Hearing Research 193 (2004) 39–50

www.elsevier.com/locate/heares

Effects of an acute acoustic trauma on the representation of avoice onset time continuum in cat primary auditory cortex

Masahiko Tomita, Arnaud J. Nore~na, Jos J. Eggermont *

Departments of Physiology and Biophysics, and Psychology, Neuroscience Research Group, University of Calgary,

2500 University Drive N.W., Calgary, Alta., Canada T2N 1N4

Received 1 December 2003; accepted 2 March 2004

Available online 24 March 2004

Abstract

Here we show that hearing loss associated with an impairment of speech recognition causes a decrease in neural temporal

resolution. In order to assess central auditory system changes in temporal resolution, we investigated the effect of an acute hearing

loss on the representation of a voice onset time (VOT) and gap-duration continuum in primary auditory cortex (AI) of the ketamine-

anesthetized cat. Multiple single-unit activity related to the presentation of a /ba/–/pa/ continuum – in which VOT was varied in 5-ms

step from 0 to 70 ms – was recorded from the same sites before and after an acoustic trauma using two 8-electrode arrays. We also

obtained data for gaps, of duration equal to the VOT, embedded in noise 5 ms after the onset. We specifically analyzed the

maximum firing rate (FRmax), related to the presentation of the vowel or trailing noise burst, as a function of VOT and gap

duration. The changes in FRmax for /ba/–/pa/ continuum as a function of VOT match the psychometric function for categorical

perception of /ba/–/pa/ modeled by a sigmoid function. An acoustic trauma made the sigmoid fitting functions shallower, and shifted

them toward higher values of VOT. The less steep fitting function may be a neural correlate of an impaired psychoacoustic temporal

resolution, because the ambiguity between /ba/ and /pa/ should consequently be increased. The present study is the first one in

showing an impairment of the temporal resolution of neurons in AI caused by an acute acoustic trauma.

� 2004 Elsevier B.V. All rights reserved.

Keywords: Noise trauma; Hearing loss; Voice onset time; Speech perception; Temporal resolution

1. Introduction

Animal vocalizations and speech in humans, are

characterized by distinct amplitude fluctuations. It hasbeen demonstrated that the auditory system makes ef-

fective use of these temporal cues for perception. In this

context, Shannon et al. (1995) showed that noise-bands

modulated with the temporal envelope of speech, pro-

ducing an acoustical signal with strongly degraded

spectral information but with preserved temporal

envelope cues, was sufficient to allow a correct identifi-

cation of consonants, vowels and words. Moreover,

* Corresponding author. Tel.: +1-403-220-5214; fax: +1-403-282-

8249.

E-mail address: eggermon@ucalgary.ca (J.J. Eggermont).

0378-5955/$ - see front matter � 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.heares.2004.03.002

amplitude fluctuation can be used by the auditory sys-

tem in auditory scene analysis. Indeed, different fre-

quency components coherently modulated may be

interpreted as being related to the same source or au-ditory object (Bregman et al., 1990).

Amplitude fluctuations can be defined as periodic (as

in vowels) or aperiodic. The voice onset time (VOT), the

interval with aspiration noise between the consonant re-

lease and voicing onset, is an aperiodic amplitude fluc-

tuation. The difference between /ba/ and /pa/ phonemes,

for instance, depends to a large extent on the difference in

VOT (Kuhl andMiller, 1978). A VOT of 0ms is generallyassociated with the perception of /ba/, whereas a VOT of

25 ms and higher results in the perception of /pa/. The

perception changes abruptly from voiced to voiceless at

this boundary (Abramson and Lisker, 1970). Interest-

ingly, this boundary does not depend on a specific human

mail to: eggermon@ucalgary.ca

40 M. Tomita et al. / Hearing Research 193 (2004) 39–50

speech mechanism and was reported in chinchillas (Kuhl

and Miller, 1978) and monkeys (Morse and Snowdon,

1975; Sinnott and Adams, 1987) at approximately the

same value. The fact that the perceptual boundary was

reported in other mammals than humans suggests thatevolution may have taken advantage of specific proper-

ties of the central nervous system to design speech-sounds

(Kuhl and Miller, 1978). VOT boundaries also shift

with place of articulation and the boundaries for /da/–/ta/

and /ga/–/ka/ are higher than for the /ba/–/pa/ distinction.

Neural correlates of these perceptual boundaries have

been investigated in the auditory nerve of chinchillas

(Sinex and McDonald, 1988, 1989), inferior colliculus ofchinchillas (Chen et al., 1996) and in the auditory cortex

of monkey (Steinschneider et al., 1995, 2003) and cat

(Eggermont, 1995a,b). Steinschneider et al. (1995) ex-

amined multi-unit activity elicited by the consonant

vowel syllable /da/ and /ta/ that varied in VOT (0, 20, 40

and 60 ms). They found responses time-locked to stim-

ulus onset and to the onset of voicing with 40 ms VOT.

The response to voicing onset was markedly diminishedwith 20 ms and was similar to that evoked by /da/ with

0 ms VOT. These results were seen in 59% penetrations,

which display the ‘‘double-on’’ response pattern.

Eggermont (1995b) studied the representation of VOT

for a /ba/–/pa/ continuum in which VOT was varied in 5-

ms steps from 0 to 70 ms. He found that the mean value

for the minimum detectable VOT, which was repre-

sented in onset responses to both the voiceless andvoiced parts of the sound, was 46 ms in adult cat.

Comparing the results obtained in awake monkeys

(Steinschneider et al., 1995, 2003) and anesthetized cats

(Eggermont, 1995a,b) suggests that the minimum VOT

is little affected by anesthesia.

The role of aperiodic amplitude fluctuations in the

understanding of speech has also been studied by using

gap-detection paradigms (Schneider and Hamstra, 1999;Snell and Frisina, 2000; Snell et al., 2002). Eggermont

(1995a, 1999, 2000) focused on the cortical representa-

tion of gaps embedded in noise. For short leading noise-

burst duration (5 ms), he found mean values for the

minimum neural gap threshold (MNGT) around 35 ms,

and similar to those obtained with /ba/–/pa/ continuum.

Hearing loss and aging, conditions known as being

associated with an impairment of speech understanding(CHABA, 1998; Humes, 1991; Abel et al., 1990), are

both linked to a decrease in temporal resolution (Moore,

1985, 1993; Gordon-Salant and Fitzgibbons, 1993; Snell

et al., 2002; Mazelova et al., 2003). Thus, impairment in

temporal resolution, as measured in a gap-detection task

(Walton et al., 1998) or evoked potentials by a VOT

continuum (Tremblay et al., 2003), may be related to the

decline in the understanding of speech. It has beendemonstrated that gap-detection thresholds were im-

paired in older subjects compared to those in younger

subjects (Snell, 1997). Consistent with this finding,

Walton et al. (1998) reported an age-related alteration of

temporal resolution of neurons in inferior colliculus. It

is suggested that the decline in temporal resolution

during aging may involve, in addition to a deficit in the

function of the cochlea, a central component. Numerousstudies have shown that a hearing loss is followed by

changes in central inhibition (Rajan, 1998, 2001; Wang

et al., 2002; Nore~na et al., 2003). However, there is noelectrophysiological study in the literature addressing

the effect of hearing loss on the cortical representation of

VOT, i.e., on the temporal resolution of the auditory

system.

In the present study, we investigated the effect of anacute hearing loss on the representation of a voice onset

time continuum (/ba/–/pa/ phonemes) in primary audi-

tory cortex. Multiple single-unit (MU) activity related to

the presentation of the /ba/–/pa/ continuum – in which

VOT was varied in 5-ms step from 0 to 70 ms – was

recorded from the same recording sites before and after

an acoustic trauma. Moreover, MU activity related to

the presentation of gaps inserted 5 ms after the onset ofthe noise (‘‘early’’ gap) in a 1-s noise-burst was also

recorded. Gap duration was varied in 5-ms step from

0 to 70 ms. The goal of our study is to gain insight into

the effects of an acute hearing loss on the cortical rep-

resentation of VOT and gaps. More generally, our study

is aimed at pointing out potential central changes after a

hearing loss that are associated with a decline in speech

understanding in humans.

2. Method

2.1. Animal preparation

The care and the use of animals reported on in this

study were approved by the Life and EnvironmentalScience Animal Care Committee of the University of

Calgary (BI-2001-021). All cats were anesthetized with

administration of 25 mg/kg of ketamine (100 mg/ml)

injected intra-muscularly followed after approximately

10 min by 20 mg/kg of pentobarbital sodium (65 mg/ml)

intra-muscularly and 0.25 ml/kg body weight of a mix-

ture of 0.1 ml acepromazine (0.25 mg/ml) and 0.9 ml

of atropine methyl nitrate (5 mg/ml) subcutaneously.Lidocain (20 mg/ml) was injected subcutaneously and a

skin flap and muscle tissue overlying right temporal

bone was removed and the skull cleared. A large screw

was cemented upside down on the skull with dental

acrylic. Two 8 mm diameter holes were trephined over

the right temporal cortex so as to expose parts of pri-

mary auditory cortex. The holes were enlarged to expose

both anterior and posterior ectosylvian sulci if needed.The dura was left intact, and the brain was covered with

light mineral oil to prevent tissue drying. Then the cat

was placed in a sound-treated room on a vibration iso-

M. Tomita et al. / Hearing Research 193 (2004) 39–50 41

lation frame, and the head was secured with the screw.

Throughout the experiment, anesthesia at a level suffi-

cient to abolish pinna reflexes, was maintained with in-

tra-muscular injections of 2–5 mg/kg per hour of

ketamine, and additional acepromazine/atropine mix-ture was administered every 2 h. Body temperature was

maintained at 37 �C with use of a temperature-con-trolled heating pad. At the end of the experiment, the

animals were killed with an overdose of pentobarbital

sodium. During the recordings the ear contra-lateral

to the speaker was filled with an ear mold substance

(Dur-a-Sil, Insta Mold Products).

2.2. Peripheral threshold estimation

Prior to acute recordings and in one-third of the

cases after the end of experiment, approximately 6 h

after the exposure, peripheral hearing sensitivity was

determined from auditory brainstem response (ABR)

threshold. For this purpose, tone pips with frequencies

of 3, 4, 6, 12, 16, 24 and 32 kHz were presented at 10/sin an anechoic room from a speaker placed 45� fromthe midline and at 55 cm distance from the cat’s head.

The sound-treated room was made anechoic for fre-

quencies above 625 Hz by covering walls and ceiling

with acoustic wedges (Sonex 300) and by covering ex-posed parts of the vibration isolation frame, equipment

and floor with wedge material as well. Calibration and

monitoring of the sound field was done using a Br€ueland Kj€aer (type 4134) microphone placed above ani-mal’s head and facing the loud speaker. ABRs were

recorded, with needle electrodes in ipsi- and contralat-

eral muscles covering both mastoids, in response to

c-function shaped tone pips with an effective (50% frompeak) duration of 15 ms. In this recording montage, the

first negative–positive component representing the

compound action potential of the auditory nerve isfrequently the largest component. The signals were

amplified between 300 and 3000 Hz using a DAM 500

(World Precision Instruments) differential amplifier and

averaged with a Br€uel and Kj€aer (type 2034) dual signalanalyzer in the signal enhancement mode. Artifact re-

jection and local lidocaine infusion were used to avoid

contamination of the ABR by muscle action potentials.

At high intensity levels 20–50 averages sufficed but atnear threshold values 200 averages were obtained

and repeated once. Step size was 10 dB, except

around threshold where it was 5 dB. Threshold was

defined as the lowest intensity that yielded a repro-

ducible response.

2.3. Acoustic stimulus presentation

Stimuli were generated using MATLAB� and trans-

ferred to the DSP boards of a TDT (Tucker Davis

Technologies) sound delivery system. Acoustic stimuli

were presented in an anechoic room from a speaker

placed 45� from the midline 55 cm distance from thecat’s left ear. Characteristic frequency (CF) and tuning

properties of individual neuron were determined withc-shape envelope tone pips. These tone pips with a half-peak-amplitude duration of 15 ms and a c-function-shaped envelope, were presented at a rate of 1/s in a

pseudo-random frequency order at fixed intensity level.

The 81 different frequencies used were equally spaced

logarithmically between 625 Hz and 20 kHz (or between

1.25 and 40 kHz) so that 16 frequencies were present per

octave. Each recording session comprised of up to ninedifferent intensity series in 10 dB steps of these tone pip

stimuli.

After the frequency–tuning properties of the cells at

each electrode were determined, the /ba/–/pa/ continuum

with VOTs ranging from 0 to 70 ms in steps of 5 ms was

presented in random order as described previously

(Eggermont, 1995b). These phonemes were generated

with a parallel/cascade Klatt-synthesizer KLSYN88a,using a 20-kHz sampling frequency. The total duration

of the stimulus was 250 ms, regardless the duration of

the VOT, and the onset noise burst was 5 ms in duration

(leading noise burst). The dominant frequency ranges

for the vowel part were F 0 ¼ 120 Hz, F 1 ¼ 700 Hz, andF 2 ¼ 1200 Hz. The fundamental frequency started at120 Hz, remained at that value for 100 ms, and then

dropped from there to 100 Hz at the end of the vowel.F 1 started at 512 Hz and increased from 25 ms to700 Hz, F 2 started at 1019 Hz and increased from 25 msto 1200 Hz, and F 3 changed in the same time span from2153 to 2600 Hz. In order to match the CF of neurons to

the spectral content of the vowel, its spectrum was

transformed so that the upper edge of the F 2 band wasabove the CF of neurons. The phonemes were presented

once every 2 s. The only parameter that changed in thecontinuum was the VOT; formant transition durations

were not altered. Twenty repetitions were presented at

each VOT value so that the entire stimulus sequence

lasted 600 s.

Gaps ranging from 5 to 70 ms in duration were placed

in two positions in wideband noise bursts of 1 s in du-

ration that were presented once per 3 s in random order.

The first position, the ‘‘early gap,’’ started 5 ms after thenoise burst onset, and the second position, the ‘‘late

gap,’’ was positioned 500 ms after noise burst onset. Only

results for the early gap will be presented as they relate to

the VOT results (Eggermont, 1999). Each gap stimulus

was presented 15 times. The noise burst used consisted of

‘‘frozen’’ noise, i.e., the pseudo-random noise sequence

was the same for all conditions. These stimuli (a con-

tinuum of /ba/–/pa/ phonemes and a stimulus detectingearly and late gap) were presented at peak intensities of

65 or 75 dB SPL and the intensity level was same before

and after the acoustic trauma.

Fig. 1. Changes in frequency tuning of MU responses recorded in one

cat. The frequency–tuning curves of each recording site obtained be-

fore and after the acoustic trauma are shown in panels (a) and (b),

respectively. Panel (c) shows the lower envelopes of all frequency–

tuning curves in panels (a) and (b). The difference of the value esti-

mated from contour lines in panel (c) are plotted at 3, 4, 6, 8, 12, 16, 24

and 32 kHz in panel (d). ABR threshold shifts obtained 5 h after the

trauma are shown as dot line in panel (d).

42 M. Tomita et al. / Hearing Research 193 (2004) 39–50

2.4. Acoustic trauma

The acoustic trauma was induced by a 1-h exposure

to a pure-tone at the maximum output level. The trauma

tone frequency (TF) was set at 5 or 6 kHz. The soundwas presented by a high-power amplifier (Samson Servo

240) and loudspeaker (Yorkville 120 W) placed 75 cm

away from the cat’s head. Measured at cat’s head, the

trauma tone fundamental frequency had a level of 115–

120 dB SPL; whereas the second and third harmonics

had a level of 25 and 38 dB lower than that of funda-

mental frequency, respectively. The pure tone exposure

was administered about 8 h after the start of the prep-aration, following typically 4 h of recording at the se-

lected recording sites. The post-exposure threshold at

CF was generally obtained with c-shape envelope tonepips between 1 and 1.5 h after the end of the exposure,

and just prior to recording responses to the VOT and

gap stimuli.

2.5. Recording and spike separation procedure

Two arrays of eight electrodes (Frederic Haer Corp.)

each with impedances between 1 and 2 MX were used.The electrodes were arranged in a 4 · 2 configurationwith inter-electrode distance within rows and columns

equal to 0.5 mm. Each electrode array was oriented with

the long axis parallel to the midline in primary auditory

cortex located in majority between anterior and pos-terior ectosylbian sulci. The arrays were manually and

independently advanced using Narishige M101 hydrau-

lic microdrives. The signals were amplified 10,000 times

using a Frederic Haer Corp. HiZx8 set of amplifiers with

filter cut-off frequencies set at 300 Hz and 5 kHz. The

amplified signals were processed by a DataWave multi-

channel data acquisition system. Spike sorting was done

off line using a semi-automated procedure based onprincipal component analysis (Eggermont, 1990) im-

plemented in MATLAB. The spike times and wave-

forms were stored. The data presented in this paper

represent only well separated single units that, because

of their regular spike wave form, likely are dominantly

from pyramidal cells, thereby minimizing potential

contributions from thalamocortical afferents or inter-

neurons (Eggermont, 1996). For statistical purposes, theseparated single-unit spike trains were added again to

form a multi-unit spike train. We have previously shown

that the CFs of neurons recorded from the same elec-

trode are similar (Eggermont, 1996).

2.6. Data analysis

To assess frequency–tuning properties, the peaknumber of action potentials in the post-stimulus time

histogram (PSTH, 5 ms bins) calculated over the first

100 ms after c-tone presentation was estimated. The

peak counts for three adjacent frequencies were then

combined, in order to reduce variability, and divided by

number of stimuli and presented as a firing rate per

stimulus. This resulted in 27 frequency bins covering 5

octaves so that the final frequency resolution for deter-mining the CF was approximately 0.2 octaves. The re-

sults were calculated per stimulus intensity, and were

combined into an intensity–frequency–rate profile from

which tuning curves, rate-intensity functions and

iso-intensity rate-frequency contours could be derived

(Eggermont, 1996) using routines implemented in

MATLAB. The frequency–tuning curve was defined for

a firing rate at 25% of the maximum peak-firing rate.This was about 10–20% above the background firing-

rate, but as the latter was dependent on the level of

stimulus-induced suppression, the tuning curve criterion

based on a percentage of peak firing rate was preferred

over that based on increase over background activity. In

order to calculate threshold shifts with the acoustic

trauma for each cat, all tuning curves for each cat were

drawn as the contour lines at 25% from maximum re-sponse as function of a frequency before and after the

acoustic trauma (Fig. 1(a) and (b)). The lower envelope

of the set of frequency–tuning curves was traced

(Fig. 1(c)) and the difference of the values at the lowest

threshold between before and after the noise exposure

was estimated as the cortical threshold shift as a func-

tion of frequency (Fig. 1(d)).

M. Tomita et al. / Hearing Research 193 (2004) 39–50 43

Post-stimulus time histograms with a 5-ms binwidth

were used to represent the recordings for the /ba/–/pa/

continuum and early stimuli. From these PSTHs both

the peak latency and the spike count were calculated.

The minimum neural VOT (MNVOT) was defined asthe minimum VOT at which neural responses are time-

locked to both the burst onset and the onset of voicing.

Similarly, the minimum neural gap threshold (MNGT)

for ‘‘early’’ gap was defined as the minimum gap du-

ration at which neural responses are time-locked to

both the leading and the trailing noise burst. The re-

sponses related to the vowel and the trailing noise

burst were analyzed only when an onset response tothe leading noise burst was clearly visible. The

MNVOT and MNGT were obtained from inspection

of the dot displays and depended on the number of

stimulus presentation and the firing rate of the neu-

rons. We presented 20 repetition for each VOT dura-

tion of a /ba/–/pa/ continuum and required at least four

time-locked spikes per 20 presentations to the onset of

voicing for assigning the MNVOT. Similarly, in gapstimuli 15 repetition for each gap duration were rep-

resented and three time-locked spikes per 15 presenta-

tions to the trailing noise burst were required to assign

the MNGT.

Moreover, the maximum firing rate (FRmax) related

to the responses for the voicing or the trailing noise

burst for each VOT or gap duration were derived from

PSTHs. In order to avoid the increase in firing rate re-lated to a late rebound response, only spikes that oc-

curred within a time window of 30 ms starting at a time

equal to the peak latency of the leading noise burst plus

VOT or gap duration were included in the analysis.

All statistical analyses were performed using Statview

5�. Illustrations and figures were prepared with MAT-

LAB� and SigmaPlot�.

3. Results

Recordings were made from the right auditory cortex

in 13 cats. For a total of 46 recording sites out of 103

that showed reliable frequency–tuning curves before and

after the noise trauma, we recorded clear MU responses

to the VOT or gap stimuli approximately 1–1.5 h beforeand approximately 1.5–2 h after the acoustic trauma.

The consonant–vowel stimuli were frequency trans-

formed so that the upper edge of the F2 band was above

the neuron’s pre-trauma CF. The study presented here

includes data collected in the same animals as those in

two other studies. Namely, the effects of an acoustic

trauma on the neural responses related to frequency

response properties were reported in Nore~na et al.(2003). Moreover, the effects of an acoustic trauma on

spontaneous activity and neural synchrony were de-

scribed in Nore~na and Eggermont (2003).

3.1. Effects of the acoustic trauma on thresholds

As described in Section 2, the cortical threshold shift

after the trauma was estimated by comparing the CFs of

103 MU recordings before and after the trauma.Fig. 1(d) shows the threshold shift after the acoustic

trauma in one cat. The solid line indicates the estimated

hearing loss from cortical MU recordings and dotted

line indicates the ABR threshold. ABR thresholds were

measured before the trauma in all the cats studied.

However, ABR thresholds measured before and after

the acoustic trauma were obtained in only five cats and

were on average 40 dB for frequencies above the traumatone (reported in Nore~na et al., 2003). In these cats, theABR threshold shift was generally larger than cortical

threshold shift, however, the shape of the threshold shift

according to frequency was similar in both ABR and

cortical data. The reason for this underestimation of the

peripheral hearing loss by CF thresholds is due to cor-

tical reorganization.

As reported before, after an acoustic trauma un-masking of new excitatory connections occurs resulting

in considerable CF shifts with little threshold increase

(Nore~na et al., 2003). This is illustrated in Fig. 2showing a response area for a particular electrode site

just before the acoustic trauma induced with a 5 kHz

tone and about 3 h after the trauma. Shown are contour

lines for the percentage of maximum response in the

PSTH obtained in a 5 ms bin in the time window of10–60 ms following tone pip onset. Contour lines are

shown for 75%, 50% and 35%, whereby the lower con-

tour is considered here as the neuron’s frequency–tuning

curve. The pre-trauma tuning curve (top) has a CF

around 10 kHz and threshold below 15 dB SPL (the

lowest sound level used). At 3 h 40 min after the expo-

sure (bottom), the CF is now about 7 kHz and the

threshold about 25 dB SPL. The response area above8 kHz that was initially present has completely disap-

peared and a new response area between 4 and 8 kHz,

and thus largely outside the original response area,

emerged. Tuning curves were also obtained at regular

intervals spanning the 4.5-h time range between those

recordings but are not shown here (for details see

Nore~na et al., 2003). Thus, the estimated hearing lossfrom cortical CFs will always be less than that estimatedfrom ABR thresholds.

Fig. 3(a) shows the ABR threshold shift in five cats as

a function of the difference between the frequency at

which ABR thresholds have been measured and the TF.

One notes that ABR threshold shift is around 40 dB in

average above the TF. Fig. 3(b) shows the cortical CF-

threshold shifts, for the recording sites that yielded re-

sponses to the /ba/–/pa/ and gap stimuli, as a function ofthe difference between the pre-trauma CF and the TF.

The cortical threshold shift is around 20 dB above the

TF and is larger for the frequency band above TF than

Fig. 3. Threshold shift after the acoustic trauma. Panel (a) shows pe-

ripheral threshold shift based on auditory brainstem response (ABR)

in five cats. Panel (b) shows the cortical threshold shift at the char-

acteristic frequency (CF) as a function of the difference between the

pre-exposure CF and trauma tone frequency (TF), respectively. The

black solid line in panel (a) shows the averaged peripheral threshold

shift. The black line in panel (b) shows the average CF-threshold shift

for those units that showed responses to the VOT or gap stimuli.Fig. 2. Pre- and post-acoustic trauma frequency response areas. The

contour lines separate response at 75% of maximum (black), 50% of

maximum (gray) and 35% of maximum. The latter contour line is

considered as the frequency–tuning curve. One observes that the post-

trauma tuning curve barely overlaps with the pre-trauma tuning curve.

44 M. Tomita et al. / Hearing Research 193 (2004) 39–50

below TF. The average threshold elevation is 13.1 dB for

MUs with CF below and above TF and significantly

different from 0 (tð43Þ ¼ 5:287, p < 0:001). As notedabove, these threshold shifts were calculated at the CFs,

which typically changed dramatically after the noise

trauma and thus do not represent the peripheral hearingloss at particular frequencies. The increase in the

thresholds at CF in the subset of 46 recordings analyzed

here was nearly identical to that in the overall set of 124

neurons analyzed previously (Nore~na et al., 2003).

3.2. Individual example

Fig. 4 shows neural responses obtained at the samerecording site for presentation of the /ba/–/pa/ contin-

uum ((a)–(c)) and the ‘‘early’’ gap stimuli ((d)–(f)) (see

Section 2). Left-hand panels ((a) and (d)) show dot

displays: each dot corresponds to a spike. The dot dis-

plays are organized vertically according to gap duration

or VOT and horizontally for time since leading noise

burst onset. Middle panels ((b) and (e)) represents the

PSTHs (the bin size is 5 ms) of the data shown in panels

(a) and (d), respectively. The oblique dotted lines indi-cate the time windows (30-ms duration) in which the

FRmax, for each gap duration or VOT, is derived.

Right-hand panels ((c) and (f)) show the FRmax, for

each gap duration or VOT, related to the presentation of

the trailing noise or the vowel.

One notes first that the responses related to the pre-

sentation of the /ba/–/pa/ continuum are similar to those

related to the presentation of the ‘‘early’’ gap stimuli. Inboth cases, neural responses are evoked at the onset of

the stimuli. Moreover, neurons are activated by the

trailing stimulus only if it is presented at more than

Fig. 4. A comparison of the responses to a /ba/–/pa/ continuum (a)–(c) and early gap (d)–(f) conditions from the same recording site. Dot displays

(left column) and PSTH (middle column) are organized vertically according to VOT or gap duration and horizontally for time since the onset of the

leading noise burst. Time windows for evaluation of the PSTHs to the trailing stimulus are selected (between dot lines) according to VOT or gap

duration and the latency of peak response for the leading noise burst. The maximum firing rate in a 5 ms bin (FRmax) in these time windows is called

the peak responses to the vowel or trailing noise burst, and plotted as a function of VOT or gap duration (right column). The fit curves in panels (c)

and (f) are calculated according to Eq. (1): Y ¼ y0 þ a=ð1þ ekðX�x50ÞÞ (see text).

M. Tomita et al. / Hearing Research 193 (2004) 39–50 45

40 ms after the leading stimulus. Namely, the MNVOT

and MNGT have similar values, around 40 ms. In thisexample, the background firing rate is very low; neurons

do not respond to the second stimulus (vowel or trailing

noise) for VOT below 40 ms.

The data were fitted by a sigmoid function

Y ¼ y0 þ a 1��

þ ekðX�x50Þ�; ð1Þ

where Y is the FRmax, X is VOT or gap duration in ms,x50 is the VOT or gap duration corresponding to the

50% point of the function, y0 is the background activity,a is the difference between maximum and minimum ofthe sigmoid function, and k is a factor defining the slopeof the function. The middle (50%) points between min-

imum value and maximum value of FRmax from the

fitting curves to obtained data (panels (c) and (f)) are

very close to the value of the minimum VOT or gap

duration compared to that derived from visual inspec-

tion in the left (dot displays) and middle panels (PSTH).Namely, MNVOT as estimated from x50 is 42 ms for

/ba/–/pa/ stimulus continuum and the MNGT is esti-

mated at 37 ms.

3.3. Group data for the /ba/–/pa/ continuum and the

‘‘early’’ gap condition

As shown in Fig. 3, the average CF-threshold shiftis larger for the frequency band above TF than that

below TF. Consequently, we divided the data of the MU

recordings into two groups based on the CF and TF.

However, we found no significantly difference (with re-spect to onset response, FRmax, MNVOT and MNGT)

between the groups with CF above and below TF, and

thus pooled the data into a single group. All the fol-

lowing results then concern the entire group.

Fig. 5 shows the distribution of MNVOT and MNGT

values before and after the acoustic trauma. One notes

that the distribution in MNVOT and MNGT values

presents a peak around 40 ms before the acoustic trau-ma. After the acoustic trauma, the distribution is more

uniform and the number of MUs with MNVOT and

MNGT above 40 ms is increased. The averaged

MNVOT is 31 ms (SD¼ 13 ms) and 35 ms (SD¼ 17 ms)before and after the acoustic trauma, respectively. The

averaged MNGT is 36 ms (SD¼ 17 ms) and 38 ms(SD¼ 18 ms) before and after the acoustic trauma, re-spectively. The right-hand panels show the distributionin changes of MNVOT and MNGT with the acoustic

trauma. The number of MUs with difference of

MNVOT and MNGT above 0 ms is larger than that

below 0 ms, although the increase of MNVOT and

MNGT with the acoustic trauma was not significantly

different using one sample t-test hypothesized mean of

zero (/ba/–/pa/: p ¼ 0:11; early gap: p ¼ 0:17).Fig. 6 shows the changes in MNVOT and MNGT as

a function of the threshold shift at CF for the /ba/–/pa/

continuum (a) and the ‘‘early’’ gap condition (c). And

panels (b) and (d) show the changes in MNVOT and

Fig. 6. Differences of MNVOT ((a) and (b)) and MNGT ((c) and (d))

before and after the acoustic trauma as a function of threshold shift at

CF (left column) and as a function of the difference between CF and

TF (right column).

Fig. 5. Distribution of MNVOT (top row) and MNGT (bottom row) values for MU recording before (left column) and after (middle

column) acoustic trauma. Note that modal values of MNVOT and MNGT are around 40 ms. After the trauma, the number of MNVOT and

MNGT values of more than 40 ms is increased. Right column shows the difference of MNVOT and MNGT between before and after

acoustic trauma. Note that the number of MUs with an increase of MNVOT and MNGT after the acoustic trauma is larger than that

showing a decrease.

46 M. Tomita et al. / Hearing Research 193 (2004) 39–50

MNGT as a function of the difference between the pre-

trauma CF and the TF for the /ba/–/pa/ continuum (b)

and the ‘‘early’’ gap condition (d). One observes a ten-

dency for the MNVOT or the MNGT to be dependent

on threshold shift, but this was only significant for

MNVOT (R2 ¼ 0:206, p ¼ 0:013). Changes in MNVOTor MNGT were not significantly correlated with the

difference between the pre-trauma CF and TF.

Fig. 7 shows the averaged PSTHs in response to the

leading noise burst. We averaged the PSTHs obtained

fromMU recordings of stimulus representation with 25–

45 ms of VOT or gap duration. For each bin, the

changes in PSTH (before vs. after the trauma) were

statistically tested (paired t test). There was no signifi-cant difference between firing rates before and after the

acoustic trauma in both the /ba/–/pa/ continuum and

‘‘early’’ gap conditions.

Fig. 8 shows the average of the normalized FRmax as

a function of VOT or (‘‘early’’) gap duration. The nor-

malized FRmax were obtained by dividing the FRmax

obtained at each VOT- or gap-duration by the highest

FRmax to the trailing stimuli over all VOTs or gapdurations. As shown for individual recordings in Fig. 4,

the largest FRmax were usually obtained for highest

values of VOT or gap duration, namely around 70 ms.

An ANOVA showed that the data were very well fitted

by a sigmoid function (Eq. (1)); pre-trauma /ba/–/pa/:

R2 ¼ 0:96, p < 0:001, post-trauma /ba/–/pa/: R2 ¼ 0:98,p < 0:001, pre-trauma early gap: R2 ¼ 0:97, p < 0:001,and post-trauma early gap: R2 ¼ 0:97, p < 0:001.

Fig. 7. Averaged PSTHs for leading noise burst before and after

acoustic trauma in /ba/–/pa/ continuum and early gap condition (±SE).

Binwidth is 5 ms. Note that PSTH after the trauma is similar to the

PSTH before the trauma for the /ba/–/pa/ continuum and early gap

conditions.

Fig. 8. Average normalized maximum firing rate for the vowel (top)

and trailing noise burst after the early gap (bottom) obtained before

(filled circles) and after (open circles) the acoustic trauma (±SE). The

sigmoid curves shown provide the best statistical fit to the data. Note

that fitted curves for both the /ba/–/pa/ continuum and the early gap

condition are shifted toward longer VOT or gap duration.

M. Tomita et al. / Hearing Research 193 (2004) 39–50 47

One notices that the shape of the fitting curves before

the trauma is similar for /ba/–/pa/ and ‘‘early’’ gap

stimuli. Below a VOT or gap duration of 20 ms, the

FRmax is low and constant and corresponds to thebackground activity (the trailing stimulus does not

evoke stimulus-locked responses). Between 30 and

50 ms, the FRmax increases abruptly as a function of

VOT and gap duration. Finally, the FRmax reaches a

plateau from values of VOT or gap duration around 50

ms. The middle point (x50: the 50% detectable point)

from minimum to maximum value of FRmax gives

similar values for MNVOT and MNGT, namely 35 and38 ms, respectively.

After the trauma, the shape of the fitting curves for

both /ba/–/pa/ continuum and ‘‘early’’ gap condition is

changed. The most striking difference compared to the

pre-trauma condition is that the fitting curves are shifted

toward longer values of VOT or gap duration for both

conditions. And the slope of the fitting function is also

less steep for /ba/–/pa/ continuum. Indeed, slope valuesderived from the fitted data were 0.017 and 0.014 for

/ba/–/pa/ continuum in the pre and post-trauma condi-

tion, respectively. For the gap condition the slope stayed

the same at 0.019. As a consequence, the estimated

MNVOT and MNGT from the fit functions are in-

creased after the trauma compared to the pre-trauma

condition: MNVOT and MNGT derived from the fitted

data for post-trauma condition were 46 and 51 ms, re-spectively. Moreover, FRmax does not reach a plateau

at high values of VOT or gap duration after the trauma.

Again, after the trauma, the shape of the fitting curves

was similar between /ba/–/pa/ and ‘‘early’’ gap stimuli.

As we described above, the shape of the fitting curve

to data obtained before the trauma consists of three

parts divided by VOT or gap duration, a short VOT or

gap duration part with mostly background activity, amoderate duration part with abrupt changes of FRmax

and a long duration plateau. In order to estimate the

change of these FRmax values with the acoustic trauma,

we divided the data into three groups by VOT or gap

duration, namely short (0–20 ms) duration group, bor-

der (25–45 ms) duration group around MNVOT or

MNGT and long (50–70 ms) duration group. The

change of FRmax was statistically tested using a re-peated measures ANOVA with repeated factor of time

of recording (before and after the acoustic trauma). For

/ba/–/pa/ continuum, there was a significant decrease of

FRmax with the acoustic trauma for the border group

(F ¼ 9:829, p ¼ 0:002) and no significant change in theother two groups. For the early gap condition, the

FRmax in the short duration group and border group

48 M. Tomita et al. / Hearing Research 193 (2004) 39–50

were also significantly decreased with the acoustic

trauma (F ¼ 12:48, p ¼ 0:0005, F ¼ 27:18, p < 0:0001,respectively).

4. Discussion

The results of the present study can be summarized as

follows. There were no significant differences in the

mean MNVOT and MNGT to the /ba/–/pa/ and gap-in-

noise stimuli. The mean MNVOT and MNGT values as

estimated from the dot-displays were not significantly

affected by the trauma. However, the acoustic traumasignificantly shifted the sigmoid functions used to fit the

average data toward higher values of VOT or gap du-

ration, and the VOT fit curve became less steep. Before

the trauma, the mean 50% points of the fit curves were

35 and 38 ms for /ba/–/pa/ and gap stimuli, respectively.

After the trauma, the mean 50% points were in-

creased to 46 and 51 ms for /ba/–/pa/ and gap stimuli,

respectively.For the pre-trauma condition, our results are the

same as those from the previous studies of Eggermont

(1995a,b, 1999): MNGT and MNVOT presented similar

mean values around 35–40 ms. Moreover, these results

have been corroborated by another study where authors

have recorded auditory evoked potentials elicited by

stop consonant–vowel syllables directly from Heschl’s

gyrus and temporal gyrus in awake humans (Steins-chneider et al., 1999). This once more underlines the

minor effect that anesthesia has on our findings. Fur-

thermore, the recordings of the gap and /ba/–/pa/ stimuli

were at most 4 h apart, the earlier recording typically

occurring at least 5 h after the pentobarbital injection.

At this time one does not expect any change resulting

from lingering barbiturate effects. We have previously

reported about the lack of changes in temporal responseproperties as a function of time after anesthesia onset

(Eggermont, 1991), and we have extensive recordings of

VOT studies in animals (Tomita et al., unpublished)

with a permanent noise trauma where we did not ob-

serve any changes that could be attributed to time after

the single pentobarbital injection at the start of the ex-

periment. Thus we conclude that the changes that we

observe are due to the intervening noise trauma.Syllables with a relatively long VOT – above 30 ms,

were associated with a clear ‘‘double-onset’’ response,

namely a neural response time-locked to the consonant

release and one to voicing onset. On the other hand,

syllables with a VOT below 30 ms generally elicited a

reduced response – if any – to voicing onset. The latter

results suggest ‘‘that the perceptual category boundary

could reflect an average of the temporal activity patternsin auditory cortex (AI) that showed a double-on re-

sponse’’ (Eggermont and Ponton, 2002). In addition, the

fact that noise stimuli and different syllables are asso-

ciated with similar MNGT or MNVOT around 35 ms in

cortex suggests that these values reflect intrinsic cortical

neural properties and are not directly related to per-

ceptual boundaries. These points are developed below.

4.1. Neurophysiological mechanisms

Eggermont (1999, 2000) argued that cortical post-

activation suppression might be the main neurophysio-

logical mechanism accounting for the MNVOT and

MNGT. Post-activation suppression may result from

after hyperpolarization as well as from feed-back or

feed-forward inhibition (Eggermont, 2000). As men-tioned by Eggermont (1999), inhibition resulting from

the involvement of GABAA receptor activation is diffi-

cult to distinguish from after hyperpolarization. They

have approximately the same time constants and both

start a few ms after the excitatory on-response of the

neurons. Regardless the mechanism, post-activation

suppression limits the temporal resolution of the audi-

tory system in preventing the occurrence of locked re-sponses to the trailing or voicing stimulus when gap

duration or VOT are short. As modeled by Eggermont

(2000), the post-activation suppression is expected to be

prominent immediately after the onset of the leading

stimulus or consonant release. This hypothesis is con-

sistent with electrophysiological data showing that

MNGT decreases as a function of leading stimulus du-

ration (Eggermont, 2000).Post-activation suppression, in constraining temporal

neural responses, may then be responsible for the be-

havioral performances, namely for the categorical per-

ceptual boundary of a speech continuum, for instance of

/ba/–/pa/ phonemes. In this context, it is noted that the

perceptual boundary – around 25 ms (Kuhl and Miller,

1978) – is about 10 ms shorter the middle point of our

normal data fitted with a sigmoid function (Fig. 8). Thisdifference could be due to the anesthesia used as it in-

creases inhibition and after hyperpolarization effects.

Moreover, gap thresholds measured by Phillips et al.

(1997) corroborate this. Namely, in a condition where

the silent gap was marked by a wide-band noise (leading

stimulus, intended to simulate a consonantal burst) and

a low frequency narrow-band noise (trailing stimulus,

intended to simulate the vowel), gap thresholds werefound to match MNGTs found in Eggermont’s (1999,

2000) and in the present study. In addition, perceptual

gap thresholds and MNGT present similar values as a

function of leading burst duration (Eggermont, 2000).

However, the broad distribution of minimum gap and

VOT thresholds also suggests that the perceptual deci-

sion is likely made on basis of the population firing

activity as reflected in Fig. 8 and not on the distributionof the threshold values. In this respect, the activity in AI

likely only provides a basis on which to set perceptual

boundaries. If the activity produced by the consonant–

M. Tomita et al. / Hearing Research 193 (2004) 39–50 49

vowel phonemes in AI is affected so will be ability to

discriminate between them.

4.2. Neural changes induced by the acoustic trauma

We have shown that the acoustic trauma impairs

temporal resolution in AI: the sigmoidal fitting function

shifted toward higher values of gap duration or VOT.

Specifically, the middle points derived from these func-

tions increased by about 10 ms (Fig. 8). Because the

acoustic trauma induced a hearing loss (Figs. 1 and 3), it

is important to separate the potential effect of the ef-

fective intensity level of stimulation on the neural re-sponses from that of changes in temporal acuity. A

decrease in the effective intensity level may have induced

the impairment in temporal resolution we observed. For

instance, Fig. 6 shows that the increase in MNVOT and

MNGT is slightly dependent on the threshold shift.

However, the unaffected FRmax values for large VOTs

suggest that the changes in temporal resolution of cor-

tical neurons after the acoustic trauma are not related tothe effective intensity level. Fig. 8 shows that the FRmax

(for gap durations or VOTs above 55 ms) is not de-

creased after the trauma; to the contrary, the largest

FRmax values are even increased. This result suggests

that, on average, neurons have roughly conserved the

same responsiveness to the trailing stimulus. Eggermont

(1995b, 1999) showed that MNGT or MNVOT could

depend on the strength of the onset response related tothe presentation of the leading stimulus or the conso-

nant release. A decrease in the onset response related to

the onset of the leading stimulus was associated with a

decrease in MNGT and MNVOT. A decrease in effec-

tive level due to hearing loss may have decreased the

neural responsiveness to the leading stimulus and con-

sequently decreased the MNGT and MNVOT. This

explanation is not corroborated by our results: the av-eraged onset responses related to the leading stimulus

are not changed after the acoustic trauma compared to

before (Fig. 7). This is likely due to the elevated sound

level (65 or 75 dB SPL) used in these experiments.

In our previous paper (Nore~na et al., 2003), weshowed that neural activity was more strongly sup-

pressed following the presentation of the tone burst for

the post-noise trauma condition than for the pre-traumacondition. This suggests that the acoustic trauma may

result in an increase of post-activation suppression, and

hence longer MNVOT and MNGT values.

4.3. Potential perceptual correlates of the central changes

that follow an acoustic trauma

Glasberg et al. (1987) showed that hearing loss wasassociated with a slower rate of recovery from forward

masking. The authors suggested that larger gap thresh-

olds in hearing-impaired subjects might be associated

with this. This result is consistent with the hypothesis,

suggested by our results (described above), that hearing

loss increases the strength of the post-activation sup-

pression. Namely, in preventing the occurrence of the

second ‘‘on-response’’ related to the presentation of thetrailing stimulus or voicing, the increased post-activa-

tion suppression could alter the temporal resolution.

We showed that the acoustic trauma changed the slope

of the fitting function (Fig. 8). Indeed, for a /ba/–/pa/

continuum, the slopes were shallower in the post-trauma

condition. Then, if the presence of the VOT in the /ba/–

/pa/ continuum is actually encoded through a population

average of the FRmax related to the presentation of thetrailing stimulus, the ambiguity between /ba/ and /pa/

should consequently be increased. It is important to note

that the slope derived from the psychometric function

(Kuhl and Miller, 1978; Strouse et al., 1998) defines the

accuracy of the categorical perception or, in other words,

the ambiguity between the two extremes. It is indispens-

able for accurate coding of a signal-like speech to mini-

mize these ambiguities. That may be the reason why theslope in the psychometric function is relatively steep. The

decreased slopes of the fitting function of our data in

the post-trauma conditionmay be a neural correlate of an

impaired psychoacoustic temporal resolution. In addi-

tion, in low signal-to-noise ratio conditions, this decrease

in temporal resolutionmay even aggravate the perceptual

performances.

This study shows for the first time that a hearing lossinduced by an acute acoustic trauma decreases the

temporal resolution of cortical neurons, both in the re-

gion of the hearing loss and up to 1.5 octave below it.

We suggest that this impairment in the ability to code

VOT might be related to an increase in the post-acti-

vation suppression.

Acknowledgements

This work was supported by the Alberta Heritage

Foundation for Medical Research, the National Sci-

ences and Engineering Research Council, the Canadian

Language and Literacy Research Network, and the

Campbell McLaurin Chair for Hearing Deficiencies.

References

Abel, S.M., Krever, E.M., Alberti, P.W., 1990. Auditory detection,

discrimination and speech processing in ageing, noise-sensitive and

hearing-impaired listeners. Scand. Audiol. 19, 43–54.

Abramson, A., Lisker, L., 1970. Discrimination along the voicing

continuum: cross-language tests. In: Proceedings of the 6th

International Congress of Phonetic Science, Prague, 1967. Aca-

demic Press, Prague, pp. 569–573.

50 M. Tomita et al. / Hearing Research 193 (2004) 39–50

Bregman, A.S., Levitan, R., Liao, C., 1990. Fusion of auditory

components: effects of the frequency of amplitude modulation.

Percept. Psychophys. 47, 68–73.

Chen, G.D., Nuding, S.C., Narayan, S.S., Sinex, D.G., 1996.

Responses of single neurons in the chinchilla inferior coliculus to

consonant–vowel syllables differing in voice-onset time. Aud.

Neurosci. 3, 179–198.

Committee on Hearing, Bioacoustics, and Biomechanics, Commission

on Behavioral and Social Sciences and Education, National

Research Council (CHABA) 1998. Speech understanding and

aging. Working Group on Speech Understanding and Aging.

J. Acoust. Soc. Am. 83, 859–895.

Eggermont, J.J., 1990. The Correlative Brain: Theory and Experience

in Neural Interaction. Springer, Berlin.

Eggermont, J.J., 1991. Rate and synchronization measures of

periodicity coding in cat primary auditory cortex. Hear. Res.

56, 153–167.

Eggermont, J.J., 1995a. Neural correlates of gap detection and

auditory fusion in cat auditory cortex. Neuroreport 6, 1645–1648.

Eggermont, J.J., 1995b. Representation of a voice onset time contin-

uum in primary auditory cortex of the cat. J. Acoust. Soc. Am. 98,

911–920.

Eggermont, J.J., 1996. How homogeneous is cat primary cortex.

Evidence from simultaneous single-unit recordings. Aud. Neurosci.

2, 76–96.

Eggermont, J.J., 1999. Neural correlates of gap detection in three

auditory cortical fields in the cat. J. Neurophysiol. 81, 2570–2581.

Eggermont, J.J., 2000. Neural responses in primary auditory cortex

mimic psychophysical, across-frequency-channel, gap-detection

thresholds. J. Neurophysiol. 84, 1453–1463.

Eggermont, J.J., Ponton, C.W., 2002. The neurophysiology of audi-

tory perception: from single units to evoked potentials. Audiol.

Neurootol. 7, 71–99.

Glasberg, B.R., Moore, B.C., Bacon, S.P., 1987. Gap detection

and masking in hearing-impaired and normal-hearing subjects.

J. Acoust. Soc. Am. 81, 1546–1556.

Gordon-Salant, S., Fitzgibbons, P.J., 1993. Temporal factors

and speech recognition performance in young and elderly listeners.

J. Speech Hear. Res. 36, 1276–1285.

Humes, L.E., 1991. Understanding the speech-understanding problems

of the hearing impaired. J. Am. Acad. Audiol. 2, 59–69.

Kuhl, P.K., Miller, J.D., 1978. Speech perception by the chinchilla:

identification function for synthetic VOT stimuli. J. Acoust. Soc.

Am. 63, 905–917.

Mazelova, J., Popelar, J., Syka, J., 2003. Auditory function in presby-

cusis: peripheral vs. central changes. Exp. Gerontol. 38, 87–94.

Moore, B.C., 1985. Frequency selectivity and temporal resolution in

normal and hearing-impaired listeners. Br. J. Audiol. 19, 189–201.

Moore, B.C., 1993. Temporal analysis in normal and impaired hearing.

Ann. N.Y. Acad. Sci. 682, 119–136.

Morse, P.A., Snowdon, C.T., 1975. An investigation of categorical

speech discrimination by rhesus monkeys. Percept. Psychophys. 17,

9–16.

Nore~na, A.J., Eggermont, J.J., 2003. Changes in spontaneous neural

activity immediately after an acoustic trauma: implications for

neural correlates of tinnitus. Hear. Res. 183, 137–153.

Nore~na, A.J., Tomita, M., Eggermont, J.J., 2003. Neural changes in

cat auditory cortex after a transient pure-tone trauma. J. Neuro-

physiol. 90, 2387–2401.

Phillips, D.P., Taylor, T.L., Hall, S.E., Carr, M.M., Mossop, J.E.,

1997. Detection of silent intervals between noises activating

different perceptual channels: some properties of ‘‘central audi-

tory’’ gap detection. J. Acoust. Soc. Am. 101, 3694–3705.

Rajan, R., 1998. Receptor organ damage causes loss of cortical

surround inhibition without topographic map plasticity. Nat.

Neurosci. 1, 138–143.

Rajan, R., 2001. Plasticity of excitation and inhibition in the receptive

field of primary auditory cortical neurons after limited receptor

organ damage. Cereb. Cortex 11, 171–182.

Schneider, B.A., Hamstra, S.J., 1999. Gap detection thresholds as a

function of tonal duration for younger and older listeners.

J. Acoust. Soc. Am. 106, 371–380.

Shannon, R.V., Zeng, F.G., Kamath, V., Wygonski, J., Ekelid, M.,

1995. Speech recognition with primarily temporal cues. Science 270,

303–304.

Sinnott, J.M., Adams, F.S., 1987. Differences in human and monkey

sensitivity to acoustic cues underlying voicing contrasts. J. Acoust.

Soc. Am. 82, 1539–1547.

Sinex, D.G., McDonald, L.P., 1988. Average discharge rate represen-

tation of voice onset time in the chinchilla auditory nerve.

J. Acoust. Soc. Am. 83, 1817–1827.

Sinex, D.G., McDonald, L.P., 1989. Synchronized discharge rate

representation of voice-onset time in the chinchilla auditory nerve.

J. Acoust. Soc. Am. 85, 1995–2004.

Snell, K.B., 1997. Age-related changes in temporal gap detection.

J. Acoust. Soc. Am. 101, 2214–2220.

Snell, K.B, Frisina, D.R., 2000. Relationships among age-related

differences in gap detection and word recognition. J. Acoust. Soc.

Am. 107, 1615–1626.

Snell, K.B., Mapes, F.M., Hickman, E.D., Frisina, D.R., 2002. Word

recognition in competing babble and the effects of age, temporal

processing, and absolute sensitivity. J. Acoust. Soc. Am. 112, 720–

727.

Steinschneider, M., Fishman, Y.I., Arezzo, J.C., 2003. Representation

of the voice onset time (VOT) speech parameter in population

responses within primary auditory cortex of the awake monkey.

J. Acoust. Soc. Am. 114, 307–321.

Steinschneider, M., Schroeder, C.E., Arezzo, J.C., Vaughan Jr., H.G.,

1995. Physiologic correlates of the voice onset time boundary in

primary auditory cortex (A1) of the awake monkey: temporal

response patterns. Brain Lang. 48, 326–340.

Steinschneider, M., Volkov, I.O., Noh, M.D., Garell, P.C.,

Howard 3rd, M.A, 1999. Temporal encoding of the voice

onset time phonetic parameter by field potentials recorded

directly from human auditory cortex. J. Neurophysiol. 82,

2346–2357.

Strouse, A., Ashmead, D.H., Ohde, R.N., Grantham, D.W., 1998.

Temporal processing in the aging auditory system. J. Acoust. Soc.

Am. 104, 2385–2399.

Tremblay, K.L., Piskosz, M., Souza, P., 2003. Effects of age and age-

related hearing loss on the neural representation of speech cues.

Clin. Neurophysiol. 114, 1332–1343.

Walton, J.P., Frisina, R.D., O’Neill, W.E., 1998. Age-related alter-

ation in processing of temporal sound features in the auditory

midbrain of the CBA mouse. J. Neurosci. 18, 2764–2776.

Wang, J., Ding, D., Salvi, R.J., 2002. Functional reorganization in

chinchilla inferior colliculus associated with chronic and acute

cochlear damage. Hear. Res. 168, 238–249.

Effects of an acute acoustic trauma on the representation of a voice onset time continuum in cat primary auditory cortexIntroductionMethodAnimal preparationPeripheral threshold estimationAcoustic stimulus presentationAcoustic traumaRecording and spike separation procedureData analysis

ResultsEffects of the acoustic trauma on thresholdsIndividual exampleGroup data for the /ba/-/pa/ continuum and the ``early'' gap condition

DiscussionNeurophysiological mechanismsNeural changes induced by the acoustic traumaPotential perceptual correlates of the central changes that follow an acoustic trauma

AcknowledgementsReferences