Departments of 1Biomedical Engineering and 2Otolaryngology-Head and Neck Surgery and Center for Hearing and Balance, Johns Hopkins University, Baltimore, Maryland 21205
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Ramachandran, Ramnarayan, Kevin A. Davis, and Bradford J. May. Single-Unit Responses in the Inferior Colliculus of Decerebrate Cats I. Classification Based on Frequency Response Maps. J. Neurophysiol. 82: 152-163, 1999. This study proposes a classification system for neurons in the central nucleus of the inferior colliculus (ICC) that is based on excitation and inhibition patterns of single-unit responses in decerebrate cats. The decerebrate preparation allowed extensive characterization of physiological response types without the confounding effects of anesthesia. The tone-driven discharge rates of individual units were measured across a range of frequencies and levels to map excitatory and inhibitory response areas for contralateral monaural stimulation. The resulting frequency response maps can be grouped into the following three populations: type V maps exhibit a wide V-shaped excitatory area and no inhibition; type I maps show a more restricted I-shaped region of excitation that is flanked by inhibition at lower and higher frequencies; and type O maps display an O-shaped island of excitation at low stimulus levels that is bounded by inhibition at higher levels. Units that produce a type V map typically have a low best frequency (BF: the most sensitive frequency), a low rate of spontaneous activity, and monotonic rate-level functions for both BF tones and broadband noise. Type I and type O units have BFs that span the cat's range of audible frequencies and high rates of spontaneous activity. Like type V units, type I units are excited by BF tones and noise at all levels, but their rate-level functions may become nonmonotonic at high levels. Type O units are inhibited by BF tones and noise at high levels. The existence of distinct response types is consistent with a conceptual model in which the unit types receive dominant inputs from different sources and shows that these functionally segregated pathways are specialized to play complementary roles in the processing of auditory information.
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The central nucleus of the inferior colliculus
(ICC) receives converging projections from most, if not all, of the
auditory nuclei in the brain stem and, in turn, provides nearly all of the input to the thalamocortical pathway (for review, Irvine
1986; Oliver and Huerta 1992
; Oliver and
Shneiderman 1991
). Afferent projections to the ICC may be
excitatory (e.g., cochlear nucleus, Oliver 1987
;
Semple and Aitkin 1980
; superior olive,
Glendenning et al. 1992
) or inhibitory (e.g., dorsal
nucleus of lateral lemniscus) (Adams and Mugnaini 1984
;
Roberts and Ribak 1987
; Shneiderman and Oliver
1989
; Shneiderman et al. 1988
, 1993
). The manner
in which these inputs combine creates functionally distinct synaptic domains within ICC and presumably basic differences in ICC response types (Aitkin and Schuck 1985
; Brunso-Bechtold et
al. 1981
; Maffi and Aitkin 1987
; Oliver
and Huerta 1992
; Oliver et al. 1997
; Roth et al. 1978
; Ryugo et al. 1981
;
Shneiderman and Henkel 1987
; Zook and Casseday
1987
).
ICC neurons have been classified previously according to excitatory
response patterns, which are mapped in the frequency domain by
recording single-unit responses to tone bursts across a range of
frequencies and sound pressure levels (Aitkin et al.
1975; Bock et al. 1972
; Ehret and
Merzenich 1988
; Fuzessery and Hall 1996
;
Ramachandran et al. 1997
; Rose et al.
1963
; Ryan and Miller 1978
; Wang et al.
1996
; Yang et al. 1992
). The resulting frequency response maps exhibit three general patterns of excitatory tuning: open
tuning curves that are V-shaped; level-tolerant tuning curves that
remain narrow with level; and upper threshold (or closed) tuning curves
that respond to a circumscribed range of stimulus frequencies and
levels. At the present time, the inhibitory influences that shape ICC
frequency response maps are not well characterized because most prior
studies were conducted with anesthetized preparations that lack the
necessary spontaneous activity for a direct demonstration of inhibition.
This report provides the first extensive evaluation of the inhibitory
patterns of ICC neurons by sampling single-unit responses in
unanesthetized, decerebrate cats. Results obtained with the decerebrate
preparation support previous observations that there are three basic
types of frequency response maps in ICC; in addition, it is shown that
these response types differ mainly in the strength and distribution of
inhibitory inputs. Frequency response maps of units with open tuning
curves exhibit no inhibition (so-called type V units); the narrow
bandwidth of units with level-tolerant tuning curves is flanked by
strong inhibition (type I units); and units with closed tuning curves
show predominantly inhibitory responses at levels above their islands
of excitation (type O units). Other basic differences between these ICC
response types include rates of spontaneous activity, shapes of BF tone
and noise rate-level functions, and relative responsiveness to
narrowband versus broadband stimuli. A companion paper (Davis et
al. 1999) describes the binaural response properties of ICC
neurons in decerebrate cats.
![]() |
METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Surgical preparation
Adult male cats (n = 8) with clean external ears and middle ears free of infection were used under institutional animal care guidelines. Cats were anesthetized with xylazine (2 mg im) and ketamine (initial dose 40 mg/kg im; supplemental doses 15 mg/kg iv) and given atropine sulfate (0.1 mg im) to minimize mucous secretions. The cephalic vein was cannulated to allow intravenous infusions of fluids, and a tracheotomy was performed to facilitate quiet breathing. A core temperature of 39°C was maintained with a regulated heating pad. A midline incision was made over the skull and the temporalis muscles reflected to visualize the top of the skull and the ear canals. A craniotomy (~1 cm wide) was performed over parietal cortex, then the underlying cortex and brain stem (between the thalamus and superior colliculus) was aspirated to create a complete supracollicular decerebration; anesthesia was discontinued. The ear canals were transected near the tympanic annulus to accept hollow ear bars for delivering closed-field acoustic stimuli. The bullae were vented with ~40 cm of PE 200 tubing to prevent build-up of pressure in the middle ear. The cat's head was secured in a stereotaxic frame. A second craniotomy was performed over occipital cortex and tissue overlying the IC was aspirated to visualize the recording site. In most subjects, complete access to the ICC required partial removal of the tentorium.
Gallamine triethiodide was administered to some cats (n = 4) to reduce brain pulsations that compromised the stability of recording. These subjects were placed on a mechanical respirator and maintained at an end-tidal C02 of 4%. Gallamine paralysis was never induced within 8 h of the initial surgical preparation to ensure that the decerebration procedure was complete (as judged by lack of voluntary movements).
The cat was euthanized with an overdose of pentobarbital sodium (26 mg/kg iv) at the end of the recording session (48-72 h after start of surgery). The brain was perfused transcardially with phosphate buffered saline and fixed with 10% formalin in phosphate buffer. Completeness of decerebration was verified by examination of the postfixed brain. Locations of electrode tracks were verified histologically from patterns of gliosis and electrolytic lesions.
Recording protocol
Electrophysiological recordings were made in a sound-attenuating chamber. Stimuli were delivered via electrostatic speakers that were coupled to the hollow ear bars; this closed acoustic system produced a uniform response (±5 dB) at frequencies from 40 Hz to 40 kHz. Acoustic calibrations were performed with a probe tube microphone that inserted into the ear bars near the tympanic membrane. All test stimuli were 200 ms in duration with 10-ms rise/fall times; presentation rates were 1 burst/s. Unit activity was recorded with platinum-iridium microelectrodes. The electrode signal was amplified 10,000-30,000 times and low-pass filtered at 5 kHz. A variable threshold Schmitt trigger was used to discriminate action potentials; spike times were recorded relative to stimulus onset. Discharge rates are described in terms of driven rates; that is, the total response rate during a stimulus presentation minus spontaneous activity. Response rates were computed during the final 150 ms of the stimulus-on interval (to reflect steady-state responses) and spontaneous rates were measured during the final 400 ms of the stimulus-off interval of each 1-s stimulation period. This choice of response window eliminated most effects of adaptation; informal analyses suggest that response magnitudes, and not the general patterns of excitatory/inhibitory responses, were affected by changes in the duration of the response window.
Electrodes were advanced dorsoventrally into the IC using a
remote-controlled hydraulic micromanipulator (Kopf Instruments, CA).
Search stimuli included 50-ms tone or noise bursts presented to the
contralateral ear. As the electrode advanced, the frequency that
excited background activity first decreased (usually from 10 to 20 kHz
at the surface to <1 kHz) and then increased. This reversal in the
sequence of BFs, observed 1-2 mm below the surface of the IC,
indicated entry into ICC (Aitkin et al. 1975;
Merzenich and Reid 1974
). In addition, units in ICC
tended to exhibit simple spikes and have sharp tuning, whereas units
outside the ICC (e.g., in dorsal cortex, ICD, or the external nucleus,
ICX) tended to respond with complex spikes, show offset responses, have
long latencies, and exhibit poor tuning.
When a single ICC unit was isolated, its BF was determined audiovisually and the following characterization protocol was initiated. Rate-level functions were obtained by sweeping the level of BF tones or broadband noise bursts over a 100-dB range. Frequency response maps were created by sweeping the frequency of tone bursts over a four octave range centered on unit BF. These frequency sweeps were presented at multiple sound levels, ranging from ~10 dB below to ~70 dB above threshold. Each frequency-intensity combination was presented once.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Results are based on 146 units from eight cats. Within this sample, 56 units were localized to the ICC based on histological reconstruction of electrode tracks; the remaining units were inferred to be in the ICC based on the sequence of BFs within tracks and the similarity of their response properties to those of histologically localized units. The photomicrograph in Fig. 1 shows one recording track through the IC; the reconstructed track is marked by a line and the electrolytic lesion is enclosed by a circle. In this track, the electrode was placed on the caudal aspect of the IC and passed through the ICD before entering the ICC. Other recovered electrode tracks passed through the ICX before entering the ICC, suggesting that the full extent of the ICC was sampled. Response properties of ICC units fell into two broad categories: units with sustained discharges (n = 134) and units that responded with a single onset spike at all frequencies and levels (n = 12). Responses of onset units are not considered further in this report.
|
Frequency response maps
Units with sustained responses were divided into three groups
based on the patterns of excitation and inhibition revealed in their
frequency response maps (Table 1).
Representative data for each unit type are shown in Fig.
2. In these plots, excitatory areas ()
are defined as stimulus conditions that elicited responses
1 SD above
spontaneous activity; similarly, inhibitory areas (
) indicate tone-driven
rates
1 SD below spontaneous activity. Type V units (top)
have a V-shaped excitatory area that widens about unit BF (vertical
line) with increasing sound levels. These units do not show inhibitory
responses to pure tones. Type I units (middle) generally
have an I-shaped excitatory area that maintains its sharp tuning at
higher levels; this level tolerant excitatory area is flanked on both
sides by wide inhibitory areas. Some (predominantly low-BF) type I
units show less pronounced inhibitory effects at lower frequencies and
thus exhibit more V-shaped excitatory areas; nonetheless, these units
are easily distinguishable from type V units based on the presence of
strong above-BF inhibition. Type O units (bottom) are
characterized by an O-shaped island of excitation around BF threshold
that gives way to inhibition at higher sound levels. Type O units may
exhibit additional excitatory areas, but the frequency location of
these responses is highly variable between units. Type V units were the
least abundant unit type in our sample (16/134); type O units were the
most prevalent (71/134).
|
|
Figure 3 shows the distribution of BFs
for the three response types. The BFs of type V units were always low
(<3 kHz), whereas the BFs of type I and type O units spanned most of
the cat's range of audible frequencies. As in previous studies
(Aitkin et al. 1975; Merzenich and Reid
1974
), BFs increased as the electrode advanced from the dorsal
to ventral limits of the ICC. Consequently, the unit types were not
distributed uniformly across the dorsoventral axis of the ICC. Type V
units typically were recorded during the initial dorsal progression of
the electrode track, and type I units were recorded more ventrally.
Type O units dominated our sample and could be found throughout the
course of most tracks.
|
ICC unit types show differences in the range of frequencies that evoke
excitatory responses. Frequency tuning was assessed by calculating each
unit's Q values (BF divided by bandwidth of the excitatory area in the
frequency response map). At 10 dB above threshold (Fig.
4A), Q10
values increase with BF until ~10 kHz, after which they remain
relatively unchanged. Most data points fall within the range of values
recorded from auditory-nerve fibers (ANFs) in our laboratory
(Calhoun et al. 1997; Miller et al.
1997
); therefore, low-level frequency tuning in the ICC appears
to be determined by peripheral processes. At 40 dB above threshold
(Fig. 4B), type V units continue to follow the tuning
properties of ANFs (data taken from Liberman 1978
),
while type O units show no excitation (i.e., nonexistent
Q40 values). The lateral inhibitory sidebands of
type I units maintain a sharply tuned excitatory area at levels well
above threshold, as indicated by their comparable Q40 and Q10 values. A few
predominantly low-BF type I units showed reduced low-frequency
inhibition, which allowed the bandwidth of their excitatory areas to
expand to produce Q40 values seen in the auditory
nerve.
|
Traditional Q measures provide an adequate description of the strictly
excitatory response areas of ANFs but fail to capture the frequency
tuning properties of the inhibitory responses that characterize type I
and O units. The bandwidths of these inhibitory tuning curves were
quantified by calculating Q40 values across the
total width of the lower and upper inhibitory response areas, assuming
that they form a single V-shaped area centered on BF. This metric is
designated the Q40 of inhibition and is plotted in Fig. 4C; for comparison, excitatory
Q40 values of ANFs also are shown. In general,
the Q40 of inhibition for both type I and type O
units was a third of ANF Q40 values. This broad
bandwidth is presumably established by the convergence of multiple
inhibitory inputs with different BFs. It is likely that these events
occur at a lower level of auditory processing because inhibitory
bandwidth is only slightly changed by pharmacological manipulations
within the ICC (Davis et al. 1999).
Responses to BF tones
ON-BF responses were characterized by obtaining rate-level functions for BF tone bursts. Figure 5A shows representative data for each unit type; for display purposes, the functions have been smoothed with a triangularly weighted moving window average. Responses of the representative type V unit climb to a maximum and then maintain a steady discharge rate at higher stimulus levels. By contrast, the rate-level function of the type O unit is strongly nonmonotonic; as the level of the tone grows, the unit exhibits a rate increase followed by strong inhibition. The type I unit also exhibits nonmonotonicity at higher tone levels, but unlike the type O unit, the neuron does not show actual inhibitory effects at BF (i.e., the decline in driven rate does not fall below levels of spontaneous activity).
|
The degree of nonmonotonicity shown by each unit can be quantified by
calculating the normalized slope of the saturated component of BF
rate-level functions (Davis et al. 1996; Young
and Voigt 1982
). This measure involves normalizing response
rates by the maximum rate of each function and then fitting a best fit
line (least squares) to the portion of the curve between the maximum (or the first sharp change in slope) and the next inflection point (or
end of the data). As illustrated on the rate-level function of the type
V unit in Fig. 5A (- - -), the best fit line included data
spanning
10 dB. Units having more negative normalized slopes exhibit
stronger nonmonotonicity.
Figure 5B shows the distribution of normalized slopes for each unit type; median values are marked (*). The typical saturating monotonicity of type V units is indicated by the clustering of values near 0. A majority of type I units displayed monotonic properties that were quite similar to those of type V units, although slopes in this response type also could be nonmonotonic (negative values). Type O units, on the other hand, exhibited exclusively nonmonotonic rate-level functions. Nonparametric statistical tests indicated that rate-level functions of type I units were significantly less nonmonotonic than those of type O units (P < 0.001, Mann Whitney U test). Type V units were excluded from these statistical analyses because of their small sample size; however, Fig. 5B clearly shows the opposite polarity of type V versus type O responses at high stimulus levels (- - -).
Rate-level functions like those in Fig. 5A also can be used to define basic trends in spontaneous rate, BF threshold, and maximum firing rate for units in each response type; these data are summarized in Table 1. As shown in Fig. 6, most type V units (10/16) exhibited spontaneous rates <5 spikes/s; whereas, most type I (38/47) and type O (63/71) units had spontaneous rates >5 spikes/s.
|
The relationship between the absolute threshold and BF of ICC units was
essentially identical to trends previously observed in the auditory
nerve (Fig. 7A), except that
the range of thresholds at a particular BF was more compressed in ICC
and tended to fall along the best threshold curve of ANFs (shown by
line) (Calhoun et al. 1997; Miller et al.
1997
). These results suggest that the lower limits of auditory
sensitivity for CNS neurons are established in the periphery. Further
support for this interpretation is provided by histograms in Fig.
7B, which plot the distribution of ICC thresholds relative
to the ANF best threshold curve at the same BF; negative values
indicate ICC units with lower thresholds than their auditory nerve
counterparts. Type V units tended to fall among the least sensitive
units; most ICC neurons in this response type (13/16) had thresholds
5 dB above the auditory nerve best threshold curve. In contrast, many
type I and type O units displayed relative thresholds that were <5 dB
above the best threshold curve of ANFs. A few of these units produced
thresholds that were
5 dB below the most sensitive auditory nerve
responses. This apparent hypersensitivity most likely reflects
differences in threshold criteria between previous ANF studies and
present methods: ANF thresholds were obtained from tuning curves based
on a 20-spikes/s rate increase criteria; ICC unit thresholds were
obtained from rate-level curves based on a rate change of 1 SD from
spontaneous rate.
|
The differing amounts of ON-BF inhibition shown by response
maps in Fig. 2, and rate-level functions in Fig. 5A can
influence the maximum driven rates elicited by BF tones. Figure
8 shows the distribution of maximum BF
tone-driven rates for each unit type. Strong inhibitory effects in type
O units were correlated with low maximum driven rates (median 34 spikes/s); while the less inhibited type I units achieved much higher
rates (median 101 spikes/s). Interestingly, some type V units reached
driven rates approaching 200 spikes/s, whereas others exhibited maximum driven rates <100 spikes/s. This latter result is somewhat surprising given the fact that type V units are the only response type to show no
obvious signs of inhibition. Two properties of type V units may account
for this variability in responsiveness to contralateral BF tones:
first, their maximum discharge rate has a tendency to shift to lower
frequencies with increasing sound level (Fig. 2), and second, many type
V units respond maximally to ipsilateral stimulation (Davis et
al. 1999). For the subset of five type V neurons that are known
to be contralaterally dominated (
in histogram), the median maximum
discharge rate for contralateral stimulation is 152 spikes/s.
|
Responses to broadband noise
Almost all ICC units (118/122) responded to bursts of broadband noise, and generally, the response functions were similar in shape to those evoked by BF tones. Figure 9A shows representative noise rate-level functions for each unit type. As for BF tones, the feature that distinguishes the noise rate-level functions of each unit type is the degree of nonmonotonicity at high stimulus levels. The function for the type V unit shows a monotonically increasing discharge rate. The noise-driven rates of the type O unit exhibit a severe nonmonotonicity that leads to strong inhibition at levels >20 dB above threshold. The noise rate-level function of the typical type I unit is monotonic and saturating.
|
The distribution of normalized slopes for each unit type is shown in Fig. 9B. Values for 11/16 type V units fall near zero, which is a reflection of the minimal inhibition that is apparent in type V response maps. The five remaining type V units showed high noise thresholds and failed to reach saturation at the highest noise levels tested (e.g., Fig. 9A); the slope value is undefined for these units and is plotted off the axis. Monotonic noise rate-level functions were seen in 32/41 type I units (units to the right of - - - ); the remaining units in this response class exhibited nonmonotonicity at higher noise levels, but discharge rates never fell to levels approaching spontaneous activity. For the majority of type O units (59/65), high levels of noise produced sufficient inhibition to reduce noise-driven rates down to or below levels of spontaneous activity; consequently, rate-level functions for this unit class tended to be highly nonmonotonic (41/65 with slopes below - - -). Nonparametric statistical tests indicated that noise rate-level functions of type O units were significantly more nonmonotonic than those of type I units (P < 0.001, Mann Whitney U test).
The magnitude of nonmonotonicity for each unit's noise and BF
tone-driven responses is compared in Fig.
10A, which plots the normalized slope of the noise rate-level function against values obtained with BF tones; the represents equal slopes. Units with undefined noise slopes (e.g., the type V unit in Fig. 9A)
are plotted above the axis. Note that the majority of the data points for all three unit types are broadly distributed along the equity line,
suggesting that the shape of a unit's noise response is largely
predictable from its tone response. In particular, a unit's noise
response is usually more monotonic than its tone response (the data
points lie above the equity line; P < 0.001, sign
test). However, some units show very different degrees of
nonmonotonicity for tones and noise. For example, some type O and type
I units produced strongly nonmonotonic rate-level functions for tones but not for broadband noise.
|
Thresholds for noise responses are plotted as a function of BF in Fig.
10B. To allow a comparison with tone thresholds (data from
Fig. 7A), noise thresholds are specified in terms of the total noise power in the bandwidth of each unit 10 dB above threshold. The overall distribution of noise thresholds shows a tendency to
decrease as BF increases, which is similar to the pattern observed in
the ICC of anesthetized cats (Ehret and Merzenich 1988).
The lowest noise thresholds were observed at BFs between 8 and 30 kHz;
whereas, very sensitive tone thresholds extended from 1 to 12 kHz.
These trends suggest a divergence of noise versus tone sensitivity at
low frequencies; however, for the majority of units, tone thresholds
provided a good estimation of noise thresholds, as indicated by the
scatterplot of tone versus noise thresholds in Fig. 10D. In
general, neurons in the type O response class were the most sensitive
units for both tones and noise. As was observed for BF tones, type V
units tended to show relatively high noise thresholds in particular
when compared with type O units with similar BFs.
Each unit's maximum discharge rate for noise is plotted against its maximum tone-driven rate in Fig. 10C. Type I units showed significantly lower discharge rates when tested with noise than with tones (P < 0.001, sign test); this reduced responsiveness is most likely due to the effects of noise energy falling into the inhibitory sidebands that surround the narrow excitatory area of type I units. Type V units, where lateral inhibitory effects are presumed to be weak or nonexistent, exhibited no significant difference in maximum driven rates for tones and noise (P > 0.05, sign test). The maximum noise-driven rates of type O units were also not statistically different from their maximum tone-driven rates (P > 0.05, sign test); however, the scatter in these data suggests groups of units that may be biased toward either broadband or narrowband stimuli. Ten type O units showed excitatory responses when tested with tonal stimuli but were inhibited (8 cases) or unresponsive (2 cases) to noise; these data are plotted along the x axis in Fig. 10C.
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The response types in this report were identified with monaural
stimulation of the contralateral ear. Other studies have proposed alternative classification systems that are based on patterns of
excitation and inhibition for binaural stimuli (Aitkin et al. 1984; Irvine 1986
for reviews). Davis et
al. (1999)
investigates the binaural response properties of ICC
neurons in decerebrate cats and describes the systematic relationships
between the two classification schemes.
Does the decerebrate preparation yield normal physiological response types?
The advantage of the decerebration procedure is that it provides a preparation for long-term stable recording without the confounding effects of anesthesia. The decerebration techniques used in this study were directed toward the thalamus, which is the primary target of ICC projections. An important concern with this surgical approach is that the decerebration procedure may have damaged the distal axons of our neural sample and therefore altered their physiological properties. One extreme consequence of axotomy is neuronal degeneration. When sections through the ICC of three cats were examined under light microscope, normal-looking cell bodies were observed throughout the inferior colliculus and gliosis was restricted to locations of electrode tracks (Fig. 1).
It is possible that axotomy altered the response properties of ICC
units without producing obvious signs of neural degeneration. Several
studies have looked at the effects of acute axotomy on neurons in
sensory ganglia (Czeh et al. 1977; Gallego et al.
1987
; for review, see Titmus and Faber 1990
) or
CNS (Faber and Zottoli 1981
; Kuwada and Wine
1981
); few changes in electrophysiological properties were
noted during time courses that extended well beyond the duration of our
recording experiments (48-72 h). Our own results support those
observations in that no obvious differences were seen in the response
properties of units studied near the beginning or end of experiments.
As a further test of the reliability of the decerebration preparation,
control experiments were conducted without decerebration in intact
barbiturate-anesthetized cats. Figure
11 provides examples of the three types
of frequency response maps that were obtained in these experiments. The
maps clearly show the basic excitatory response properties that
constitute the classification system proposed in Fig. 2. That is, units
with open tuning curves (top) have V-shaped excitatory areas
that are only as sharp as ANFs at 40 dB re threshold
(Q40 = 0.7); level tolerant units
(middle) have I-shaped excitatory areas that maintain their
sharp tuning at higher levels (Q40 = 3.2 > ANFs); and upper threshold units (bottom) have O-shaped
islands of excitation around BF threshold that give way at higher
levels to regions of little-or-no response. The upper threshold unit in
Fig. 11 also shows a region of excitation above BF; this feature is
shared, in part, by the type O unit in Fig. 2. The conspicuous absence
of inhibitory responses for the level tolerant and upper threshold
units is a side effect of anesthesia. Other electrophysiological
studies of ICC units in anesthetized cats show similar distributions of
the shape and bandwidth of excitatory response areas (Aitkin et
al. 1975; Ehret and Merzenich 1988
;
Ramachandran et al. 1997
; Wang et al.
1996
).
|
Another detrimental effect of the decerebration procedure is that it
disrupts descending inputs to the inferior colliculus. Anatomic studies
suggest that the ICC of the cat receives only sparse projections from
the thalamus and little, or no, direct input from the cortex
(Adams 1980; Andersen et al. 1980
; for a review, see Huffman and Henson 1990
; Oliver and
Huerta 1992
). Effects of de-efferentation are best assessed by
comparing our results to the limited ICC data that have been obtained
in intact, awake preparations. Excitatory tuning characteristics in the
ICC of decerebrate cats are very similar to results obtained in awake bats (Yang et al. 1992
) and primates (Ryan and
Miller 1978
). The distribution of spontaneous rates in
decerebrate cats (Fig. 6) also matches that seen in awake cats
(Bock and Webster 1974
; Bock et al. 1972
)
and primates (Ryan and Miller 1978
). Patterns of inhibition resemble those observed in awake primates (Ryan and Miller 1978
). Although these earlier studies did not address
the relationship between response pattern and BF, it is interesting that all examples of units with broad tuning at higher levels had low
BFs, whereas the level tolerant units and the upper threshold units had
high BFs. Our population of type V units was confined to frequencies
<3 kHz, whereas the type I and type O units were more prevalent at
higher frequencies (Fig. 3). The good agreement of results obtained in
awake and decerebrate animals lends additional support to the
assumption that our surgical procedure did not introduce fundamental
changes in ICC discharge patterns.
Origins of the unit types
Neurons in the ICC can be grouped into two major anatomic classes
based on somatic shape and dendritic distribution. Disc-shaped cells
have compact dendritic fields that are oriented along the path of
lemniscal fibers; consequently, these cells are thought to receive
inputs from afferent neurons with similar BFs. Stellate cells have more
broadly distributed dendrites that presumably integrate inputs across a
wide range of BFs (Oliver and Morest 1984). A similar
morphological dichotomy has been described for two classes of stellate
cells (called planar and radiate) in the anteroventral cochlear nucleus
(AVCN) (Doucet and Ryugo 1997
). Intracellular studies
suggest that AVCN planar neurons are sharply tuned in frequency and
show sustained responses to tones, whereas radiate neurons are broadly
tuned and show onset responses (Smith and Rhode 1989
).
If a similar morphological/physiological correlation holds in ICC, then
sustained responses may reflect the activity of disc-shaped cells,
whereas onset responses may originate from stellate cells. This
interpretation is given some credence by the independent observation
that anatomically defined stellate cells and physiologically defined
onset-responders represent a small minority of ICC neurons.
Given the assumption that sustained responses in the ICC originate from
the same population of disc-shaped cells, it is likely that functional
differences between the three unit types reflect heterogeneity in their
inputs. Principal cells in the medial superior olive (MSO) lateral
superior olive (LSO), and cochlear nuclei have unique response patterns
and are primary sources of innervation for the ICC. When anterograde
tracers are injected into LSO and dorsal cochlear nucleus (DCN)
(Oliver et al. 1997), labeled axons from the two sources
remain segregated in some regions of the ICC but overlap at other
locations. This topographical organization suggests that synaptic
domains within the ICC are created to perform the dual role of relaying
information from individual sources of input to higher processing
levels and integrating information across auditory nuclei
(Semple and Aitkin 1981
). From this perspective, responses of disc-shaped cells may be expected to reflect a mixture of
the physiological properties of their ascending inputs and may even be
essentially identical to their inputs within regions of the ICC where
anterograde labeling indicates little convergence (Oliver and
Huerta 1992
).
Several lines of experimental evidence lead to the conjecture that type
V response patterns are strongly shaped by inputs from the MSO. Type V
units have broad excitatory areas without overt inhibition, low
spontaneous discharge rates, and relatively high thresholds. Of the
primary sources of input for inferior colliculus, these characteristics
most closely match the response properties of MSO units
(Goldberg and Brown 1969; Guinan et al. 1972
). More than 50% of the MSO is devoted to the processing
of frequencies <4 kHz; type V units in the ICC also reflect a strong bias toward low-frequency sensitivity. Injections of anterograde autoradiographic tracers in dorsal MSO (i.e., the low-frequency half of
the nucleus) produce tonotopically organized labeling in dorsal-lateral
ICC (Aitkin and Schuck 1985
; Henkel and Spangler 1983
); in the present study, type V units were recorded
primarily at low BFs during the initial dorsal penetration of electrode tracks. Although most MSO neurons project to the ipsilateral ICC, binaural response properties are established before this one-sided distribution of MSO input. Like MSO neurons, type V units in the ICC
show exclusively excitatory responses when stimulated with BF tones in
either ear (EE) and are more sensitive to binaural than to monaural
stimuli (Davis et al. 1999
).
Type I units have high BFs and level tolerant excitatory areas that are
flanked by inhibitory areas. These physiological characteristics suggest that the LSO is the primary source of ascending inputs for
neurons of this response type (Caird and Klinke 1983;
Guinan et al. 1972
; Tsuchitani and Boudreau
1966
), although the significant projection from stellate cells
in the contralateral AVCN cannot be discounted as an additional or
alternate source of ascending input (Adams 1979
, 1983
;
Cant 1982
; Osen 1972
; Ryugo et al.
1981
; Shofner and Young 1985
). Rate-level
functions of type I units show a mixture of monotonic and nonmonotonic
responses that also is seen in the LSO of decerebrate cats
(Brownell et al. 1979
). In addition, type I units are
excited by contralateral tones but inhibited by ipsilateral tones (EI)
(Davis et al. 1999
); neurons with these so-called
EI binaural properties are found mainly in LSO (Guinan
et al. 1972
; Tsuchitani and Boudreau 1969
).
Type O units have high BFs and receptive fields dominated by inhibition
at high levels. These properties suggest that the inputs to this unit
type are derived from the DCN (Spirou and Young 1991;
Young and Brownell 1976
). The ICC receives direct inputs
from pyramidal and giant cells in the DCN (Adams 1979
; Osen 1972
; Ryugo et al. 1981
). Like type
O units in ICC, the DCN projection neurons have frequency response maps
that show consistent BF excitation only at stimulus levels near
threshold (the so-called type IV unit response). As a result, highly
nonmonotonic rate-level functions are observed for both type IV and
type O units if BF tones are used as stimuli. Although the binaural
properties of the DCN have not been studied in great detail
(Mast 1970
, 1973
), type IV units can show both
EE and EI response patterns, which are also a property of type
O units (Davis et al. 1999
). Despite these basic
similarities, the preponderance of type O units in our sample is at
odds with the relatively minor DCN projection to ICC (Adams
1979
). This apparent discrepancy suggests either a sampling
bias (e.g., type O units are recorded from the larger disc-shaped
neurons) (Oliver and Morest 1984
) or that DCN afferents ramify widely in ICC (Oliver et al. 1997
).
Alternatively, these data suggest that type O unit properties are also
created at the level of the ICC by a suitable convergence of excitatory
and inhibitory inputs (Yang et al. 1992
).
Signal processing in the ICC
The existence of three physiologically distinct response types
(with sustained responses) suggests that ICC neurons may be specialized
for different aspects of signal processing. Because type V units have
low BFs and predominantly excitatory response patterns, their inputs
are hypothesized to have origins in the MSO. Acoustic information is
carried from the auditory nerve to the MSO through a pathway that is
defined by its powerful synaptic coupling (Schwartz 1972,
1992
). It has been suggested that this is the primary channel
for transmitting the timing of spike discharge rates to higher levels
of processing (Irvine 1986
, 1992
). Consequently, type V
units may play an important role in auditory behaviors that require
faithful transmission of temporal information such as the encoding of
stimulus location based on interaural time difference cues
(Kuwada and Yin 1983
; Yin et al. 1986
,
1987
).
Level-tolerant tuning curves like those shown by type I units have been
found at multiple levels in the bat auditory system and have been
proposed as a mechanism for maintaining frequency selectivity at high
sound levels (Olsen and Suga 1991; Suga and Tsuzuki 1985
; Yang et al. 1992
). In this
interpretation, lateral inhibition attaches a specific frequency to a
neuron by preventing expansion of excitatory bandwidth. Lateral
inhibition also may make type I units ideal candidates for detecting
narrowband signals in noisy environments either by reducing the
effective bandwidth of masking noise at higher stimulus levels (the
power spectrum model of masking) (Patterson and Moore
1986
) or by reducing the responsiveness to wideband relative to
narrowband stimuli. One potential difficulty in the latter case,
however, is that the noise-evoked activation of the extensive
inhibitory sidebands of the unit could result in complete inhibition of
the excitatory response to the narrowband stimulus. Type I units do not
appear to suffer from this drawback as their noise-driven rates are
almost never reduced to zero (Fig. 10C), and preliminary
observations suggest that type I units do provide the best
representation of BF tones in noise (Ramachandran et al.
1997
). In the visual system, retinal ganglion cells use a
receptive field organization similar to type I units to reduce visual
noise (Werblin 1974
). Like their presumed LSO inputs,
type I units are also sensitive to interaural level differences
(Davis et al. 1999
).
Type O responses are similar to the upper threshold units that have
been described throughout the auditory system of bats (Suga and
Manabe 1982; Yang et al. 1992
); these units have
been characterized previously as "feature detectors" because they
respond under such a restricted set of stimulus conditions. It also has been suggested that units with strongly nonmonotonic rate-level functions may be specialized for intensity coding because unlike monotonic units they do not respond to an upwardly unlimited range of
stimulus levels (Brugge and Merzenich 1973
;
Phillips et al. 1985
; Suga and Manabe
1982
). To represent all discriminable sound levels, it is
expected that a population of level-tuned units would show different
"best intensities," which sum to encompass the full range of
hearing. Our results do not support this interpretation in that all
type O thresholds fell within 30 dB of the most sensitive neural
thresholds. If the basic similarities between type O response pattern
in ICC and type IV responses in DCN reflect a direct input, type O
units also may be involved in sound localization. Young and his
colleagues have shown that DCN type IV units respond selectively to
peaks and notches in the spectrum of broadband sounds (Nelken and Young 1994
; Young et al. 1992
). In the
normal behavioral repertoire of the cat, these sharp spectral features
are created by the filtering properties of the pinnae (Rice et
al. 1992
) and must be present for accurate localization of
complex sounds (Huang and May 1996
).
The similarity of ICC response types to those of lower-level brain stem
neurons, including their sensitivity to interaural level differences
described by Davis et al. (1999), suggests that the
parallel pathways evident in the cochlear nucleus and superior olive
may remain segregated at the level of the inferior colliculus. Although
the systems described in this paper resemble those at lower levels, the
presence of convergent inhibitory inputs in the colliculus, such as
projections from the dorsal nucleus of the lateral lemniscus, and the
fact that there are several ascending inputs to the colliculus that
were not considered earlier, such as the periolivary nuclei and nucleus
of the lateral lemniscus, argue that the inferior colliculus is not
merely relaying an unmodified representation from below. Some of the
modifications in the ascending representation involve binaural
properties (e.g., McAlpine et al. 1998
), some involve
temporal sensitivity (Delgutte et al. 1998
;
Langner and Schreiner 1988
; Litovsky
1998
), and others remain to be revealed by studies of complex
natural stimuli.
![]() |
ACKNOWLEDGMENTS |
---|
The authors thank E. D. Young and D. Peruzzi for comments on an earlier version of the manuscript and A. Palmer and an anonymous reviewer for critical review of the paper. The authors thank P. P. Taylor for support with histology and figure production.
This work was supported by National Institute of Deafness and Other Communication Disorders Grants DC-00979 and DC-00023.
![]() |
FOOTNOTES |
---|
Address for reprint requests: B. J. May, Dept. of Otolaryngology-HNS, Johns Hopkins University, 720 Rutland Ave./505 Traylor Research Bldg., Baltimore, MD 21205.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Received 19 October 1998; accepted in final form 23 March 1999.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|