Download as a PDF

Ear & Hearing:Volume 20(1)February 1999pp 33-44
Cortical, Auditory, Event-Related Potentials in Response to Periodic and
Aperiodic Stimuli with the Same Spectral Envelope
Martin, Brett A.; Boothroyd, Arthur
Program in Speech and Hearing Sciences (B.A.M., A.B.), Graduate Center, City
University of New York, New York, New York; and Department of SpeechLanguage-Hearing Sciences (B.A.M.), Hofstra University, Hempstead, New York.
Address for correspondence: Brett A. Martin, Ph.D., Program in Speech and
Hearing Sciences, Graduate Center, City University of New York, 33 West 42
Street, New York, NY 10036.
Received March 17, 1998; accepted September 28, 1998
Abstract
Objective: 1) To determine whether the N1-P2 acoustic change complex is
elicited by a change of periodicity in the middle of an ongoing stimulus, in the
absence of changes of spectral envelope or rms intensity. 2) To compare the
N1-P2 acoustic change complex with the mismatch negativity elicited by the
same stimuli in terms of amplitude and signal to noise ratio.
Design: The signals used in this study were a tonal complex and a band of
noise having the same spectral envelope and rms intensity. For elicitation of
the acoustic change complex, the signals were concatenated to produce two
stimuli that changed in the middle (noise-tone, tone-noise). Two control stimuli
were created by concatenating two copies of the noise and two copies of the
tone (noise-only, tone-only). The stimuli were presented using an onset-toonset interstimulus interval of 3 sec. For elicitation of the mismatch negativity,
the tonal complex and noise band stimuli were presented using an oddball
paradigm (deviant probability = 0.14) with an onset-to-onset interstimulus
interval of 600 msec. The stimuli were presented via headphones at 80 dB
SPL to 10 adults with normal hearing. Subjects watched a silent video during
testing.
Results: The responses to the noise-only and tone-only stimuli showed a
clear N1-P2 complex to the onset of stimulation followed by a sustained
potential that continued until the offset of stimulation. The noise-tone and tonenoise stimuli elicited an additional N1-P2 acoustic change complex in
response to the change in periodicity occurring in the middle. The acoustic
change complex was larger for the tone-noise stimulus than for the noise-tone
stimulus. A clear mismatch negativity was elicited by both the noise band and
tonal complex stimuli. In contrast to the acoustic change complex, there was
no significant difference in amplitude across the two stimuli. The acoustic
change complex was a more sensitive index of peripheral discrimination
capacity than the mismatch negativity, primarily because its average
amplitude was 2.5 times as large.
Conclusions: These findings indicate that both the acoustic change complex
and the mismatch negativity are sensitive indexes of the neural processing of
changes in periodicity, though the acoustic change complex has an advantage
in terms of amplitude. The results support the possible utility of the acoustic
change complex as a clinical tool in the assessment of peripheral speech
perception capacity.
Introduction
The general purpose of the project, of which the present study forms part, is to
develop electrophysiological tests of auditory speech discrimination capacity.
Such tests would contribute to the objective evaluation of subjects who, for
reasons of age and hearing loss, lack the auditory, linguistic, and/or cognitive
prerequisites for behavioral speech perception tests (Boothroyd, 1991; Tyler,
1993). Considerable success has been reported from the use of cochlear
implants in children who have profound, prelingually acquired, deafness
(Geers & Moog, 1994). As a result, there is increasing pressure to implant
children with lesser degrees of hearing loss. This pressure creates an acute
need for reliable, objective tests of auditory speech perception capacity that
remain valid when used with very young, prephonological, hearing-impaired
children. The availability of such tests would help reduce the possibility that
children with severe hearing losses will be given implants that provide no more
auditory capacity than could have been obtained with properly fitted hearing
aids.
Most of the previous work on this topic has involved mismatch negativity
(Näätänen, Gaillard, & Mäntysalo, 1978). Mismatch negativity is a negativegoing, cortical, event-related potential that occurs in response to an occasional
deviant acoustic stimulus in a train of repeated standard stimuli. It occurs even
when the subject is paying no attention to the stimuli and is believed to reflect
registration of a difference between the deviant stimulus and the memory trace
that has been developed by the standard stimuli-hence its potential use as an
index of peripheral auditory resolution (see Näätänen, 1992; Näätänen &
Kraus, 1995, for a review). Mismatch negativity has been shown to be elicited
by small changes in stimulus frequency, intensity, duration, and spatial
location (Näätänen, Paavilainen, Alho, Reinikainen, & Sams, 1987;
Näätänen, Paavilainen, & Reinikainen, 1989; Paavilainen, Karlsson,
Reinikainen, & Näätänen, 1989; Sams, Paavilainen, Alho, & Näätänen,
1985). It is also sensitive to changes in the spectral and temporal
microstructure of complex stimuli (Näätänen, Schröeger, Karakas,
Tervaniemi, & Paavilainen, 1993; Schröeger, 1994; Schröeger, Näätänen,
& Paavilainen, 1992). Mismatch negativity can, therefore, be used to
demonstrate the capacity for discrimination among speech sounds, both
synthetic (Aaltonen, Paavilainen, Sams, & Näätänen, 1992; Kraus, McGee,
Sharma, Carrell, & Nicol, 1993; Martin, Kurtzberg, & Stapells, Reference
Note 1; Sams, Aulanko, Aaltonen, & Näätänen, 1990). and natural
(Sandridge & Boothroyd, 1996).
There are, however, potential limitations to the use of mismatch negativity as a
clinical tool in a pediatric population. The amplitude of mismatch negativity is
low in relation to the background EEG activity (Picton, 1995), which, in young
children, is likely to be higher than in adults. In addition, there is a practical
limit to the amount of response averaging that can be done to increase signal
to noise ratio. This limit occurs for two reasons. First, there is a limit to the time
that a young child will remain immobile and cooperative. Second, the number
of deviant stimuli can only be a small proportion of the total, which limits the
amount of signal averaging that can be done to improve signal to noise ratio
for the deviant response. Even in cooperative adults (Ritter, personal
communication) and children (Kurtzberg, Vaughan, Kreuzer, & Fliegler,
1995) with normal hearing the mismatch negativity potential cannot always be
reliably demonstrated. For these reasons, mismatch negativity may not be the
most appropriate tool for clinical evaluation of auditory speech discrimination
capacity in young children with hearing loss.
There is, however, another event-related potential that might serve in this
application, namely the N1-P2 complex that occurs in response to acoustic
change during an ongoing stimulus. The N1-P2 complex is best known as an
onset response (Hillyard & Picton, 1978; Näätänen, 1992; Näätänen &
Picton, 1987; Onishi & Davis, 1968; Pantev, Eulitz, Hampton, Ross, &
Roberts, 1996). It is affected by changes of stimulus parameters. For
example, the amplitude is greater, and the response occurs earlier, as
stimulus amplitude increases (Davis & Zerlin, 1996; Keidel & Spreng, 1965),
and low-frequency sounds produce larger responses than do high-frequency
sounds of the same intensity (Antinoro & Skinner, 1968; Picton, Woods, &
Proulx, 1978). Unfortunately, these differences are not reliable or sensitive
enough to serve as an index of peripheral auditory resolution.
It has been shown, however, that the N1-P2 complex can be elicited by
changes in an ongoing acoustic stimulus. We refer to this event-related
potential as the acoustic change complex. It has been shown to occur in
response to intensity and frequency modulations in sustained tones (Clynes,
1969; Jerger & Jerger, 1970; Näätänen & Picton, 1987; Spoor, Timmer, &
Odenthal, 1969; Yingling & Nethercut, 1983). It has also been shown to
occur in response to simple syllables at the transition from consonantal
segment to vocalic segment (Hari, 1991; Imaizumi, Mori, Kiritani, &
Yumoto, 1996; Kaukoranta, Hari, & Lounasmaa, 1987; McGee, Kraus,
King, Nichol, & Carrell, 1996; Ostroff, Martin, & Boothroyd, 1998;
Steinschneider, Arezzo, & Vaughan, 1982). If the auditory cortex produces a
measurable response to an acoustic change during an ongoing stimulus, it
follows that the different acoustic patterns, occurring before and after the
change, are represented by different patterns of excitation that are manifested
at every level of processing from the cochlea through the cortex. Such a
response would index the arrival, at a cortical level, of potentially discriminable
information. The acoustic change complex may provide the desired index of
speech discrimination capacity if it is elicited, assuming that it is elicited by the
types of acoustic change that are involved in phonological contrasts.
We have previously demonstrated that the acoustic change complex is elicited
by the transition from fricative to vowel in the naturally produced syllable "say"
(Ostroff et al., 1998). In this example, however, the acoustic change involved
intensity, spectral envelope, and periodicity. In a subsequent study, we have
found that the acoustic change complex can be produced solely by a change
of spectral envelope. The stimuli consisted of synthetic 3-formant vowels with
a change of either first or second formant frequency at the temporal midpoint
(Ostroff, Martin, & Boothroyd, Reference Note 2).
The primary purpose of the present study was to determine whether the
acoustic change complex is elicited by a change of periodicity only, in the
absence of changes of spectral envelope or intensity. We also wished to
compare the acoustic change complex with mismatch negativity, in terms of
amplitude and sensitivity. It should be noted that "sensitivity" is a general term
that takes into account the signal to noise ratio of the recordings (i.e.,
amplitude of the waveforms, the interstimulus interval, and the number of
sweeps). Additionally, sensitivity is determined by the ability to show
statistically significant potentials that index the encoding of small changes in
periodicity in individual subjects.
Method
Subjects
Ten adults (three men, seven women), aged 23 to 41 yr (mean = 31 yr)
participated. All subjects had normal hearing sensitivity (thresholds ≤25 dB HL
from 250 through 8000 Hz bilaterally) and no history of neurological disorder.
Stimuli
The contrasting signals used in this study were a tonal complex and a band of
noise having the same spectral envelope and rms level. The tonal complex
was created by combining a series of equal-amplitude pure tones whose
frequencies were integral multiples of 100 Hz and passing the result through a
band-pass filter with edge frequencies of 600 and 1400 Hz. The noise band
was created by passing white noise, with a constant level per cycle, through
the same filter. After filtering, the rms intensity levels of the two signals were
equated. Both signals were trimmed to 400 msec. The first and last 10 msec
were then shaped with raised cosine functions to minimize spectral splatter at
onset and offset.
Two stimuli were created for elicitation of the acoustic change complex by
concatenating the tone and noise signals with a 10 msec overlap. Each
stimulus was, therefore 790 msec long. The acoustic change began at 390
msec and lasted for 10 msec. The resulting stimuli will be referred to as tonenoise and noise-tone. In addition, two control stimuli were created by
concatenating two copies of the noise and two copies of the tone. The
resulting stimuli will be referred to as noise-only and tone-only, respectively.*
Figure 1 shows the short-term rms amplitude of the two signals (integrated
over 50 msec), their waveforms, and their spectra.†
Figure 1. The top portion of the figure shows the rms amplitude of the tone
and noise signal integrated over 50 msec. The middle portion shows the
waveform of the tone-noise stimulus. The bottom portion shows the spectra of
the two signals. The spectra are superimposed to show the similarity of their
bandpaths.
The two stimuli were matched for spectral envelope and rms intensity. Their
fine spectra, however, necessarily differ. The aperiodic noise stimulus has a
continuous spectrum, whereas the periodic tonal stimulus has a line spectrum.
Note, also, that although the rms intensity is matched for the two stimuli, there
are small intensity fluctuations for the aperiodic stimulus that are not seen for
the periodic stimulus.
For elicitation of the mismatch negativity, the tone and noise stimuli were
shortened to have durations of 150 msec, with 10 msec raised-cosine shaping
of onsets and offsets.
All stimuli were digitized at 12 bits and 22050 samples per second. After A/D
conversion, they were presented to subjects binaurally, via EAR-3A insert
earphones at levels of 80 dB SPL.
EEG Recordings
Five EEG channels were recorded from surface electrodes placed at Fz, Cz,
Pz, A1 and A2. The EEG channels were referenced to an electrode at the tip
of the nose (Vaughan & Ritter, 1970). A sixth channel to monitor vertical eye
movements and eye blinks was recorded from electrodes placed above and
below the right eye. An electrode at Fpz served as ground. Electrode
impedances were maintained below 5000 Ohms.
The EEG channels were amplified (gain = 20,000 except for the EOG channel
where gain = 5000), and filtered (0.01 to 100 Hz, 6 dB/octave). Single-trial
event-related potential waveforms were baseline corrected (across the entire
sweep duration), an ocular artifact reduction algorithm was applied (Semlitch,
Anderer, Schuster, & Presslich, 1986), and the data were further filtered
(0.1 to 30 Hz, 12 dB/octave). When activity in any channel (except the EOG
channel) exceeded ± 100 µV for a single sweep, data from all channels were
rejected. Additionally, averages were baseline corrected (-100 to 0 msec).
Acoustic Change Protocol
Stimuli were presented with an onset-to-onset interval of 3 sec. They were
presented in 10 blocks, in which each of the four stimuli (tone-noise, noisetone, noise-only, and tone-only) was heard 30 times. Thus, each stimulus was
presented a total of 300 times. Stimulus order in each block was randomized
but with the provision that no single stimulus could occur twice in succession.
Subjects ignored the stimuli and watched a silent, captioned video during
testing.
The amplified EEG signals were digitized at 341 Hz over a 1501 msec (512
point) window, beginning 100 msec before stimulus onset. Responses to the
two experimental and two control stimuli were averaged separately to give four
tracings for each of the 10 subjects. Each individual tracing was based on 300
sweeps (minus the number of artifact rejects). Stimulus presentation time was
15 minutes for each of the four stimuli (excluding breaks).
Mismatch Protocol
Stimuli were presented with an onset-to-onset interval of 600 msec. They were
presented in four blocks, two with tone as standard and noise as deviant and
two with noise as standard and tone as deviant. In each block, the standard
occurred 600 times and the deviant 100 times for a deviant probability of 0.14.
Stimulus presentation order within a run was not randomized, but presentation
order across conditions and subjects were. Subjects were instructed to ignore
the stimuli and watch a silent, captioned video during testing.
The amplified EEG signals were digitized at 890 Hz over a 575 msec (512
point) window beginning 100 msec before stimulus onset. Responses to each
stimulus as both standard and deviant were averaged separately to give four
tracings for each of the 10 subjects. The two standard tracings were each
based on 1200 sweeps (minus the number of artifact rejects). The two deviant
tracings were each based on 200 sweeps (minus the number of artifact
rejects). Total stimulus presentation time was 28 minutes (excluding breaks).
Data Analysis
Response windows were developed using the group mean waveforms to aid
in response identification and measurement in the acoustic change and
mismatch negativity data from individual subjects. The response windows
were ± 50 msec with reference to the latency of the peak (N1, P2, or mismatch
negativity) in the group mean waveforms.
Results
Acoustic Change Complex
The data shown here are for recordings at Cz only-this being the site that
usually gives the largest N1-P2 complex (Vaughan & Ritter, 1970).
Group Means • Figure 2 (top) shows the group mean waveforms obtained in
response to the tone-noise stimulus and to the tone-only control stimulus.
Vertical lines show standard errors of the mean at each sampling point. Note
that these standard errors are derived from the 10 individual average values
and reflect the contributions of both within-and between-subject variance.
Figure 2. The top portion of the figure displays group mean waveforms
obtained in response to the tone-noise (thick line) and to tone-only stimuli
(thin line). The bottom portion of the figure displays group mean
waveforms obtained to the noise-tone (thick line) and tone-only (thin line)
stimuli. Vertical lines indicate the intersubject standard errors of the mean
at each sampling point.
It will be seen that the tone-only stimulus produced a classical pattern with a
clear N1-P2 complex at onset, followed by a sustained negativity, and a return
to baseline after offset (N1-P2 components to the offset of stimulation were
quite variable and will not be addressed in this paper). The tone-noise
stimulus produced the same pattern but with a second N1-P2 complex
occurring after, and presumably in response to, the acoustic change. There
was no evidence of an N1-P2 complex at the midpoint of the response to the
tone-only stimulus.
The onset responses in the group mean waveforms to the tone-noise stimuli
have peak-to-peak amplitudes of 8.0 µV with latencies of 102 msec and 185
msec for N1 and P2, respectively. The peak-to-peak amplitude of the acoustic
change complex is 5.5 µV with latencies (re change onset) of 135 msec and
245 msec, respectively.
Note that the 5.5 µV peak-to-peak amplitude of the acoustic change complex
is some 10 times the standard error of the mean for a single sample. This
response cannot be attributed to random variability either within or between
subjects.
Figure 2 (bottom) shows the group mean waveforms, and standard errors of
the means, obtained in response to the noise-tone stimulus and to the noiseonly control stimulus. The noise-only stimulus produced the N1-P2 complex at
onset, followed by a sustained negativity, and a return to baseline after
stimulus offset. The noise-tone stimulus produced the same pattern but with a
second N1-P2 complex occurring after the acoustic change. This complex was
followed by increased negativity, lasting until stimulus offset. There is no
evidence of an N1-P2 complex at the midpoint of the response to the noiseonly stimulus.
The onset potentials in the responses to the noise-tone stimuli have peak-topeak amplitudes of 7.3 µV with latencies of 99 msec and 190 msec for N1 and
P2, respectively. The peak-to-peak amplitude of the acoustic change complex
is 1.8 µV with latencies (re change onset) of 152 msec and 252 msec,
respectively.
Although the 1.8 µV peak-to-peak amplitude of the acoustic change complex
is smaller than that observed for the tone-noise stimulus, it is still some three
times the standard error of the mean for a single sample. This response
cannot be attributed to random variability.
Scalp Distribution • Figure 3 displays the grand mean waveforms for the
tone-noise and noise tone conditions. The data from Fz, Cz, Pz, A1, and A2
are displayed. Both the N1-P2 complex to stimulus onset and the acoustic
change complex show largest amplitude at Cz, with smaller amplitudes at Fz,
and at Pz. As would be expected, the polarity of the responses inverts at the
earlobe electrode sites (Vaughan & Ritter, 1970).
Figure 3. The grand mean waveforms obtained to the tone-noise stimulus
(top) and noise-tone stimulus (bottom) are shown at electrode sites Fz, Cz,
Pz, A1, and A2.
Individual Data • Figure 4 shows the waveforms obtained from each of the
10 subjects, together with the group mean waveforms from Figures 2 and 3.
The windows show the expected locations of the change responses and are
based on the group data (±50 msec). By visual inspection, all subjects show
an N1-P2 complex at noise and tone onset. Nine of the 10 subjects show a
clear acoustic change complex in response to the tone-noise stimulus, the
exception being subject 8. Six or seven appear to show an identifiable change
complex in response to the noise-tone stimulus.
Figure 4. The waveforms obtained from each of the 10 subjects at Cz to
the acoustic change from tone-noise and noise-tone are displayed
together with the group mean waveforms. The windows show the
expected locations of the acoustic change complex based on the group
mean data (±50 msec).
To obtain a more objective estimate of the presence of a demonstrable
change complex in individuals, the individual peak-to-peak amplitudes, within
the expected response window, were compared with the individual standard
errors of the differences between means at two sampling points. These
standard errors were obtained by first pooling standard error estimates for a
single mean, within the response window, and then multiplying by the square
root of 2. The results are shown in Table 1, together with individual t values,
i.e., the ratios of the N1-P2 difference to the standard error of the difference.
Taking t = 2.0 as a criterion for significance at the 5% level, all subjects show
a change complex for the tone-noise stimulus and seven show one for the
noise-tone stimulus. Moreover, all 10 subjects meet a more stringent criterion
of t = 2.7 (1% significance) for the tone-noise change. Only four subjects meet
this criterion for the noise-tone change.
TABLE 1. Peak-to-peak N1-P2 amplitudes of the acoustic change complex, average
standard errors of the difference between means at two sampling points, number of
accepted sweeps (n), and the ratio between amplitude and standard error (t).
Mismatch Negativity
The data shown here are for recordings at Fz only-this being the site that
usually gives the largest Mismatch (Näätänen, 1992).
Group Means • Figure 5 (top) shows the group mean waveforms obtained in
response to the tone as the standard stimulus and the tone as the deviant
stimulus. Vertical lines show standard errors of the mean at each sampling
point, derived from the 10 individual average values. It will be seen that there
is a clear mismatch negativity with a peak amplitude around 1.4 µV. Peak
latency in relation to stimulus onset is approximately 125 msec. Peak
amplitude of the mismatch negativity is some two times the standard error of
the difference between the two tracings.
Figure 5. The top left panel displays the group mean waveforms obtained
to the tone stimuli presented as standards and as deviants (left). The top
right panel shows difference waveforms obtained by subtracting responses
to stimuli presented as standards from responses to the same stimuli
presented as deviants. Vertical lines show standard errors of the mean at
each sampling point, derived from the 10 individual average values. The
lower panels show the same data for the noise stimuli.
Figure 5 (bottom) shows the group mean waveforms, and standard errors of
the means, obtained in response to the noise as standard stimulus and the
noise as deviant stimulus. As in the top portion of the figure, there is a clear
mismatch negativity. The peak amplitude is around 1.2 µV and the peak
latency is approximately 138 msec. The peak amplitude of the mismatch is 1.5
times the standard error of the difference between the two tracings.
Scalp Distribution • The mismatch negativity is displayed in the grand mean
difference waveforms shown in Figure 6. Data for the noise condition and the
tone condition are displayed at Fz, Cz, Pz, A1 and A2. The mismatch
negativity shows a broad frontocentral scalp distribution, with largest
amplitudes at Fz and Cz. The waveform polarity inverts at the earlobe
electrode sites.
Figure 6. The grand mean difference waveforms from electrode sites Fz, Cz,
Pz, A1, and A2 are shown for the noise stimulus (top) and the tone stimulus
(bottom).
Individual Data • Figure 7 shows the two difference waveforms obtained from
each of the 10 subjects, together with the group mean waveforms from Figure
5. The windows show the expected locations of the mismatch responses and
are based on the group data (±50 msec). By visual inspection, most of the
subjects appear to show an identifiable mismatch for both the noise band and
the tone complex.
Figure 7. Difference waveforms obtained from each of the 10 subjects at Fz in response to
the tone and noise stimuli are displayed together with the group mean waveforms. The
windows show the expected locations of the acoustic change complex based on the group
mean data (±50 msec).
Table 2 shows individual mismatch amplitudes together with standard errors
of the difference between two tracings and the ratio of amplitude to standard
error (t). The standard errors are derived as follows. For each trace, the data
within the expected time window of the mismatch are pooled to give an
average standard error. The standard errors for the standard and deviant
traces then are combined to derive the standard error of a difference- taking
into account the different number of accepted samples for the two. All 10
subjects meet the t = 2.0 criterion (5% significance) for the noise band, and
eight meet it for the tone complex. Eight subjects meet the more stringent t =
2.7 criterion (1% significance) for the noise band, and five subjects meet this
criterion for the tone complex.
TABLE 2. Amplitudes of the mismatch negativity (MMN), average standard
errors of the difference between means at a single sampling point, number of
accepted sweeps (n), and the ratio between the two (t) for two stimuli and 10
subjects.
Comparison of Response Amplitudes.
Individual amplitudes from Tables 1 and 2 were examined using a repeatedmeasures analysis of variance. The two repeated measures were Potential
(Acoustic change or Mismatch), and First stimulus or, in the case of Mismatch,
the only stimulus (Tone or Noise). The results are shown in Table 3. The main
effects of Potential and First stimulus are highly significant, as is the
interaction between the two. Post hoc analysis, using the least-significantdifference test, suggests that the large acoustic change response to the tonenoise stimulus is responsible for all three effects. Its amplitude is significantly
greater than the other three amplitudes (p < 0.0003), which do not differ
significantly from one another (p > 0.05). These findings are illustrated in
Figure 8, which shows the four mean amplitudes, together with their standard
errors. (Note that these standard errors are derived from four separate withincell estimates of variance rather than from the pooled final error term of the
analysis of variance).
TABLE 3. Repeated-measures analysis of variance in the individual amplitude
data of Tables 1 and 2. Independent variables are potential (acoustic change
complex versus mismatch negativity) and first stimulus (tone or noise).
Figure 8. Mean amplitudes for the acoustic change complex and mismatch
negativity are displayed along with standard errors derived from four
separate within-cell estimates of variance. In a analysis of variance, the
amplitude of the acoustic change complex to the tone-noise stimulus was
found to be significantly greater than in the other conditions.
Discussion
These data show that the acoustic change complex can be elicited by a
change from tonal complex to narrow-band noise, or the reverse, in the
absence of a change of rms amplitude or spectral envelope. The tone-only
and noise-only stimuli produced no evidence of an acoustic change complex
at the midpoint. This last finding supports the conclusion that the effects
observed with the tone-noise and noise-tone stimuli were not artifacts of the
concatenate-overlap procedure used to produce the stimuli.
The N1-P2 complex to changes from noiseband to tonal stimuli (and vice
versa) have been documented previously (Kaukoranta et al., 1987; Mäkelä,
Hari, & Leinonen, 1988). However, the previous studies did not control for
spectral envelope and rms amplitude. Additionally, the transition between the
noise and tonal stimuli was not carefully controlled. As a result, it cannot be
determined whether the N1-P2 was elicited by the change in periodicity, or by
the relatively abrupt intensity rise of the second stimulus.
The N1-P2 amplitudes in the change complex were less than those elicited at
stimulus onset. Moreover, the latencies of the N1 and P2 peaks in the group
mean data (around 175 and 250 msec re change onset) were higher than
those observed in the onset complex (around 100 and 190 msec re stimulus
onset). Nevertheless, the patterns are similar, suggesting that the underlying
processes are similar. This last is not a particularly radical suggestion. The
onset response is, itself, a change response, except that the change is from
silence to sound. The potential recorded at the surface of the head reflect a
change in the amount and/or synchrony of neural excitation as the effects of
the change are registered at the level of the auditory cortex (Coles & Rugg,
1996). The presence of similar potentials at the change from tone to noise, or
noise to tone, suggest similar changes in the amount and/or synchrony of
neural excitation.
The acoustic change complex observed in this study probably reflects the
spatio-temporal organization of cortical excitation along a periodic-aperiodic
continuum. This organization may involve the amount of phase-locking
occurring in the auditory pathways (Phillips, 1993, 1995). Other possibilities
should, however, be considered. Although the two stimuli were matched in
terms of gross spectral envelope, the fine structures of their spectra inevitably
differed. Similarly, although the stimuli were matched for overall rms level, the
narrow-band noise inevitably contained short-term amplitude fluctuations that
were absent in the tonal complex. Note, also, that the noise waveform was
"frozen," i.e., the identical waveform was reproduced in every presentation. A
brief amplitude increase soon after the onset of the noise could have
contributed to repeatable changes in the amount of neural excitation. Before
definitive statements can be made about the acoustic and neural origins of the
observed change response, further studies are needed, preferably involving
multiple recording sites to permit the location of primary regions of
synchronous excitation (Scherg & Von Cramon, 1986; Vaughan, Weinberg,
Lehman, & Okada, 1986).
Despite the cautions just raised, there is some support for the notion of
periodic-aperiodic coding at the cortical level in the data reported here. This
support comes from the observation that the group mean response to the
tone-only stimulus produced substantially more sustained negativity (around
2.0 µ V) than did the response to the noise-only stimulus (around 0.3 µV).
Moreover, after the tone-noise transition, the sustained negativity fell.
Similarly, after the noise-tone transition, the sustained negativity rose. These
observations are difficult to reconcile with the notion that the change
complexes are simply the result of amplitude fluctuation in the frozen noise.
A striking feature of the acoustic change findings is their sensitivity to the
direction of the acoustic change. The tone-to-noise transition produced an N1P2 amplitude of around 5 µV whereas the noise-to-tone transition produced
only 2 µV. The reality of this difference is supported by the analysis of
variance in the individual data. One possible explanation is that the noise
stimulus excites a larger cortical volume than does the tone stimulus. That is,
that there are neuronal cells that respond to the noise that were not activated
by the specific frequencies present in the periodic stimulus. If the cells
responding to the tone were completely contained within the volume of cells
responding to the noise, one would expect an onset response from newly
excited cells when the stimulus changes from tone to noise, but no effect
when the change is in the opposite direction. If the two unequal volumes
overlap, but not completely, one might expect the kind of directional sensitivity
observed here. To interpret amplitude differences solely in terms of volume
excitation, however, is an over simplification. Synchrony of firing also
influences amplitude. It is also possible that there is some type of interaction
between the response to the first portion of the stimulus and the response to
the second, such as nonsimultaneous masking or inhibition. Once again, more
detailed mapping of the sources of the change potential will be needed to
determine the exact source of the sensitivity to direction on the acoustic
change.
At first sight, the directional sensitivity observed in the change response does
not appear to occur in the mismatch data. Note, however, that the mismatch
reported here was for tone as standard versus tone as deviant and noise as
standard versus noise as deviant. During data collection, the subject actually
heard either tone as standard and noise as deviant or the reverse. From the
group mean data of Figure 5, it is apparent that the peak amplitude of the
mismatch between tone as standard and noises as deviant was close to zero.
In contrast, the peak amplitude of the mismatch between noise as standard
and tone as deviant was around 2.5 µV. There is, therefore, a directional
sensitivity in the mismatch data. The fact that the pattern is in the opposite
direction to that which would be predicted from the change data presumably
reflects the different processes involved in eliciting the two potentials.
The findings of this study support the conclusion that the acoustic change
complex offers a more sensitive index of peripheral discrimination capacity
than does mismatch negativity in terms of the signal to noise ratio of the
recordings. In the individual data, the average of the two peak-to-peak change
amplitudes was some 2.5 times the average of the two mismatch amplitudes.
A further advantage of the change protocol is that every stimulus contributes
to improving signal to noise ratio.§ In contrast, in the mismatch protocol, only
1/7 of the stimuli contribute to improving signal to noise ratio for the deviant
stimulus and it is this signal to noise ratio that dominates the standard error of
the difference between the standard and deviant traces. This disadvantage for
the mismatch protocol is partially offset by the higher total number of stimulus
presentations made possible by the lower interstimulus interval. It is possible,
however, to use a shorter interstimulus interval in the change protocol. As
shown in the Appendix, these opposing effects tend to cancel when using the
protocols employed in the present study. The difference in amplitude is,
therefore, the primary determinant of the difference in sensitivity, which is
2.5:1 if using the average of the two change complexes and 3.5:1 if using only
the larger of the two.
The amplitude advantage seen for the change protocol did not result in an
advantage in sensitivity over the mismatch protocol in terms of the number of
subjects with statistically significant responses. In fact, the two protocols were
equally sensitive in indexing the ability of individual subjects to encode a
change in periodicity. Specifically, there were 10 subjects with a statistically
significant (p ≤ 0.05) acoustic change component in the tone-noise condition,
and there were seven in the noise-tone condition. For the mismatch negativity,
eight subjects had a statistically significant response for the tone condition and
10 for the noise condition. This finding may be due, in part, to the fact that the
periodicity change was large and easily apparent. Perhaps waveforms elicited
by finer acoustic changes will show more of a difference between the acoustic
change complex and the mismatch negativity in terms of statistical
significance.
It is important to note, however, that any conclusions about relative sensitivity
of the two protocols, based on the present study, apply only to the extreme
periodicity change used here. It remains to be seen whether similar
advantages for the acoustic change complex are present for other acoustic
contrasts. Of particular interest will be the responses to changes of spectral
envelope. Sensitivity to such changes has a major influence on speech
perception capacity.
The present findings support the possible utility of the acoustic change
complex as a clinical tool in the assessment of peripheral speech perception
capacity in young, prephonological, hearing-impaired children. Before further
progress can be made, however, it must be established that the complex can
be elicited in young children despite the known immaturity of their cortical
obligatory potentials. And, if the technique is to be used with pediatric
implantees, the issue of stimulus artifact from the ongoing electrical stimulus
will need to be addressed. Even if the acoustic change complex proves
impractical as a clinical tool in this application, its continued investigation may
contribute to increased understanding of the spatio-temporal organization of
cortical excitation in response to changing sound patterns.
Acknowledgments:
This work was supported by NIH-NIDCD grant #5P50DC00178. Portions of
this paper were presented at the 21st Midwinter Meeting of the Association for
Research in Otolaryngology, St. Petersburg, Florida, February 15, 1998.
References
Aaltonen, O., Paavilainen, P., Sams, M., & Näätänen, R. (1992). Event-related
brain potential and discrimination of steady-state vowels within and between
phoneme categories: A preliminary study. Scandinavian Journal of Logopedics
and Phoniatrics, 17, 102-112.
Antinoro, F., & Skinner, P. H. (1968). The effects of frequency on the auditory
evoked response. Journal of Auditory Research, 8, 119-123.
Boothroyd, A. (1991). Assessment of speech perception capacity in profoundly
deaf children. American Journal of Otology, 12(Suppl.), 67-72.
Clynes, M. (1969). Dynamics of vertex evoked potentials: The R-M brain
function. In E. Donchin & D. B. Lindsley (Eds.), Average evoked potentials:
Methods, results, and evaluations (NASA SP-191) (pp. 363-374). Washington,
DC: U.S. Government Printing Office.
Coles, M. G. H., & Rugg, M. D. (1996). Event-related brain potentials: An
introduction. In M. D. Rugg & M. G. H. Coles (Eds.), Electrophysiology of mind.
Event-related brain potentials and cognition (pp. 1-23). New York: Oxford
University Press.
Davis, H., & Zerlin, S. (1996). Acoustic relations of the human vertex potential.
Journal of the Acoustical Society of America, 39, 109-116.
Geers, A. E., & Moog, J. S. (1994). ). Effectiveness of cochlear implants and
tactile aids for deaf children: The sensory aids study at Central Institute for the
Deaf. Volta Review, 96, 1-232.
Hillyard, S. A., & Picton, T. W. (1978). ON and OFF components in the auditory
evoked potential. Perception and Psychophysics, 24, 391-398.
Hari, R. (1991). Activation of the human auditory cortex by speech sounds. Acta
Otolaryngology. Supplement, 491, 132-138.
Imaizumi, S., Mori, K., Kiritani, S., & Yumoto, M. (1996). Neural representation of
concurrent sounds: A magnetoencephalographic study. In I. Hashimoto, Y. C.
Okada, & S. Ogawa (Eds.), Visualization of information processing in the human
brain: Recent advances in MEG and functional MRI (EEG Suppl. 47) (pp. 191197). Amsterdam: Elsevier Science.
Jerger, J., & Jerger, S. (1970). Evoked responses to intensity and frequency
change. Archives of Otolaryngology, 91, 433-436.
Kaukoranta, E., Hari, R., & Lounasmaa, O. V. (1987). Responses of the human
auditory cortex to vowel onset after fricative consonants. Experimental Brain
Research, 69, 19-23.
Kraus, N., McGee, T., Sharma, A., Carrell, T., & Nicol, T (1993). Mismatch
negativity event-related potential to speech stimuli. Ear and Hearing, 14, 158164.
Keidel, W. D., & Spreng, M. (1965). Neurophysiological evidence for the Stevens
power function in man. Journal of the Acoustical Society of America, 38, 191195.
Kurtzberg, D., Vaughan, H. G., Kreuzer, J., & Fliegler, K. Z. (1995).
Developmental studies and clinical application of mismatch negativity: Problems
and prospects. Ear and Hearing, 16, 104-116.
Mäkelä, J. P., Hari, R., & Leinonen, L. (1988). Magnetic responses of the human
auditory cortex to noise/square wave transitions. Electroencephalography and
Clinical Neurophysiology, 69, 423-430.
McGee, T., Kraus, N., King, C., Nicol, T., & Carrell, T. D. (1996). Acoustic
elements of speechlike stimuli are reflected in surface recorded responses over
the guinea pig temporal lobe. Journal of the Acoustical Society of America, 99,
3606-3614.
Näätänen, R. (1992). Attention and brain function. Hillsdale, New Jersey:
LawrenceErlbaum Associates.
Näätänen, R., Gaillard, A. W. K., & Mäntysalo, S. (1978). Early selective attention
effect on evoked potential reinterpreted. Acta Psychologica, 42, 313-329.
Näätänen. R., & Kraus, N. (1995). Special issue: Mismatch negativity as an index
of central auditory function. Ear and Hearing, 16, 1-146.
Näätänen, R., Paavilainen, P., Alho, K., Reinikainen, K., & Sams, M. (1987). The
mismatch negativity to intensity changes in an auditory stimulus sequence. In R.
Johnson, J. W. Rohrbaugh, & R. Parasuraman (Eds.), Current trends in eventrelated brain potential research (EEG Suppl. 40) (pp. 125-131). Amsterdam:
Elsevier.
Näätänen, R., Paavilainen, P., & Reinikainen, K, (1989). Do event-related
potentials to infrequent decrements in duration of auditory stimuli demonstrate a
memory trace in man? Neuroscience Letters, 107, 347-352.
Näätänen, R., & Picton, T. W. (1987). The N1 wave of the human electric and
magnetic response to sound: A review and an analysis of the component
structure. Psychophysiology, 24, 375-425.
Näätänen, R., Schröeger, E., Karakas, S., Tervaniemi, M., & Paavilainen, P.
(1993). Development of a memory trace for a complex sound in the human brain.
Neuroreport, 4, 503-506.
Onishi, S., & Davis, H. (1968). Effects of duration and rise time of tone bursts on
evoked V potentials. Journal of the Acoustical Society of America, 44, 582-591.
Ostroff, J., Martin, B. A., & Boothroyd, A. (1998). Cortical evoked responses to
spectral change with a syllable. Ear and Hearing, 19, 290-297.
Pantev, C., Eulitz, C., Hampton, S., Ross, B., & Roberts, L. E. (1996). The
auditory evoked "off" response: Sources and comparison with the "on" and the
"sustained" responses. Ear and Hearing, 17, 255-265.
Paavilainen, P., Karlsson, M. L., Reinikainen, K., & Näätänen, R. (1989).
Mismatch negativity to change in spatial location of an auditory stimulus.
Electroencephalography and Clinical Neurophysiology, 73, 129-141.
Phillips, D. P. (1993). Representation of acoustic events in primary auditory
cortex. Journal of Experimental Psychology: Human Perception and
Performance, 19, 203-216.
Phillips, D. P. (1995). Central auditory processing: A view fro auditory
neuroscience. American Journal of Otology, 16, 338-352.
Picton, T. W. (1995). The neurophysiological evaluation of auditory
discrimination. Ear and Hearing, 16, 1-5.
Picton, T. W., Woods, D. L., & Proulx, G. B. (1978). Human auditory sustained
potentials. II. Stimulus relationships. Electroencephalography and Clinical
Neurophysiology, 45, 198-210.
Sams, M., Aulanko, R., Aaltonen, O., & Näätänen, R. (1990). Event-related
potentials to infrequent changes in phonetic features of synthetic CV stimuli.
Journal of Cognitive Neuroscience, 2, 344-357.
Sams, M., Paavilainen, P., Alho, K., & Näätänen, R. (1985). Auditory frequency
discrimination and event-related potentials. Electroencephalography and Clinical
Neurophysiology, 62, 437-448.
Sandridge, S., & Boothroyd, A. (1996). Using naturally produced speech to elicit
the mismatch negativity. Journal of the American Academy of Audiology, 7, 105112.
]
Scherg, M., & Von Cramon, D. (1986). Evoked dipole source potentials of the
human auditory cortex. Electroencephalography and Clinical Neurophysiology,
65, 344-360.
Schröeger, E. (1994). An event-related potential study of sensory representations
of unfamiliar tonal patterns. Psychophysiology, 31, 175-181.
Schröeger, E., Näätänen, R., & Paavilainen, P. (1992). Event-related potentials
reveal how non-attended complex sound patterns are represented by the human
brain. Neuroscience Letters, 146, 183-186.
Semlitch, H. V., Anderer, P., Schuster, P., & Presslich, O. (1986). A solution for
reliable and valid reduction of ocular artifacts applied to the P300 ERP.
Psychophysiology, 23, 695-703.
Spoor, A., Timmer, F., & Odenthal, D. W. (1969). The evoked auditory response
(EAR) to intensity modulated and frequency modulated tones and tone bursts.
International Audiology, 8, 410-415.
Steinschneider, M., Arezzo, J., & Vaughan, H. G. (1982). Speech evoked activity
in the auditory radiations and cortex of the awake monkey. Brain Research, 252,
353-365.
Tyler, R. S. (1993). Speech perception by children. In R. S. Tyler (Ed.), Cochlear
implants: audiological foundations (pp. 191-256). San Diego: Singular Publishing
Group.
Vaughan, H. G., & Ritter, W. (1970). The sources of auditory evoked responses
recorded from the human scalp. Electroencephalography and Clinical
Neurophysiology, 28, 360-367.
Vaughan, H. G., Weinberg, H., Lehman, D., & Okada, Y. (1986). Approaches to
defining the intracranial generators of event-related electrical and magnetic
fields. In W. C. McCallum, R. Zapolli, & F. Denoth (Eds.), Cerebral
psychophysiology studies in event-related potentials (EEG Suppl. 38) (pp. 505544). Amsterdam: Elsevier.
Yingling, C. D., & Nethercut, G. E. (1983). Evoked responses to frequency
shifted tones: Tonotopic and contextual determinants. International Journal of
Neuroscience, 22, 107-118.
]
Reference Notes
1 Martin, B. A., Kurtzberg, D., & Stapells, D. R. (in press). The effects of
decreased audibility produced by high-pass noise masking on N1 and the
mismatch negativity to speech sounds /ba /and /da/. Journal of Speech and
Hearing Research.
2 Ostroff, J., Martin, B. A., & Boothroyd, A. (in preparation). Cortical evoked
responses to spectral change within ongoing vowels.
Appendix
Relative standard errors of the amplitudes of the acoustic change complex
and mismatch negativity.
We begin with four premises:
The underlying noise in the EEG signal is determined by the standard
deviation (d) of repeated voltage samples at a given temporal location in the
sampling window.
The noise in an averaged waveform is equal to d divided by the square root of
the number of sweeps (n) contributing to the average. That is:
Equation A1
The value of d is independent of the test protocol. i.e
Equation A2
The standard error of a difference between two averages is the square root of
the sum of the squares of the two standard errors. That is:
Equation A3
From (1), (2), (3), and (4): where na and nb are the number of sweeps
contributing to the averages a and b, respectively.
Equation A4
Consider, first, the acoustic change complex. Let: nac = the number of sweeps
contributing to the averaged waveform and eac = the standard error of the
difference between two means.
From equation (4):
Equation A5
Consider, next, the mismatch negativity. Let: nmm = the total number of
sweeps contributing to two averaged waveforms, p = the proportion of n
contributing to the deviant waveform, (1-p) = the proportion of n contributing to
the standard waveform, and emm = the standard error of the difference
between two means.
From equation (4):
Equation A6
From (5) and (6), the ratio between the two standard errors is or
Equation A7
In the present study, the inter-onset intervals for the mismatch and acoustic
change protocols were 600 and 3000 msec, respectively. In a given testing
time, therefore, five times as many stimuli could be presented in the mismatch
as in the change protocol. Within the mismatch protocol, the proportion of
deviants was 1/7, or 0.14. Substituting these values in equation (7):
Equation A8
It will be seen from Equation (7) that the ratio is determined by two opposing
factors. The first is the relative number of stimuli presented in the two
protocols. This factor favors the mismatch protocol. The second factor is the
division of the stimuli in the mismatch protocol into standards and deviants
and the small proportion of the latter. This factor favors the change protocol.
Under the conditions of the present study, these factors effectively cancel
each other.
*
An alternative approach would have been to create 790 msec of the tonal
complex and noise band and to shape the onsets and offsets. The
concatenation approach was chosen for two reasons: first, to ensure that any
change complex was not an artifact of the concatenation-overlap procedure;
second, to ensure that the second half of the noise-only stimulus had the
same waveform as the second half of the tone-noise stimulus.
† The spectra are superimposed to show the similarity of their band paths.
The spectral peaks of the tonal stimulus were corrected by - 13 dB. The peaks
of the spectra of the tonal stimuli are higher than those of the noise stimulus
because the energy is integrated over a smaller time frame.
‡If the standard error of the mean at a single sampling pint is represented by
se, the standard error of a difference between two independent means is se ×
√2. Peak-to-peak differences of 1.98 and 2.67 times this value may be taken
to be significantly different from zero at the 5% and 1% levels, respectively. It
could be argued that these criteria are too lax because the N1 and P2 peaks
are not selected at random. On the other hand, one could also argue for the
use of a less conservative 1-tail test, based on the known morphology of the
complex. Additionally, the means (N1 and P2) are not really independent.
Therefore, the standard error of the difference may be less than is estimated
here. In fact, application of the criteria developed here produces conclusions
that are in close agreement with judgments based on visual inspection of the
averaged waveforms. Moreover, these criteria can be applied uniformly across
subjects.
§In Clinical application it would be unnecessary to include the tone-only and
noise-only stimuli.
© 1999 Lippincott Williams & Wilkins, Inc.