A comparison of AP and ABR tuning curves in the guinea pig

Heurivg

Reseurch, 25 (1987)

193

193-204

Elsevier

HRR

00850

A comparison

of AP and ABR tuning curves in the guinea pig* Carolyn J. Brown? and Paul J. Abbas

Depc~~~tt

o[ Speech Pathoiogv and Audioiogv. Uniuer.sit.v of lowu.lowuCity, IA 5.??4_‘. U.S.4

(Received

10 March

1986: accepted

IO August

1986)

AP and ABR tuning curves were measured using a forward masking paradigm in guinea pigs with chronically implanted electrodes. Measurements were made before exposure to wide-band noise and at several intervafs after exposure. The noise exposure was sufficient to produce temporary threshold shifts up to 60 dB lasting several days. Results showed similar reductions in Q,,,, tip-to-tail ratio and slope of simultaneously recorded AP and ABR tuning curves as a function of threshold shift following noise exposure. Tuning curves based on changes in response latency are presented as well as tuning curves based on changes in response amplitude. AP and ABR latency tuning curves showed similar form and changes with hearing loss as amplitude tuning curves. The similarities between AP and ABR tuning characteristics provide evidence that the ABR is sensitive enough to peripheral changes to be useful as a tool to study auditory frequency selectivity. The similarities between amplitude and latency tuning characteristics suggest that information regarding frequency selectivity of the auditory system can be obtained using response latency as well as response amplitude. brainstem,

whole nerve action

potential,

tuning

curve, forward

Introduction The whole nerve action potential (AP) and the auditory brainstem response (ABR) have become widely used tools in the diagnosis of auditory pathology and in the assessment of the sensitivity and function of the cochlea. Both evoked potentials are summed neural responses to acoustic stimulation characterized by a series of differentially recorded peaks. The AP is typically recorded from the round window and consists of a series of two negative peaks (N, and N2) occurring within the first few milliseconds following acoustic stimulation. The first peak, N,, has been shown to reflect activity at the level of the auditory nerve (Kiang et al., 1976). When recorded differentially between vertex

i

Corrwpondence address: Wendell ing Clinic, no. 305. University of U.S.A. * Preliminary results of this study meeting of the Acoustical Society 1985.

037%5955/87/$03.50

Johnson Speech and HearIowa, Iowa City, IA 52242, were presented at the Fall of America, Nashville, TN,

6 19X7 Elsevier Science Publishers

masking,

guinea

pig

and mastoid electrodes the scalp-recorded ABR is characterized by a series of five to seven positive peaks (three to four in animals) that occur within six to ten milliseconds following acoustic stimulation. The first peak of the ABR has been shown to be essentially a far-field recording of the AP (Spire et al., 1982). The later peaks of the ABR include contributions from higher levels of the auditory tract than the auditory nerve (Lev and Sohmer, 1972; Buchwald and Huang, 1975; Achor and Starr, 1980a, b). A number of studies have used these evoked potentials in a simultaneous or forward masking paradigm in order to obtain information about the tuning characteristics of auditory neurons (Dallos and Cheatham, 1976; Eggermont, 1977; Mitchell. 1976; Harris 1979; Abbas and Gorga, 1981; Harrison et al., 1981; Dolan et al., 1985a, b). Three studies have been published to date making direct comparison of AP and ABR tuning characteristics. Mitchell and Fowler (1980) using a simultaneous masking paradigm, found no significant differences among tuning characteristics of the AP, wave I or wave III of the ABR in normal

B.V. (Biomedical

Division)

194

guinea pigs. Dolan et al. (1985a) obtained AP and ABR tuning curves using a forward masking paradigm in gerbils with normal evoked potential thresholds and compared these tuning curves to single fiber tuning curves. Measures of Q,,, tipto-tail ratio and high frequency slope of these three types of tuning curves (AP, ABR, neural) were very similar. Klein and Mills (1981a, b) used normally hearing human subjects and a simultaneous masking paradigm to collect tuning curves of the first and fifth peaks of the ABR as well as psychophysical tuning curves. They found wave V and the psychophysical tuning curves to be similar to each other, while the wave I tuning curve, representing the peripheral response, was more broadly tuned. A follow-up study (Klein and Mills, 1981b) showed that exposure to a bandpass noise sufficient to produce temporary threshold shifts of approximately 30 dB resulted in a loss of tuning for all three responses (wave I, wave V, and psychophysical). The purpose of the present study was to investigate changes in tuning of the AP and ABR in guinea pig using a forward masking paradigm as a function of threshold shift post-exposure. The intent was to ascertain the degree to which the most robust wave of the ABR (wave III in guinea pig) can be used to predict changes in frequency selectivity that are correlated with changes in sensitivity as measured by the AP tuning curve. Because ABR amplitude can be relatively noisy in humans and in unanesthetized subjects, a second goal of this study was to determine whether information relative to changes in frequency selectivity could be obtained by measuring changes in response latency with masking rather than changes in response amplitude. Methods Chronic AP and ABR electrodes were implanted into four normal, pigmented guinea pigs weighing between 500 and loo0 g following a technique outlined by Aran and Erre (1979). A combination of ketamine (30 mg/kg) and Rompun (6 mg/kg) was administered intramuscularly to anesthetize the animals prior to surgery and for subsequent recording sessions. Supplementary doses of the anesthetic were used as necessary and

lidocaine was administered topically. During surgery and the subsequent recording sessions, body temperature was monitored rectally and maintained at approximately 38*C by a warm water circulating pad placed under the animal. The electrode assembly consisted of a modified four prong female adapter to which three lengths of insulated silver wire were soldered. A dorsal incision was made from a point anterior to the vertex along the sagittal suture to a point 5-10 mm anterior to the inion. A second incision was made laterally from this point to the auditory bulla. The adapter was cemented to the top of the skull with dental acrylic. One electrode was attached to a stainless steel screw at the vertex. A second electrode was placed in the soft tissue just posterior to the bulla. The third was placed on the bony ridge just inferior to the round window. The wound was then irrigated with normal saline and sutured. The animal was allowed to recover and was given a course of antibiotics (oxytetracyline; 25 mg/kg) for a period of one to two weeks following surgery. Subsequent recording sessions were nontraumatic and required that the animals be anesthetized only so that they would tolerate placement of the earphone/probe assembly in the external auditory meatus for the time necessary to make the recordings. During the recording sessions a probe tube was inserted into the external auditory meatus under visual inspection and acoustic stimuli were delivered via a Koss HVX dynamic earphone sealed to the probe assembly. Sound pressure levels were measured with a calibrated Knowles DBA microphone placed within approximately 5 mm of the tympanic membrane. The animal was contained within a double-wall, sound-treated chamber throughout the recording sessions. The probe stimulus was a 10 or 14 kHz tone burst with a 1 ms rise-fall time (10% to 90% points) and a 10 ms plateau (measured at 50% points). AP threshold curves for guinea pigs show a broad region of maximum sensitivity that ranges from approximately 4 to 16 kHz (Cazals et al., 1980; Syka and Popelar, 1980). A probe frequency of either 10 or 14 kHz was chosen for each animal in this study, depending on which threshold was lowest. The masker was a tone burst (10 ms rise-fall and 50 ms duration) that was systemati-

195

tally varied in frequency from 4000 to 17000 Hz. A forward masking paradigm was used with the silent interval between the masker and the probe set to 10 ms. Stimuli were presented at a rate of 5/s. Both probe and masker were high-pass filtered at 2000 Hz and attenuated independently before being mixed and transduced. The AP and ABR were recorded simultaneously in two separate, identical channels. The differentially recorded signals were passed through two stages of amplification providing a total gain of 100 000 and were bandpass filtered between 300 and 3000 Hz. The AP and ABR signals were sampled at a rate of 20000/s on each channel. Five hundred twelve sweeps were averaged for each response. Data were analyzed off-line using a separate analysis program. The amplitude of the AP was measured from the first negative peak (N,) to the following positive peak (P,). Wave III of the ABR response was measured between the third positive peak and the following negative excursion. Latency measures are reported relative to onset of the tone burst at the tympanic membrane. At the onset of each experimental session AP and ABR probe thresholds were measured at 8, 10, 14, 18 and occasionally at 4 kHz. Threshold was defined as the lowest intensity level at which a response was still visible. The probe level was set to 15 dB relative to the ABR threshold. AP and ABR thresholds were typically within 10 dB of each other with AP thresholds being slightly lower than the ABR thresholds. Responses to the probe signal alone (no masker) were collected as controls prior to each masking series. A masker was introduced and increased in level in 10 dB steps until the response to the probe was depressed more than 50% relative to the unmasked condition. During off-line analysis the amplitude of each response was normalized to the control response (Abbas and Gorga, 1981) so that comparisons could be made between ears and across conditions. Both amplitude and latency tuning curves were constructed using a piecewise linear interpolation. Amplitude tuning curves were plotted as the level of masker which caused a 40% decrement in normalized amplitude of response to the probe as a function of masker frequency. Latency tuning curves were plotted as the level of masker which

resulted in a 0.1 ms increase in probe response latency as a function of masker frequency. A 3 h recording session was necessary to obtain a set of 6 to 9 point amplitude and latency tuning curves. After surgery, a pre-exposure tuning curve was obtained for each animal. The animal was then exposed (while awake) to a white noise 110-112 dB SPL for 6-8 h in a sound-treated booth. The first tuning curves were obtained 6612 h post-exposure and then in 24-48 h intervals until recovery from TTS was complete. Thresholds were obtained at the onset of each experimental session allowing consideration of changes in the tuning characteristics of the AP and ABR as a function of the amount of threshold shift for the same animal. We present data from four animals, one of which was subsequently re-exposed two weeks following complete recovery of thresholds from the initial exposure, thus, data from five series of tuning curves are presented. Results

Examples of AP and ABR tuning curves obtained for one animal are shown in Fig. l(A-F). The masker level necessary to achieve a 40% decrement in normalized amplitude as a function of masker frequency is plotted for both the AP and the ABR tuning curves. Probe frequency and level are indicated by the square symbol on each tuning curve. The pre-exposure tuning curve is shown in Fig. 1A. Fig. lB-F shows tuning curves obtained sequentially for a period of seven days following exposure as indicated in the figure legend. At 12 h following exposure, tuning is very broad (Fig. 1B). As threshold recovers to pre-exposure levels, tip level drops, the sharpness of tuning around the tip returns and the tip-to-tail ratio becomes larger. Similarities between the AP and ABR tuning curves are apparent. A series of latency tuning curves obtained preand post-exposure is shown in Fig. 2(A-F). The level of masker necessary to cause an increase in response latency of 0.1 ms relative to the control latency is plotted as a function of masker frequency. Probe frequency and level are indicated by the square symbol on each tuning curve. These latency tuning curves were obtained from the same animal whose amplitude tuning curves are shown

100

100

60

80

60

eo

3

3

4 ii! 5 5 rd

40

l

40

AP (8.81

20

20

B

A 0

0

100

B

I

80

60-

60

w

a

a

0: 10

12

14

te

e

4

e

10

12

14

ie

t6

00

80.

3

* AP 0.8) 0 ABR

0 A8R (10.3)

iii

a '

'O-

. AP

40

0.1)

o ABR n.01

?

0 AP (7.0) o ABR (7.8)

20

20-

D

E

Of 4

12

14

16

4 10

0

WIZ)

i4 ie

4

&

&

(kllr)

Fig. 1. AP and ABR tuning curves obtained for one animal are shown. The tuning curves in (A) were collected prior to exposure. (B-F) Tuning curves collected at 12,36,60,96 and 168 h post-exposure, respectively. The square symbol represents probe frequency and level. In (F) the probe frequency is 14 kHz and the probe level is 23 dB SPL. The value of Qro calculated for each tuning curve is shown in parentheses.

in Fig. l(A-F). The pre-exposure latency tuning curve is shown in Fig. 2A. Following exposure these latency tuning curves show the same trends toward broader tuning and smaller tip-to-tail ratios with increasing threshold shift as their amplitude counterparts (follow Fig. 2B-F). A similar set of pre- and post-exposure ampli-

tude tuning curves was obtained for each of the 5 exposures. Latency tuning curves were obtained for four of the five exposures. Two linear regression equations were computed for each tuning curve. One used the three points of the function just above and including the tip of the tuning curve and the other used the three points of the

80

* AP .(13.3) _.--0 a_S?_(El@.

a

0

e

10

MASKER F~E~UE~CY

12

14

(kHz1

18

4

8

8

10

12

14

MASKER FREQUENCY kHd

15

4

6

8

10

12

14

Ill 18

MASKER FREQUENCY [k&z)

Fig. 2. AP and ABR latency tuning surves obtained for one animal are shown. The latency tuning curves shown in {A) were obtained prior to exposure. (B-F) Latency tuning curves collected at 32, 36. 60, 96, and 168 h post-exposure. respectively. The square symbol represents probe frequency and level. The value of QtO calculated for each tuning curve is shown in parentheses.

function just below and including the tip of the tuning curve. Q,, was calculated by dividing the tip frequency by the bandwidth of the best fit regression lines at a level 10 dB above the level at the tip. These regression equations were also used to calculate slope of the high and low frequency portions of the tuning curves. Fig. 3 shows AP and ABR Qlo as a function of

the magnitude of threshold shift for all five exposure series. This figure shows QIO values calculated from both amplitude (filied and open squares) and latency jfilled and open circles) tuning curves. QxO values obtained for AP tuning curves are indicated by filled symbols. Open symbols are used to represent Qro values of the ABR tuning curves. QIo values obtained for both am-

0 ABR LATENCY

0 0. 1

0

/

10

20

I

/

I

30

40

60

:. I

00

70

90

THRESHOLD SHIFT (dB) Fig. 3. Qlo as a function of threshold shift is shown for five exposure series. Filled symbols represent AP &, data. Open symbols represent ABR Qlo data. Filled and open squares represent data from amplitude tuning curves. Filled and open circles represent data from latency tuning curves. Points below the zero line indicate cases where Q,, could not be measured (see text).

plitude and latency tuning curves collected with 0 dB threshold shift (including both pre-exposure and after complete recovery) range from 7 to 16. These Q,, values are similar both in absolute magnitude and range to those reported by Harrison et al. (1981) for forward masked AP tuning curves in guinea pigs with normal AP thresholds for probe frequencies from 10 to 18 kHz. Q,, is shown to decrease as a function of magnitude of the threshold shift. At threshold shifts greater than 20 dB, although there are fewer datum points, Q,, values are substantially lower ranging from 1.5 to 9. In three cases Q,, was not measurable because tuning became so broad that the tip could not be identified (Fig. 1B; ABR is one example). These data are plotted below the Q,, scale in Fig. 3 for illustration. They were not included in subsequent regression calculations. Data from both AP and ABR, amplitude and latency tuning curves, appear to show no differences. Paired-cumparison r-tests were performed pairing simultaneously collected AP and ABR Q,, values for both amplitude and latency tuning curves. No significant differences between AP and ABR Q,, values were found for either

type of tuning curve. AP and ABR Q,, data were then combined and linear regression analysis was performed on the Q,, versus threshold shift data shown in Fig. 3. Both amplitude and latency tuning curves show Qlo values that decrease significantly with increasing threshold shift (amplitude: F&46) = 42.97, P < 0.01; latency: F(1,40) = 38.8, P < 0.01). Additionally, the equation obtained describing the regression of Qlo on threshold shift for the amplitude tuning curves is not significantly different from the equation obtained by regressing the Q,, values of the latency tuning curves on threshold shift. Fig. 4 shows tip-to-tail ratios for the same AP and ABR tuning curves as a function of threshold shift post-exposure. In order to limit the time each recording session required and to insure that the tuning curves would be relatively well defined around the tip, fewer points were obtained on the lower frequency portion of the tuning curve. Consequently, the two slope, low frequency tails generally reported in the literature for neural tuning curves were not always well defined. Since the thresholds at 4000 and 8000 Hz were typically very similar and since several tuning curves did not have a 4000 Hz point, tip-to-tail ratio was --___ ooII

w

9 AP: AMPLilUf_lE L: ABR: AMPLilUDE 0 AP: LATEECY 0 ABR LATENCY

Y

U. /

0

u

/

10

20

30

40

90

a0

70

90

THRESHOlDSHlFf (da Fig. 4. Tip-to-tail ratios are shown as a function of threshold shift for five exposure series. Filled symbols represent AP data. Open symbols represent ABR data. Filled and open squares represent data from amplitude tuning curves. F&d and open circles represent data from latency tuning curves. Points below the zero line indicate cases where the tip-to-tail ratio could not be calculated (see text).

199

arbitrarily defined as the difference in dB between the level of an 8000 Hz masker and the level at the tip frequency necessary to produce a criterion change in response. A 10000 Hz probe rather than a 14000 Hz probe was used for one animal and in this case the tip-to-tail ratio was calculated using a 6000 Hz rather than 8000 Hz masker. Both amplitude and latency tip-to-tail ratios for threshold shifts of 0 to 20 dB range from 45 to 85 dB and decrease to less than 25 dB at threshold shifts greater than 30 dB. In three cases the tuning was so broad that the tip of the tuning curve could not be identified and the tip-to-tail ratios could not be calculated. These data are plotted below the zero axis on Fig. 4 for illustrative purposes but were not included in the statistical analyses. Cases in which the maximum level of the 8000 or 6000 Hz masker produced less than a 40% decrement in normalized amplitude of response to the probe or less than a 0.1 ms increase in response latency are not included in Fig. 4 or the statistical analyses. Paired-comparison t-tests were used to compare simultaneously collected AP and ABR tipto-tail ratios for both amplitude and latency tuning curves. Tip-to-tail ratios of amplitude tuning curves were significantly larger for the AP than the ABR (t(43) = 3.72, P < 0.01). AP and ABR tip-to-tail ratios obtained from amplitude tuning curves were treated separately and the tip-to-tail ratios were regressed on threshold shift using a linear model. Both AP and ABR tip-to-tail values were found to decrease significantly with increasing threshold shift (AP: F(1,22) = 85.3, P < 0.01; ABR: F(1,20) = 49.0, P < 0.01). Slope and Y intercept of the AP and ABR regression equations were compared using r-tests. The slopes of the two functions were not significantly different from each other. The Y intercept of the AP response was significantly larger than the Y intercept of the ARR response (t(22) = 2.9, P < 0.01). These results suggest that similar changes in tuning of the two responses (as measured by tip-to-tail ratio) occur as a function of threshold shift following acoustic trauma. Latency tuning curves did not show significant differences between AP and ABR tip-to-tail ratios. These latency data were combined, and tip-to-tail ratio was regressed on threshold shift using a linear model. Tip-to-tail values obtained from latency tuning curves de-

creased significantly with increasing threshold shift (F(1,36) = 163.3, P -c 0.31). No significant differences were found between the equation describing the regression of latency tip-to-tail ratios on threshold shift and either of the equations describing the regression of AP amplitude tip-totail ratio on threshold shift or the regression of ABR amplitude tip-to-tail ratio on threshold shift. These results further suggest that similar changes in tuning occur for both amplitude and latency measures. If the AP and ABR data are analyzed separately for either the amplitude or the latency tuning curves and Q,, of each response is regressed on threshold shift, the two resulting equations are not significantly different from each other or from the regression equation calculated using the combined AP and ABR data. In addition, both the amplitude and latency tuning curves show variability of the AP around the regression line that is similar to the variability of the ABR around the regression line as judged by comparison of the standard error of both estimates. Similarities between the standard error of the estimate of AP Q,, around the regression line and the standard error of the estimate of the ABR QlOaround the regression line suggest that AP and ABR data can be used with equal precision to measure changes in tuning that occur with temporary threshold shift. Fig. 5 shows the slopes (dB/oct) of the high frequency portion of the tuning curves (masker frequency greater than or equal to the probe frequency) as a function of threshold shift. Fig. 6 shows the slopes (dB/oct) of the low frequency portion of the tuning curves (masker frequency less than or equal to the probe frequency) also as a function of threshold shift. In both of these figures filled symbols indicate AP data while open symbols indicate ABR data. Filled and unfilled squares represent data from amplitude tuning curves, filled and unfilled circles represent data from latency tuning curves. Paired-comparison r-tests failed to indicate a significant difference between AP and ABR data for either the slope of the high frequency portion of the tuning curves shown in Fig. 5 or the low frequency slope data shown in Fig. 6. Linear regression analyses were computed on the combined AP and ABR data. The regression line

200

600

iii

P

9

n AP:

AMPUTUDE

j AER

. 600-

13

3

. AMPLiTUDE

0 AP:

P

LATENCY

0 ABR, LATENCY

0

POOa

Y 83oo-

1

G .

is 1

zoo-

P I

loo-

!

P 0’

I

0

,

10

20

30

40

60

1

60

70

80

THRESHOLD SHIFT (dB) Fig. 5. Slope of the best fit regression line of the high frequency portion of the tuning curve is shown as a function of threshold shift. Filled symbols indicate AP data. Open symbols indicate ABR data. Filled and open squares indicate data from amplitude tuning curves. Filled and open circles indicate data from latency tuning curves.

fitting the high frequency data shown in Fig. 5 decreased significantly with increasing threshold shift for both amplitude and latency tuning curves (amplitude: F(1,45) = 20.4, P < 0.01;latency: F&40) = 27.2, P -c0.01). The reduction in slope of the low frequency portion of the tuning curves with increasing threshold shift was also significant

~1

l !3 :

.

c

_I .

:

Id

4 2 l” B

. AP: AMPLITUDE n ABR: AMWfUDE

-300

. AP: LATENCY I

30

0 ABR: LATENCY

-I

-400 A

, 0

/ IO

1 20

I 30

/ 40

I 60

/ 60

/

m

I

80

THRESHOLD SHIFT (dB) Fig. 6. Slope of the best fit regression line of the low frequency portion of the tuning curve is shown as a function of threshold shift, Filled symbols indicate AP data. Open symbols indicate ABR data. Filled and open squares indicate data from amplitude tuning curves. Filled and open circles indicate data from latency tuning curves.

(amplitude: F(1,45) = 20.9, P < 0.01;latency: F(1,40) = 36.9, P -c0.01). The reduction in slope of the high and low frequency sides of the tuning curves is consistent with a reduction in Qlo. The results of this study show ABR tuning characteristics which correspond closely with AP tuning characteristics as measured by Q,, and tip-to-tail ratios. Following exposure both responses showed similar reductions in the sharpness of tuning and tip-to-tail ratios with increases in the amount of threshold shift. Similar reductions of AP and ABR measures of PI0 and tip-totail ratio with threshold shift were found for tuning curves constructed on the basis of either changes in response amplitude or response latency with masking. Discussion The AP and ABR (along with other electrophysiologic responses) have been used to assess the effects of noise exposure at central versus peripheral levels of the auditory system. Comparison of changes in sensitivity and the time course of recovery of peripherally generated responses, such as the AP and the cochlear microphonic (CM), with more centrally generated responses (for example, from the cochlear nucleus, inferior colliculus or wave V of the ABR) following noise exposure reveals differences which appear to be dependent on the exposure paradigm used. Results of studies using long duration noise exposures sufficient to produce asymptotic temporary threshold shifts suggest that the disorder associated with TTS is primarily a cochlear phenomenon (Salvi, 1976; Benitez et al., 1972). However, shorter or lower level noise exposures have resulted in greater threshold shift and slower courses of recovery for units from the cochlear nucleus and inferior colliculus than for the more peripherally generated AP and CM (Salvi, 1976; Babighian et al., 1975). Adaptation studies have also revealed differential effects of an adapter stimulus at central and peripheral levels. The amplitude of wave V of the ABR (representing central activity) has been shown to be more resistant to adaptation than the amplitude of wave 1 of the ABR or the AP (representing more peripheral activity). Wave V also shows greater changes in

201

latency than does N, in the presence of an adapting stimulus (Thornton and Coleman, 1975; Kevanishvili and Lagidze, 1979; Tietze and Gobsch, 1980; Kramer and Teas, 1982). Given the potentially different effects of noise exposure and adaptation at a peripheral versus central level, it may not be unreasonable to expect differences in tuning characteristics of the AP and ABR. The present data suggest otherwise. In Figs. 3 and 4 both amplitude and latency tuning curves show a decrease in Q,, and tip-to-tail ratios as a function of the amount of threshold shift measured following exposure. AP and ABR tuning characteristics as measured by Q,, were not significantly different from each other regardless of the amount of threshold shift present. Similar findings of reduced Qlo and tip-to-tail ratios for AP and single fiber tuning curves in impaired ears have been reported by Van Heusden and Smoorenburg (1981) and by Harrison et al. (1981). Harrison et al. (1981) obtained Q,, values for normally hearing guinea pigs which ranged from 6 to 12 for probe frequencies between 10 and 18 kHz which is consistent with the values in Fig. 3. For threshold shifts of greater than 30 dB Harrison et al. (1981) reported Q10 values for forward masked AP tuning curves that range from 2 to 4, also consistent with the AP and ABR Qu, values reported here for animals with threshold shifts of similar magnitude. In contrast to these studies, Gorga and Abbas (1981) reported forward masked AP tuning curves in normal and acoustically traumatized cats with permanent threshold shifts. Their data showed no difference in Q,, between the normal and impaired animals. There is also evidence that single fiber and psychophysical tuning curves may retain sharply tuned tip regions with hearing impairment (Dallos et al., 1977; Ryan et al., 1979). The mechanism responsible for producing temporary threshold shifts and that responsible for creating permanent threshold shifts may be quite different and may contribute to the apparent discrepancy between these earlier results and those of the present study. Certainly the influence of changing probe level on these data should not be ignored. At the largest threshold shifts probe level was high enough that the low Q,, values may reflect the spread of

excitation of the probe on the basilar membrane rather than or in addition to changes in the underlying tuning characteristics of a particular subset of fibers responding to the probe. However, because the pre-exposure thresholds of the animals used in this study were low (ranging from 0 to 22 dB SPL), the probe levels used, even with threshold shifts as large as 50-60 dB, were less than 75580 dB SPL. Within this range probe level has been shown to have little effect on tuning characteristics (Dallos and Cheatham, 1976; Saunders et al., 1980; Gorga and Abbas, 1981). One finding that seems to be consistent in the literature is that tip-to-tail ratio is reduced in the impaired ear. The data presented in Fig. 4 show that when threshold shifts are minimal (O-10 dB), tip-to-tail ratios range from 45 to 85 dB. These tip-to-tail ratios are of the same order of magnitude as tip-to-tail ratios reported previously for AP and ABR tuning curves obtained from animals with normal evoked potential thresholds (Eggermont, 1977; Gorga and Abbas, 1981; Van Heusden and Smoorenburg, 1981; Gorga et al., 1983; Shepard and Abbas, 1983; Dolan et al., 1985a). The AP tip-to-tail ratios for amplitude tuning curves were significantly larger than the ABR tip-to-tail ratios; however, both responses showed similar reductions in tip-to-tail ratio with threshold shift (see Fig. 4). Several investigators have reported tip-to-tail ratios in impaired ears similar to those shown in Fig. 4 (Eggermont. 1977; Gorga and Abbas, 1981; Van Heusden and Smoorenburg, 1981; Shepard and Abbas, 1983). Regardless of whether or not changes in frequency selectivity near the tip are seen, the effectiveness of maskers with frequencies less than that of the probe increases with loss of evoked potential sensitivity. Several previous studies have investigated differences between AP and ABR tuning curves. Mitchell and Fowler (1980) using a simultaneous masking paradigm and animals with normal evoked potential thresholds, found no significant differences in tuning among the AP, the first and third waves of the ABR. Dolan et al. (1985a), using a forward masking paradigm, also found similar AP and ABR tuning characteristics for gerbils with normal response thresholds. The results of the present study, while not directly com-

parable with these studies due to differences in probe frequency and masking paradigm, support this finding and extend it to suggest that the tuning characteristics of the AP and ABR show similar changes following temporary threshold shifts due to acoustic trauma. Klein and Mills (198la, b) measured changes in AP and ABR tuning with threshold shift using human subjects. They reported simultaneously masked ABR wave I, ABR wave V, and psychophysical tuning curves for five human subjects both before and after exposure to a narrow band of noise of sufficient level to produce 20 to 30 dB temporary threshold shifts. Only three maskers with frequencies lower than the probe frequency were used to define the low frequency portion of the tuning curves, and two maskers with frequencies higher than the probe frequency were used to define the high frequency portion of the tuning curves. They found all three types of tuning curves to be slightly wider following noise exposure. For two of their subjects the wave I tuning curve exhibited greater changes post-exposure than the wave V tuning curve due primarily to changes in the high frequency portion of the tuning curve. That is, for a given masker level, higher masker frequencies were necessary post-exposure to obtain threshold estimates than were necessary prior to exposure, implying that there may be a reduction in the high frequency slope of the tuning curve after exposure. The data in Figs. 5 and 6 show no significant difference between slope of the AP and the ABR for either the steeper high frequency portion of the tuning curves or the more shallow low frequency portion of the tuning curves. In the present data both sides of the tuning curves showed a reduction in slope with threshold shift indicating that the reduction in sharpness of tuning is due to a reduction in the slope of both sides of the tuning curve rather than the high frequency portion as Klein and Mills’ (1981b) human data may suggest. The present data demonstrate that the frequency selectivity of the AP and ABR are similar to each other in normal and impaired ears suggesting that the mechanism responsible for determining frequency selectivity lies peripheral to or at the generation site of the AP. The changes in frequency selectivity measured indicate that with loss of

evoked potential sensitivity the tuning of the subset of fibers that respond to the probe becomes broader resulting in widening of the tuning curves with increasing threshold shifts. These data also suggest that the ABR, although a central measure, is sensitive enough to reflect these peripheral changes. The interpretation of amplitude based AP or ABR tuning curves in a forward-masking paradigm is relatively straightforward (Abbas and Gorga, 1981). Tuning curves based on changes in stimulus latency are more complex. The presence of a masker also affects the latency in response to a probe. Generally, a masker will reduce the neural response to a probe. Such a reduction in neural responsiveness could increase the time required to reach threshold, and therefore, could increase response latency throughout the auditory pathway. This increase in response latency should be relatively independent of masker frequency. Additionally, a masker may reduce the responsiveness of fibers with high characteristic frequency resulting in an increase in response latency due to increased travel time along the cochlear partition. If that is the case, one may expect high frequency maskers to reduce the number of high frequency fibers that contribute to the response to the probe thereby increasing the response latency. Maskers lower in frequency than the probe may have little effect on response latency. Certainly both of these factors (as well as others not mentioned) may affect the latency of response to a probe in the presence of a masker. The present study was designed to examine changes in amplitude of response with masking. Unfortunately, neither the masker intensity step size or the intensity range over which the masker was varied was sufficient to allow the issue of the origin of latency changes with masking to be addressed directly. Msller (1985) presents forward masked AP data that show changes in latency that are greater for maskers above probe frequency than for maskers below probe frequency. This may be consistent with a theory that latency changes are determined primarily by travel time along the basilar membrane. Further research needs to be done to examine the origins of these changes in latency with masking and to consider whether changes in response latency can be used

203

to infer masking patterns and place of excitation on the basilar membrane. The latency tuning curves presented in this paper are similar in shape and show trends with threshold shift that are virtually identical to their amplitude counterparts. The implication of these data is that information regarding frequency selectivity of the auditory system can be obtained using response latency rather than response amplitude. This finding may be significant as it relates to work with unanesthetized human subjects where amplitude of the ABR is known to be variable. The purpose of the present study was to investigate the degree to which the most robust wave in the ABR response can be used to assess changes in frequency selectivity due to changes in peripheral integrity/function. After acoustic trauma sufficient to cause threshold shifts that were significant but temporary in nature, a systematic decrease in frequency selectivity (Q,,) was observed due to a reduction in the slope of both the low and high frequency portions of the tuning curve and a decrease in the tip-to-tail ratio. No systematic differences were noted between AP and ABR tuning characteristics as a function of threshold shift and both types of evoked potentials showed similar trends. These data are in good agreement with Mitchell and Fowler (1980) and Dolan et al. (1985a) showing similar AP and ABR frequency specificity and suggesting that the mechanism for determining frequency specificity lies peripheral to or at the generation site of the AP response. In addition, the finding that. following noise exposure, both the AP and ABR provide similar estimates of frequency selectivity, suggests that the ABR is sensitive enough to peripheral changes to be useful as a tool to study auditory frequency selectivity. Acknowledgements

Part of this research was supported by a grant from the Deafness Research Foundation. The authors wish to thank Dr. C. Fowler for her helpful comments regarding a previous version of this paper and Dr. E. Luschei for his assistance with electrode design and implantation techniques.

References A&as. P. and Gorga. M. (1981) AP response in forward masking paradigms and their relationship to responses of auditoq-nerve fibers. .I. Acoust. Sec. Am. 69. 402-499 Achor. L. and Starr. A. (19XOa) Auditory brainstem response.\ in the cat. I. Intracranial and extracranial recordings. Electrocncephalogr. Clin. Neurophysiol. 4X. 154-173. Achor. L. and Starr. A. (19XOb) Auditory hralmtem rekponxs in the cat. II. Effects of lesions. Electroencephalogr. Clin. Neurophysiol. 4X. 174-190. Aran. J. and Erre. J. (lY79) Long-term recording of cochlea-neural potentials in the guinea pig. In: Auditory Investigations: The Scientific and Technological Hasi&. pp. 233-261. Editor: H. Beagley. Oxford. New York Bahighian. <;.. Moushegian. G. and Rupert. A (1975) Central auditory fatigue. Audiology 14. 72-X3. Benitcz, L., Eldredge. D. and Templer. J. (lY72) Temporaw threshold shifts in chinchilla: Electrophwiological cornlates. J. Acouht. Sot. Am 52. 1115~1123 Buchwald. J. and Huang. C. (197.5) Far-field acoustic rcsponaes: OrIgina in the cat. Sclcncc lXY, 3X2-3X4. Cazals. Y.. Aran. J. and Hawkins, J. (1YXO) Threshold &cation at high frequencies of the auditon neme action potential in acute versus chronic recordings in guinea pigs. Hear. Res 2. 95-10’). Dallw P. and Cheatham. M. (1976) Compound action potcntial (AP) tuning curve&. J. Acoust. Sec. Am 5Y. 5Y1-SY7, Dallas, P.. Ryan. A., Harris, D.. M&ice. T. and Ozdnmar. 0. (1977) Cochlear frequency aelcctlvity in the presence of hair cell damage. In: Psychophysics and Phwiologv of Hearing. pp. 249%25X. Editors: E. Eva”\ and J W~lwn. 4cadernlc Prcsh. London. Dolan, T., Mills. J. and Schmiedt. R. (1YXSa) A comparlaon of brainstem. whole-nerve AP and single-fiber tuning tune\ in the gcrbll: Normatwc data. Hear. Reb. 17. 759-266 Dolan. T.. Mills. J. and Schmicdt. R. (lYX5h) Brainstcm, Hhole-nerve AP and Gngle-fiber suppression in the gerbil: Normative data. Hear. Res. 1X. 203%210. Eggermont. J. (1977) Compound action potcntlal tumng curws m normal and pathological human cars. J. Acou\t Sot. Am. 62. 1247-1251. Gxga, M. and Abbas. P. (19X1) Forward-masked AP tuning curves in normal and in acoustic& traumatircd car\. J. Acouat. Sot. Am. 70. 1322-1330. Gorga. M.. McGee. J.. Walsh, II.. Javel. E. and Farlev.
activity in the auditory nerve. In: Electrocochleography, pp. 95-115. Editors: R. Ruben, C. Elberling and Ct. Salomon. University Park, Baltimore. Klein, A. and Mills, J. (1981a) Physiological (waves I and V) and psychophysical tuning curves in human subjects. J. Acoust. Sot. Am. 69. 760-768. Klein, A. and Mills, J. (1981b) Physiological and psychophysical measures from humans with temporary threshold shift. J. Acoust. Sot. Am. 70, 1045-1053. Kramer, S. and Teas, D. (1982) Forward masking of auditory (N,) and brainstem (wave V) responses in humans. J. Acoust. Sot. Am. 72, 795-803. Lev, A. and Sohmer, H. (1972) Sources of averaged neural responses recorded in animal and human subjects during cochlear audiometry (electrocochleogram). Arch. Klin. Exp. Ohren Nasen Kehlkopfheilkd. 201. 29-90. Mitchell, C. (1976) Frequency specificity of the N, potential from the cochlear nerve under various stimulus conditions. J. Aud. Res. 16, 247-255. Mitchell, C. and Fowler. C. (1980) Tuning curves of cochlear and brainstem responses in the guinea pig. J. Acoust. Sot. Am. 68, 8966900. Moller, A. (1985) Origin of latency shift of cochlear nerve potentials with sound intensity. Hear. Res. 17, 177-189. Ryan, A., Dallos, P. and McGee, T. (1979) Psychophysical tuning curves and auditory thresholds after haircell damage in the chinchilla. J. Acoust. Sot. Am. 66, 370-377. Salvi, R. (1976) Central components of the temporary threshold shift. In: Effects of Noise on Hearing, pp. 247-262. Edi-

tors: D. Henderson, R. Hamemik. D. Dosanjh and J Mills. Raven Press, New York. Saunders, J., Dolgin, K. and Lowry, L. (19X0) The maturation of frequency selectivity in C57BL/6J mice studied with auditory evoked response tuning curves. Brain Res. 1X7, 69-79. Shepard, N. and Abbas, P. (1983) Compound action-potential tuning curves in normal and acoustically traumatized cats. Ann. Otol. Rhinol. Laryngol. 92, 496-503. Spire, J., Dohrman, G. and Prieto. P. (1982) Correlation of brainstem evoked response with direct acoustic nerve potential In: Clinical Applications of Evoked Potentials in Neurology, pp. 159-167. Editors: J. Courjon. F. Manguiere and M. Revol. Raven Press, New York. Syka. J. and Popelar, J. (1980) Hearing threshold shifts from prolonged exposure to noise in guinea pigs. Hear. Res. 3, 205-213. Thornton, A. and Coleman, M. (1975) The adaptation of cochlear and brainstem auditory evoked potentials in humans. Electroencephalogr. Clin. Neurophysiol. 39, 399-406. Tietze. G. and Gobsch, H. (1980) Measurements of adaptation of brainstem responses, In: Cochlear and Brainstem Evoked Response Audiometry and Electrical Stimulation of the VIII Nerve, pp. 91-96. Editors: M. Hoke. Ci Kauffmann and E. Babbert. Stand. Audiol. Suppl. 11. Van Heusden, E. and Smoorenburg. G. (1981) Eighth-nerve action-potential tuning curves in cats before and after inducement of an acute noise trauma. Hear. Res 5. 25-4X.