Behavioural Brain Research 156 (2005) 95–103
Research report
Auditory training improves neural timing in the human brainstem
Nicole M. Russoa,b,c,∗ , Trent G. Nicola,c , Steven G. Zeckerc ,
Erin A. Hayesa,b,c , Nina Krausa,b,c,d,e
a
e
Auditory Neuroscience Laboratory, Northwestern University, 2240 Campus Drive, Evanston, IL 60208, USA1
b Institute for Neuroscience, Northwestern University, 2240 Campus Drive, Evanston, IL 60208, USA
c Department of Communication Sciences, Northwestern University, 2240 Campus Drive, Evanston, IL 60208, USA
d Department of Neurobiology and Physiology, Northwestern University, 2240 Campus Drive, Evanston, IL 60208, USA
Department of Otolaryngology, Northwestern University, Northwestern University, 2240 Campus Drive, Evanston, IL 60208, USA
Received 13 January 2004; received in revised form 16 April 2004; accepted 11 May 2004
Available online 19 June 2004
Abstract
The auditory brainstem response reflects neural encoding of the acoustic characteristic of a speech syllable with remarkable precision. Some
children with learning impairments demonstrate abnormalities in this preconscious measure of neural encoding especially in background noise.
This study investigated whether auditory training targeted to remediate perceptually-based learning problems would alter the neural
brainstem encoding of the acoustic sound structure of speech in such children. Nine subjects, clinically diagnosed with a language-based
learning problem (e.g., dyslexia), worked with auditory perceptual training software. Prior to beginning and within three months after
completing the training program, brainstem responses to the syllable /da/ were recorded in quiet and background noise. Subjects underwent
additional auditory neurophysiological, perceptual, and cognitive testing. Ten control subjects, who did not participate in any remediation
program, underwent the same battery of tests at time intervals equivalent to the trained subjects.
Transient and sustained (frequency-following response) components of the brainstem response were evaluated. The primary pathway
afferent volley – neural events occurring earlier than 11 ms after stimulus onset – did not demonstrate plasticity. However, quiet-to-noise
inter-response correlations of the sustained response (∼11–50 ms) increased significantly in the trained children, reflecting improved stimulus
encoding precision, whereas control subjects did not exhibit this change. Thus, auditory training can alter the preconscious neural encoding
of complex sounds by improving neural synchrony in the auditory brainstem. Additionally, several measures of brainstem response timing
were related to changes in cortical physiology, as well as perceptual, academic, and cognitive measures from pre- to post-training.
© 2004 Elsevier B.V. All rights reserved.
Keywords: Auditory brainstem response; Neural timing; Plasticity; Speech; Auditory training; Frequency-following response; Reading disability
1. Introduction
This study addresses several questions: Is there plasticity
in the neural encoding of sound in the human auditory
brainstem? If so, is this manifested in a way that can be
readily measured? Can the brainstem representation of
speech-sound structure in children with learning disabilities
be altered by work with a commercially available auditory
training regimen?
∗
1
Corresponding author. Tel.: +1 847 491 2465; fax: +1 847 491 2523.
E-mail address:
[email protected] (N.M. Russo).
http://www.communication.northwestern.edu/csd/research/brainvolts.
0166-4328/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.bbr.2004.05.012
Auditory training has been shown to alter the neural encoding of sound structure at the cortical level.
Cortical plasticity has been established in both animals
[4,7,11,15,22,31,36,42,43] and humans [25,39]. Cortical
changes have accompanied perceptual learning of nonnative speech sounds in adults [59] and improved auditory perception in children with learning problems [25,61].
However, neural plasticity is not necessarily restricted to
the cortex. The auditory cortex receives sensory input
via the thalamocortical loop and there is a precedent for
subcortical plasticity from a number of animal studies
[3,10,12–14,21,24,27–30,35,45–47,56,62]. In general, it is
thought that subcortical plasticity is short-term. However,
96
N.M. Russo et al. / Behavioural Brain Research 156 (2005) 95–103
once conditioned, the association of a sound with a meaning causes long-term cortical changes. Furthermore, it has
been suggested that there is an interaction between ascending auditory pathways and the descending corticofugal system, as well as interactions with the amygdala and
basal forebrain [56]. A positive feedback loop involving
lateral inhibition modulates subcortical and cortical activity. The extent to which plasticity at subcortical regions
directly influences the cortex or vice versa has yet to
be determined. Whatever the mechanism, current research
supports a relationship between cortical and subcortical
plasticity.
Subcortical plasticity in the medial geniculate body
(MGB), which synapses directly onto auditory cortex,
occurred with classical conditioning in rats [10]. These
experience-dependent changes persisted for 45 days. Later,
using a guinea pig model, Edeline and Weinberger extensively investigated plasticity in the dorsal (MGd), ventral
(MGv), and medial (MGm) divisions of the MGB in response
to associative cardiac conditioning to specific frequencies.
Each area of the MGB experienced changes in receptive field
properties after only a short conditioning period. Changes
in the nonprimary pathway (MGd) were resilient and persisted at the one-hour post-test session [12], while changes
in the primary pathway (MGv) were susceptible to decay after 1 h [13]. Both short- and long-term changes were seen
in the MGm, reflecting the broad- and fine-tuned bandwidth
variation of cells in this area [14]. Edeline and Weinberger
concluded that the subdivisions of the MGB act in conjunction with each other and that the significance of the stimulus
affects the duration of the change.
Plasticity in the cochlear nucleus has been demonstrated
in a decerebrate preparation using a cat as the animal model
[3,27,28]. This basic paradigm resulted in the expression
of habituation and spontaneous recovery in the cells of the
cochlear nucleus in response to repetitive stimulation. Alterations in neural connectivity following cochlear ablation
demonstrate plasticity in even lower subcortical structures.
Unilateral cochlear removal in ferrets produced changes in
the number of neurons projecting to the contralateral inferior
colliculus [46,47]. Ablation of the cochlear nucleus in rats
resulted in new patterns of synaptic connections within the
brainstem [29,30]. Illing et al. [30] further explored brainstem plasticity after cochlear ablation in rats. They observed
plasticity in the superior olivary complex, ventral and dorsal cochlear nucleus, and inferior colliculus via the increased
presence of GAP-43, which is abundant during synaptogenesis both in development and remodeling.
Plasticity at the level of the inferior colliculus has been
observed in barn owls. Behavioral changes in sound localization following filtering or ear occlusion were accompanied by
changes in auditory space maps within the inferior colliculus
[21,35,45].
These animal studies showed changes occurring in receptive fields, space maps, and synaptic activity and connectivity at the first levels of sensory processing. In the only
human study to our knowledge, Khalfa et al. investigated
the modulation of auditory periphery by higher cortical regions in epileptic patients following resection surgery [32].
They were able to demonstrate reciprocal relationships between changes in the medial olivocochlear system and auditory cortex. Using transiently evoked otoacoustic emission
recordings and equivalent attenuation calculations, they were
able to assess effects of the surgery on the medial olivocochlear nucleus. Specifically, they showed evidence for corticofugal influence on the medial olivocochlear nucleus and
associated changes in speech processing both in quiet and
noise.
Thus, considerable evidence suggests that neuronal activity occurring in the human auditory midbrain may be dynamic. The current study was designed to investigate plasticity in the physically intact auditory system by capitalizing on
the ability to quantify temporal changes using evoked potentials. A strength of evoked potentials is their use in quantifying neural synchrony and timing in the encoding of complex
stimuli, such as speech.
Specific aspects of the sound structure are maintained
and reflected in the neural code of the auditory brainstem
[51]. The brainstem response to a speech sound consists of
two components, the onset and the frequency-following response (FFR), which represent transient and sustained processes, respectively. Transient responses, with precision on
the order of tenths of milliseconds, represent primarily the
response to discrete events in the stimulus, such as the
stimulus onset and the successive modulations caused by
the vibration of the vocal folds. Sustained response components last for the duration of a periodic stimulus and reflect the overall integrity of the response with respect to the
stimulus.
A speech syllable can be divided into transient and sustained portions – consonants and vowels – that share some
characteristics with the brainstem response components.
Consonants are rapid, transient, and generally aperiodic features of speech; they are represented by the transient components of the brainstem response and are easily disrupted by
noise. Vowels are periodic, sustained signals; they are represented by the sustained features of the brainstem response,
are generally much larger in amplitude than consonants and
are more resistant to noise.
Stop consonants are difficult to perceive, especially for
people with learning disabilities [2,6,49]. Children with
language-based learning problems often exhibit deficits in
auditory perception and the neural encoding of speech sounds
at both cortical and brainstem levels [5,20,38,52,57,64,68],
especially when background noise is introduced [1,33,63].
Commercial auditory training programs have been developed
to provide remediation for auditory perception and related
learning deficits [8,18,19,44,57]. The physiological consequence of this kind of training is little understood. Thus, testing children before and after undergoing such training offers
an ideal opportunity to examine neural plasticity at the level
of the auditory brainstem.
N.M. Russo et al. / Behavioural Brain Research 156 (2005) 95–103
Normative values and test-retest reliability for the brainstem measures in this study have been established and provide
a means for determining the degree to which brainstem responses may be expected to change over time [33,51]; preto post-training changes that exceed test–retest changes can
be attributed to auditory training. Moreover, because not all
children benefit in the same way from training programs, it is
important to determine what pre-training neurophysiological
measures are markers for successful training.
Speech-evoked brainstem activity was obtained before
and after children with learning disabilities participated in
a commercial auditory training regimen. Both transient and
sustained components of the brainstem response to the syllable /da/ presented in quiet and in background noise were
assessed. Relationships of brainstem measures to improvements in cognitive, perceptual, and academic achievement
tests were also explored.
2. Materials and methods
2.1. Subjects and training regimen
Nineteen children, 8–12 years old, were included in this
study. All of the subjects were native English speakers, with
normal IQ (≥85 on Brief Cognitive scale or Test of Nonverbal Intelligence; range 85–135), and had normal hearing thresholds at or below 20 dB HL for octaves from 500
to 4000 Hz. The experimental group comprised nine children with learning disabilities (LD) based on diagnoses by
outside professionals (child psychologists, neurologists, etc.)
and their performance on study-internal measures of learning
and academic achievement (see Section 2.3). Consent and assent were obtained from the parent(s) or legal guardian(s) and
the children. The Institutional Review Board of Northwestern
University approved all research.
Children in the experimental group participated in 35–40
independently supervised one-hour sessions of Earobics [9]
over an 8-week period. Earobics is a commercial auditory
training program that provides training through interactive
computer games of phonological awareness, auditory processing, and language processing skills. The stimuli are presented in quiet and background noise, with both visual and
auditory feedback. Children listen to sounds while playing interactive, animated computer games; they match sounds (indicating alike or different) by clicking the computer mouse on
appropriate pictures or sound representations they hear. Earobics is a two-step program; Step 1 has six interactive games
covering phonological awareness and processing, while five
games comprise Step 2, which further develops the skills
trained in Step 1 and concentrates more on language processing skills to help individuals better interpret spoken and written language [9]. Children in the experimental group underwent auditory neurophysiological and perceptual/cognitive
testing prior to and within three months following completion of the training program.
97
The control group underwent the same battery of tests
at a time interval equivalent to the trained subjects, but did
not participate in the training program. This group (n = 10)
consisted of both normal learning (NL; n = 5) and LD (n = 5)
children who met the same inclusion criteria as those in the
experimental group.
2.2. Neurophysiological testing
Auditory brainstem and cortical evoked potentials were
evaluated in response to the speech syllable /da/ presented in
quiet and background noise.
2.2.1. Auditory brainstem response
The brainstem response was elicited by the synthesized
[34] speech stimulus /da/ (Fig. 1, top). The stimulus duration
was 40 ms. Randomly alternating polarities were presented
(Neuroscan, Stim, Compumedics) to the right ear through
an insert earphone (ER-3, Etymotic Research) at 80 dB SPL
with a 51 ms inter-stimulus interval. The syllable was presented in two conditions, quiet and with white Gaussian background noise (+5 dB SNR). The response was differentially
recorded from Cz-to-ipsilateral earlobe, with the forehead as
ground. Three thousand sweeps per polarity were collected
(Neuroscan, Scan, Compumedics) in each noise condition.
The sampling rate was 20,000 Hz and responses were online filtered from 100 to 2000 Hz. Trials with activity greater
than 35 V were online rejected. Responses to alternating
polarity stimuli were added together to create a mainly neu-
Fig. 1. Stimulus waveform and grand averages of those subjects whose
quiet-to-noise inter-response correlations improved. The stimulus has been
shifted to align peaks present in the stimulus with their corresponding response waveform peaks; this shift accounts for a time delay introduced by
the amount of time required for the sound to traverse the ear canal to the
brainstem. Peaks are labeled, the onset response is bracketed and the FFR
is underlined in the quiet response. Waveforms show that the improvement
of correlations can be attributed to more accurate encoding of the signal in
noise, rather than a change in quiet.
98
N.M. Russo et al. / Behavioural Brain Research 156 (2005) 95–103
ral response [23]. Throughout the testing session, the children
watched a video of their choice and listened to the soundtrack
at less than 40 dB SPL in the non-test ear.
2.2.1.1. Transient response. The brainstem response to /da/
consists of six major transient peaks (A–F) following the familiar I–V series. These peaks represent synchronized neural
activity in response to the phonetic/acoustic characteristics of
the speech syllable and represent peaks within the stimulus
with remarkable precision. Peaks V, A, C, and F are the most
reliable peaks in the response, exhibiting small latency variability and excellent detectability in all subjects [33]. These
peaks (Fig. 1) were evaluated both in terms of timing (latency)
and magnitude (amplitude). The VA complex was further analyzed by interpeak latency, area, amplitude, and slope. A
wavelet-denoising technique derived from Qian Quiroga and
Garcia [48] was used to aid in determining peak latencies and
amplitudes of responses elicited in noise.
2.2.1.2. Sustained response. The sustained FFR component
of the response (11.5–46.5 ms) (Fig. 1) was evaluated both
by magnitude and timing measures. Magnitude was evaluated in two ways. RMS amplitude was calculated over the
FFR epoch. The amplitude of the spectral component encompassing the fundamental frequency of the stimulus (F0
= 103–121 Hz) was measured by fast Fourier transformation analysis. Timing also was assessed in two ways, using
a cross-correlation technique. Stimulus-to-response correlations were measured, using the 10–40 ms portion of the stimulus, and the highest correlation achieved within a response lag
of 6–9 ms was obtained. Quiet-to-noise inter-response correlations were also analyzed over a response range of 10–40 ms,
with a noise response lag of up to 2 ms. Specific details of
the methods and normative values are discussed elsewhere
[5,33,51].
2.2.2. Analysis of plasticity
Plasticity in physiological measures in trained subjects
was defined as changes in the neurophysiological response
that exceeded those observed in the untrained control subjects. Differences between groups were measured using a
repeated measures analysis of variance with test session as
the within-subject factor and training group as the betweensubject factor. Post-hoc tests were done to establish in which
group the significant changes occurred. A criterion of P <
0.05 was used. For all statistical analyses involving Pearson
correlations, Fisher’s transformation was used to convert rvalues to z′ -scores.
2.3. Cortical response
Cortical responses to the speech stimulus /da/ presented
at 80 dB SPL in quiet and noise (0 dB SNR) were recorded.
The interstimulus interval was 590 ms. The sampling rate
was 2000 Hz and responses were online filtered from 0.05
to 100 Hz. Cortical activity was recorded from Cz, with
a nasal reference and the forehead as ground. Eyeblink
was monitored with bipolar supraorbital-to-lateral canthus
electrodes. P2N2 amplitudes, latencies, and quiet-to-noise
inter-response correlations were measured. Similar to the
technique for analyzing the inter-response correlations for
the brainstem response, the cortical response to the sound
presented in quiet was cross-correlated with the response
recorded in background noise. The correlation was calculated over the 100–350 ms range and the highest correlation
value achieved within a 50 ms lag was obtained [25]. Spearman correlations were used to identify relationships between
brainstem and cortical response measures.
2.4. Perceptual and cognitive abilities testing
At both the test and retest sessions, subjects underwent
a series of tests that quantified their perceptual and cognitive abilities. Subjects were evaluated on measures of auditory processing (Incomplete Words, Memory for Words,
Sound Blending, Listening Comprehension [67]), mental
abilities (Brief Cognitive Scale [66]), and academic achievement (Word Attack [67], Reading and Spelling [65]). Other
measures of auditory perception included speech discrimination in quiet and in background noise (just-noticeable difference scores along a synthesized /da-ga/ continuum differing
in F3 onset frequency, as determined by Parameter Estimation
Sequence Tracking [58]), speech identification (perception of
Sentences in Noise [1]), and temporal resolution (Backward
Masking). These measures have been described in detail elsewhere [1,38,68].
Spearman correlations were used to identify relationships
between the brainstem response and cognitive and perceptual measures. For these analyses, if a subject showed a
decrease on a perceptual/cognitive score upon retest, their
“improvement” was coded as zero to diminish the impact of
outliers.
3. Results
3.1. Stability of brainstem measures over time: control
group
Test–retest data were collected from control subjects who
did not undergo auditory training. NL and LD controls were
combined into one control group because the degree of
change of test–retest measures was equivalent between the
NL and LD controls. Two-tailed, paired t-tests were conducted to establish changes that could be expected to occur
over a 3–6-month time interval. These comparisons revealed
that most /da/-elicited brainstem measures are stable over
time. No significant differences were found in brainstem measures obtained in quiet, with the exceptions of VA interpeak
amplitude and slope (both, P < 0.02). Onset response amplitude is known to be variable [53], so this was not a compelling
change. Onset amplitudes were thus omitted from analysis of
N.M. Russo et al. / Behavioural Brain Research 156 (2005) 95–103
training effects. In background noise, onset responses are often attenuated to a great extent and sometimes eliminated
[51]. Therefore, these responses were not evaluated for effects of training. Peaks C and F, however, remained robust
and were resistant to background noise. All FFR measures
and transient response peaks C and F remained stable in quiet
and background noise over the test–retest time interval.
3.2. Effects of training on brainstem measures:
experimental group
Measures of onset response timing did not change in
the experimental group. There was no evidence of trainingassociated changes in responses occurring earlier than 11 ms.
Quiet-to-noise inter-response correlations of the FFR increased significantly for the experimental group after training, but not for the control group (RMANOVA interaction,
Fapprox (1,17) = 6.67, P < 0.02; post-hoc one-tailed paired ttest, P < 0.02 and P > 0.25, respectively). Specifically, seven
of the nine trained subjects showed this increase (Fig. 2).
Increased quiet-to-noise inter-response correlations indicate
that timing characteristics of the stimulus became encoded
more precisely after training.
In order to discern whether an improvement in either the
response in quiet or noise contributed more to the overall improvement in the quiet-to-noise inter-response correlations,
grand average waveforms were compared (Fig. 1). Visual inspection of these waveforms suggests that responses in quiet
were stable while clearer definition of noise response components emerged following training.
Fig. 2. Improved neural timing in noise. Quiet-to-noise inter-response correlations of trained (left) and control subjects (right). Subcortical changes in
the brainstem response occurred in trained subjects, as evidenced by seven
of the nine subjects with increased inter-response correlations, while control subjects did not change (z′ score conversion; RMANOVA interaction,
Fapprox (1,17) = 6.67, P < 0.02; post-hoc one-tailed paired t-test, P < 0.02 and
P > 0.25, respectively). Increased correlations are indicative of more similarity between quiet and noise responses, suggesting improved encoding in
noise.
99
To quantify this observation, partial correlation analyses,
controlling for values of pre-training stimulus-to-response
correlations in quiet, were performed. A strong and significant relationship was found between inter-response and
stimulus-to-response in noise correlations (partial correlation, 0.55, one-tailed P < 0.01). However, the stimulusto-response in quiet showed no such relationship with the
inter-response correlations. This confirmed the presumption that an improvement of neural timing in noise made
the greater contribution to the overall increase in interresponse correlations. Following training, the overall morphology of the waveform for the response in noise more
closely resembled the response in quiet, and thus the
stimulus.
Additionally, trained subjects showed better wave C peak
definition and a later latency in noise, unlike control subjects (RMANOVA interaction, Fapprox (1,17) = 7.24, P < 0.02;
post-hoc one-tailed paired t-test, P < 0.01 and P > 0.25, respectively). These changes may have contributed to the improved correlation between the quiet and noise responses.
The training-associated change in peak C latency in noise
was likely a consequence of post-training sharpening of the
wave. As can be seen in the pre-training grand averaged waveform (Fig. 1), the region around peak C (approx. 19 ms) was
very broad; the peak latency was not clearly identified. Furthermore, the standard deviation of peak C latency in noise
decreased post-training. Thus, after training, as the peak became more pronounced, the judgment of its latency became
more precise.
3.2.1. Relationships between subcortical and cortical
measures
A study conducted by Hayes et al. [25] showed changes in
response timing and magnitude in noise following auditory
training. This was manifested by increases in cortical quietto-noise inter-response correlations and P2N2 amplitudes in
noise. Relationships between improvements at both the brainstem and cortical levels were explored. Increases in brainstem quiet-to-noise inter-response correlations were significantly associated with cortical P2N2 amplitude increases in
noise (Spearman’s ρ = 0.70, one-tailed P < 0.03). Increases
in stimulus-to-brainstem response in noise correlations were
associated with increases in cortical amplitudes (Spearman’s
ρ = 0.67, one-tailed P < 0.03). Overall, improved subcortical timing was associated with improvements in the cortical
response.
3.2.2. Relationships between brainstem responses and
behavior
Relationships between training-related brainstem response changes and changes in perceptual and cognitive
measures were examined. Additionally, pre-training brainstem response indicators of behavioral improvement were
sought.
Children in the trained group demonstrated significant
gains on the Incomplete Words, Auditory Processing, and
100
N.M. Russo et al. / Behavioural Brain Research 156 (2005) 95–103
Sentences-in-Noise tests. Although brainstem changes were
not directly related to improvements on those particular tests,
gains in Listening Comprehension were related to changes in
the brainstem response in noise. Decreases in the RMS amplitude of the FFR in noise were accompanied by improved
Listening Comprehension scores (Spearman’s ρ = −0.88, P
< 0.002), a measure of auditory processing. No other significant relationships were found between brainstem response
changes and changes on the other tests of perceptual and cognitive abilities.
3.2.3. Brainstem response markers of training success
Particular pre-training brainstem measures marked children who demonstrated significant training-associated gains
in auditory processing and speech discrimination in noise.
Children with later peak F latencies in noise demonstrated
improvements on Incomplete Words (Spearman’s ρ = −0.90,
P < 0.001), while children with larger peak F amplitudes
in noise showed improvements in /da-ga/ discrimination in
noise (Spearman’s ρ = −0.84, P < 0.005).
4. Discussion
Measures of transient onset response timing were stable
over time and resistant to the effects of training; they did
not change in either experimental or control subjects. Auditory training did alter sustained response timing. Brainstem response quiet-to-noise inter-response correlations, as
well as FFR peak C latency in noise, differed between
test sessions in children who received training, but not in
control subjects. Training did not alter sustained response
magnitudes.
4.1. Improved neural timing in noise
Auditory training appears to alter the brainstem response
to speech sounds. Specifically, neural encoding became more
resistant to the deleterious effects of background noise following training. Increases in quiet-to-noise inter-response
correlations represent greater timing precision in the FFR
in noise after training.
Certain assumptions can be made about the nature of plasticity within the auditory brainstem based on the latency
ranges over which changes did and did not occur following
auditory perceptual training. Onset responses to /da/, occurring within the first 11 ms post-stimulus onset, were relatively
stable over time and were also unaffected by training. Thus,
a response that arises exclusively from the primary afferent
volley did not demonstrate plasticity; neural events occurring so early in the processing of auditory stimulation may
be hardwired. Unlike the onset response, the FFR element
of the brainstem response was found to be dynamic. Auditory training altered the neural encoding of the harmonic,
periodic aspects of sound occurring 12–40 ms post-stimulus
onset.
4.2. Where do the changes occur?
Isolating the precise source of neural plasticity in the auditory brainstem cannot be accomplished with far-field recordings, although the time frame of the plasticity provides considerable information regarding the likely neuroanatomical
contributions. Because no changes occurred earlier than the
first 12 ms post-stimulus onset, it is plausible that the inferior
colliculus itself is the locus of plastic activity [37]. However,
it is also possible that plasticity at sites peripheral to the inferior colliculus may be contributing to the plasticity shown
in this study. Some studies have shown effects of attention
on cochlear activity [17,40,41]. Galbraith et al. [16] suggests
that such short latency attentional effects may affect the FFR
component of the brainstem response to vowels. Therefore
modulation of cochlear hair cells might influence early processing within the superior olivary complex and thus alter the
activity of the inferior colliculus. Hoormann et al. [26] also
corroborates the concept of early attentional modulation of
the FFR.
Cortical feedback can also induce plasticity within the
FFR. Intracranial recordings in human auditory cortex have
observed activation as early as 12 ms in response to clicks and
tone bursts. Steinschneider et al. [54] reported a similar time
frame in response to a /da/ syllable. Given that the initial
response in cortex occurs at such a short latency, it can be
theorized that cortical feedback may regulate neural activity
as early as the timeframe seen in the present study (e.g., within
the first 30 ms). While the cortex may modulate activity, the
locus of plasticity is not likely rostral to the inferior colliculus
since the MGB and auditory cortex do not phase-lock at rates
as fast as fundamental and first formant frequencies [50,60].
The corticofugal descending system is critical in manipulating signal encoding via positive feedback or lateral inhibition mechanisms [55]. Once trained or conditioned, egocentric selection [69,70] allows for the cortex to recognize
the behavioral significance of an acoustic stimulus and then
fine-tune its own input by altering the sound representation
at lower levels. Specifically, the cortex modulates subcortical
areas that encode basic stimulus features and thus improves
subsequent cortical representation. Even a short-term subcortical change, lasting 1–3 h, is sufficient to influence long-term
cortical changes [13,14,55,70]. Although it is still unknown
precisely how corticofugal modulation is initiated, the evidence remains that subcortical regions are malleable with
training and that modulation may occur in multiple domains
(frequency, time, etc.).
To our knowledge training-associated neural plasticity at
the level of the brainstem in humans has not been previously identified. However, extensive animal research, as reviewed above, has demonstrated regions of plasticity at subcortical levels. Classical conditioning, auditory deprivation,
and cochlear ablation studies support the idea that plasticity does occur subcortically and may affect cortical processing directly. Alternatively, cortical and subcortical activity
may modulate each other through corticofugal loops. The
N.M. Russo et al. / Behavioural Brain Research 156 (2005) 95–103
aforementioned animal studies together with the present work
demonstrate plasticity in the auditory brainstem and support
the notion that early sensory processing is malleable.
4.3. Behavioral ramifications
The relationships between brainstem changes and behavioral measures supports the idea that pre-conscious alteration
of the brainstem response affects auditory perception. Gains
in Listening Comprehension were related to a reduction in
the sustained response RMS amplitude in noise. During the
prestimulus period, RMS amplitudes did not change between
test sessions (paired t-test, P = 0.34), indicating that this reduction was confined to stimulus-evoked activity and not an
overall reduction in physiological noise due to factors such
as subject state or electrode impedance. Clearly, noisy listening environments impair perception. Subjects without extraneous noise in their brainstem response, as suggested by
lower RMS amplitudes and sharper peak definition, were
able to more accurately decipher what they heard, as evidenced by improved Listening Comprehension scores. A
more precise brainstem response in noise may benefit the
listener by providing a more accurate representation of the
acoustic characteristics of the stimulus. This study also suggested that particular pre-training brainstem response measures in both quiet and noise may be related to improvements
in measures of auditory processing and speech discrimination.
Clinicians and parents might be able to streamline their
children’s training programs based on information gained
from pre-training speech-evoked brainstem response screening. This study and other related work from our laboratory
[25,33] indicate that children with delayed brainstem timing
are particularly likely to profit from auditory training. Thus,
brainstem response screening may serve as a means to identify children for auditory training rehabilitation. Eventually,
one might envision designing a training regimen tailored to
a child’s particular needs.
Not all children who went through auditory training
demonstrated neurophysiological changes at the level of the
brainstem. The amount of time between finishing training
and returning for neurophysiological testing did not affect
the outcome. The two subjects who did not show improved
neural timing were in the middle of the group with respect
to test–retest interval. Thus, the elapsed time appeared not
to influence whether or not the subject exhibited timing improvements in the brainstem response. Because behavioral
improvements could occur in the absence of neurophysiological changes, these changes may be sufficient, but are not
entirely necessary for behavioral gains. However, a considerably larger population needs to be assessed before the “sufficient versus necessary” question can be answered definitively. It is possible that those children who showed no
changes in brainstem activity had deficits that were not addressed by the training they received. Alternatively, those
subjects’ learning and auditory perception problems may not
101
have stemmed from an auditory encoding deficiency at the
brainstem level. Future longitudinal investigations may determine whether longer training sessions (hours per day or
number of weeks) or repeated training sessions spread out
over multiple 8-week periods would, in fact, alter the brainstem responses in the children who did not show physiological changes in this study. Extended research may fill in
the gaps pertaining to the rigidity of the onset of the brainstem response to training. Follow-up testing would offer further information about the resilience of the neurophysiological and corresponding perceptual and behavioral effects of
training.
Pre-conscious modification of sensory processing, prior
to cognitive processing, may help overcome higher level
deficits. Previous research [25] showed that children who
went through Earobics training experienced changes in cortical responses to speech syllables, including accelerated maturation of the response, larger amplitudes, and improved
quiet-to-noise inter-response correlations. The relationship
between subcortical and cortical improvements reported here
suggest that alterations in the brainstem response could have
contributed to a more intact neural representation of sound at
the cortical level.
4.4. Extensions of this work
This work demonstrated the existence of plasticity at the
level of the human auditory brainstem and that auditory training can improve neural timing in response to sounds. There
are broad-reaching implications. Previous work has shown
that specific measures of the brainstem response can serve
as biological markers that can identify a subset of languageimpaired children with encoding deficits [6,33,64]. Consequently, the brainstem response to speech could be used in
early detection of children “at risk” for these learning problems and who may benefit from auditory training. Thus, remediation can begin before children reach school age. Regardless of the age of identification and remediation, any changes
in the brainstem response may be used as an objective monitor of auditory training success.
Although the children in this study underwent a general
mode of auditory training, effects were transferable among
sounds, since it was associated with alteration of the response
to the laboratory test syllable /da/. Even so, one can imagine
greater success of training programs that target specific difficulties or encoding deficits. For example, training via cue
enhancement is used in other auditory training programs.
The brainstem response employed here could be informative regarding effects of different forms of auditory training.
Moreover, auditory training could be targeted at enhancing
specific acoustic characteristics that are not encoded accurately at the brainstem. Finally, this experimental approach
can be applied to other populations in which perceptual learning relevant to language and communication are of interest
(e.g., second-language learning, aging, cochlear implant recipients, autistic individuals, etc.).
102
N.M. Russo et al. / Behavioural Brain Research 156 (2005) 95–103
4.5. Conclusions
Neural encoding of sound in the human brainstem appears
to be modified by auditory training. This study used measures of timing and magnitude of the brainstem response to
identify possible mechanisms of brainstem plasticity. In addition, measures of brainstem plasticity were discovered to be
associated with perceptual and cognitive changes. The conclusions drawn from this data set complement results drawn
from cortical and subcortical animal and human studies that
indicate learning-associated plasticity in the auditory pathway. Moreover, this study provides evidence that commercially available auditory training can alter the preconscious
neural encoding of sound by improving neural synchrony in
the human auditory brainstem.
The National Institute of Health NIDCD R01-01510 supported this research.
Acknowledgements
We would like to thank the children who participated in
this study and their families. We would also like to acknowledge Brad Wible, Pam Horstmann, and Erika Skoe who assisted with the data collection and processing.
References
[1] Bradlow AR, Kraus N, Hayes E. Speaking clearly for children with
learning disabilities: sentence perception in noise. J Speech Lang
Hear Res 2003;46(1):80–97.
[2] Bradlow AR, Kraus N, Nicol TG, McGee TJ, Cunningham J,
Zecker SG, et al. Effects of lengthened formant transition duration on discrimination and neural representation of synthetic CV
syllables by normal and learning-disabled children. J Acoust Soc
Am 1999;106(4(Pt 1)):2086–96.
[3] Buchwald JS, Humphrey GL. Response plasticity in cochlear nucleus of decerebrate cats during acoustic habituation procedures. J
Neurophysiol 1972;35(6):864–78.
[4] Cansino S, Williamson SJ. Neuromagnetic fields reveal cortical
plasticity when learning an auditory discrimination task. Brain Res
1997;764(1–2):53–66.
[5] Cunningham J, Nicol T, Zecker S, Kraus N. Speech-evoked
neurophysiologic responses in children with learning problems:
development and behavioral correlates of perception. Ear Hear
2000;21(6):554–68.
[6] Cunningham J, Nicol T, Zecker SG, Bradlow A, Kraus N. Neurobiologic responses to speech in noise in children with learning
problems: deficits and strategies for improvement. Clin Neurophysiol 2001;112(5):758–67.
[7] Diamond DM, Weinberger NM. Role of context in the expression
of learning-induced plasticity of single neurons in auditory cortex.
Behav Neurosci 1989;103(3):471–94.
[8] Diehl S. Listen learn? A software review of Earobics. Lang Speech
Hear Serv Schools 1999;30:108–16.
[9] Earobics. http://www.earobics.com/. Evanston, IL: Cognitive Concepts.
[10] Edeline JM, Neuenschwander-el Massioui N, Dutrieux G. Discriminative long-term retention of rapidly induced multiunit changes
in the hippocampus, medial geniculate and auditory cortex. Behav
Brain Res 1990;39(2):145–55.
[11] Edeline JM, Pham P, Weinberger NM. Rapid development of
learning-induced receptive field plasticity in the auditory cortex.
Behav Neurosci 1993;107(4):539–51.
[12] Edeline JM, Weinberger NM. Subcortical adaptive filtering in the
auditory system: associative receptive field plasticity in the dorsal
medial geniculate body. Behav Neurosci 1991;105(1):154–75.
[13] Edeline JM, Weinberger NM. Thalamic short-term plasticity in the
auditory system: associative returning of receptive fields in the ventral medial geniculate body. Behav Neurosci 1991;105(5):618–39.
[14] Edeline JM, Weinberger NM. Associative retuning in the thalamic
source of input to the amygdala and auditory cortex: receptive field
plasticity in the medial division of the medial geniculate body. Behav Neurosci 1992;106(1):81–105.
[15] Edeline JM, Weinberger NM. Receptive field plasticity in the
auditory cortex during frequency discrimination training: selective retuning independent of task difficulty. Behav Neurosci
1993;107(1):82–103.
[16] Galbraith GC, Bhuta SM, Choate AK, Kitahara JM, Mullen Jr TA.
Brain stem frequency-following response to dichotic vowels during
attention. NeuroReport 1998;9(8):1889–93.
[17] Giard MH, Collet L, Bouchet P, Pernier J. Auditory selective attention in the human cochlea. Brain Res 1994;633(1–2):353–6.
[18] Gillam RB, Crofford JA, Gale MA, Hoffman LM. Language change
following computer-assisted language instruction with Fast ForWord
or Laureate Learning Systems software. Am J Speech Lang Pathol
2001;10(3):231–47.
[19] Gillam RB, Loeb DF, Friel-Patti S. Looking back: a summary of
five exploratory studies of Fast ForWord. Am J Speech Lang Pathol
2001;10(3):269–73.
[20] Godfrey JJ, Syrdal-Lasky AK, Millay KK, Knox CM. Performance
of dyslexic children on speech perception tests. J Exp Child Psychol
1981;32(3):401–24.
[21] Gold JI, Knudsen EI. A site of auditory experience-dependent plasticity in the neural representation of auditory space in the barn owl’s
inferior colliculus. J Neurosci 2000;20(9):3469–86.
[22] Gonzalez-Lima F, Scheich H. Neural substrates for tone-conditioned
bradycardia demonstrated with 2-deoxyglucose. II. Auditory cortex
plasticity. Behav Brain Res 1986;20(3):281–93.
[23] Gorga MP, Abbas PJ, Worthington DW. Stimulus calibrations in
ABR measurements. In: Jacobson JT, editor. The Auditory Brainstem Response. San Diego: College-Hill Press; 1985. p. 49–62.
[24] Halas ES, Beardsley JV, Sandlie ME. Conditioned neuronal responses at various levels in conditioning paradigms. Electroencephalogr Clin Neurophysiol 1970;28(5):468–77.
[25] Hayes EA, Warrier CM, Nicol TG, Zecker SG, Kraus N. Neural
plasticity following auditory training in children with learning problems. Clin Neurophysiol 2003;114(4):673–84.
[26] Hoormann J, Falkenstein M, Hohnsbein J. Effect of selective attention on the latency of human frequency-following potentials. NeuroReport 1994;5(13):1609–12.
[27] Huang CM, Buchwald JS. C comparison of acoustic nerve and
cochlear nucleus responses during acoustic habituation. Brain Res
1979;173(3):562–6.
[28] Huang CM, Buchwald JS. Changes of acoustic nerve and cochlear
nucleus evoked potentials due to repetitive stimulation. Electroencephalogr Clin Neurophysiol 1980;49(1–2):15–22.
[29] Illing RB, Forster CR, Horvath M. Evaluating the plasticity potential of the auditory brain stem nucleus in the rat. Am J Otol
1997;18(Suppl. 6):52–3.
[30] Illing RB, Horvath M, Laszig R. Plasticity of the auditory brainstem:
effects of cochlear ablation on GAP-43 immunoreactivity in the rat.
J Comp Neurol 1997;382(1):116–38.
[31] Irvine DR, Rajan R, Brown M. Injury- and use-related plasticity in
adult auditory cortex. Audiol Neurootol 2001;6(4):192–5.
[32] Khalfa S, Bougeard R, Morand N, Veuillet E, Isnard J, Guenot M,
et al. Evidence of peripheral auditory activity modulation by the
auditory cortex in humans. Neuroscience 2001;104(2):347–58.
N.M. Russo et al. / Behavioural Brain Research 156 (2005) 95–103
[33] King C, Warrier CM, Hayes E, Kraus N. Deficits in auditory brainstem pathway encoding of speech sounds in children with learning
problems. Neurosci Lett 2002;319(2):111–5.
[34] Klatt D. Software for cascade/parallel formant synthesizer. J Acoust
Soc Am 1980:67971–5.
[35] Knudsen EI, Zheng W, DeBello WM. Traces of learning in
the auditory localization pathway. Proc Natl Acad Sci USA
2000;97(22):11815–20.
[36] Kraus N, Disterhoft JF. Response plasticity of single neurons in
rabbit auditory association cortex during tone-signalled learning.
Brain Res 1982;246(2):205–15.
[37] Kraus N, McGee T. Electrophysiology of the human auditory system. In: Popper AN, Fay RR, editors. The Mammalian Auditory
Pathway: Neurophysiology, vol. 2 New York: Springer-Verlag; 1992
(p. 335–403).
[38] Kraus N, McGee TJ, Carrell TD, Zecker SG, Nicol TG, Koch DB.
Auditory neurophysiologic responses and discrimination deficits in
children with learning problems. Science 1996;273(5277):971–3.
[39] Kraus N, McGee TJ, Koch DB. Speech sound perception and learning: biologic bases. Scand Audiol Suppl 1998:497–517.
[40] Meric C, Collet L. Visual attention and evoked otoacoustic emissions: a slight but real effect. Int J Psychophysiol 1992;12(3):233–5.
[41] Meric C, Collet L. Differential effects of visual attention on spontaneous and evoked otoacoustic emissions. Int J Psychophysiol
1994;17(3):281–9.
[42] Merzenich M, Wright B, Jenkins W, Xerri C, Byl N, Miller S, et
al. Cortical plasticity underlying perceptual, motor, and cognitive
skill development: implications for neurorehabilitation. Cold Spring
Harb Symp Quant Biol 1996:611–8.
[43] Merzenich MM, Grajski K, Jenkins W, Recanzone G, Perterson B.
Functional cortical plasticity. Cortical network origins of representational changes. Cold Spring Harb Symp Quant Biol 1991:55873–
87.
[44] Merzenich MM, Jenkins WM, Johnston P, Schreiner C, Miller SL,
Tallal P. Temporal processing deficits of language-learning impaired
children ameliorated by training. Science 1996;271(5245):77–81.
[45] Miller GL, Knudsen EI. Adaptive plasticity in the auditory thalamus
of juvenile barn owls. J Neurosci 2003;23(3):1059–65.
[46] Moore DR. Plasticity of binaural hearing and some possible
mechanisms following late-onset deprivation. J Am Acad Audiol
1993;4(5):277–84.
[47] Moore DR. Auditory brainstem of the ferret: long survival following cochlear removal progressively changes projections from
the cochlear nucleus to the inferior colliculus. J Comp Neurol
1994;339(2):301–10.
[48] Quian Quiroga R, Garcia H. Single-trial event-related potentials with
wavelet denoising. Clin Neurophysiol 2003;114(2):376–90.
[49] Reed MA. Speech perception and the discrimination of brief auditory cues in reading disabled children. J Exp Child Psychol
1989;48(2):270–92.
[50] Rouiller E, Ribaupierre YD, Ribaupierre FD. Phase-locked responses to low-frequency tones in the medial geniculate-body. Hear
Res 1979;1(3):213–26.
[51] Russo N, Musacchia G, Nicol T, Zecker S, Kraus N. Brainstem
responses to speech syllables. Clin Neurophysiol 2004; in press.
[52] Schulte-Koerne G, Deimel W, Bartling J, Remschmidt H. The
role of phonological awareness, speech perception, and auditory
[53]
[54]
[55]
[56]
[57]
[58]
[59]
[60]
[61]
[62]
[63]
[64]
[65]
[66]
[67]
[68]
[69]
[70]
103
temporal processing for dyslexia. Eur Child Adolesc Psychiatry
1999;8:III/28–34.
Starr A, Don M. Brain potentials evoked by acoustic stimuli. In:
Picton T, editor. Human event-related potentials. Handbook of electroencephalography and clinical neurophsyiology, vol. 3 New York:
Elsevier; 1988 (p. 97–158).
Steinschneider M, Volkov IO, Noh MD, Garell PC, Howard III MA.
Temporal encoding of the voice onset time phonetic parameter by
field potentials recorded directly from human auditory cortex. J
Neurophysiol 1999;82(5):2346–57.
Suga N, Gao E, Zhang Y, Ma X, Olsen JF. The corticofugal
system for hearing: recent progress. Proc Natl Acad Sci USA
2000;97(22):11807–14.
Suga N, Xiao Z, Ma X, Ji W. Plasticity and corticofugal modulation
for hearing in adult animals. Neuron 2002;36(1):9–18.
Tallal P, Miller SL, Bedi G, Byma G, Wang X, Nagarajan
SS, et al. Language comprehension in language-learning impaired
children improved with acoustically modified speech. Science
1996;271(5245):81–4.
Taylor MM, Creelman CD. PEST: efficient estimates on probability
functions. J Acoust Soc Am 1967;41:782–7.
Tremblay K, Kraus N, McGee T. The time course of auditory perceptual learning: neurophysiological changes during speech–sound
training. NeuroReport 1998;9(16):3557–60.
Wallace MN, Rutkowski RG, Shackleton TM, Palmer AR. Phaselocked responses to pure tones in guinea pig auditory cortex. NeuroReport 2000;11(18):3989–93.
Warrier CM, Johnson KL, Hayes EA, Nicol T, Kraus N. Learning
impaired children exhibit timing deficits and training-related improvements in auditory cortical responses to speech in noise. Exp
Brain Res 2004; in press.
Weinberger NM. Dynamic regulation of receptive fields and maps
in the adult sensory cortex. Annu Rev Neurosci 1995;18:129–58.
Wible B, Nicol T, Kraus N. Abnormal neural encoding of repeated
speech stimuli in noise in children with learning problems. Clin
Neurophysiol 2002;113(4):485–94.
Wible B, Nicol T, Kraus N. Atypical brainstem representation of
onset and formant structure of speech sounds in children with
language-based learning problems. Biol Psychiatry 2004; in press.
Wilkinson G. Wide range achievement test-3. Wilmington, DE: Jastak; 1993.
Woodcock R, Johnson M. Woodcock–Johnson Psycho-Educational
Battery—revised. Tests of cognitive ability. Allen, TX: DLM Teaching Resources; 1977.
Woodcock R, Johnson M. Woodcock–Johnson Psycho-Educational
Battery—revised. Tests of cognitive ability. Allen, TX: DLM Teaching Resources; 1989.
Wright BA, Lombardino LJ, King WM, Puranik CS, Leonard CM,
Merzenich MM. Deficits in auditory temporal and spectral resolution in language-impaired children. Nature 1997;387(6629):176–8.
Yan W, Suga N. Corticofugal modulation of time-domain processing
of biosonar information in bats. Science 1996;273:1100–3.
Zhang Y, Suga N. Modulation of responses and frequency tuning of
thalamic and collicular neurons by cortical activation in mustached
bats. J Neurophysiol 2000;84(1):325–33.