Attention improves auditory performance in noisy environments by either enhancing the processing of task-relevant stimuli (“gain”),
suppressing task-irrelevant information (“sharpening”), or both. In the present study, we investigated the effect of focused auditory
attention on the population-level frequency tuning in human auditory cortex by means of magnetoencephalography. Using complex
tions, we observed that focused auditory attention caused not only gain, but also sharpening of frequency tuning in human auditory
The ability to encode specific sounds in noisy environments is
important in daily life. In most day-to-day situations, we are
by paying attention to them. The processing might be tuned by
both enhancement of neural responses corresponding to task-
relevant stimuli (gain) and suppression of task-irrelevant neural
activities (sharpening). Despite extensive research, the tuning ef-
fects of attention in the human auditory cortex remain to be
exactly determined (Alain and Arnott, 2000).
A functional magnetic resonance imaging (fMRI) study
(Murray and Wojciulik, 2004) of the visual system showed that
attention not only increased the activation (gain), but also en-
hanced the selectivity of the neural population representing an
attended object. In a manner comparable with the visual system,
auditory attention might cause an overall increase in auditory
topic map, contributing to finer neural population-level coding
for attended sound signals (sharpening), as illustrated in Figure
have also been investigated using fMRI and other neuroimaging
techniques (Grady et al., 1997; Benedict et al., 1998; Murray and
Wojciulik, 2004). Previous electroencephalography (EEG) (Hill-
yard et al., 1973; Picton and Hillyard, 1974) and magnetoen-
cephalography (MEG) (Woldorff et al., 1993) studies also ob-
served that focused auditory attention increased the N1 auditory
response, a component thought to originate in lateral aspects of
Heschl’s gyrus and the posterior temporal plane (Pantev et al.,
1995; Eggermont and Ponton, 2002). A sharpening effect of at-
tention, however, has not yet been shown in this area.
The effect of the efferent (top-down) neural system on fre-
quency tuning has been studied at single-neuron level. Polley et
down task-dependent processes controlled cortical reorganiza-
tion in adult rats. Between two groups of rats, the authors used
identical auditory stimuli, but different attention tasks. The re-
in the reorganization of primary and secondary auditory cortex
and, thus, attention focused on frequency cues might also inten-
sify efferent neural inputs and expand the representation of the
target frequency range within the cortical tonotopic map.
Based on the aforementioned results, the major goal of the
present study was to investigate gain as well as sharpening effects
of focused auditory attention on the population-level frequency
tuning in human auditory cortex by means of MEG. We posit
that attention might strengthen not only excitatory neural con-
contribute to finer frequency tuning and better auditory
Participants were 13 healthy subjects between 22 and 28 years of age
(seven females; mean, 24.2 years) with no history of psychological or
neurological disorders. All subjects were right handed (assessed with the
Edinburgh Handedness Inventory) and their hearing thresholds were
within norms for the frequency range of 250–8000 Hz, as tested by
means of clinical pure tone audiometry. Participants gave written in-
Correspondence should be addressed to Dr. Christo Pantev, Institute for Biomagnetism and Biosignalanalysis,
Westfalian Wilhelms-University Muenster, Malmedyweg 15, 48149 Muenster, Germany. E-mail:
TheJournalofNeuroscience,September26,2007 • 27(39):10383–10390 • 10383
formed consent for participation in the study
Ethics Commission of the Medical Faculty,
University of Muenster.
Experimental design and stimuli. To evaluate
gain and sharpening effects of attention, we
presented a test stimulus (TS) either indepen-
dently or simultaneously with four different
band-eliminated noises (BENs) queued in a
random sequence. Neuronal activities evoked
by BEN and TS can be divided into three cate-
gories: activity evoked exclusively by BEN, ac-
tivity evoked exclusively by TS, or activity
evoked by both BEN and TS (Fig. 1B, light
gray, dark gray, black areas, respectively). Ac-
tivities of auditory neurons, which can be acti-
vated by both BEN and TS, decrease with BEN
becoming wider and/or with frequency tuning
becoming sharper. Thus, the diminution of
enlargement of areas activated solely by TS
(Fig. 1B, dark gray areas) illustrate improved
population-level frequency tuning. In this
study, neurons in areas of overlap (Fig. 1B,
black areas) could be activated by both BEN
and TS, but in fact they were already activated
by BEN when TS was presented (Fig. 2A).
reflects the activity of the neural group acti-
vated solely by TS onset (Fig. 1B, dark gray
areas). If the effect of attention were gain only,
in the case of each BEN condition (Fig. 1B,
dark gray areas) compared with no BEN (Fig.
between active versus distracted listening con-
ditions. In contrast, if attention sharpened the
population-level frequency tuning, the ratio
would become larger in the active compared
with the distracted listening condition. Thus,
the combination BEN plus TS allows us to
measure population-level frequency tuning of
the alert human auditory cortex by means of
The TS was a 40 Hz 100% amplitude-
modulated tone (12.5 ms rise and fall time)
with a carrier frequency of 1000 Hz and a du-
ration of 0.7 s. The sound onset asynchrony
between two subsequent TS was 3.0 s. In 10%
of the trials, the TS deviated in structure from
the standard TS. In these “deviant” trials, the
TS contained a silent period of 50 ms duration
(“temporal gap,” 12.5 ms fall and rise time)
starting randomly at 50, 100, 150, 200, 250,
served to control for subject’s compliance. Deviant stimuli could not be
analyzed appropriately because of low signal-to-noise ratio as a result of
few trials and contamination by artifacts, and were therefore excluded
from additional analysis.
BENs were prepared as follows: spectral frequency bands with
widths of either 20 Hz (BEN20), 40 Hz (BEN40), 80 Hz (BEN80), or
160 Hz (BEN160) centered around the 1000 Hz TS carrier frequency
(Fig. 2B) were eliminated from 8000 Hz low-pass filtered white noise
(corresponding to the upper frequency limit of the sound delivery
system). All BENs (duration of 3.0 s with 12.5 ms rise and fall times)
were presented starting 2.0 s before TS onset and ending 0.3 s after
files and presented via Presentation (Neurobehavioral Systems, Al-
bany, CA). Between BENs, there were silent intervals of 40 ms dura-
tion because of time delays produced by the sound presentation sys-
tem. Frequency tags of 18000 Hz (which were not perceivable by the
the sound stimulation. SRM-212 electrostatic earphones (Stax,
Saitama, Japan) were used as transducers. Sounds were delivered
through 60 cm silicon tubes with an inner diameter of 5 mm and
terminating at silicon earpieces fitting to the subject’s ears. The hear-
ing threshold for the TS was determined for each individual and for
each ear at the beginning of the MEG session. The TS was presented
binaurally at an intensity of 35 dB above individual sensation level;
corresponding sound pressure levels varied between 41 and 51 dB
B1–B4, The figures illustrate the relationship of neural activities elicited by BEN and TS as predicted by the different attention
models. Light gray areas represent neural activities exclusively elicited by BEN, and dark gray areas represent neural activities
broad spectral notch; right diagrams illustrate BENs with narrow spectral notch. Of note, the neural activities in both gain and
BEN and wide BEN differ between models: for B3and B4, ratios are much closer to 1 compared with B1and B2, reflecting the
10384 • J.Neurosci.,September26,2007 • 27(39):10383–10390Okamotoetal.•AttentionImprovesFrequencyTuning
(mean ? SD, 47.7 ? 3.7). The power of all BENs, which were also
presented binaurally, was 15 dB larger than TS power. In each session,
To investigate the effects of attention, we contrasted two different
attentional conditions per subject: active listening and distracted listen-
as quickly as possible with their left or right index finger (randomized
between subjects) for each detection of deviant TS. During distracted
listening, no task was required; subjects watched a silent movie of their
choice. The movie served to distract attention from the auditory modal-
ity. We decided not to present the silent movie during active listening
because, despite clear instructions, the movie might have distracted at-
tention from the auditory stimuli which, in turn, might have resulted in
states (Suzuki et al., 2005). The sound stimulation was identical between
the two sessions, which were performed on different days. Session order
was balanced across subjects.
Data acquisition and analysis. Auditory evoked fields (AEFs) were
measured with a helmet-shaped 275-channel whole-head neurogra-
diometer (Omega; CTF Systems, Coquitlam, British Columbia, Can-
ada) in a silent magnetically shielded room. Participants were com-
fortably seated upright. Head position was fixed with pads and
subjects were instructed not to move. Alertness and compliance were
verified via video monitoring. The measured magnetic response fields
were digitally sampled at a rate of 600 Hz. Epochs of data elicited by
standard TS, including a 300 ms pre-TS-onset interval and a 400 ms
post-TS-onset interval, were averaged selectively for each BEN con-
than three picotesla. The evoked field source locations and orienta-
tions were determined in a head-based Cartesian coordinate system
with the origin at the midpoint of the mediolateral axis ( y-axis),
which joined the center points of the en-
trances to the ear canals (positive toward the
left ear). The posterior–anterior axis (x-axis)
ran between nasion and origin, the inferior–
superior axis (z-axis) ran through the origin
perpendicularly to the x–y plane.
The N1m response is known to be gener-
ated in a relatively focused cortical area (pos-
terior temporal plane and lateral aspects of
Heschl’s gyrus) (Pantev et al., 1995; Egger-
mont and Ponton, 2002). We estimated N1m
source locations and orientations by means
of two single equivalent current dipoles (one
for each hemisphere) based on the no-BEN
condition using a spherical head model and
assuming identical locations and orienta-
tions for the BEN conditions, because a pre-
vious MEG study (Sams and Salmelin, 1994)
showed that simultaneously presented BENs
did not influence the calculated locations
and orientations of the N1m components
elicited by the test tone. For analysis of the
N1m component, the averaged magnetic field
signals were 30 Hz low-pass filtered initially fol-
ms prestimulus interval. The cortical sources
Initially, the time point of maximal global field
all sensors around 100 ms after stimulus onset,
before the peak was used for spatiotemporal
hemisphere of each subject was fixed in its loca-
tion and orientation, and the source strengths
condition (BEN160, BEN80, BEN40, and
BEN20). The maximal N1m source strengths were calculated in time win-
dows between 75 and 175 ms (no-BEN condition), 125 and 225 ms (BEN
160 condition), and 150 and 250 ms (BEN80, BEN40, and BEN20 condi-
tions), respectively. The estimated N1m locations with respect to each axis
active vs distracted; hemisphere: left vs right). In this article, the p values
provided for repeated-measures ANOVA results are Greenhouse–Geisser
generate the so-called “auditory steady-state response” (Makela and Hari,
was not possible to clearly extract this response for the BEN conditions
For the evaluation of the sharpening effect of attention on the
population-level frequency tuning, the maximal source strength of
the N1m elicited by the TS for each BEN condition in each hemi-
sphere was normalized with respect to the maximal N1m source
strength in the no-BEN condition for each subject and each hemi-
sphere individually. Normalization was used to reduce the impact of
the typically observed interindividual and intersession variability in
N1m source strength. The normalization procedure was not applied
to N1m latency given that the variability of latency among subjects
source strengths and latencies were then evaluated by repeated-
measures ANOVA using three factors (BEN type: BEN160, BEN80,
BEN40, and BEN20; hemisphere: left vs right; attention: active vs
distracted). Post hoc comparisons were performed using Bonferroni–
Dunn’s multiple-comparisons correction (significance threshold,
p ? 0.0083). In addition, non-normalized maximal N1m source
strength was similarly analyzed because the hypothesized “gain” ef-
fect of attention would get lost in normalized data.
of its 15 dB lower amplitude compared with BEN power. B, Amplitude spectra of the 3 s BENs measured at the earpiece. The
eliminated bandwidths are 20 Hz (BEN20), 40 Hz (BEN40), 80 Hz (BEN80), and 160 Hz (BEN160). The center frequency of the
Okamotoetal.•AttentionImprovesFrequencyTuningJ.Neurosci.,September26,2007 • 27(39):10383–10390 • 10385
To evaluate the deviant detection performance of the subjects, we con-
ducted additional behavioral measurements in a third session. These
measurements took place in the MEG room and therefore stimulation
devices, stimuli, and experimental parameters were identical to the “ac-
tive” MEG session with the exception of likelihood of stimulus appear-
ance: both standard and deviants appeared with a probability of 50%.
Stimulus order was pseudorandomized; each deviant stimulus was pre-
deviant trials for each BEN condition. Participants were instructed to
press a button with their right index finger as quickly as possible when
detecting a deviant stimulus. Reaction time and error rate (misses plus
ANOVA (BEN type: BEN160, BEN80, BEN40, and BEN20) and post hoc
comparisons were performed using Bonferroni–Dunn’s multiple-
comparisons correction (significance threshold, p ? 0.0083).
Theoretically, the possibility that participants have changed the
amount of allocated attention based on BEN condition cannot com-
pletely be ruled out. In this study, subjects might have paid more atten-
ior might have been reflected by larger attentional effects for narrower
compared with wider BENs. To rule out this possibility, just after the
termination of the behavioral measurement, we investigated whether
asking the participants whether she/he had noticed that different BENs
had been presented. Ten participants did not notice this at all, but three
asked to rank BENs via button press [button 1 (easy) to button 4 (diffi-
cult)]. Each BEN was presented 15 times for 3 s in randomized order.
Clearly identifiable AEFs were obtained from all subjects in all
conditions. After artifact rejection, a number of 156–180 (mean,
170) trials remained in each condition to be used for AEF aver-
laid on the MRI of one representative subject are displayed in
Figure 3 (supplemental Fig. 1, movies 1, 2, available at www.
jneurosci.org as supplemental material). Clear dipolar patterns
of-fit of the underlying dipolar model for dipole estimation was
in a range of 91.8–98.2% (mean ? SD, 95.8 ? 1.78%), confirm-
ing the adequacy of the chosen equivalent current dipole
approach. Figure 4 displays the group-averaged dipole locations
the 95% confidence interval limits of the relative differences
around the distracted listening condition.
The repeated-measures ANOVA applied to the dipole source
for hemisphere in the posterior–anterior dimension (x-axis;
F(1,12)? 5.8; p ? 0.05), the mediolateral dimension ( y-axis;
F(1,12)? 6.3; p ? 0.05), and the inferior–superior dimension
(z-axis; F(1,12)? 20.5; p ? 0.001). There was no significant inter-
action or main effect of attention. Hence, the estimated source
locations of the neural activities measured differed slightly be-
tween hemispheres regardless of whether the subjects focused
tion between hemispheres most likely reflects anatomical hemi-
spheric differences (Rademacher et al., 2001). At first glance, no
significant source location difference between active and dis-
tracted conditions seems to be inconsistent with previous fMRI
results revealing significantly larger cortical activations during
focused auditory attention (Grady et al., 1997; Benedict et al.,
1998), until it is considered that the dipole fit approach only
not the extent of activated areas. Thus, the extent of activated
neural areas may have differed between the two sessions, but the
centers of gravity of the neural responses were not significantly
This figure demonstrates a clear N1m response peaking at ?100
in the BEN conditions are delayed and show smaller peaks com-
pared with the no-BEN condition.
latencies for the left and right hemispheres and each BEN condi-
6. The repeated-measures ANOVA applied to the normalized
N1m source strengths resulted in significant main effects of BEN
type (F(3,36)? 22.8; p ? 0.0001), hemisphere (F(1,12)? 7.4; p ?
0.019), and attention (F(1,12)? 19.4; p ? 0.001) as well as a
5.3; p ? 0.014). Post hoc comparisons showed significant differ-
ences between BEN160 and BEN80 ( p ? 0.0003), BEN160 and
BEN40 ( p ? 0.0001), BEN160 and BEN20 ( p ? 0.0001), and
BEN80 and BEN20 ( p ? 0.003).
Moreover, because there was no significant interaction of
hemisphere with any factor, we collapsed data across hemi-
spheres and calculated planed comparisons (paired two-tailed t
tests; Bonferroni–Dunn-corrected significance threshold, p ?
0.0127) between active and distracted attentional states on nor-
0.1067 s. Red areas represent inward flows of magnetic fields from the brain, whereas blue
10386 • J.Neurosci.,September26,2007 • 27(39):10383–10390Okamotoetal.•AttentionImprovesFrequencyTuning
malized N1m source strength within each BEN condition to di-
rectly compare active and distracted conditions. The results
showed significant differences between BEN160-active and
BEN160-distracted ( p ? 0.008), BEN80-active and BEN80-
distracted ( p ? 0.0003), BEN40-active and BEN40-distracted
( p ? 0.0001) as well as between BEN20-active and BEN20-
distracted ( p ? 0.0001). Furthermore, we compared estimated
linear slopes of change for active and distracted conditions by
means of paired t test. The results show that the slope for dis-
tracted was significantly steeper than the slope for active (t(12)?
2.84; p ? 0.015).
indicated that attention as well as BEN type and hemisphere sig-
nificantly influenced the strength of the neural activities mea-
sured. Crucially, effects of attention and BEN type were not in-
dependent from each other, but interacted: the effect of auditory
focused attention increases with narrowing BEN.
The repeated-measures ANOVA applied to N1m latency re-
vealed significant main effects of attention (F(1,12)? 8.1; p ?
0.016) and BEN type (F(3,36)? 63.7; p ? 0.0001), but no signifi-
cant interaction between factors. Significant differences between
parisons. Hence, both attention and BEN type influenced the
timing of the neural activities measured, whereas the timing did
applied to the non-normalized maximal N1m source strengths
resulted in significant main effects for attention (F(1,12)? 61.2;
p ? 0.0001) and BEN type (F(3,36)? 23.0; p ? 0.0001). Thus,
ing compared with distracted listening. Attention as well as BEN
type influenced the strength of the neural activities measured.
This gain effect caused by focused auditory attention did not
differ between hemispheres.
Error rates (false alarms plus misses) became larger and reaction
times became longer with narrowing BENs as shown in Figure 7.
The repeated-measures ANOVA applied to error rate showed a
significant main effect of BEN type (F(3,36)? 103.1; p ? 0.0001),
and post hoc comparisons revealed significant differences be-
tween all BEN types ( p ? 0.0005). Also, the repeated-measures
of BEN type (F(3,36)? 46.5; p ? 0.0001), and again post hoc
comparisons revealed significant differences between all BEN
types ( p ? 0.001) except for BEN40 vs BEN80 ( p ? 0.024).
who had noticed differences between BENs were unable to rank
Estimated source locations of N1m. Localization of the N1m sources in the y–x
Grand-averaged source strength waveforms. The top graph displays grand-
Normalized N1m source strengths and latencies. The graphs display the group
Okamotoetal.•AttentionImprovesFrequencyTuningJ.Neurosci.,September26,2007 • 27(39):10383–10390 • 10387
them properly [mean, 1 (easy) to 4 (difficult) ? SD: BEN20,
2.62 ? 0.41; BEN40, 2.64 ? 0.60; BEN80, 2.69 ? 0.15; BEN160,
2.76 ? 0.65). Hence, participants were not able to identify the
different BENs reliably.
Moreover, to verify relationships between behavioral and
electro-neurophysiological responses, we performed additional
correlation analyses. For MEG variables (normalized N1m
source strength during active listening and N1m latency during
active listening) we obtained the means per BEN condition
(BEN160, BEN80, BEN40, and BEN20) across hemispheres and
obtained the means per BEN condition across subjects. Product-
icant relationships between normalized N1m source strength ac-
tive and reaction time (r ? ?0.961; p ? 0.039), N1m latency
active and reaction time (r ? 0.969; p ? 0.031), as well as a
0.913; p ? 0.087).
The present study experimentally confirmed the hypothesis that
AEFs elicited by TS depend on the type of simultaneously pre-
sented BEN as well as the subject’s attentional state. The results
showed that N1m responses were significantly larger during ac-
tive compared with distracted listening, especially when BENs
with relatively narrow eliminated frequency bands were pre-
sented. Identical auditory stimuli (BENs and TS) were used dur-
ing active and distracted listening conditions; the attentional
auditory inputs alone cannot explain the significant differences.
Our results strongly suggest that focused auditory attention im-
pacts the generators of the N1m, possibly via top-down neural
In this study, we have successfully investigated neural
population-level frequency tuning by means of MEG. Each BEN
activated a neural population overlapping with the population
corresponding to the TS. The degree of overlap differed between
BENs (Fig. 1B, black areas); in the case of narrow BENs, fewer
neurons were newly activated by the delayed TS onset (Fig. 1B,
elicited by TS onset represents the number of newly activated
neurons, which in turn reflects population-level frequency tun-
ing, as has been shown by Sams and Salmelin (1994). Using a
distracted listening condition, we replicated these results show-
tantly, in addition, we were able to demonstrate that the effect of
BEN type significantly differed as a function of attention.
Previous EEG studies (Hillyard et al., 1973; Picton and Hill-
yard, 1974) showed significantly increased N1 responses during
focused auditory attention. The authors suggested that focused
auditory attention could modulate neural activities at an early
Naatanen (1982) argued that the overlapping “processing nega-
tivity,” a component of endogenous origin characterized by a
source differing from N1 source (Woods and Clayworth, 1987),
caused the N1 amplitude enlargement. In the present experi-
tracted conditions were not observed, a result that has already
been found previously (Fujiwara et al., 1998). Therefore, it is
likely to assume that the N1m enlargement was caused by mod-
ulation of neural activities affecting transmission, analysis, and
representation of stimulus information in the auditory pathway
(Hansen, 1990). Moreover, normalized N1m source strength
showed significant differences between the two attentional con-
BEN type: the normalized N1m source strength difference be-
tween active and distracted listening conditions became system-
atically larger with decreasing spectral notch (Fig. 6). These re-
sults imply that focused auditory attention did not only amplify
neural activity, but also sharpened the frequency tuning in the
A series of previous psychoacoustical studies (Schlauch and
Hafter, 1991; Hafter et al., 1993; Hubner and Hafter, 1995) also
supports this hypothesis. It could be demonstrated that reducing
frequency uncertainty by presenting frequency cues leads to
sharpened frequency tuning and improved tone-detection per-
formance. In the present study, we used one single TS fixed in
frequency for all conditions. Hence, frequency uncertainty was
of neurophysiological filtering by focused auditory attention
around the test stimulus frequency.
Inhibitory neural interactions might play an essential role for
this sharpening effect. Previous studies have shown that the clas-
sical lateral inhibition concept (von Be ´ke ´sy, 1967; Suga, 1995;
10388 • J.Neurosci.,September26,2007 • 27(39):10383–10390Okamotoetal.•AttentionImprovesFrequencyTuning
Pantev et al., 2004; Okamoto et al., 2005, 2007) can account for
sharpening of frequency tuning in the central auditory system.
Afferent neural inputs consist not only of excitatory, but also of
broadly tuned inhibitory inputs, which suppress surrounding
neural activities resulting in improved spectral contrast. How-
ever, a cotuned excitatory and inhibitory neural model based on
previous single-neuron studies (Wehr and Zador, 2003; Oswald
et al., 2006) can also explain the sharpening effect. Those studies
demonstrated that frequency tuning curves of excitatory and in-
hibitory inputs are similar. However, inhibitory inputs follow
excitatory inputs with a few milliseconds delay. Cotuned neural
activities can improve temporal coding by shortening the dura-
attentional enhancement of cotuned neural activities could also
improve frequency tuning.
In the present study, the inhibitory system, intensified by fo-
cused auditory attention, might have sharpened the population-
level frequency tuning via the top-down auditory pathway. As a
consequence, neurons corresponding to edge frequencies of the
BENs might have been less activated by the initial part of the
resulting in increased N1m source strength during active com-
To summarize, the intensified inhibitory system in the auditory
cortex may explain the sharpening effect observed in response to
Single-cell recording studies revealed that frequency tuning
can be modulated by learning-induced plasticity in the inferior
auditory cortex (Weinberger et al., 1984; Ohl and Scheich, 1996;
Fritz et al., 2005), and secondary auditory cortical fields (Dia-
mond and Weinberger, 1984). Suga et al. (2002) showed that
electrical stimulations in auditory cortex could cause expanded
or compressed reorganization in this area and also subcortical
auditory nuclei via efferent inputs. The tuning curves of neurons
tant sounds (expanded reorganization, gain) or away from those
auditory nuclei via the efferent auditory pathway.
However, the comparison between plasticity of frequency
tion of population-level frequency tuning of human auditory
cortical responses might be inappropriate (Ohl and Scheich,
effects on both P1m and N1m, but the effect was larger on N1m.
An fMRI-study also showed that the mesial part of the human
auditory cortex is a stimulus-driven area that was always acti-
vated by a sound stimulus regardless of the subject’s state of at-
tention, whereas the activation of the lateral auditory cortex de-
pended on the state of attention regardless of sound properties
(Petkov et al., 2004). Although the lowest stage on which atten-
tional sharpening occurs remains to be determined, these results
indicated that attentional modulation of population-level neural
activities might take place mainly on the cortical level. Thus, it is
reasonable to assume that in the present study attention mainly
modulated the inhibitory neural conductance of the lateral audi-
in normalized N1m source strength, which was larger for the left
hemisphere during both active and distracted listening. These
laterality effects cannot be explained by the mere presence of the
left hemisphere may be dominant for processing requiring fine
temporal resolution (Zatorre and Belin, 2001), and temporal co-
herence seems to be crucial for the segregation of target sounds
from nontarget sounds (Barbour and Wang, 2002). Other au-
thors (Poeppel, 2003; Boemio et al., 2005) have suggested the
“asymmetric sampling in time” hypothesis. This model suggests
that the left auditory cortex dominantly exploits short temporal
tex exploits longer ones (150–250 ms). In the present study, in-
formation from short temporal integration windows would be
important for detection of the TS onset inside the BENs. Thus,
left hemisphere plays a more important role for monitoring and
analyzing auditory signals in noisy environments.
One can hypothesize that subjects may allocate more or less
attention to solving a task depending on its difficulty. In this
study, subjects were unable to strategically adjust the degree of
attention allocated before TS onset based on BEN type. More-
over, transient and sustained AEFs elicited by the initial part of
the different BENs did not differ systematically, but were gener-
ally larger in case of focused auditory attention. Sustained AEFs
evoked by BENs overlapped with neural activities elicited by TS.
tained neural activities elicited by BENs (Fig. 1B, black and light
gray areas) and were able to successfully extract neural responses
elicited by the test stimulus only (dark gray areas).
In conclusion, this study has shown for the first time that
focused auditory attention cannot only amplify neural activities
in general (gain effect), but also can sharpen population-level
frequency tuning in the human auditory cortex, possibly via the
inhibitory system. Auditory cortical neurons seem to be influ-
inputs on excitatory and inhibitory neural networks within the
human auditory cortex result in enhanced and sharpened
population-level neural responses, which are reflected by the
N1m response during focused auditory attention.
Alain C, Arnott SR (2000) Selectively attending to auditory objects. Front
Barbour DL, Wang X (2002) Temporal coherence sensitivity in auditory
cortex. J Neurophysiol 88:2684–2699.
(1998) Functional neuroimaging of attention in the auditory modality.
Boemio A, Fromm S, Braun A, Poeppel D (2005) Hierarchical and asym-
metric temporal sensitivity in human auditory cortices. Nat Neurosci
Diamond DM, Weinberger NM (1984) Physiological plasticity of single
neurons in auditory cortex of the cat during acquisition of the pupillary
conditioned response: II. Secondary field (AII). Behav Neurosci
Edeline JM, Weinberger NM (1992) Associative retuning in the thalamic
source of input to the amygdala and auditory cortex: receptive field plas-
EggermontJJ,PontonCW (2002) Theneurophysiologyofauditorypercep-
tion: from single units to evoked potentials. Audiol Neurootol 7:71–99.
Engelien A, Schulz M, Ross B, Arolt V, Pantev C (2000) A combined func-
Okamotoetal.•AttentionImprovesFrequencyTuningJ.Neurosci.,September26,2007 • 27(39):10383–10390 • 10389
tional in vivo measure for primary and secondary auditory cortices. Hear Download full-text
Fritz J, Elhilali M, Shamma S (2005) Active listening: task-dependent plas-
FujiwaraN,NagamineT,ImaiM,TanakaT,ShibasakiH (1998) Roleofthe
primary auditory cortex in auditory selective attention studied by whole-
head neuromagnetometer. Brain Res 7:99–109.
Gao E, Suga N (1998) Experience-dependent corticofugal adjustment of
midbrain frequency map in bat auditory system. Proc Natl Acad Sci USA
Grady CL, Van Meter JW, Maisog JM, Pietrini P, Krasuski J, Rauschecker JP
(1997) Attention-related modulation of activity in primary and second-
ary auditory cortex. NeuroReport 8:2511–2516.
Hafter ER, Schlauch RS, Tang J (1993) Attending to auditory filters that
were not stimulated directly. J Acoust Soc Am 94:743–747.
Hansen JC (1990) Processing negativity—comparison process or selective
processing. Behav Brain Sci 13:242.
Hillyard SA, Hink RF, Schwent VL, Picton TW (1973) Electrical signs of
selective attention in the human brain. Science 182:177–180.
Hubner R, Hafter ER (1995) Cuing mechanisms in auditory signal detec-
tion. Percept Psychophys 57:197–202.
Lennartz RC, Weinberger NM (1992) Frequency-specific receptive-field
plasticity in the medial geniculate-body induced by pavlovian fear condi-
tioning is expressed in the anesthetized brain. Behav Neurosci
Makela JP, Hari R (1987) Evidence for cortical origin of the 40 Hz auditory
evoked response in man. Electroencephalogr Clin Neurophysiol
Murray SO, Wojciulik E (2004) Attention increases neural selectivity in the
human lateral occipital complex. Nat Neurosci 7:70–74.
Naatanen R (1982) Processing negativity—an evoked-potential reflection
of selective attention. Psychol Bull 92:605–640.
Naatanen R, Gaillard AW, Mantysalo S (1978) Early selective-attention ef-
fect on evoked-potential reinterpreted. Acta Psychologica 42:313–329.
Ohl FW, Scheich H (1996) Differential frequency conditioning enhances
spectral contrast sensitivity of units in auditory cortex (field Al) of the
alert Mongolian gerbil. Eur J Neurosci 8:1001–1017.
Ohl FW, Scheich H (2005) Learning-induced plasticity in animal and hu-
man auditory cortex. Curr Opin Neurobiol 15:470–477.
Okamoto H, Kakigi R, Gunji A, Kubo T, Pantev C (2005) The dependence
of the auditory evoked N1m decrement on the bandwidth of preceding
notch-filtered noise. Eur J Neurosci 21:1957–1961.
Okamoto H, Kakigi R, Gunji A, Pantev C (2007) Asymmetric lateral inhib-
itory neural activity in the auditory system: a magnetoencephalographic
study. BMC Neurosci 8:33.
Oswald AMM, Schiff ML, Reyes AD (2006) Synaptic mechanisms underly-
ing auditory processing. Curr Opin Neurobiol 16:371–376.
Pantev C, Bertrand O, Eulitz C, Verkindt C, Hampson S, Schuierer G, Elbert
T (1995) Specific tonotopic organizations of different areas of the hu-
man auditory cortex revealed by simultaneous magnetic and electric re-
cordings. Electroencephalogr Clin Neurophysiol 94:26–40.
Pantev C, Roberts LE, Elbert T, Ross B, Wienbruch C (1996) Tonotopic
organization of the sources of human auditory steady-state responses.
Hear Res 101:62–74.
Pantev C, Okamoto H, Ross B, Stoll W, Ciurlia-Guy E, Kakigi R, Kubo T
(2004) Lateral inhibition and habituation of the human auditory cortex.
Eur J Neurosci 19:2337–2344.
Petkov CI, Kang X, Alho K, Bertrand O, Yund EW, Woods DL (2004) At-
tentional modulation of human auditory cortex. Nat Neurosci
Picton TW, Hillyard SA (1974) Human auditory evoked potentials. II. Ef-
fects of attention. Electroencephalogr Clin Neurophysiol 36:191–199.
Poeppel D (2003) The analysis of speech in different temporal integration
windows: cerebral lateralization as “asymmetric sampling in time.”
Speech Commun 41:245–255.
PolleyDB,SteinbergEE,MerzenichMM (2006) Perceptuallearningdirects
auditory cortical map reorganization through top-down influences.
J Neurosci 26:4970–4982.
Zilles K (2001) Probabilistic mapping and volume measurement of hu-
man primary auditory cortex. NeuroImage 13:669–683.
Ross B, Borgmann C, Draganova R, Roberts LE, Pantev C (2000) A high-
precision magnetoencephalographic study of human auditory steady-
state responses to amplitude-modulated tones. J Acoust Soc Am
Sams M, Salmelin R (1994) Evidence of sharp frequency tuning in the hu-
man auditory cortex. Hear Res 75:67–74.
Schlauch RS, Hafter ER (1991) Listening bandwidths and frequency uncer-
tainty in pure-tone signal detection. J Acoust Soc Am 90:1332–1339.
Suga N (1995) Sharpening of frequency tuning by inhibition in the central
auditory system: tribute to Yasuji Katsuki. Neurosci Res 21:287–299.
Suga N, Xiao Z, Ma X, Ji W (2002) Plasticity and corticofugal modulation
for hearing in adult animals. Neuron 36:9–18.
SuzukiJ,NittonoH,HoriT (2005) Levelofinterestinvideoclipsmodulates
Tan AY, Zhang LI, Merzenich MM, Schreiner CE (2004) Tone-evoked ex-
neurons. J Neurophysiol 92:630–643.
von Be ´ke ´sy G (1967) Sensory inhibition. Princeton, NJ: Princeton UP.
WehrM,ZadorAM (2003) Balancedinhibitionunderliestuningandsharp-
ens spike timing in auditory cortex. Nature 426:442–446.
WeinbergerNM,HopkinsW,DiamondDM (1984) Physiologicalplasticity
of single neurons in auditory cortex of the cat during acquisition of the
pupillary conditioned response. I. Primary field (AI). Behav Neurosci
FE (1993) Modulation of early sensory processing in human auditory
cortex during auditory selective attention. Proc Natl Acad Sci USA
Woods DL, Clayworth CC (1987) Scalp topographies dissociate N1 and Nd
Zatorre RJ, Belin P (2001) Spectral and temporal processing in human au-
ditory cortex. Cereb Cortex 11:946–953.
10390 • J.Neurosci.,September26,2007 • 27(39):10383–10390 Okamotoetal.•AttentionImprovesFrequencyTuning