymodulated (FM) ‘‘CF-like’’ call (Kanwal et al. 1994).] Domestic cats may communicate with quasi-CF sounds and
may occasionally feed on rodents that emit CF and/or quasiCF sounds (Ehret 1987; Sales and Pye 1974). It is, however,
not yet known whether level-tolerant fine frequency analysis
is important to cat’s auditory behavior in nature, although
the cat’s central auditory system contains level-tolerant neurons. On the contrary, the mustached bat emits CF-FM biosonar pulses (orientation sounds). The CF components of the
biosonar pulses are used for Doppler-shift measurements:
relative target velocity and insect wing beats (Gaioni et al.
1990; Goldman and Henson 1977; Henson et al. 1987; Johnson et al. 1974; Schnitzler 1970). Doppler-shift measurements are based on frequency analysis. Its auditory system
is highly specialized for the fine frequency analysis of the
CF components in the biosonar signals, in particular the
2099
2100
N. SUGA, Y. ZHANG, AND J. YAN
quency (RF) of each bat was measured before surgery and at the
beginning of each experiment (Fig. 1). The CF2 RF is the frequency of the CF2 component of the biosonar pulse emitted by the
bat at rest. The CF2 RF was used to normalize best frequencies
(BFs) of individual neurons, as explained later. After an intramuscular injection of the neuroleptanalgesic Innovar-Vet (Fentanyl,
0.08 mg/kg body wt and Droperidol, 4 mg/kg body wt), the
temporal muscles of the bat were reflected laterally to expose the
dorsal part of the skull. The surface of the skull was scraped, and
a 1.5-cm-long stainless steel post was attached to it with cyanoacrylate glue and dental cement. The surgical wound was treated with
antibiotic ointment (Furacin) and local anesthetic (Lidocaine) and
was sutured. Prednisolone (corticosteroid hormone) was administered before and 2 h after the surgery to reduce possible surgical
shock. The bat was kept separated in a humidity- and temperaturecontrolled room during recovery and thereafter. Electrophysiological experiments began 3–4 days after the surgery.
The reference electrode was a vinyl-lacquer-coated tungsten wire
with a tip diameter of Ç20 mm. It was placed on the dura mater
at the dorsocaudal part of the cerebrum. A five-barreled carbonfiber microelectrode (Armstrong-James and Millar 1979) was used
for the recording of action potentials from single neurons in the
MGB and also for the microiontophoretic injections of a g-aminobutyric acid-A (GABAA ) receptor antagonist, BMI (5 mM, pH
3.0, Sigma) and/or an NMDA receptor antagonist, D-2-amino-5phosphonovalerate (APV, 50 mM, pH 7.7, Genosys Biotech). A
glass tube at the center contained a 10-mm-diam carbon fiber. The
carbon fiber was etched to be Ç3 mm at the tip and was used to
record action potentials. This tube was surrounded by five glass
tubes (barrels). The overall diameter of the multibarreled electrode
Recording of action potentials and iontophoretic injections
of drugs
An unanesthetized mustached bat was placed in a Styrofoam
restraint that was suspended with an elastic band at the center of
an echo-attenuated soundproof room maintained at 31–327C. The
metal post glued on the skull was locked into a metal rod to
immobilize the bat’s head. Holes (100–200 mm diam) were made
in the skull with a needle just before electrode placement.
The multibarreled electrode was inserted into the MGBv through
the auditory cortex and then the hippocampus. Action potentials
of single MGBv neurons were then recorded. The recording sites
of thalamic DSCF neurons were determined to be within the MGBv
on the basis of their positions on the frequency map of the MGB.
Previous anatomic and physiological studies of the MGB have
shown that the thalamic DSCF neurons are clustered in the dorsoanterior portion of the MGBv, a region that projects to the DSCF
area of the primary auditory cortex (Olsen 1986) and that receives
ascending input from the dorsoposterior division of the central
nucleus of the inferior colliculus (Olsen 1986; Wenstrup et al.
1994). To avoid contamination of action potentials discharged by
more than one neuron, a time-amplitude window discriminator
(BAK Electronic, model DLS-1) was used.
For an iontophoretic injection of a drug, an electric current was
applied to a drug-filled barrel with a constant current source (Medical Systems, model BH-2). The timing of the injection was controlled manually during monitoring of responses of a single thalamic DSCF neuron. Automatic current balancing was performed
through the saline-filled barrel. Injection current was /50 nA for
BMI and 040 to 050 nA for APV. Retention current was set at
010 nA for BMI and /10 nA for APV. The effect of BMI (increase
in response magnitude and/or background discharges) was observed in °1 min after the onset of the injection, and further
change in neural activity was not observed during a 12-min BMI
application. The drug effect apparently stabilized in °1 min. The
duration of the BMI injection during the frequency-amplitude (FA)
scan was set at 12 min, because the drug effect required 1 min to
reach a steady state and because the FA scan, to obtain a frequencytuning curve, required 11 min.
Acoustic stimuli
FIG . 1. Schematized sound spectrograms of a biosonar pulse (solid
lines) of the mustached bat and its echo (dashed lines). CF1 –CF4 , 1st–
4th of a constant-frequency (CF) component; DS, Doppler shift; FM1 –
FM4 , 1st–4th harmonics of a frequency-modulated (FM) component.
Arrows: difference in frequency between the pulse CF2 and its Dopplershifted echo CF2 .
/ 9k0f$$ap31
J864-5
Two types of acoustic stimuli were delivered to the bats: 30-ms
CF tones (tone bursts) to measure frequency-tuning curves and
paired CF-FM sounds, similar to harmonics in a pulse-echo pair,
to study facilitative responses (Fig. 1). The durations of the CF
and FM components in the CF-FM sounds were 20 and 3.0 ms,
respectively. The rise/fall time of the CF tones and CF-FM sounds
was 0.5 ms. These acoustic stimuli were delivered by a condenser
loudspeaker placed 74 cm in front of the bat. The repetition rate
of these stimuli was either 5.0 or 6.7 per second, similar to that of
biosonar pulses emitted by the bat in a cruising phase of echolocation.
Single CF tones were used to measure the BF, minimum threshold (MT), and frequency-tuning curve of a thalamic neuron. Paired
CF-FM sounds were used to evoke the facilitative response of a
neuron and to determine whether the neuron was qualified to be
categorized as a thalamic DSCF neuron, i.e., a counterpart to a
cortical DSCF neuron that is sharply tuned to Ç61 kHz (Suga and
Manabe 1982) and shows a facilitative response to paired CF-FM
sounds similar to a combination of the first harmonic (H1 ) of a
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
Electrodes for recording of action potentials and for
iontophoretic injections of drugs
was 15–20 mm at the tip of the barrels, which was 20–40 mm
proximal to the tip of the carbon-fiber electrode. Three of the five
barrels were filled with drugs: BMI and/or APV. The remaining
two barrels were filled with isotonic saline solution (pH 7.0).
These barrels were used as either a ground or a balance electrode.
SHARPENING OF FREQUENCY TUNING IN AUDITORY THALAMUS
Data acquisition
Once action potentials of a single thalamic DSCF neuron were
recorded with a multibarreled electrode, the data acquisition was
started. The data acquired consisted of 1) the measurements of BF,
best amplitude (BA), upper threshold, and MT of the neuron; 2)
recording of the neuron’s response to the D scan; and 3) continuous
recording of the activity of the neuron while the FA scan was
repeatedly delivered.
When the single-unit recording was stable and the bat was quiet,
the response of the neuron to the FA or D scan was recorded
before, during, and after BMI and/or APV injections. (BMI or
APV was iontophoretically injected into the recording site, as already described.) The responses of a neuron to the acoustic stimuli
of the FA or D scan were continuously monitored as rasters and
an array of peristimulus time (PST) histograms displayed on a
computer monitor screen. The shapes of action potentials were
continuously monitored by comparing incoming action potentials
with the template of an action potential recorded at the beginning
of the continuous recording with a digital storage oscilloscope
(Tektronix 2211). The data were used for off-line analysis as far
as the shape of action potentials matched the template.
Data processing
Rasters and PST or PST cumulative (PSTC) histograms were
plotted to display the responses of a single neuron to the D
/ 9k0f$$ap31
J864-5
and/or FA scans. The binwidth of PST or PSTC histograms was
1.0 ms. Each histogram displayed the sum of the responses of a
single neuron to an identical stimulus repeatedly delivered 50 times
for the D scan or 5 times for the FA scan. A frequency-tuning
curve was plotted with the use of the responses to the FA scans
displayed as rasters.
The criterion for the determination of a threshold for an excitatory response was the boundary between a block showing one
response (action potential) per five trials and a block showing
no response per five trials. In unanesthetized animals, a response
magnitude and a background discharge rate of thalamic neurons
fluctuated to some extent. Therefore the threshold was determined
with additional criteria as described below. If a given block showed
at least one response per five trials, it was interpreted to be within
an excitatory area. If a given block showed no response per five
trials, but its adjacent blocks, away from the excitatory area,
showed at least one response per five trials, the given block is
interpreted to be within the excitatory area. Applying the above
criteria, excitatory tuning curves were plotted by visually inspecting not only the response in a given block, but also the responses in its adjacent eight blocks. Excitatory tuning curves based
on the data obtained by 0.25-kHz, 5.0-dB steps were then smoothed
by hand (e.g., Fig. 4).
The criterion for the determination of a threshold for an inhibitory response was the boundary between a block showing a decrease of õ70% in background discharges and a block showing a
decrease of ú70% in background discharges. However, additional
criteria were also applied for determining a threshold for an inhibitory response for the following reasons. 1) The rate of background
discharges was low, 0.8–6.8 per second (3.27 { 1.52 per s,
mean { SD; N Å 30 neurons), so that the number of background
discharges per block (5 sweeps per block; 200 1 5 ms per block)
was usually too small and widely fluctuated to obtain a reliable
value for a decrease. 2) Like the excitatory response, the inhibitory
response of a thalamic neuron fluctuated to some extent in unanesthetized animals. Therefore an inhibitory area was mapped by examining not only a given block, but also the adjacent eight blocks.
If the mean number of discharges in a given block and the adjacent
eight blocks was õ70% of the background rate, the given block
was interpreted as part of an inhibitory area. Inhibitory tuning
curves were smoothed by hand, as were excitatory tuning curves
(e.g., Fig. 4).
The effect of BMI on thalamic DSCF neurons was evaluated
by comparing the frequency-tuning curves, PST histograms, and
stimulus amplitude-impulse count functions (hereafter called amplitude-response functions) obtained before, during, and after the
drug injection. If a tuning curve broadened by the drug application
more than two blocks (i.e., ú500 Hz) in the FA scan and returned
(recovered) more than one block toward the original tuning curve,
the broadening was considered to be significant. A difference in
distribution of quality factors (Q values) or bandwidths of frequency-tuning curves between the peripheral, thalamic, and cortical
neurons was statistically tested with the Kolmogorov-Smirnov test
for goodness of fit.
RESULTS
Sharp frequency tuning and combination sensitivity of
thalamic DSCF neurons
Stable recordings of action potentials before, during, and
after a BMI injection were made from 30 thalamic DSCF
neurons. Like cortical DSCF neurons (Suga and Manabe
1982), these thalamic DSCF neurons were easily identified
because of their very sharp frequency tuning to a sound at
Ç61 kHz (i.e., CF2 frequency of biosonar signals). The
frequency-tuning curves of these thalamic DSCF neurons
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
pulse and the second harmonic (H2 ) of an echo (Fitzpatrick et al.
1993).
Sinusoidal waves were produced by a voltage-controlled function generator (Wavetek 134). The frequency of the waves was
varied manually or by a computer. A CF tone was generated by
shaping the output of the voltage-controlled function generator
with a custom-built electronic switch. The FM component of a
CF-FM sound was produced by applying a linear voltage ramp to
the function generator. The voltage ramp was synchronized with
the electronic switch. The amplitudes of these sounds were varied
with a decade attenuator (Hewlett-Packard, model 350D) or a
custom-built digital attenuator that was computer controlled. When
the frequency and amplitude of a CF tone were controlled by an
IBM 286 computer and a Modular Instruments hardware interface,
these parameters were varied in a certain way called the FA scan.
In the FA scan, frequency (F) scan consisted of 33 time blocks,
in 32 blocks of which frequency was changed in 0.25-kHz steps
from 56.00 to 63.75 kHz. In the 33rd (last) block, no stimulus was
presented. The duration of each block was 200 ms, so that the
duration of the F scan was 6,600 ms. The F scan was repeated five
times at a given amplitude with a 50-ms silent period. After every
five F scans, the amplitude of a CF tone was changed in 5-dB
steps from 106 to 6 dB SPL [amplitude (A) scan], so that the FA
scan consisted of 33 1 20 blocks. The duration of each FA scan
was 11 min.
To study facilitative responses of a thalamic neuron to paired
CF-FM sounds, the time interval between two CF-FM sounds in
a pair was varied manually or by the computer. When it was
computer controlled, it was varied in a certain way called a delay
(D) scan. The D scan consisted of 13 time blocks: pulse alone
(P), 10 pulse-echo pairs with different echo delays (0–36 ms in
4-ms steps), echo alone (E), and no stimulus (N) (Fig. 3). P
and E stimuli, respectively, consisted of H1 and H2 , the essential
harmonics to evoke the facilitative response of thalamic DSCF
neurons (Fig. 1). The N condition was used to count background
discharges. The duration of each block was 150 ms, so that
the duration of the D scan was 1,950 ms. The silent period between D scans was 50 ms. The silent period between the last E
stimulus in the D scan and the first P stimulus in the next D scan
was ¢314 ms.
2101
2102
N. SUGA, Y. ZHANG, AND J. YAN
had either a very sharp triangular shape (Fig. 2A), a pencil
shape, or a spindle shape (Fig. 2B), and most of them were
level tolerant. The frequency-tuning curves of peripheral
neurons with a BF of Ç61 kHz are narrow and triangular
in shape, but not level tolerant (Fig. 2, CN fibers) (Suga
and Jen 1977; Suga et al. 1975). The thalamic DSCF neurons
were much more sharply tuned in frequency than those peripheral neurons, as further documented throughout the present paper.
Like cortical DSCF neurons (Fitzpatrick et al. 1993),
these thalamic DSCF neurons showed a facilitative response
to the combination of H1 of a synthesized pulse and H2 of
a synthesized echo. In other words, the interesting response
properties of cortical DSCF neurons were due to those carried up by thalamic DSCF neurons. The essential combinations of signal elements in the H1 and H2 to evoke the facilitative response of the thalamic DSCF neurons were pulse CF1
and echo CF2 , or pulse FM1 and echo CF2 . In Fig. 3A, a
neuron shows a response to echo H2 stimulus (E) but not
to pulse H1 stimulus (P). When these stimuli were paired
and E was delayed from P, the neuron showed a facilitative
response at a 24-ms E delay (Fig. 3A1, asterisk). At the 24ms E delay, pulse FM1 was followed by echo CF2 with a
1.0-ms silent period. It was observed that the responses of
thalamic DSCF neurons to acoustic stimuli often consisted
of two components: fast and slow. The slow component
was usually more facilitated than the fast component in the
response to a paired P-E stimulus (Fig. 3), and was more
affected by BMI and APV than the fast component (Figs.
9, 10, 12, and 13). The responses to P, E, and P-E pair with
/ 9k0f$$ap31
J864-5
a 24-ms E delay are shown in Fig. 3, A2–A4, with a time
resolution higher than those in Fig. 3A1. In Fig. 3A4, the
fast and slow components are identifiable in the facilitative
response.
Figure 3 B shows another example of the responses of a
thalamic DSCF neuron that did not respond to P stimulus
but to E stimulus. As these two stimuli were paired, the
neuron showed a facilitative response to the pair, in which
E delayed from P by 0–4 or Ç24 ms. The facilitation for
0- to 4-ms E delays was due to the combination of CF1 and
CF2 , and that for a 24-ms E delay was due to the combination
of FM1 and CF2 . Figure 3, B2–B4, shows the responses to
P, E, and the simultaneous delivery of P and E with a time
resolution higher than those in Fig. 3B1. The fast and slow
components of the facilitative response are identifiable in
Fig. 3, B3 and B4. Thalamic DSCF neurons were broadly
tuned to echo delays, showing a single peak (Fig. 3A1) or
two peaks (Fig. 3B1) in delay tuning curves measured with
P-E pairs. Elimination of echo FM2 from the pairs had no
effect on the facilitative response, but the elimination of
pulse FM1 and echo CF2 abolished the facilitative response.
Elimination of pulse CF1 reduced the facilitative response
of most neurons. The essential elements of the P-E pair for
the facilitation were pulse CF1 , pulse FM1 , and echo CF2 .
Effect of BMI on thalamic DSCF neurons
Of the 30 thalamic DSCF neurons studied, BMI evoked
no noticeable change in the frequency-tuning curves of 19
neurons, but produced broadening in the frequency-tuning
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
FIG . 2. Inhibitory transmitter antagonist (bicuculline methiodide, BMI) broadens frequency-tuning curves of thalamic
Doppler-shifted CF (DSCF) neurons. Excitatory and inhibitory frequency-tuning curves of 2 thalamic DSCF neurons ( A and
);
B) were measured before, during, and after a 50-nA, 12-min application of 5.0 mM BMI. Con.: control condition (
excitatory frequency-tuning curve measured before the BMI application. Area inside the curve is the excitatory (response)
area. Inh.-con. (shaded area): inhibitory (response) area measured before the BMI application. In this area, background
discharges were inhibited by single tone stimuli. Curve surrounding the inhibitory area is the inhibitory tuning curve. BMI
● ): excitatory frequency-tuning curve measured during BMI application. Rec.: recovery condition (- - -); excitatory
(●
frequency-tuning curve measured after BMI application. CN fiber (- - -): average excitatory frequency-tuning curve of 18
cochlear nerve fibers that were tuned to 60.5–61.5 kHz. Crosses and open circles indicate the best amplitudes (BAs) to
excite the neurons before and during the BMI application, respectively. Averaged frequency-tuning curve of 18 cochlear
nerve fibers was obtained by Suga and Jen (1977). MGB, medial geniculate body.
SHARPENING OF FREQUENCY TUNING IN AUDITORY THALAMUS
2103
curves of the remaining 11 neurons: broadening toward both
low and high frequencies in 6 neurons, broadening only
toward high frequencies in 3 neurons, and broadening only
toward low frequencies in 1 neuron. In the remaining one
neuron, broadening was accompanied by a 10-dB lowering
in an overall tuning curve, so that its sharpness (Q value)
did not change.
The 11 thalamic DSCF neurons that showed a change
in a frequency-tuning curve for a BMI application had an
inhibitory frequency-tuning curve or curves (inhibitory area
or areas). In Fig. 4A, for example, the neuron was tuned to
61.04 kHz and showed the best response to a 66- to 71-dB
SPL, 60.25-kHz tone burst. The response to the tone burst
at BF (61.04 kHz) was weak, although it was low in threshold, 01 dB SPL. It had a large inhibitory area that was
located at frequencies ú60.25 kHz and ú66 dB SPL. In this
inhibitory area, background discharges were inhibited by
single tone stimuli. Because of this inhibition, the neuron
showed upper thresholds for sounds between 60.75 and
62.50 kHz. A small inhibitory area was also located at
57–58 kHz and 66–86 dB SPL. When BMI was applied to
this neuron, background discharges did not change significantly in rate, but excitatory responses to sounds, in particu-
/ 9k0f$$ap31
J864-5
lar at 66–96 dB SPL, increased. The best response was
evoked by a 91-dB SPL, 60.50-kHz tone burst. The inhibitory areas disappeared and excitatory responses were evoked
by the tone bursts that previously inhibited background discharges. As a result, the upper thresholds disappeared and the
frequency-tuning curve became particularly broader toward
frequencies higher than the BF of 61.04 kHz (Fig. 4B). The
broadening in the frequency-tuning curve and the increase
in the response magnitude to frequencies within the original
excitatory area indicate that the inhibitory areas were greatly
overlapped with the excitatory area. The excitatory frequency-tuning curve became narrower 22 min after BMI
application was stopped (Fig. 4C).
Figure 2 shows the frequency-tuning curves of two thalamic DSCF neurons; the curves showed broadening toward
both low and high frequencies when BMI was applied to
the neurons. The level-tolerant frequency-tuning curve of
the neuron in Fig. 2A (Con.: control condition) is triangular
and is sandwiched between inhibitory areas, whereas the
level-tolerant frequency-tuning curve of the neuron in Fig.
2B (Con.) is spindle shaped and is engulfed by an inhibitory
area. These tuning curves, including their tip portion, are
much narrower than those of peripheral neurons. When BMI
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
FIG . 3. Facilitative responses of 2 thalamic
DSCF neurons. Peristimulus time (PST) histograms
display the responses of 2 thalamic DSCF neurons
(A and B) to acoustic stimuli. The acoustic stimuli
were the delay (D) scan ( A1 and B1), pulse alone
(A2 and B2), echo alone (A3 and B3), and pulseecho pair with 24- or 0-ms echo delay (A4 and B4).
The D scan consisted of pulse stimulus (P), 10
pulse-echo paired stimuli (P-E) with echo delays of
0–36 ms, echo stimulus (E), and no stimulus (N).
Asterisk in A and B: best delay to evoke a facilitative response. P and E stimuli simulated the 1st
) and the 2nd
harmonic (H1 ) of P (A2 and B2,
harmonic (H2 ) of E (A3 and B3, - - -), respectively.
Each harmonic consisted of a 20-ms CF and a 3.0ms frequency-modulated (FM) component. Parameters of acoustic stimuli: CF1 , 29.09 kHz, 83 dB
SPL, CF2 , 62.35 kHz, 68 dB SPL (A); CF1 , 29.03
kHz, 80 dB SPL, CF2 , 61.79 kHz, 60 dB SPL (B).
In FM1 and FM2 , frequency swept downward by
6.0 and 12.0 kHz from the frequencies of CF1 and
CF2 , respectively. Vertical dashed lines in A3, B3,
A4, and B4: boundary or point of overlap between
fast and slow components of the response to E or
P / E.
2104
N. SUGA, Y. ZHANG, AND J. YAN
was applied to these neurons, their inhibitory areas changed
into the excitatory areas. As a result, the excitatory frequency-tuning curves became broader. The amount of broadening was small at the tip portion of the tuning curves, but
large at high sound pressure levels. However, the tuning
curves broadened by BMI were still much narrower than
those of peripheral neurons. The shape of the tuning curves
returned to those in the control condition Ç25 min after
BMI application was stopped (Rec.: recovery condition).
These data clearly indicate that the frequency-tuning curves
sharpened by prethalamic auditory nuclei are further sharpened by lateral inhibition occurring in the MGB. BMI increased the magnitude of the responses of these neurons to
tone bursts and their BAs.
Figure 5 shows the frequency-tuning curves of two thalamic DSCF neurons; the curves showed broadening toward
either low or high frequencies when BMI was applied to the
/ 9k0f$$ap31
J864-5
neurons. The tip portion of these curves (Con.) is similar to
that of peripheral neurons, but their skirt portion is much
narrower than that of peripheral neurons. An inhibitory area
was located at frequencies either higher (Fig. 5A) or lower
(Fig. 5B) than the neuron’s BF. When BMI was applied to
these neurons, their inhibitory areas changed into excitatory
areas, so that their excitatory frequency-tuning curves became broader. However, the BMI-broadened frequency-tuning curves were still much narrower than those of peripheral
neurons. These data again indicate that the frequency-tuning
curves sharpened by prethalamic auditory nuclei are further
sharpened by lateral inhibition occurring in the MGB. BMI
increased not only the magnitude of the response to tone
bursts of these neurons, but also their BAs.
All the 19 neurons that showed no noticeable change in
the frequency-tuning curve for a BMI application showed
background discharges of 1.2–6.8 per second (3.78 { 1.88
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
FIG . 4. Effect of inhibitory transmitter
antagonist (BMI) on the auditory response
of a thalamic DSCF neuron. Rasters: action
potentials discharged by a single thalamic
DSCF neuron before (A), during (B), and
after (C) a 50-nA, 12-min application of 5.0
mM BMI. The frequency and amplitude of
a 30-ms tone burst were controlled by a computer (see METHODS ).
SHARPENING OF FREQUENCY TUNING IN AUDITORY THALAMUS
2105
per s, N Å 19) and ON responses. The neurons had no noticeable inhibitory area in which background discharges were
inhibited by single tone stimuli. When BMI was applied to
the neurons, their background discharges and/or magnitude
of their responses increased by 18.8 { 26.4% (N Å 19) and
71.7 { 46.2% (N Å 19), respectively. These data indicate
that the neurons’ level-tolerant frequency tuning was created
by prethalamic auditory nuclei and transmitted up to the
neurons. Two examples of frequency-tuning curves that were
unaffected by BMI are shown in Fig. 6.
The curves in Fig. 6 labeled ‘‘Con.’’ for control condition
show the excitatory frequency-tuning curves of two thalamic
DSCF neurons measured before a BMI application: one is
opened at the upper portion (A) and the other is closed (B).
Both of the curves, including their tip portion, are narrower
than the tuning curves of peripheral neurons. The tuning
curves measured during and after a BMI application are
indicated by ‘‘BMI’’ and ‘‘Rec’’ (recovery condition), respectively. These three curves (Con., BMI, and Rec.) are
identical. This indicates not only that the tuning curves were
sharpened in prethalamic auditory nuclei, but also that our
recording and measurement of tuning curves were stable and
repeatable during the data acquisition. BMI increased the
magnitude of the responses of those neurons to tone bursts,
but did not change their BAs (Fig. 6, ^ ).
The broadening in a frequency-tuning curve evoked by
BMI was usually small at low sound pressure levels, but
large at high sound pressure levels (Figs. 2 and 5). This
FIG . 6. Inhibitory transmitter antagonist (BMI) had no effect on excitatory frequency-tuning curves of 2 thalamic DSCF
neurons (A and B). See legend to Fig. 2.
/ 9k0f$$ap31
J864-5
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
FIG . 5. Inhibitory transmitter antagonist (BMI) broadens frequency-tuning curves of thalamic DSCF neurons. Excitatory
and inhibitory frequency-tuning curves of 2 thalamic DSCF neurons ( A and B) were measured before, during, and after a
50-nA, 12-min application of 5.0 mM BMI. See legend to Fig. 2.
2106
N. SUGA, Y. ZHANG, AND J. YAN
sharpening of frequency tuning that occurred in prethalamic
auditory nuclei. The mean difference is 0.05, 0.25, 3.50, and
6.79 kHz, respectively, at 10, 30, 50, and 70 dB above MT.
The bandwidths at 50 and 70 dB above MT are much narrower
in all 30 thalamic neurons than those of peripheral neurons
regardless of whether or not thalamic neurons were affected
by BMI (Fig. 7, Bc and Bd). The difference in bandwidth at
30 dB above MT is small, but significant (P õ 0.05; Fig.
7Bb). The difference in bandwidth at 10 dB above MT is
not significant (P ú 0.05; Fig. 7Ba).
Are background discharges different between neurons that
showed broadening in frequency tuning with BMI and
those that did not?
The rate of background discharges was 1.2–6.8 per second (3.78 { 1.88 per s) for the 19 neurons showing no BMI
broadening and 0.8–5.1 per second (2.45 { 1.29 per s) for
the 11 neurons showing BMI broadening. The difference in
background discharge rate between the two groups of neurons is significant (t-test, P õ 0.05). During BMI applica-
FIG . 7. Change in bandwidth (BW) of frequencytuning curves of thalamic DSCF neurons evoked by
an inhibitory transmitter antagonist, BMI (5.0 mM,
50 nA, 12 min). Aa–Ad: changes in bandwidth of
tuning curves at 10, 30, 50, and 70 dB above minimum
threshold (MT) evoked by BMI. B: changes in bandwidth of the tuning curves evoked by BMI (white
bars) and differences in bandwidth between the tuning
curves of the thalamic DSCF neurons (before BMI
application) and those of peripheral neurons (black
bars). Bandwidths were measured at 10, 30, 50, and
70 dB above MT (Ba–Bd). Differences between
white and black bars indicate the amount of sharpening that occurred in prethalamic auditory nuclei. Mean
difference in frequency between white and black bars
is given in Ba–Bd.
/ 9k0f$$ap31
J864-5
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
was a general rule. However, the amount of change in a
frequency-tuning curve varied from neuron to neuron, so
that all the tuning curve data obtained from the 30 thalamic
DSCF neurons were pooled, and changes in the bandwidths
of the frequency-tuning curves were measured at 10, 30, 50,
and 70 dB above MT (Fig. 7A). Broadening in a frequencytuning curve at 10 dB above MT was very small and was
ú500 Hz only in 1 of the 30 neurons. However, broadening
was noticeable at 30 dB above MT in six of the neurons. It
was prominent at 50 and 70 dB above MT in 8 and 11
neurons, respectively. As shown in Figs. 2 and 5, BMI could
not, however, broaden any of the frequency-tuning curves
of the thalamic DSCF neurons to be the same as the tuning
curves of peripheral neurons.
Figure 7B shows the distribution of the amounts of BMIinduced changes in the frequency-tuning curves of the 30
thalamic DSCF neurons (white bars) and that of differences
in bandwidth between thalamic (without BMI) and peripheral
frequency-tuning curves (black bars) at four stimulus levels:
10, 30, 50, and 70 dB above MT. The difference in distribution
between the white and black bars indicates the amount of
SHARPENING OF FREQUENCY TUNING IN AUDITORY THALAMUS
tions, the rate of background discharges was 1.3–12.5 per
second (4.40 { 2.74 per s, N Å 19) and 1.0–11.2 per second
(3.51 { 3.02 per s, N Å 11), respectively, for the two groups
of neurons. Percent change in discharge rate was 039–102%
(18.8 { 26.4%, N Å 19) and 025–120% (24.2 { 33.8%,
N Å 11), respectively, for the two groups of neurons. The
change in background discharge rate evoked by BMI does
not differ between the above two groups of neurons (t-test,
P ú 0.10). There was no clear link between a BMI-induced
broadening in frequency tuning and a change in background
discharges.
2107
and cortical neurons (P õ 0.01 and 0.05 for Fig. 8, B and
C, respectively). Q10dB values show no significant difference
between peripheral and thalamic neurons (P ú 0.05), but a
significant difference between thalamic and cortical neurons:
cortical neurons are sharper than thalamic neurons (Fig. 8A;
P õ 0.01). The above comparisons indicate that further
sharpening of frequency tuning takes place even in the DSCF
area of the auditory cortex, and that the sharpening in the
DSCF area occurs even at the tip portion of frequency-tuning
curves.
Changes in response pattern and BA evoked by BMI
Differences in sharpness of frequency tuning between
thalamic and cortical DSCF neurons
FIG . 8. Distributions of quality factors (Q10dB ,
Q30dB , and Q50dB values) of peripheral (AN, shaded
bars), thalamic DSCF (MGB, black bars), and cortical DSCF (AC, white bars) neurons. Best frequencies of all these neurons were between 61.0 and 62.0
kHz. Infinity (` ): neuron had a closed tuning curve.
[Peripheral and cortical data were obtained by Suga
and Jen (1977) and Suga and Manabe (1982), respectively.]
/ 9k0f$$ap31
J864-5
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
To examine whether the frequency-tuning curves of cortical DSCF neurons are narrower or broader than those of
thalamic DSCF neurons, the quality factors (QndB value) of
the cortical DSCF neurons measured by Suga and Manabe
(1982) are plotted together with those of thalamic DSCF
neurons in Fig. 8. The quality factors of the Ç61-kHz-tuned
peripheral neurons measured by Suga and Jen (1977) were
also plotted in the figure for comparison. A QndB value is
defined as a BF divided by a bandwidth at n decibels above
MT. Here, n is either 10, 30, or 50. Q30dB and Q50dB values
are respectively larger in the order of peripheral, thalamic,
As shown in Fig. 4, BMI evoked an increase in magnitude
of an excitatory response (number of impulses per stimulus)
and/or a change of an inhibitory response into an excitatory
response. These changes were not due to an overall increase
in discharge rate of a neuron, but specific to the responses
to sound stimuli at particular frequencies and amplitudes.
Figure 9 shows the amplitude-response functions of a thalamic DSCF neuron obtained before, during, and after a BMI
application. Before the BMI application, the neuron showed
the best response to a 48-dB SPL, 60.50-kHz tone burst, a
weak inhibitory response to the tone burst at 81–100 dB
SPL, and an upper threshold at 81 dB SPL. During the
BMI application, the neuron showed almost no change in
2108
N. SUGA, Y. ZHANG, AND J. YAN
background discharges, MT, and responses to 28- to 45-dB
SPL tone bursts, but a drastic change in responses to 50- to
100-dB SPL tone bursts. The BA to excite the neuron became 65 dB SPL, and both the upper threshold and inhibitory
responses disappeared. The prominent increase in response
was due to multiple discharges. The PST histograms of Fig.
9B show the responses to 30-dB SPL (Ba) and 70-dB SPL
(Bb) tone bursts of the same neuron as in Fig. 9A. The
response to the 70-dB SPL sound consists of a fast component and a slow component. The vertical dashed lines in
Fig. 9B indicate the boundary between these two components. BMI increased both the fast and slow components of
the response, but its effect was much larger on the slow
component than on the fast component (B2b). Therefore the
large increase in impulse count shown in Fig. 9A was mainly
due to the change in the slow component.
Figure 10 shows an additional example of the amplituderesponse functions of a thalamic DSCF neuron obtained before, during, and after a BMI application. This neuron
showed no sign of inhibitory response, but BMI prominently
increased its excitatory response without changing both MT
and upper threshold (Fig. 10A). This increase in response
was mainly due to the change in the slow component of
/ 9k0f$$ap31
J864-5
the response (Fig. 10B). The BA of the neuron increased
by Ç10 dB.
In Figs. 9 and 10, BMI increased the BAs of the neurons.
The effect of BMI on BAs varied from neuron to neuron.
In some neurons, a frequency-tuning curve broadened, but
a BA did not change. In some others, a frequency-tuning
curve did not change, but a BA changed. Figure 11 shows
the relationship between BAs measured with and without BMI.
Shaping response patterns by NMDA and GABA
To obtain an insight into synaptic interactions shaping the
auditory responses of thalamic neurons, BMI and/or APV
was iontophoretically applied to thalamic DSCF neurons.
The responses of all the 24 neurons studied were affected
by BMI and APV without exception. However, there were
some variations in the amount of the drug’s effects, as explained below.
Responses of single thalamic DSCF neurons were recorded to H1 of a pulse (P stimulus) and/or H2 of an echo
(E stimulus). Each stimulus consisted of the CF and FM
components (See Fig. 1). The PST histograms in Fig. 12,
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
FIG . 9. Effect of inhibitory transmitter antagonist (BMI)
on the amplitude-response function (A) and auditory responses (B) of a thalamic DSCF neuron. Con., BMI, Rec.:
amplitude-response functions obtained with a 30-ms, 60.50kHz tone burst before, during, and after a 50-nA, 12-min
application of 5.0 mM BMI. Shaded portion of ‘‘Con.’’: decrease in background discharges evoked by a single tone stimulus. Auditory responses displayed by the PST histograms in
B were evoked by the 30-dB SPL (Ba) or 70-dB SPL (Bb),
60.50-kHz tone bursts delivered 15 times. Note that the response to the 70-dB SPL tone burst consists of a fast and a
slow component. Vertical dashed line: boundary between
these two components.
SHARPENING OF FREQUENCY TUNING IN AUDITORY THALAMUS
FIG . 11. BAs of thalamic DSCF neurons measured before and during
a 50-nA, 12-min application of 5.0 mM BMI. Filled and open circles:
data obtained from neurons that showed broadening or no broadening,
respectively, of a frequency-tuning curve for BMI. Data points on oblique
line: no change in BA was evoked by BMI.
The PSTC histograms of Fig. 13 show the responses of
three thalamic DSCF neurons studied with APV and BMI.
In A and B, BMI dramatically enhanced the slow component
of the auditory responses of the neurons with little effect on
its fast component, and APV reduced the BMI-enhanced
slow component with little effect on the fast component.
(The boundary or the point of overlap between the fast and
slow components is indicated by the arrows in A and B.) In
C, however, BMI enhanced the whole response of the neuron, and APV reduced the whole BMI-enhanced response.
The boundary between the fast and slow components, if any,
was unclear in the response of this neuron.
The differential effect of APV on the fast and slow components of the BMI-enhanced auditory responses of thalamic
DSCF neurons was found in 17 neurons of the 24 studied.
In these neurons, the reduction of the response evoked by
APV was 3–32% (16.7 { 10.4%) for the fast component,
but was 38–71% (52.6 { 10.6%) for the slow component.
The reduction of the slow component was 2.0–16.3
(5.09 { 4.97) times larger than that of the fast component.
The data obtained with the drugs clearly indicate that the
response properties of thalamic DSCF neurons are shaped
in the MGB through the interaction between inhibition mediated by GABA and excitation/facilitation mediated by nonNMDA and NMDA.
DISCUSSION
FIG .
10. Effect of inhibitory transmitter antagonist (BMI) on the amplitude-response function (A) and auditory responses (B) of a thalamic DSCF
neuron. The amplitude-response functions were obtained with a 30-ms,
61.34-kHz tone burst. The auditory responses displayed by the PST histograms were to 26-dB SPL (Ba) or 76-dB SPL (Bb), 61.34-kHz tone bursts.
Note that the response to the 76-dB SPL tone burst consists of a fast and
a slow component. See legend to Fig. 9.
/ 9k0f$$ap31
J864-5
How to compare the sharpness of frequency tuning of
central auditory neurons with that of peripheral auditory
neurons
The sharpness of a tuning curve has been expressed by a
QndB value that is a BF divided by a bandwidth at n decibels
above MT (n dB bandwidth). When Kiang et al. (1965)
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
A–C, display the responses of a thalamic DSCF neuron to
P, E, and P-E stimuli, respectively. The response of the
neuron to E stimulus consisted of a fast and a slow component (B1). (The point of overlap between the 2 components
is indicated by a vertical dashed line.) When BMI was applied to the neuron, the slow component increased, but the
fast component was unchanged (B2). When APV was applied during the BMI application, this increased portion of
the slow component disappeared (B3). After termination of
the APV application, but continuation of the BMI application, only the slow component increased (B4). When the
BMI application was terminated, the slow component became very small (B5). These observations indicate that the
slow component of the excitatory response is mediated by
NMDA, and that the NMDA-mediated response is controlled
(suppressed) by GABA. The drug effects described above
were more prominent on the facilitative response to a P-E
stimulus than on the response to the E stimulus alone. In
Fig. 12C, the slow component of the facilitative response of
the neuron to P-E stimulus (C1) was dramatically increased
by BMI (C2). This increased portion of the facilitative response was eliminated by APV (C3).
2109
2110
N. SUGA, Y. ZHANG, AND J. YAN
introduced a Q10dB value, they wrote: ‘‘an objective measure
of sharpness of tuning used by engineers is the quantity
Q defined as center frequency/bandwidth at 3 dB above
threshold. A somewhat similar measure may be useful here.
A Q for tuning curves may, for convenience, be defined as
BF/10 dB bandwidth’’ (page 88). It should be noted that
a Q10dB value was defined on an arbitrary basis without any
consideration of whether sharpness of frequency tuning of
peripheral and central neurons is expressed adequately or
best by this value.
The frequency-tuning curve of a peripheral neuron is
somewhat triangular, whereas that of a central neuron shows
a large variation in its shape. It can be either pencil shaped,
triangular shaped (e.g., Figs. 2A and 6A), spindle shaped
(e.g., Figs. 2B and 6B), bowl shaped, or multipeaked. If a
tuning curve is exactly triangular in shape, its sharpness can
be appropriately expressed by a single value, e.g., a Q10dB
value. If it is not, a Q10dB value related only to the tip portion
of a tuning curve is simply inadequate to describe the overall
sharpness of the tuning curve.
Peripheral and central auditory neurons commonly respond to sounds over a 100-dB range. Their responses are
usually weak to a sound within 20 dB above MT. The amplitude-response functions of central auditory neurons are frequently nonmonotonic. There are no data indicating that the
neurons’ responses and the widths of their tuning curves at
10 dB above MT are usually more important than those at
other sound pressure levels. Therefore there is no reason to
use only a Q10dB value to express the sharpness of frequency
tuning.
/ 9k0f$$ap31
J864-5
Suga and Tsuzuki (1985) pointed out that the choice of
parameters characterizing tuning curves should be contingent on the problem to be discussed. To discuss sharpening,
a change in the skirt portion of a tuning curve (e.g., Q50dB )
should be mainly considered, not the tip portion, because
the tip portion is sharp at the periphery without inhibition.
To discuss broadening, on the other hand, the tip portion of
a tuning curve (e.g., Q10dB ) should be mainly considered,
not the skirt portion, because the skirt portion is broad at
the periphery without excitatory integration.
The frequency-tuning curves of peripheral neurons commonly show a deflection point at Ç40 dB above MT where
the slopes of the passive (1st) and active (2nd) filters join
(e.g., Evans 1972). Therefore Q30dB and Q50dB values or 30and 50-dB bandwidths may be used to determine whether
the passive and/or active portions of a tuning curve are
sharpened by inhibition in the central auditory system.
In cats, frequency-tuning curves with Q30dB and/or Q50dB
values much larger than those of peripheral neurons have
been obtained at different levels of the central auditory system (Abeles and Goldstein 1970; Evans and Nelson 1973;
Godfrey et al. 1975; Greenwood and Maruyama 1965; Katsuki 1958, 1959a; Phillips et al. 1985; Schreiner and Mendelson 1990; Schreiner and Sutter 1992; Spirou and Young
1991; Sutter and Schreiner 1991; Young and Brownell
1976). Therefore the cat’s central auditory system has a
mechanism for the sharpening of frequency tuning, and this
mechanism drastically sharpens the skirt of a tuning curve.
The functional significance of the sharpening of frequency
tuning is discussed later.
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
FIG . 12. Effects of an inhibitory transmitter antagonist (BMI) and an excitatory transmitter antagonist (APV, D-2-amino-5-phosphonovalerate) on auditory responses of a thalamic DSCF neuron. Responses to acoustic stimuli are displayed as PST
histograms. Vertical dashed lines in B and C: point
of overlap between the fast and slow components in
the response. The acoustic stimuli were a pulse alone
(A), echo alone (B), and pulse-echo pair (C). The
)
pulse (P) and echo (E) stimuli were H1 (A1,
of the pulse and H2 (B1, - - -) of the echo, respectively. Each harmonic consisted of a 20-ms CF and
a 3.0-ms FM component. CF1 : 29.47 kHz, 42 dB
SPL; CF2 : 61.89 kHz, 29 dB SPL. In FM1 and FM2 ,
frequency swept downward by 6.0 and 12 kHz from
the frequencies of CF1 and CF2 , respectively. 1–5
in A–C: control condition, BMI application, BMI
and APV applications, BMI application, and recovery condition, respectively.
SHARPENING OF FREQUENCY TUNING IN AUDITORY THALAMUS
2111
plete, because all high-threshold neurons may become level
tolerant. Because MT differs from neuron to neuron and
frequency-tuning curves are measured up to 100 dB SPL,
we may propose the following equation: level tolerance
õ4a(100 dB SPL 0 MT)/70 dB. Here, a is a 10-dB bandwidth of a corresponding peripheral tuning curve. Division
by 70 dB applies the criterion that ‘‘a level-tolerant tuning
curve must have a bandwidth less than 4a at least up to 70
dB above MT’’ to all neurons regardless of MT. The reason
to choose 70 dB above MT is that a 70-dB bandwidth can
be measured in most neurons (e.g., Suga and Manabe 1982;
Suga and Tsuzuki 1985). According to this definition, the
tuning curves shown in Figs. 2, 5A, and 6 are level tolerant,
but the curves shown in Figs. 4 and 5B are not, although
they are much sharper than a peripheral tuning curve.
FIG . 13. Effects of an inhibitory (BMI) and an excitatory (APV) transmitter antagonist on auditory responses of 3 thalamic DSCF neurons ( A–
C). The responses to acoustic stimuli delivered 100 times are displayed as
peristimulus time cumulative (PSTC) histograms. Paired acoustic stimulus
consisted of H1 of the pulse stimulus (CF component Å Ç29 kHz) and H2
of the echo stimulus (CF component Å Ç61 kHz). A–C, top left corner:
schematized sound spectrogram of the paired stimulus. BMI was 1st applied
to the neurons, and then APV was applied without stopping the BMI application. Arrows in A and B: boundary between the fast and slow components
in the response. The fast component was not affected by the drugs, but the
slow component was. In C, the presence of the fast and slow components
was unclear.
Definition of level tolerance
For auditory physiologists, ‘‘sharp’’ tuning curves mean
the tuning curves with high Q10dB values. Different from
frequency-tuning curves at the periphery, a frequency-tuning
curve sharpened by inhibition has a narrow bandwidth even
at high stimulus levels, but has a Q10dB value similar to that of
peripheral neurons. To distinguish it from sharp frequencytuning curves of peripheral neurons, the term level-tolerant
sharp frequency tuning was introduced by Suga (1977) and
Suga and Manabe (1982). For simplicity, we call it a leveltolerant tuning curve and a neuron showing it a level-tolerant
neuron. However, it has not yet been defined how narrow a
bandwidth of a tuning curve should be at all stimulus levels
to be called level-tolerant. We may propose that level tolerance applies for a tuning curve that has a bandwidth narrower
than 4 times a 10 dB bandwidth of a ‘‘corresponding’’ peripheral tuning curve regardless of sound pressure levels.
The reason to choose 4 times a 10 dB bandwidth is that the
frequency-tuning curves of peripheral neurons commonly
show a deflection point at Ç40 dB above MT where the
sharply tuned active filter joins the broadly tuned passive
filter (e.g., Evans 1972), and that lateral inhibition sharpens
at least this broadly tuned portion (i.e., the skirt portion) of
the tuning curves. However, the above definition is incom-
/ 9k0f$$ap31
J864-5
Auditory nuclei and the auditory cortex consist of several
subdivisions, and each subdivision contains neurons with
different response properties (e.g., Suga 1990). Therefore
neurons with identical BFs should be pooled for a comparison of sharpness of frequency tuning between peripheral and
central neurons, and a division of a nucleus or the cortex
studied for the comparison should be identified. Therefore
in the present study we focused on neurons that are directly
involved in fine frequency analysis to extract velocity information carried by Doppler-shifted echoes at Ç61 kHz.
All thalamic DSCF neurons studied had frequency-tuning
curves that were much sharper than those of peripheral neurons. The frequency-tuning curves of 37% of these thalamic
neurons became broader when BMI was iontophoretically
applied to the neurons, but the BMI-broadened curves were
still much narrower than those of peripheral neurons (Figs.
2, 5, and 7). The quality factors of cortical DSCF neurons
are slightly higher than those of thalamic DSCF neurons
(Fig. 8). BMI injections into the DSCF area of the primary
auditory cortex broadened the frequency-tuning curves of
58% of cortical DSCF neurons studied and eliminated the
upper thresholds of 39% of the neurons (Sawaki et al. 1992).
These data indicate the following three important facts. 1)
Sharpening of frequency tuning is extensively performed
in prethalamic auditory nuclei. 2) Additional sharpening of
frequency tuning takes place in 37% of thalamic neurons
by GABA-mediated inhibition. 3) Sharpening of frequency
tuning by GABA-mediated inhibition also takes place in the
auditory cortex.
In the inferior colliculus of the mustached bat, 42% of
neurons studied showed broadening of a frequency-tuning
curve for BMI (Yang et al. 1992). In the cochlear nucleus of
the mustached bat, a level-tolerant tuning curve sandwiched
between inhibitory tuning curves was found (Suga et al.
1975). On the basis of all the data summarized above, it
may be concluded that sharpening of frequency tuning by
inhibition takes place at different levels of the central auditory system, and that progressive sharpening by inhibition
creates very sharp level-tolerant frequency tuning of some
neurons, as Katsuki et al. (1958) described. This does not
mean that all frequency-tuning curves are progressively
sharpened. It is possible that level-tolerant tuning created
at one level is not further sharpened. Perhaps it is worth
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
Sharpening of frequency tuning by inhibition occurs at
different levels
2112
N. SUGA, Y. ZHANG, AND J. YAN
mentioning that the central auditory system also creates
broadly frequency-tuned neurons, as reported in numerous
papers (e.g., in cats, Aitkin and Webster 1972; in little brown
bats, Suga 1965a,b).
Three types of frequency-domain processing of auditory
signals
Level-tolerant neurons and level-tolerant fine frequency
discrimination
The acoustic behavior of mustached and horseshoe bats
indicates that they are highly specialized for processing velocity information that is carried by the long CF components
of a biosonar pulse and its Doppler-shifted echo (Gaioni et
al. 1990; Henson et al. 1987; Schnitzler 1968, 1970; Schuller
et al. 1974). In the DSCF and CF/CF areas of the auditory
cortex of the mustached bat, the great majority of neurons
tuned to the frequencies of these CF components shows
level-tolerant tuning that is created by lateral inhibition
(Suga and Manabe 1982; Suga and Tsuzuki 1985; Suga et
al. 1979). The sharper the level-tolerant tuning curve, the
closer the best inhibitory frequencies to the best excitatory
or facilitatory frequencies (Suga and Tsuzuki 1985). Fine
analysis of velocity information (Doppler shifts evoked by
relative target motion and insect wing beats) is directly related to fine frequency analysis, so that level-tolerant tuning
is directly related to level-tolerant fine frequency discrimination. The DSCF area of the primary auditory cortex shows
a very fine frequency representation (Suga and Jen 1976;
Suga and Manabe 1982; Suga et al. 1987). Inactivation of
this area with muscimol evokes a clear deficit of the bat’s
ability to perform a fine frequency discrimination (Riquimaroux et al. 1991).
Big brown bats frequently emit long ‘‘shallow FM’’
sounds in the search phase of echolocation. In these sounds,
frequency slowly sweeps from 30 to 20 kHz (Master et al.
1991). The great majority of collicular neurons tuned to
these frequencies shows level-tolerant tuning (Casseday and
Covey 1992). The long shallow FM sound is suited for
/ 9k0f$$ap31
J864-5
Fast and slow components of auditory responses of
thalamic DSCF neurons
In rats, neurons in the dorsal division of the MGB show
a large NMDA-mediated depolarization to an electrical stimulation of the brachium of the inferior colliculus, whereas
neurons in the MGBv seldom show it. However, some
MGBv neurons show a noticeable NMDA response when
GABA-mediated inhibition is eliminated by picrotoxin (Hu
et al. 1994). In the mustached bat, thalamic DSCF neurons
are located in the MGBv (Olsen 1986; Wenstrup et al.
1994). These thalamic DSCF neurons are combination sensitive (Fig. 3) and show an NMDA-mediated auditory response that is mostly inhibited by GABA. When BMI is
applied to the neurons, the NMDA-mediated response (i.e.,
the slow component of the response) becomes prominent
(Figs. 12 and 13). Therefore the data obtained from mustached bats and rats are similar to each other. However,
there appears to be a quantitative difference between them,
because the auditory responses of thalamic DSCF neurons
were reduced by APV without exception. This difference
may be due to the combination sensitivity of the thalamic
DSCF neurons.
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
The frequency-domain processing of auditory signals in
the central auditory system may be divided ‘‘neuroethologically’’ into three types: 1) processing of CF and quasi-CF
sounds by neurons with level-tolerant sharp frequency tuning, 2) processing of spectral patterns of broadband sounds
such as complex sounds with many harmonics and noise
bursts by neurons with excitatory (or facilitatory) and inhibitory areas and by neurons particularly sensitive to broadband
sounds, and 3) processing of sounds changing in spectrum
such as FM sounds by neurons with excitatory (or facilitatory) and inhibitory areas and by neurons particularly sensitive to FM sweeps. These three types of frequency-domain
processing are mutually related and overlap. Which type of
signal processing is more prominent than the others depends
on the properties of biologically important sounds such as
those produced by a given species and by its prey and predators (e.g., Casseday and Covey 1992; Fuzessery 1994; Suga
and Manabe 1982; Suga et al. 1983). The processing of CF
or quasi-CF sounds is directly related to the problem of
whether the central auditory system has a mechanism for
the sharpening of frequency tuning of neurons.
target detection at long distances. The sharper the tuning,
the larger the signal-to-noise ratio, i.e., the better the target
detection ability.
Rodents emit long CF and quasi-CF sounds. When a baby
mouse is isolated from its mother, for example, it emits such
sounds (Sales and Pye 1974). In the inferior colliculus of
mice, many level-tolerant neurons have been found (Ehret
and Moffat 1985). The long CF or long quasi-CF sounds
are suited to detecting and discriminating the individuals
producing them. Level-tolerant tuning is ideal for this
purpose.
The data obtained from animals that frequently use CF or
quasi-CF sounds for the detection and discrimination of a
signal source and/or target motion indicate that level-tolerant neurons are directly related to the properties of the species-specific CF or quasi-CF sounds and species-specific behavior utilizing these sounds. Therefore it is our contention
that one of the important functions of level-tolerant neurons
is level-tolerant fine frequency discrimination.
Many central auditory neurons have both excitatory and
inhibitory frequency-tuning curves. Their responses to complex sounds depend on the balance between excitation and
inhibition. Therefore, ‘‘the analysis of complex sounds is
not performed in such a way that single neurons with narrow
excitatory areas always respond when certain components of
complex sounds fall into the excitatory areas. Many neurons
require a certain structure in the complex sound to be excited
by it’’ (Suga 1968). Level-tolerant neurons are not exceptional in this respect. They would also be involved in the
processing of complex sounds. It is also our contention that
the processing of auditory signals by central auditory neurons should be evaluated in relation to biologically important
sounds.
Inhibition plays different roles in processing auditory information in the frequency, amplitude, time, and binaural
domains. Therefore inhibition does not necessarily create
level-tolerant tuning (e.g., Caspary et al. 1994; Evans and
Zhao 1993; Palombi and Caspary 1992; Vater et al. 1992;
Yang et al. 1992).
SHARPENING OF FREQUENCY TUNING IN AUDITORY THALAMUS
We are grateful to the Jamaican Ministry of Agriculture for permitting
the export of mustached bats from Jamaica. We also thank J. F. Olsen and
A. Kadir for comments on this article.
Our research on bats has been supported by research grants from the
National Institute on Deafness and Other Communicative Disorders
(DC-00175), Office of Naval Research (N00014-90-J-1068), and Human
Frontier Science Program.
Address for reprint requests: N. Suga, Dept. of Biology, Washington
University, 1 Brookings Dr., St. Louis, MO 63130.
Received 26 December 1995; accepted in final form 18 December 1996.
REFERENCES
/ 9k0f$$ap31
J864-5
areas of auditory neurons in the cochlear nucleus. J. Neurophysiol. 28:
863–892, 1965.
GRINNELL, A. D. The neurophysiology of audition in bat: intensity and
frequency parameters. J. Physiol. Lond. 167: 38–66, 1963.
HENSON, O. W., JR ., BISHOP, A., KEATING, A., KOBLER, J. B., HENSON,
M. M., WILSON, B., AND HANSEN, R. Biosonar imaging of insects by
Pteronotus p. parnelli, the mustached bat. Natl. Geogr. Res. 3: 82–101,
1987.
HU, B., SENATAROV, V., AND MOONEY, D. Lemniscal and non-lemniscal
synaptic transmission in rat auditory thalamus. J. Physiol. Lond. 497:
217–231, 1994.
JOHNSON, R. A., HENSON, O. W., JR ., AND GOLDMAN, L. R. Detection of
insect wing beats by the bat, Pteronotus parnellii (Abstract). J. Acoust.
Soc. Am. 55, Suppl.,: 53, 1974.
KANWAL, J. S., MATSUMURA, S., OHLEMILLER, K. K., AND SUGA, N. Analysis of acoustic elements and syntax in communication sounds emitted by
the mustached bat. J. Acoust. Soc. Am. 96: 1229–1254, 1994.
KATSUKI, Y. Neural mechanism of auditory sensation in cats. In: Sensory
Communication, edited by W. A. Rosenblith. Cambridge, MA: MIT
Press, 1961, p. 561–583.
KATSUKI, Y., SUMI, T., UCHIYAMA, H., AND WATANABE, T. Electric responses of auditory neurons in cat to sound stimulation. J. Neurophysiol.
21: 569–588, 1958.
KATSUKI, Y., WATANABE, T., AND MARUYAMA, N. Activity of auditory
neurons in upper levels of brain of cat. J. Neurophysiol. 22: 343–359,
1959a.
KATSUKI, Y., WATANABE, T., AND SUGA, N. Interaction of auditory neurons
in response to two sound stimuli in cat. J. Neurophysiol. 22: 603–623,
1959b.
KIANG, N.Y.S., WATANABE, T., THOMAS, E. C., AND CLARK, L. F. Discharge
patterns of single fibers in the cat’s auditory nerve. MIT Res. Monogr.
35: 1–154, 1965.
LEPPLESACK, H. J. Funktionelle Eizenschaften der Hörbahn im Feld L des
Neostriatum candale des Staren. J. Comp. Physiol. 88: 271–320, 1974.
MASTER, W. M., JACOBS, S. C., AND SIMMONS, J. A. The structure of echolocation sounds used by the big brown bat Eptesicus fuscus: some consequences for echo processing. J. Acoust. Soc. Am. 89: 1402–1413, 1991.
METZNER, W. AND RADTKE-SCHULLER, S. The nuclei of the lateral lemniscus
in the rufous horseshoe bat, Rhinolophus rouxi. J. Comp. Physiol. A Sens.
Neural Behav. Physiol. 160: 395–411, 1987.
OLSEN, J. F. Functional Organization of the Medial Geniculate Body of the
Mustached Bat (Ph.D. thesis). St. Louis, MO: Washington Univ., 1986.
OLSEN, J. F. AND SUGA, N. Combination-sensitive neurons in the medial
geniculate body of the mustached bat: encoding of relative velocity information. J. Neurophysiol. 65: 1254–1274, 1991.
O’NEILL, W. E. Responses to pure tones and linear FM components of the
CF-FM biosonar signal by single units in the inferior colliculus of the
mustached bat. J. Comp. Physiol. 157: 797–815, 1985.
PALOMBI, P. S. AND CASPARY, D. M. GABAA receptor antagonist bicuculline alters response properties of posteroventral cochlear nucleus neurons.
J. Neurophysiol. 67: 738–746, 1992.
PHILLIPS, D. P., ORMAN, S. S., MUSICANT, A. D., AND WILSON, G. F. Neurons in the cat’s primary auditory cortex distinguished by their responses
to tones and wide-spectrum noise. Hear. Res. 18: 73–86, 1985.
POLLAK, G., HENSON, O. W., JR ., AND NOVICK, A. Cochlear microphonic
audiograms in the ‘‘pure tone’’ bat Chilonycteris parnellii parnellii. Science Wash. DC 176: 66–68, 1972.
RINVIK, E., OTTERSEN, O. P., AND STORM-MATHISEN, J. Gamma-aminobutyrate-like immunoreactivity in the thalamus of the cat. Neuroscience 21:
781–805, 1987.
RIQUIMAROUX, H., GAIONI, S. J., AND SUGA, N. Cortical computational maps
control auditory perception. Science Wash. DC 251: 565–568, 1991.
RÜBSAMEN, R. AND SCHÄFER, M. Ontogenesis of auditory fovea representation in the inferior colliculus of the Sri Lankan rufous horseshoe bat,
Rhinolophus rouxi. J. Comp. Physiol. A Sens. Neural Behav. Physiol.
167: 757–769, 1990.
SAITOH, I. AND SUGA, N. Long delay lines for ranging are created by
inhibition in the inferior colliculus of the mustached bat. J. Neurophysiol.
74: 1–11, 1995.
SALES, G. AND PYE, D. Ultrasonic Communication by Animals. New York:
Wiley, 1974, p. 230–233.
SAWAKI, S., SHIMOZAWA, T., BUTMAN, J. A., AND SUGA, N. Effects of
inhibitory amino acid antagonists on frequency tuning in the auditory
cortex of the mustached bat. Assoc. Res. Otolaryngol. Abstr. 420: 141,
1992.
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
ABELES, M. AND GOLDSTEIN, M. H., JR . Functional architecture in cat primary auditory cortex: columnar organization and organization according
to depth. J. Neurophysiol. 33: 172–187, 1970.
AITKIN, L. M. AND WEBSTER, W. R. Medial geniculate body of cat: organization and responses to tonal stimuli of neurons in ventral division. J.
Neurophysiol. 35: 365–380, 1972.
AOKI, E., SEMBA, R., KEINO, H., KATO, K., AND KASHIWAMATA, S. Glycinelike immunoreactivity in the rat auditory pathway. Brain Res. 442: 63–
71, 1988.
ARMSTRONG-JAMES, M. AND MILLAR, J. Carbon fiber microelectrodes. J.
Neurosci. Methods 1: 279–287, 1979.
BODENHAMER, R. D. AND POLLAK, G. Response characteristics of single
units in the inferior colliculus of mustache bats to sinusoidally frequency
modulated signals. J. Comp. Physiol. 153: 67–69, 1983.
CALFORD, M. B., WEBSTER, W. R., AND SEMPLE, M. M. Measurements of
frequency selectivity of single neurons in the central auditory pathway.
Hear. Res. 11: 395–401, 1983.
CASPARY, D. M., BACKOFF, P. M., FINLAYSON, P. G., AND PALOMBI, P. S.
Inhibitory inputs modulate discharge rate within frequency receptive
fields of anteroventral cochlear nucleus neurons. J. Neurophysiol. 72:
2124–2133, 1994.
CASSEDAY, J. H. AND COVEY, E. Frequency tuning properties of neurons in
the inferior colliculus of an FM bat. J. Comp. Physiol. A Sens. Neural
Behav. Physiol. 319: 34–50, 1992.
CONDON, C. J., WHITE, K. R., AND FENG, A. S. Processing of amplitudemodulated signals that mimic echoes from fluttering targets in the inferior
colliculus of the little brown bat, Myotis lucifugus. J. Neurophysiol. 71:
768–784, 1994.
EHRET, G. Left hemisphere advantage in the mouse brain for recognizing
ultrasonic communication cells. Nature Lond. 325: 249–251, 1987.
EHRET, G. AND MOFFAT, A.J.M. Inferior colliculus of the house mouse. II.
Single unit responses to tones, noise and tone-noise combinations as a
function of sound intensity. J. Comp. Physiol. 156: 619–635, 1985.
EVANS, E. F. The frequency response and other properties of single fibers
in the guinea-pig cochlear nerve. J. Physiol. Lond. 226: 263–287, 1972.
EVANS, E. F. AND NELSON, P. E. The responses of single neurons in the
cochlear nucleus of the cat as a function of their location and the anesthetic state. Exp. Brain Res. 17: 402–427, 1973.
EVANS, E. F. AND ZHAO, W. Varieties of inhibition in the processing and
control of processing in the mammalian cochlear nucleus. Prog. Brain
Res. 97: 117–126, 1993.
FITZPATRICK, D. C., KANAWAL, K. S., BUTMAN, J. A., AND SUGA, N. Combination-sensitive neurons, in the primary auditory cortex of the mustached
bat. J. Neurosci. 13: 931–940, 1993.
FUZESSERY, Z. M. Response selectivity for multiple dimensions of frequency sweeps in the pallid bat inferior colliculus. J. Neurophysiol. 72:
1061–1079, 1994.
FUZESSERY, Z. M. AND FENG, A. S. Frequency selectivity in the anuran
auditory midbrain: single unit responses to single and multiple tone stimulation. J. Comp. Physiol. A Sens. Neural Behav. Physiol. 146: 471–484,
1982.
GAIONI, S. J., RIQUIMAROUX, H., AND SUGA, N. Biosonar behavior of mustached bats swung on a pendulum prior to cortical ablation. J. Neurophysiol. 64: 1801–1817, 1990.
GODFREY, D. A., KIANG, N.Y.S., AND NORRIS, B. E. Single unit activity in
the dorsal cochlear nucleus of the cat. J. Comp. Neurol. 162: 169–284,
1975.
GOLDMAN, L. J. AND HENSON, O. W., JR . Prey recognition and selection
by the constant frequency bat, Pteronotus p. parnellii. Behav. Ecol. Sociobiol. 2: 411–419, 1977.
GREENWOOD, D. D. AND MARUYAMA, N. Excitatory and inhibitory response
2113
2114
N. SUGA, Y. ZHANG, AND J. YAN
/ 9k0f$$ap31
J864-5
tion in the auditory cortex of the mustached bat. J. Neurophysiol. 47:
225–255, 1982.
SUGA, N., NIWA, H., TANIGUCHI, I., AND MARGOLIASH, D. The personalized
auditory cortex of the mustached bat: adaptation for echolocation. J.
Neurophysiol. 58: 643–654, 1987.
SUGA, N., O’NEILL, W. E., KUJIRAI, K., AND MANABE, T. Specificity of
combination-sensitive neurons for processing of complex biosonar signals
in the auditory cortex of the mustached bat. J. Neurophysiol. 49: 1573–
1626, 1983.
SUGA, N., O’NEILL, W. E., AND MANABE, T. Harmonic-sensitive neurons
in the auditory cortex of the mustached bat. Science Wash. DC 203: 207–
274, 1979.
SUGA, N. AND SCHLEGEL, P. Coding and processing in the auditory systems
of FM-signal-producing bats. J. Acoust. Soc. Am. 54: 174–190, 1973.
SUGA, N., SIMMONS, J. A., AND JEN, P. H. S. Peripheral specialization for
fine analysis of Doppler-shifted echoes in the auditory system of the
‘‘CF-FM’’ bat Pteronotus parnellii. J. Exp. Biol. 63: 161–192, 1975.
SUGA, N. AND TSUZUKI, K. Inhibition and level-tolerant frequency tuning
in the auditory cortex of the mustached bat. J. Neurophysiol. 53: 1109–
1145, 1985.
SUTTER, M. L. AND SCHREINER, C. E. Physiology and topography of neurons
with multipeaked tuning curves in cat primary auditory cortex. J. Neurophysiol. 65: 1207–1226, 1991.
VATER, M., HABBICHT, H., KÖ SSL, M., AND GROTHE, B. The functional
role of GABA and glycine in monaural and binaural processing in the
inferior colliculus of horseshoe bat. J. Comp. Physiol. A Sens. Neural
Behav. Physiol. 171: 541–553, 1992.
WENSTRUP, J. J., LARUE, D. T., AND WINER, J. A. Projections of physiologically defined subdivisions of the inferior colliculus in the mustached bat:
targets in the medial geniculate body and extrathalamic nuclei. J. Comp.
Neurol. 346: 207–263, 1994.
WENTHOLD, R. J. AND HUNTER, C. Immunocytochemistry of glycine and
glycine receptors in the central auditory system. In: Glycine Neurotransmission, edited by O. P. Ottersen and J. Storm-Mathisen. New York:
Wiley, 1990, p. 391–416.
WINER, J. A. AND LARUE, D. T. Anatomy of glutamic acid decarboxylase
(GAD) immunoreactive neurons and axons in the rat medial geniculate
body. J. Comp. Neurol. 278: 47–68, 1988.
WINER, J. A., WENSTRUP, J. J., AND LARUE, D. T. Patterns of GABAergic
immunoreactivity define subdivisions of the mustached bat’s medial geniculate body. J. Comp. Neurol. 319: 172–190, 1992.
YANG, L., POLLAK, G. D., AND RESLER, C. GABAergic circuits sharpen
tuning curves and modify response properties in the mustache bat inferior
colliculus. J. Neurophysiol. 68: 1760–1774, 1992.
YOUNG, E. D. AND BROWNELL, W. E. Responses to tones and noise of single
cells in dorsal cochlear nucleus of unanesthetized cats. J. Neurophysiol.
39: 282–300, 1976.
08-27-97 15:12:36
neupal
LP-Neurophys
Downloaded from http://jn.physiology.org/ by 10.220.33.6 on June 14, 2017
SCHNITZLER, H. U. Die Ultraschall-Ortungslaute der Hufeisen-Fledermause
(Chiroptera-Rhinolophidae) in verschiedenen Orientierungs-situationen.
Zool. Vergl. Physiol. 57: 376–408, 1968.
SCHNITZLER, H. U. Echoortung bei der Fledermaus Chilonycteris rubiginosa. Zool. Vergl. Physiol. 68: 25–38, 1970.
SCHREINER, C. E. AND MENDELSON, J. R. Functional topography of cat primary auditory cortex: distribution of integrated excitation. J. Neurophysiol. 64: 1442–1459, 1990.
SCHREINER, C. E. AND SUTTER, M. L. Topography of excitatory bandwidth
in cat primary auditory cortex: single-neuron versus multiple-neuron recordings. J. Neurophysiol. 68: 1487–1502, 1992.
SCHULLER, G., BEUTER, K., AND SCHNITZLER, H. U. Response to frequencyshifted artificial echoes in the bat Rhinolophus ferrumequinum. J. Comp.
Physiol. 89: 275–286, 1974.
SPIROU, G. A. AND YOUNG, E. D. Organization of dorsal cochlear nucleus
type IV unit response maps and their relationship to activation by bandlimited noise. J. Neurophysiol. 66: 1750–1768, 1991.
SUGA, N. Single unit activity in cochlear nucleus and inferior colliculus of
echolocating bats. J. Physiol. Lond. 172: 449–474, 1964a.
SUGA, N. Recovery cycles and responses to frequency modulated tone
pulses in auditory neurons of echolocating bats. J. Physiol. Lond. 175:
50–80, 1964b.
SUGA, N. Analysis of frequency modulated sounds by neurons of echolocating bats. J. Physiol. Lond. 179: 26–53, 1965a.
SUGA, N. Functional properties of auditory neurons in the cortex of echolocating bats. J. Physiol. Lond. 181: 671–700, 1965b.
SUGA, N. Analysis of frequency-modulated and complex sounds by single
auditory neurons of bats. J. Physiol. Lond. 198: 51–80, 1968.
SUGA, N. Classification of inferior collicular neurons of bats in terms of
responses to pure tones, FM sounds and noise bursts. J. Physiol. Lond.
200: 555–574, 1969.
SUGA, N. Responses of inferior collicular neurons of bats to tone bursts
with different rise times. J. Physiol. Lond. 217: 159–177, 1971.
SUGA, N. Amplitude-spectrum representation in the Doppler-shifted-CF
processing area of the auditory cortex of the mustached bat. Science
Wash. DC 196: 64–67, 1977.
SUGA, N. Cortical computational maps for auditory imaging. Neural Networks 3: 3–21, 1990.
SUGA, N. Sharpening of frequency tuning by inhibition in the central auditory system: tribute to Yasuji Katsuki. Neurosci. Rec. 21: 287–299, 1995.
SUGA, N. AND JEN, P.H.S. Disproportionate tonotopic representation for
processing species-specific CF-FM sonar signals in the mustache bat
auditory cortex. Science Wash. DC 194: 542–544, 1976.
SUGA, N. AND JEN, P.H.S. Further studies on the peripheral auditory system
of ‘‘CF-FM’’ bats specialized for fine frequency analysis of Dopplershifted echoes. J. Exp. Biol. 69: 207–232, 1977.
SUGA, N. AND MANABE, T. Neural basis of amplitude-spectrum representa-