Neuroscience Research 67 (2010) 334–340

Contents lists available at ScienceDirect

Neuroscience Research
journal homepage: www.elsevier.com/locate/neures

Music playing and memory trace: Evidence from event-related potentials

Keiko Kamiyama a,b , Kentaro Katahira a,c,d , Dilshat Abla a , Koji Hori b , Kazuo Okanoya a,c,∗
a
Laboratory for Biolinguistics, Brain Science Institute, RIKEN, Japan
b
Department of Psychology, Faculty of Letters, Rikkyo University, Japan
c
Japan Science Technology Agency, ERATO, Okanoya Emotional Information Project, Japan
d
Department of Complexity Science and Engineering, Graduate School of Frontier Sciences, The University of Tokyo, Japan

a r t i c l e i n f o a b s t r a c t

Article history: We examined the relationship between motor practice and auditory memory for sound sequences to
Received 9 December 2009 evaluate the hypothesis that practice involving physical performance might enhance auditory memory.
Received in revised form 30 March 2010 Participants learned two unfamiliar sound sequences using different training methods. Under the key-
Accepted 12 April 2010
press condition, they learned a melody while pressing a key during auditory input. Under the no-key-
Available online 18 April 2010
press condition, they listened to another melody without any key pressing. The two melodies were
presented alternately, and all participants were trained in both methods. Participants were instructed
Keywords:
to pay attention under both conditions. After training, they listened to the two melodies again without
Music
Absolute pitch
pressing keys, and ERPs were recorded. During the ERP recordings, 10% of the tones in these melodies
Memory deviated from the originals. The grand-average ERPs showed that the amplitude of mismatch negativity
ERP (MMN) elicited by deviant stimuli was larger under the key-press condition than under the no-key-
MMN press condition. This effect appeared only in the high absolute pitch group, which included those with
a pronounced ability to identify a note without external reference. This result suggests that the effect of
training with key pressing was mediated by individual musical skills.
© 2010 Elsevier Ireland Ltd and the Japan Neuroscience Society. All rights reserved.

1. Introduction dition. The practicing group received additional practice sessions

It is important for musicians to memorize a passage of music
as many pianists and solo players need to perform long passages
from memory. According to Macmillan (2004), musicians play in
the absence of reading the scores in order to enrich the quality
of music rather than to demonstrate an ability to memorize. For
example, memorization promotes a better understanding of the
music and enables players to concentrate on the sounds produced.
Musicians must, however, exert great effort to achieve these ben-
efits and might therefore repeatedly practice a tune or listen to
recorded music. It remains unclear, however, whether playing or
listening represents the more efficient method for memorizing a
series of notes. Lahav et al. (2005) investigated the effect of train-
ing methods on piano performances by training non-musicians to
play a simple piece of piano music by ear. After training, partic-
ipants were assigned to one of three groups and attended three
additional sessions over a period of one week. The piano-listening
group passively listened to the same music they had practiced. The
nature-listening group listened to nature sounds as a control con-
∗ Corresponding author at: Laboratory for Biolinguistics, Brain Science Institute,
RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan. Tel.: +81 48 467 7502;
fax: +81 48 467 7503.
E-mail address: okanoya@brain.riken.jp (K. Okanoya).
regarding the initial musical piece using a keyboard. Finally, all par-
ticipants engaged in a performance test in which they played the
entire piece of music, and the performance was assessed by mea-
suring note/timing errors. The accuracy of the performances was
greatest in the practicing group, followed by the piano-listening
group. Fewer note/timing errors were committed by those in the
practicing and piano-listening groups than by those in the nature-
listening group. These results suggested that passively listening to
a recently learned passage of music might maintain motor perfor-
mance. However, the data also indicated that piano listening was
inferior to actual piano practice in terms of the ability to maintain
accuracy of performance. In the piano-listening group, accuracy
was initially acquired but was lost, subsequently, to some extent
suggesting that active playing and passive listening differentially
affected the memory and performance of music passages.
Lahav et al. (2007) showed how passive piano listening and
active piano playing stimulated different brain areas that could
exert different influences on the memory of musical pieces. They
trained non-musicians to play a piece of piano music by ear and, fol-
lowing five days of training, fMRI scans were carried out. Although
the participants did not perform any movements, the frontoparietal
motor-related network and Broca’s area were activated when they
listened to the same piece of music used during training. When the
participants listened to a different piece of music consisting of the
same notes as the music used for training, the motor-related net-

0168-0102/$ – see front matter © 2010 Elsevier Ireland Ltd and the Japan Neuroscience Society. All rights reserved.
doi:10.1016/j.neures.2010.04.007
 K. Kamiyama et al. / Neuroscience Research 67 (2010) 334–340 335

work was, however, activated to a lesser degree and Broca’s area demonstrated absolute pitch (Zatorre, 2003). Scores on this test
remained inactivate. Lahav et al. (2007) claimed that activation of were used to divide participants into two groups, and the EEG data
the motor-related area was facilitated by the connection between obtained for these groups were compared using several types of
the sound of one note and the movement involved in pressing analyses.
one key. They also suggested that Broca’s area was associated with
learning the relationship between the sequence of piano tones and 2. Materials and methods
the corresponding sequence of key-press movements, and that this
area supported the unconscious simulations and predictions of the 2.1. Participants
subsequent action and sounds.
Consistent with the suggestion of Lahav et al. (2005, 2007), piano Six men (age range: 20–29 years, mean: 22.05 years) and 14
training using key-press movements and passive listening most women (age range: 20–29 years, mean: 22.33 years) were paid
likely rely on certain mechanisms that facilitate the prediction of for participating in this experiment. All participants had received
forthcoming sounds, but these seem to function in different ways. formal piano lessons for at least two years (Table 1).
In addition, playing a piece of music might lead to the develop- All participants had normal hearing and no history of neurolog-
ment of strong connections between motor sequences and sound ical disease. Only one male was left-handed. All procedures were
sequences, thus allowing the prediction of subsequent sounds to approved in advance by the RIKEN Ethics Committee, and all par-
become easier when listening to music. No evidence that these ticipants gave written informed consent before the experiment.
training methods actually affected the prediction of musical sound
sequences has, however, been put forward. In order to clarify how 2.2. Stimuli
these different training methods might affect memory for pas-
sages of music, we used electroencephalograms (EEG) to examine Two types of unfamiliar sound sequences (Melody A and Melody
whether training involving key pressing enhances auditory mem- B) were presented alternately during the training and the EEG
ory. recoding phases. Each melody consisted of 16 notes using 5 piano
In the present study, participants learned two unfamiliar pieces keys: C4 (261.1 Hz), D4 (293.7 Hz), E4 (329.6 Hz), F4 (349.6 Hz), and
of music under two different conditions. In the key-press condition, G4 (392.0 Hz). Each note lasted 750 ms, and each sequence lasted
participants played the piano by imitating the piano music pre- 12 s.
sented. Under the no-key-press condition, they simply listened to
the musical sequence without pressing any keys. The brain activity
2.3. Procedure
of the participants was then measured while they listened to the
same extracts of music again. During the EEG measurement, 10%
2.3.1. Absolute pitch test
of the tones within each melody deviated from those that were
An absolute pitch (AP) test, in which 3 octave tones (36 pure
originally presented. We hypothesized that the key-press training
tones) were presented randomly, was conducted at the beginning
would enhance the auditory memory of an unfamiliar melody. This
of this experiment. Each pure tone was presented for 300 ms at
being the case, the deviant stimuli would thus induce a larger mis-
intervals of 4500 ms. Following each tone, participants identified
match negativity (MMN) under the key-press condition than under
which notes they heard by pressing the appropriate piano keys
the no-key-press condition.
(Roland, D-900), but participants did not receive auditory feedback.
MMN is an index of the unconscious detection of error. It has
Thirty-six different tones were presented in one session, and a total
been reported that the strength of an auditory memory trace of
of three sessions were conducted. Participants were classified into
a standard stimulus that was developed during an oddball task
high AP and low AP groups according to the rates at which they
was reflected in the MMN amplitude. As the frequency of standard
responded correctly during the three sessions. This classification
stimuli was increased, the MMN amplitude increased (Baldeweg
was used for later analysis of EEG data.
et al., 2004; Haenschel et al., 2005). In addition, an MMN study
using a recognition-masking paradigm found that MMN ampli-
2.3.2. Training
tude increased when the memory traces of standard stimuli were
Melodies A and B were presented alternately through ear-
strengthened by eliminating the masking effect (Winkler et al.,
phones. Participants were asked to play one sequence on the piano
1993). MMN can also be used to index the training-related improve-
(Roland, D-900) (key-press condition) but to refrain from playing
ment in discrimination ability (Näätänen et al., 2001). The MMN
another sequence (no-key-press condition) (Fig. 1). The assign-
amplitude has been increased using discrimination training (Kraus
ment of melodies to each condition and the order of presentation
et al., 1995) as well as long-term experience (Chandrasekaran et
were determined randomly per participant. Under the key-press
al., 2007). In view of such findings, it is probable that piano perfor-
condition, participants listened to one melody and simultaneously
mance, which might strengthen the representation of piano tones
reproduced the melody on a keyboard without a musical score. The
(Katahira et al., 2008), would promote the encoding of melodies and
would enhance the MMN component of EEG recordings in response
to deviant tones.
Pantev et al. (2009) found that musical training enhanced MMN
amplitude for deviant tones in non-musicians. It has not, however,
been established how musical talent and life-long experience might
influence the training effect. In this study, we focused on abso-
lute pitch in relation to the musical ability of individuals. Accurate
absolute pitch should facilitate the retrieval of auditory categorical
representations in the absence of playing the piano. Is key-press
training, therefore, effective for those possessing absolute pitch?
In the context of the assumption that people with absolute pitch
can quickly and effortlessly identify the precise position of a tone
Fig. 1. During training, participants alternately listened to Melodies A and B. They
on the scale without reference to any other tone, we conducted played one melody on a keyboard (key-press condition) and simply listened to
an absolute pitch test to examine the degree to which participants another melody (no-key-press condition).
336 K. Kamiyama et al. / Neuroscience Research 67 (2010) 334–340
Table 1
The age for beginning private piano lessons and the continuous period of the lessons are shown for AP groups. The initial age for beginning piano lessons was significantly
lower in the high AP group than in the low AP group. We found a significant tendency toward longer continuous periods of piano lessons in the high AP group than in the
low AP group.
High AP group
Beginning age Continuous period
1 28
3 26
3 24
3 13
4 8
4 6
5 24
5 16
5 15
5 10
5 7
6 12
Average 4.08 15.75
S.D. 1.38 7.85
auditory feedback of piano tones was provided through earphones.
Under the no-key-press condition, participants placed their hands
on their laps and were instructed to hold them still. While listening
to and playing the melodies, participants looked at a fixation point,
“+,” placed directly in front of their eyes.
In each session, Melody A and Melody B were presented a total
of 40 times. The key-press condition included 20 trials and was
divided into four blocks. Training continued until the performances
of participants were more than 95% correct, and they pressed the
piano keys within 100 ms of the presentation of stimuli in each of
the four blocks. This procedure involved 2–5 sessions.
2.4. EEG recording
After training, Melody A and Melody B were presented again
during EEG recording. The melodies alternated 18 times in every
session. Within each melody, 10% of the tones were shifted up or
down to the neighboring tones within the C-major scale using a
Musical Instrument Digital Interface program. For instance, C was
changed to D, and F was changed to E on each occasion. Partici-
pants were asked simply to listen to the melodies during the three
sessions.
EEGs were recorded by a 64-channel Ag–Cl electrode cap (10–20
system) using the Scan 4.3 acquisition system (SynAmps; Neu-
roScan), with a 0.15–30 Hz band-pass filter and a sampling rate of
500 Hz. A reference electrode was placed over the left mastoid, and
electrodes on the bilateral earlobes were referenced on later analy-
sis. The impedances on all electrodes was measured and confirmed
to be less than 5 k. Vertical and horizontal electrooculograms
(EOGs) were recorded simultaneously to eliminate contamination
of EEG data by eye movements.
2.5. Data analysis
EEG data were analyzed using EDIT 4.3 (NeuroScan, Inc.). ERP
waveforms were time-locked to the onset of each tone involving a
time period of 700 ms, including a 100-ms pre-stimulus baseline.
To eliminate artifacts caused by eye movements, EEG data were
rejected off-line whenever the amplitude exceeded ±100 ␮V, and
the remaining data were averaged per stimulus type. Four types of
stimuli were used: standard stimuli under the key-press condition,
deviant stimuli under the key-press condition, standard stimuli
under the no-key-press condition, and deviant stimuli under the
no-key-press condition. Following this procedure, all trials and
channels were scanned manually for any additional disturbances.
To account for the topographical distribution of the ERPs in the
Low AP group
Beginning age Continuous period
3 17
4 13
5 12
5 6
6 5
7 11
11 9
13 2
6.75 9.38
3.49 4.87
statistical analysis, the scalp surface was divided into seven topo-
graphical regions (Abla et al., 2008; Abla and Okanoya, 2009), each
corresponding to three electrodes: middle anterior (Fp1, Fz, Fp2);
middle central (FCz, Cz, CPz); middle posterior (P3, Pz, P4); left ante-
rior (F3, FC3, F7); right anterior (F4, FC4, F8); left posterior (CP3, TP7,
P7); and right posterior (CP4, TP8, P8). To evaluate the MMN com-
ponent, the average amplitudes at 170–210 ms after stimulus onset
were analyzed by ANOVA (˛ level = 0.05). The time window for
measuring MMN was defined as between 20 ms before and 20 ms
after the peak amplitude in the grand-average waveform (Otten et
al., 2000). The mean amplitude data were subjected to a repeated-
measures ANOVA with stimulus type (standard stimuli under the
key-press condition, deviant stimuli under the key-press condition,
standard stimuli under the no-key-press condition, deviant stimuli
under the no-key-press condition) and electrode site (seven ROIs)
as within-subject factors and AP group (high AP and low AP) as a
between-subjects factor. Additional ANOVAs were used to assess
specific differences in the mean amplitudes under each condition
and in each AP group.
3. Results
3.1. Absolute pitch test
The mean rate of correct responses in the absolute pitch test was
42.54% (S.D. = 26.99). Participants were classified into two groups
based on scores from the AP test. The high absolute pitch group
(high AP group) consisted of 12 participants who scored higher than
40% on the absolute pitch test, and the low absolute pitch group
(low AP group) consisting of 8 participants who had scored under
20% (Fujisaki and Kashino, 2002) (Fig. 2). The mean scores were
61.10% (S.D. = 17.69) for the high AP group, and 14.70% (S.D. = 2.90)
for the low AP group. The scores were significantly higher in the
high AP group than in the low AP group [t(18) = 7.288, P < 0.0001].
A negative correlation was also found between the absolute pitch
test scores and the age at which participants began piano lessons
[r = −0.525] (Fig. 2). In addition, the age at which private piano
lessons began was significantly lower in the high AP group than
in the low AP group [t(18) = 2.403; P = 0.027] (Table 1). The contin-
uous period devoted to lessons tended to be significantly longer in
the high AP group than in the low AP group [t(18) = 2.040, P = 0.056].
3.2. Training
Participants participated in an average of 2.35 (S.D. = 0.754)
training sessions. Those in the high AP group participated in an
 K. Kamiyama et al. / Neuroscience Research 67 (2010) 334–340 337

of 41.74 ms (S.D. = 7.834) was observed between key press and the
onset of presented stimuli. The mean correct response rates were
99.32% (S.D. = 0.014) in the high AP group and 98.44% (S.D. = 0.014)
in the low AP group. The mean time lag between the onset of
the stimulus and the corresponding key press was 39.87 ms
(S.D. = 8.574) in the high AP group and 44.55 ms (S.D. = 6.004) in
the low AP group. We found no significant differences between AP
groups in correct response rates [t(18) = 1.396; P = 0.180] or time lag
[t(18) = 1.335; P = 0.199] between stimulus onset and performance.

3.3. ERP results

The mean ERP amplitude at 170–210 ms after stimulus onset

Fig. 2. A negative correlation between absolute pitch test scores and the age of
beginning piano lessons was observed. () High AP group and (×) low AP group.
average of two training sessions (S.D. = 0), and those in the low
AP group participated in an average of 2.875 training sessions
(S.D. = 0.754). For the final training session, the grand-average cor-
rect response rate was 98.97% (S.D. = 0.014), and a mean time lag
revealed a significant main effect of stimulus type (standard stimuli
under the key-press condition, deviant stimuli under the key-press
condition, standard stimuli under the no-key-press condition, and
deviant stimuli under the no-key-press condition) [F(3, 54) = 11.94;
P < 0.0001] and electrode site [F(6, 108) = 100.87; P < 0.0001]. We
also found a significant stimulus type × ROIs interaction [F(18,
324) = 4.86; P < 0.0001] and a stimulus type × ROIs × group inter-
action [F(18, 324) = 2.99; P < 0.0001].
We conducted post-hoc analyses for the mean amplitudes FCz,
Cz and CPz for the 170–210 ms window in order to identify differ-
ences among ERP amplitudes by stimulus type in each AP group. In
addition, the mean amplitudes for the left ROIs (left anterior and
left posterior) and right ROIs (right anterior and right posterior)
were tested for the analysis of laterality.

Fig. 3. Grand-average ERPs for electrodes FCz, Cz, and CPz. (A) In the high AP group (n = 12), the amplitudes of MMN for deviant tones under the key-press condition were
significantly larger than were those under the no-key-press condition. (B) In the low AP group (n = 8), we found no significant differences in the MMN amplitudes under
the key-press and the no-key-press conditions. Brown: standard stimuli under the key-press condition, red: deviant stimuli under the key-press condition, black: standard
stimuli under the no-key-press condition, blue: deviant stimuli under the no-key-press condition.
338 K. Kamiyama et al. / Neuroscience Research 67 (2010) 334–340
Fig. 4. Topographical differences in deviant and correct tones at 188 ms. (A) The
high AP group under the key-press condition, (B) the high AP group under the no-
key-press condition, (C) the low AP group under the key-press condition, and (D) the
low AP group under the no-key-press condition. White dots represent each electrode
site, and the white dot within the gray circle represents electrode FCz.
3.3.1. High AP group
In the high AP group, the deviation of the melodies induced
MMN under the key-press condition and the no-key-press con-
dition (Fig. 3A). The MMN was most prominent for electrodes
located in the middle central and right hemisphere in the key-
press condition (Fig. 4A and B). A repeated-measures ANOVA for
ERP amplitudes of 170–210 ms after stimulus onset revealed no
significant difference between conditions in the amplitudes for
the standard stimuli [F(1, 11) = 1.52; P = 0.24]. The negative com-
ponents for the deviant stimuli were significantly larger than
those for the standard stimuli under the key-press condition
[F(1, 11) = 13.51; P < 0.005]and the no-key-press condition [F(1,
11) = 8.02; P < 0.05]. MMN amplitudes for the deviant stimuli in the
key-press condition were also significantly larger than those under
the no-key-press condition [F(1, 11) = 9.58; P = 0.01] (Fig. 3A).
The MMN amplitudes were significantly different between
left and right hemispheres under the key-press condition [F(1,
11) = 5.30; P < 0.05]. Laterality was not, however, significant under
the no-key-press condition [F(1, 11) = 0.27; P = 0.61].
3.3.2. Low AP group
MMN was also observed for deviant stimuli under both the
key-press [F(1, 7) = 4.41; P = 0.07] and no-key-press [F(1, 7) = 10.42;
P = 0.01] conditions in the low AP group in the 170–210 ms window
(Figs. 3B, 4C and D). However, we found no significant differences
in the MMN amplitudes for the deviant stimuli between the key-
press and no-key-press conditions [F(1, 7) = 0.01; P = 0.92]. There
was additionally no significant difference in the amplitudes for the
standard stimuli under the key-press and no-key-press conditions
[F(1, 7) = 0.16; P = 0.70]. Laterality was also not significant under
either of these conditions [P > 0.05].
4. Discussion
4.1. MMN amplitude reflects the efficiency of key-press training
We found that learning a piece of music by pressing keys
induced larger MMN amplitudes than did training without pressing
keys during listening to a passage of music (Fig. 3A). The MMN com-
ponent reflected error-detection activity, and the MMN amplitude
was affected by the strength of the auditory representation of stan-
dard stimuli (Baldeweg et al., 2004; Haenschel et al., 2005; Winkler
et al., 1993). Therefore, large MMN amplitudes in EEG recordings
might imply that the representation of standard stimuli was firmly
established during training, presumably because key-press training
promoted the formation of auditory memory traces for the piece
of music in question. Indeed, the MMN was also induced under
the no-key-press condition because participants learned a musical
piece through training. However the MMN amplitude was smaller
in the no-key-press condition than the key-press condition, indi-
cating that no-key-press training might be less effective than actual
key-press training.
This interpretation is consistent with the conclusions of Katahira
et al. (2008). In their first experimental condition, participants were
asked to play the keyboard using a musical score, and auditory feed-
back was sometimes altered during the musical performance. In the
second condition, participants listened to the music without play-
ing the keyboard, and a deviant tone was occasionally presented.
Using EEG data, they compared brain activity under conditions 1
and 2 and found ERP component N210 in the first experimental
condition only in those participants who had experienced early and
extensive piano training. They speculated that the negative com-
ponent was induced during the keyboard performance partially
because performing strengthened the imagery of auditory feed-
back. In the present study, pressing keys may also have enhanced
the representation of auditory feedback and could have led to the
enhancement of auditory memory for learned music.
We cannot rule out the possibility, however, that auditory mem-
ory was enhanced simply because the auditory input during the
key-press condition was twice as intense as the no-key-press con-
dition. In the key-press condition, participants received auditory
feedback in addition to hearing the presentation of tones, whereas
they only listened to the presentation of the stimuli during the no-
key-press condition. Therefore, it is plausible that this difference
was reflected in the formation of auditory memory traces. Par-
ticipants were, however, asked to press the keys simultaneously
with the onset of auditory stimuli. Although the sound intensity
in the key-press condition differed from that in the no-key-press
condition, little difference with respect to the content of auditory
information was involved. Therefore, the diversity of auditory input
can be ruled out as the cause of the MMN increase during the key-
press condition.
If key pressing does promote auditory memory, this needs to
be explained. As Lahav et al. (2007) argued, it is plausible that the
auditory–motor network is involved in this process. Training using
key pressing might connect the representations involved in motor
sequences and sound sequences, which could facilitate the retrieval
of the memory of sound sequences.
4.2. Training efficiency changes depending on individual musical
experience
Key-press training increased MMN amplitudes only in the high
AP group (Fig. 3). The question, however, arises as to why no sig-
nificant differences in MMN amplitudes were found between the
key-press and no-key-press conditions in the low AP group?
The quality of the auditory–motor network might vary
greatly depending on individual musical experience. Indeed, the
 K. Kamiyama et al. / Neuroscience Research 67 (2010) 334–340
auditory–motor network can be formed through training (Lahav et
al., 2007), and it is highly probable that this network is stronger
in those who have received extensive piano training since early
childhood. The high AP group included participants with such
experience (Table 1, Fig. 2), so that the representation of piano
sounds may already have been connected to key-press movements
in this case. In addition, these connections might have facilitated
the development of new connections between sequences of finger
movements and sequences of piano sounds.
The model relating to speech proposed by Warren et al. (2005)
would account for qualitatively diverse auditory–motor networks.
According to this model, the correspondence between sound
and performance is acquired by a template-matching algorithm.
More specifically, the posterior superior temporal plane (pos-
terior STP), which processes auditory representations, and the
prefrontal cortex, pre-motor cortex, and motor cortex, which pro-
cess motor programs, transfer information bi-directionally. The
series of acoustic changes in an extended auditory signal would be
encoded as a sequence of discrete units of auditory information in
the posterior STP via matching with an auditory template. An action
would be recognized as a segmented sequence of spectro-temporal
motor representations in a way analogous to the process used for
auditory information. This type of encoding would facilitate linking
auditory and motor sequences by reducing the number of computa-
tions required for the auditory–motor mapping process. Warren et
al. (2005) suggested that this model could explain the way speech
is learned, but they also considered that auditory–motor trans-
formation mechanisms might be engaged in playing a musical
instrument. Peretz and Coltheart (2003) also implied that audi-
tory segregation of sound mixtures into distinct sound sources first
occurs in an acoustic analysis module whose domain encompasses
all auditory stimuli, including musical stimuli and speech.
When this model is applied to piano tones and key-press per-
formance, the smallest unit of the piano sound template might
be a sound with a fixed frequency. In addition, the smallest unit
of the motor template might be a movement that involves press-
ing one key primarily with one finger. We have referred to these
phenomena as the smallest units because it might be possible to
combine such units to form larger ones. Templates are formed
through experience (Warren et al., 2005), and the auditory–motor
pathway might also be strengthened by experience (Baumann et
al., 2007). In the high AP group, the well-formed auditory–motor
networks might result in the facilitation of connections between
each finger and each tone and between representations of auditory
and motor sequences. If participants had acquired just one-to-one
connections during training, key-press and no-key-press training
methods should have induced equally large MMN amplitudes. It
has been hypothesized that learning sequential connections as well
as one-to-one correspondence promotes error detection in the lis-
tening phase. In the low AP group, the sound template might be
weak or obscure, or the pathway connecting motor and auditory
representations might not be sufficiently specialized for playing
the piano. In the case of language, the sound template for native
language is formed during early childhood. Although one can learn
new sound templates for other languages, this becomes difficult
after a sensitive period. Perhaps a critical age for acquiring the
categorical perception of pitch also exists (Zatorre, 2003; Bahr et
al., 2005). Furthermore, an innate aptitude also affects the capac-
ity to acquire this ability (Zatorre, 2003; Bahr et al., 2005). Thus,
it might have been impossible for people in the low AP group to
obtain a stable template for piano tones during the training phase.
Consequently, we assumed that musical experience affected the
formation of the auditory–motor network that is specialized for
musical performance, and that the diversity in this network might
be reflected in the effect of key-press training for learning a piece
of music.
339
4.3. Future prospects
We analyzed the results according to the data obtained from the
absolute pitch test, but simply attributing the difference between
AP groups in terms of their ability regarding absolute pitch may be
inappropriate because absolute pitch constitutes a trivial example
of musical aptitude. Moreover, an absolute pitch test using pure
tones was not able to measure the special case of absolute pitch
that was specialized for piano sounds (cf. Pantev et al., 2001; Bahr
et al., 2005). We additionally observed stronger error-detection-
related activities in the right than in the left hemisphere, though
the music was played by the right hand. This result is consistent
with Pantev et al. (2009), who claimed that laterality effects appear
to occur at a level where melody is represented and that is inde-
pendent of the particular hand used. The way in which key-press
training promotes the formation of auditory memory traces, as well
as the processes by which musically specialized audio–motor net-
works are formed, are questions that need to be addressed in future
research.
5. Conclusions
In this study, we showed that error-detection activities during
listening to simple auditory sequences were promoted by practic-
ing sequences of pressing keys. The influence of key-press training
appeared only to occur in the high AP group, which included 10
participants with notably high scores on an absolute pitch test. We
suggested that the musically specialized auditory–motor networks
in these participants supported the formation of new connections
between the representation of key presses and piano notes that
facilitated the prediction of forthcoming sounds.
Acknowledgment
This work was supported in part by a Grant-in-Aid for Scientific
Research (B) (No. 20300096).
References
Abla, D., Katahira, K., Okanoya, K., 2008. On-line assessment of statistical learning
by event-related potentials. Journal of Cognitive Neuroscience 20, 952–964.
Abla, D., Okanoya, K., 2009. Visual statistical learning of shape sequences: an ERP
study. Neuroscience Research 64, 185–190.
Bahr, N., Christensen, C.A., Bahr, M., 2005. Diversity of accuracy profiles for absolute
pitch recognition. Psychology of Music 33 (1), 58–93.
Baldeweg, T., Klugman, A., Gruzelier, J., Hirsch, S.R., 2004. Mismatch negativity
potentials and cognitive impairment in schizophrenia. Schizophrenia Research
69, 203–217.
Baumann, S., Koeneke, S., Schmidt, C.F., Meyer, M., Lutz, K., Jancke, L., 2007. A net-
work for audio–motor coordination in skilled pianists and non-musicians. Brain
Research 1161, 65–78.
Chandrasekaran, B., Krishnan, A., Gandour, J.T., 2007. Experience-dependent neural
plasticity is sensitive to shape of pitch contours. Cognitive Neuroscience and
Neuropsychology 18 (3), 1963–1967.
Fujisaki, W., Kashino, M., 2002. The basic hearing abilities of absolute pitch posses-
sors. Acoustical Science and Technology 23 (2), 77–83.
Haenschel, C., Vernon, D.J., Dwivedi, P., Gruzelier, J.H., Baldeweg, T., 2005. Event-
related brain potential correlates of human auditory sensory memory-trace
formation. Journal of Neuroscience 25, 10494–10501.
Katahira, K., Abla, D., Masuda, S., Okanoya, K., 2008. Feedback-based error monitor-
ing process during musical performance: an ERP study. Neuroscience Research
61, 120–128.
Kraus, N., McGee, T., Carrell, T.D., King, C., Tremblay, K., Nicol, T., 1995. Central audi-
tory system plasticity associated with speech discrimination training. Journal of
Cognitive Neuroscience 7 (1), 25–32.
Lahav, A., Boulanger, A., Schlaug, G., Saltzman, E., 2005. The power of listen-
ing: auditory–motor interactions in musical training. Annals of the New York
Academy of Science 1060, 189–194.
Lahav, A., Saltzman, E., Schlaug, G., 2007. Action representation of sound: audiomo-
tor recognition networks while listening to newly acquired actions. The Journal
of Neuroscience 27, 308–314.
Macmillan, J., 2004. Successful memorizing. Piano Professional (September), 6–8.
Näätänen, R., Tervaniemi, M., Sussman, E., Paavilainen, P., Winkler, I., 2001. ‘Primitive
intelligence’ in the auditory cortex. TRENDS in Neuroscience 24 (5), 283–288.
340 K. Kamiyama et al. / Neuroscience Research 67 (2010) 334–340
Otten, L.J., Alain, C., Picton, T.W., 2000. Effects of visual attentional load on auditory
processing. Neuroreport 11 (4), 875–880.
Pantev, C., Roberts, L.E., Schulz, M., Engelien, A., Ross, B., 2001. Timbre-specific
enhancement of auditory cortical representations in musicians. Neuroreport 12,
169–174.
Pantev, C., Lappe, C., Herholz, S.C., Trainor, L., 2009. Auditory-somatosensory integra-
tion and cortical plasticity in musical training. Annals of the New York Academy
of Science 1169, 143–150.
Peretz, I., Coltheart, M., 2003. Modularity of music processing. Nature Neuroscience
6 (7), 688–691.
Warren, J.E., Wise, R.J., Warren, J.D., 2005. Sounds do-able: auditory–motor trans-
formations and the posterior temporal plane. TRENDS in Neurosciences 28,
636–643.
Winkler, I., Reinikainen, K., Näätänen, R., 1993. Event-related brain potentials
reflect traces of echoic memory in humans. Perception & Psychophysics 53 (4),
443–449.
Zatorre, R.J., 2003. Absolute pitch: a model for understanding the influence of genes
and development on neural and cognitive function. Nature Neuroscience 6,
692–695.