Does familiarity facilitate the cortical processing of music sounds?

Abstract

Automatic cortical sound discrimination, as indexed by the mismatch negativity (MMN) component of the auditory evoked potential, is facilitated for familiar speech sounds (phonemes). In musicians as compared to non-musicians, an enhanced MMN has been observed for complex non-speech sounds. Here, musically trained subjects were presented with sequences of either familiar (tonal) or structurally matched unfamiliar (atonal) triad chords, both with either fixed or randomly transposed chord root pitch. The MMN elicited by deviant chords did not differ for familiar and unfamiliar triad sequences, and was undiminished even to unfamiliar deviant sounds which were consciously undetectable. Only subsequent attention-related components indicated facilitated cognitive processing of familiar sounds, corresponding to higher behavioral detection scores.

Full text

Page 1

Does familiarity facilitate the cortical processing
of music sounds?
Georg Neuloh1,2,CAand Gabriel Curio1
1Neurophysics Group,Departmentof Neurologyand Clinical Neurophysiology,Campus Benjamin Franklin,Charite ¤ -University Medicine Berlin,
Hindenburgdamm 30,12200 Berlin;2Departmentof Neurosurgery,Friedrich-Wilhelm-Universit?t, Sigmund-Freud-Strasse25, 53105 Bonn,Germany
CACorresponding Author:neuloh@ukb.uni-bonn.de
Received5 September 2004; accepted 23 September 2004
Automatic cortical sound discrimination, as indexed by the mis-
match negativity (MMN) component of the auditory evoked po-
tential, is facilitated for familiar speech sounds (phonemes). In
musicians as compared to non-musicians, an enhanced MMN has
been observed for complex non-speech sounds. Here, musically
trained subjects were presented with sequences of either familiar
(tonal) or structurally matched unfamiliar (atonal) triad chords,
both with either ¢xed or randomly transposed chord root pitch.
The MMN elicitedby deviantchords didnotdi¡er for familiar and
unfamiliar triad sequences, and was undiminished even to unfami-
liar deviant sounds which were consciously undetectable. Only
subsequent attention-related components indicated facilitated
cognitive processing of familiar sounds, corresponding to higher
behavioral detection scores. NeuroReport 15:2471^2475 ? c 2004
LippincottWilliams& Wilkins.
Key words:Mismatchnegativity; Music; Sensorydiscrimination;Tonal triads
INTRODUCTION
Cortical processing of acoustic stimuli can be probed non-
invasively by auditory evoked potentials (AEP). The
mismatch negativity (MMN) is an AEP component which
indicates that in a stimulus series a new auditory event
deviates from a sensory memory trace representing invar-
iant features of preceding acoustic standard stimuli [1]. This
memory trace encodes not only simple stimulus properties
such as pitch, loudness, or duration, but also higher-order
attributes of complex stimuli, such as spectral composition,
frequency ratios, dynamic or pitch contour of sequential
tone patterns, and combinations of such features [2–8]. The
MMN process is automatic as it operates without attention
directed toward the auditory stimulus stream [9]. In turn, it
can trigger an involuntary shift of attention towards the
deviant event; this shift is reflected by subsequent evoked
potential components, such as the N2b-P3a complex and the
P3 component [10]. Typically, these components indicate
conscious novelty processing.
The mismatch paradigm permits to investigate the
discriminative processing of domain-specific complex sti-
muli, such as speech (phonemes) and music sounds [5,11–
14]. MMN results obtained for speech sounds revealed that
sensory processing is facilitated specifically for phonemes
from the listener’s native language [3]. In the music domain,
the MMN to non-speech sound mismatch has been found
enhanced when comparing musicians and non-musicians
[7,8,15,16]. It is an interesting further question to what
extent facilitation by long-term familiarity is a general
feature of auditory mismatch processing, e.g., whether it
holds also for musical sounds, in particular for highly
familiar triad chords of the Western tonal system which can
be considered as units akin to phonemes of a Western
listener’s native musical language. Furthermore, it is not
clear whether a discrimination performance better for
familiar as compared to unfamiliar complex stimuli would
be preceded in general by a higher mismatch brain
response.
Here, we tested the hypothesis that long-term familiarity
of complex non-speech sounds facilitates their sensory
mismatch processing. Evoked potentials were recorded
from musically trained subjects who were presented with
sound mismatches in sequences of either highly familiar or
unfamiliar but structurally matched music chords. A
behavioral detection task allowed to correlate discrimina-
tion performance and mismatch responses.
MATERIALS AND METHODS
Twenty-two subjects (nine females, 13 males; median age 28
years; 21 right-handed) were paid to participate in this
study. All subjects had extensive musical training on up to
four musical instruments, beginning at a median age of
7 years, for a median period of 16.5 years, and listened to
Western tonal music for a median of 15h/week. Stimuli
were synthesized with the Turbosynth software (Digidesign,
Daly City, CA, USA), generated at 20kHz sampling rate,
and replayed binaurally from a DAT-tape via headphones at
B70dB SPL. Each stimulus consisted of three simultaneous
tones (triads), each composed of 16 harmonics with
exponentially decaying amplitudes, to ease sensory and
cognitive discrimination [4]. The resulting chords had a rich,
AUDITORYANDVESTIBULARSYSTEMS
NEUROREPORT
0959-4965? c Lippincott Williams& Wilkins
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Vol 15 No 16 15 November 2004
2471

Page 2

organ-like timbre. Stimulus duration was 1000ms including
25ms rise and fall times. A silent period of 1000ms
separated every two stimuli to permit aware assessment of
the sound category. Oddball chord sequences were designed
as detailed below, with a probability of pB0.2 for deviant
chords and 3–5 standard chords intervening between any
two deviants. Each sequence consisted of 433 stimuli and
was presented in three parts of 5min each to counteract
fatigue. In experiment 2, subjects performed a two-alter-
native forced-choice detection task, and discriminative
performance was calculated by Grier’s non-parametric
unbiased sensitivity index [17].
EEG activity was recorded using a 0.1–70Hz analog
bandpass and digitized at 250Hz sampling rate from Ag/
AgCl electrodes at the Fz, Cz, Pz, T3, T4 positions of the
international 10-20 system, from both mastoids (Ml, Mr) and
from frontolateral positions at 1/3 of the distance between
Fz and the mastoids (Fl and Fr). The reference electrode was
placed at the tip of the nose. Impedances were kept below
5kO. The baseline was defined by the mean amplitude
within 50ms before stimulus onset. Electro-oculogram
readings from the upper nasal vs the lower temporal right
orbital rim provided the artifact rejection criterion (770mV
within ?500 to +1500ms relative to stimulus onset). EEG
responses were digitally filtered off-line (30Hz low-pass,
FIR). Sub-averages were calculated for standards and
deviants, in experiment 2 grouped also according to
discrimination performance criteria (Fig. 2), and compared
in a multifactorial design as detailed in the results section.
Amplitudes were calculated as mean values within 50ms
time windows centered around subcomponent peak laten-
cies in the grand average waveforms. To enhance the signal-
to-noise ratio for the analysis of the MMN subcomponent,
which reverses its polarity at recording sites inferior to the
supratemporal plane [1], re-referenced amplitudes were
calculated by subtracting the averaged values at the
mastoids from those at the non-mastoid leads (e.g., Fz-
(M1+M2)/2). Analyses for later subcomponents (N2b, P3),
which do not show corresponding phase reversal character-
istics, were performed without re-referencing.
Experiment 1:
which consisted of two blocks. In the first (tonal) block the
standard stimuli were C-major triad
C¼261Hz, 4+3 semitone intervals, i.e., minor over major
third) and the deviants were C-minor triad chords (C-Eb-G,
3+4 semitones, i.e., major over minor third). In the second
(atonal) block, C-major and C-minor triads were replaced
with structurally matched chords which, instead of major
and minor thirds, were built of tritone and perfect fourth
intervals, i.e., Bb-E-A (6+5 semitones, i.e., perfect fourth
over tritone) replacing standard C-major chords, and Bb-Eb-
A (5+6 semitones, i.e., tritone over perfect fourth) replacing
deviant C-minor triads. Thus, the upper and lower tones of
tonal as well as atonal triads remained unchanged between
standards and deviants within a block, whereas a change
from E to Ebin the middle triad tone marked the occurrence
of a deviant sound in both blocks. In contrast to the
ubiquitous major/minor triads, the atonal chords do not
have a particular harmonic function in traditional Western
tonal music and do rarely occur in this context. Although
they may be employed more frequently in the extended
tonality of 20th century classic and modern jazz music, they
Eight subjects participated in experiment 1,
chords (C-E-G,
remain highly unusual to the traditionally trained musician.
Tonal and atonal blocks were presented in randomized
order. Subjects were instructed to listen attentively; no
specific task was to be performed.
Experiment 2:
2 where the chord root notes were varied in pseudo-random
order between C#and D’ on a well-tempered chromatic
scale (C¼261Hz). Accordingly, tonal chord sequences
consisted of major triads as standards and minor triads as
deviants, and atonal chord sequences consisted of tritone/
fourth triads as standards and fourth/tritone triads as
deviants. In addition, the triad sequences were arranged
such that no frequency memory traces from tone repetitions
between consecutive triad chords could emerge. Thus, there
was no simple frequency mismatch as in experiment 1, but a
complex frequency ratio difference between deviants and
standards. Subjects were instructed to discriminate stan-
dards and deviants in a forced-choice finger tapping task
within the 1s silent interval following the stimulus.
Twenty subjects participated in experiment
RESULTS
Experiment 1:
difference (deviant minus standard) waveforms at Fz for the
tonal block and the atonal block. A typical MMN was
elicited in both blocks, i.e., irrespective of tonality. A
significant MMN phase reversal was found between Fz
and mastoids in the 150–200ms poststimulus time window,
with negative amplitudes at Fz (p¼0.002 in block 1, p¼0.003
in block 2) and positive amplitudes at the mastoids
(po0.001 at M1 and p¼0.001 at M2 in block 1, p¼0.002 at
M1 and p¼0.021 at M2 in block 2). A three-way ANOVA
with the factors familiarity (familiar tonal chords in block 1
vs unfamiliar atonal chords in block 2), deviance (deviants
vs standards), and distribution (recording electrode posi-
tions) showed a significant deviance main effect (F(1,7)¼
109.343, po0.001) and a trend for a distribution?deviance
interaction (F(7,1)¼139.8, p¼0.065), but did not reveal any
familiarity main effect or interaction. A separate ANOVA
including only frontolateral leads to test for a possible
laterality effect did not show any significant right/left
difference. Thus, a typically distributed MMN was found
which did not differ between familiar and unfamiliar sound
sequences, and did not show any significant left-right
difference.
The difference curves in Fig. 1a,b display an N2b-like
negativity (peaking within the 250–300ms window after the
MMN) only in the tonal block 1, with no corresponding
negativity in the atonal block 2. A two-way left-right?
deviance ANOVA (including only the frontolateral record-
ing sites with the expectedly highest N2b amplitude),
revealed a nearly significant deviance effect in this time
window for block 1 (F(1,7)¼5.431, p¼0.053), but no right/
left main effect or interaction.
Following the N2b in block 1, and the MMN in block 2, a
positive deflection peaking within 350–400ms poststimulus
is visible in the difference curves of both blocks, corre-
sponding to a P3a component. A three-way familiarity?
deviance?distribution ANOVA revealed only a trend for a
deviance effect (F(1,7)¼5.223, p¼0.056); however, a signifi-
cant familiarity? deviance interaction was found (F(1,7)¼
8.598, p¼0.022). Therefore, two-way deviance?distribution
Figure 1a,b depicts standard, deviant, and
2472
Vol 15 No 16 15 November 2004
NEUROREPORT
G.NEULOHAND G.CURIO
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Page 3

ANOVAs were calculated for each block separately, which
showed a significant deviance main effect restricted to the
tonal block 1 (F(1,7)¼6.619, p¼0.037), whereas the corre-
sponding positivity did not reach significance in block 2.
Taken together, deviant triads elicited a distinct N2b/P3a
complex only in the familiar chord sequence.
Experiment 2:
standard) waveforms at Fz for both the tonal and atonal
block. A typical MMN with reversed polarity between Fz
and the mastoids in the corresponding 175–225ms time
window was elicited in both blocks, with a slightly earlier
peak latencyin block1.
deviance? distribution ANOVA (calculated for re-refer-
enced values in the 175–225ms window) showed a
significant deviance main effect (F(1,19)¼6.457, p¼0.02)
and a deviance? distribution interaction (F(7,13)¼8.358,
p¼0.001), but no effect involving familiarity, despite the
seemingly larger MMN amplitude in block 2. Correspond-
ing to experiment 1, a separate ANOVA including only
frontolateral leads showed no significant laterality effect.
Thus, a typically distributed MMN was elicited in both
blocks which did not differ significantly for familiar and
unfamiliar chords.
In contrast to the MMN, a typical large, parietally
maximal P3 positivity occurred only in block 1, where the
majority of subjects scored high in the detection paradigm
(Fig. 1d), whereas only a slight positive deflection was
elicited in block 2. Correspondingly, a three-way ANOVA
for the time-window 500–1000ms showed a significant
familiarity? deviance interaction (F(1,19)¼4.498, p¼0.047)
along with a deviance? distribution interaction (F(7,13)¼
6.74, p¼0.002).
Figure 1c depicts difference (deviant minus
Athree-way familiarity?
A scatter diagram of Grier’s non-parametric unbiased
sensitivity index [17] for the forced-choice detection para-
digm in both blocks (Fig. 2a) shows that majorities of
subjects scored high for tonal (continuous ellipse) and low
(dotted ellipse) for atonal sounds. Figure 2b displays the re-
referenced difference waveform at Fz from block 2
(unfamiliar chords) for all subjects, and superimposed
subaverages for the good and poor performers in the
discrimination task of this block, revealing a similar MMN
in all groups. The corresponding differences waveforms of
block 2 at Pz are superimposed in Fig. 2c, expectedly
displaying a P3 component only for the small group of good
performers. A separate two-way deviance? distribution
ANOVA was performed with only the correctly detected
standards and the deviants falsely classified as standards
(missed deviants), which still showed a deviance main effect
(F(1,19)¼6.249, p¼0.022), as well as a deviance?distribution
interaction (F(7,13)¼2.87, p¼0.048). Even with the analysis
restricted to the misses in the majority subgroup of 15
subjects with a close-to-chance performance in the detection
task in order to eliminate negative response bias (truly
undetectable misses, Fig. 2d), there was still both, a
significant deviance main effect (F(1,14)¼12.919, p¼0.003)
and a deviance? distribution interaction (F(7,8)¼3.854,
p¼0.039). Thus, a typical MMN was elicited by missed,
and even undetectable, deviants in the detection task.
DISCUSSION
The present study found a typical sensory mismatch
response in sequences of tonal and atonal music chords: In
experiment 1 the MMN reflects both an absolute frequency
mismatch and a frequency ratio mismatch, whereas in
MMN
MMN
MMN
N2b
P3a
P3a
P3b
−6
−4
−2
0
2
−2
0
2
4
−6
−4
−2
−2
0
2
4
0
2
4
0 200 400 ms0 200 400 ms
00 200400 ms500 1000 ms
µV
µV
µV
µV
(a)
(c)
(b)
(d)
Fig.1. Grand-average standard (.......), deviant (^^^^^), and di¡erence (deviant-standard F) waveforms of the tonal (a) and atonal (b) conditions of
experiment1.While the MMN is of comparable amplitude and latency in both blocks, there is a distinct N2b only in the tonal condition, followed by a
larger P3a in thisblock.Di¡erencewaveforms of the tonal (F) and atonal (.......) blockof experiment 2 at Fz (c) and Pz (d) show a seeminglylarger, and
earlier MMN componentin the atonal condition, but a signi¢cant P3b componentonlyin the tonalcondition.
Vol 15 No 16 15 November 2004
2473
DISCRIMINATIONOFFAMILIARMUSIC SOUNDS
NEUROREPORT
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Page 4

experiment 2 only a frequency ratio mismatch was available
to generate the MMN response, notably with an amplitude
lower than in experiment 1. In contrast to the hypothesis put
forward in the introduction, the MMN amplitude was not
enhanced for triads drawn from familiar musical categories.
Only the subsequent cognitive components indicated a
facilitated novelty response for familiar sounds as opposed
to structurally matched, but unfamiliar sounds; this was
reflected also by the behavioral discrimination performance.
Conspicuously, an undiminished sensory mismatch re-
sponse was generated also by missed deviant atonal chords,
even in subjects who were in general unable to consciously
discriminate these highly unfamiliar triads.
In previous studies it was consistently shown that the
MMN depends on musical proficiency [7,8,15,16], and intact
musical competence [18]. The present results are not in
conflict with these findings, which seem to reflect a general
improvement of auditory sensory discrimination in musi-
cians and musically competent subjects, and this effect is not
restricted to the domain of tonal music [7,8,15,16,19]. Here,
the comparison was not between musicians and non-
musicians, but between specific familiar tonal and unfami-
liar atonal sounds in musically trained subjects (although
subjects could be grouped according to their discriminative
performance as depicted in Fig. 2).
The fact that no MMN amplitude facilitation was found
for highly familiar music sounds is, however, in contrast to
previous results in the language domain. Phoneme dis-
crimination and categorization as reflected by the MMN
depend on long-term familiarity with speech sounds, i.e.,
they are facilitated through language-specific phonetic
representations which develop during the first few months
of life, or during acquisition of a foreign language (for
review see [11]). The exposure to speech very early in life
may contribute to this language-specificity of sensory sound
processing [3,20], as compared to the discrimination of
music sounds in the present group of subjects who had their
musical training at a later age.
The lack of correlation between discrimination efficiency
and novelty-detection related AEP components on the one
hand, and the MMN on the other hand in experiment 2
(Fig. 2), where an identical MMN was elicited in all
performance groups, and even by undetectable deviant
chords, is in contrast to previous studies where a positive
correlation of MMN amplitude and detection performance
has been found [6,21,22]. However, the improved detection
performance in these studies was typically related to an
improved sensory discrimination of small physical differ-
ences, e.g., with hearing recovery and increased MMN in
cochlear implant users [22], rather than due to an improved
analytical discrimination of more familiar complex sounds,
as it occurred with tonal chords in the present study.
From the perspective of music processing, it is a
remarkable finding that an undiminished MMN was
elicited for unfamiliar triads even by deviances which were
undetectable. Such consciously unnoticed neuronal infor-
mation about changes in chord category (e.g., in harmonic
modulation) might be effective in subconscious musical
influences on mood modulation, both in tonal and non-tonal
music. The mechanism for gating to conscious perception
can be studied when operating at the output pathways of
the auditory cortex mismatch detector, e.g., with the
induction of cognitive components related to rule-based
processing of Western music [23].
The apparent lack of a familiarity effect on sensory chord
processing in the present study might be due to a weak,
rather than an absent amplitude facilitation for tonal
triads, hidden by a low signal-to-noise ratio. However,
this would still be in contrast with findings in sensory
speech processing where the facilitating effect of phoneme
1.0
1.0
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.6
0.5
0.5
Tonal condition
Atonal condition
−2
2
0
−2
−2
2
2
0
0
0
0
200
400 ms
0 500200 400 ms 1000 ms
4
6
µV
µV
µV
(a)
(d)
(b)(c)
Fig. 2. Grier’sA ’non-parametricdetectionscores[17]forbothtonalandatonalconditionsareplottedin(a),majoritiesofsubjectsscoringhighfor tonal
(continuous ellipse), and low (dotted ellipse) for atonal sounds. Re-referenced grand-average di¡erence curves at Fz from the atonal block (b) with a
similar MMN for all subjects (F), good (^^^^), andpoor (.......) performersin the detection task.Di¡erence curves at Pz for the samegroups (c),with
a signi¢cant P3b de£ection only for thegroupwith a high detection score.Re-referenced di¡erence curves at Fz only for missed deviants and correctly
detectedstandardsin the atonalblock (d), with anundiminishedMMN for all subjects (F) andparticularly for thepoor performers (.......) in the detec-
tion task.
2474
Vol 15 No 16 15 November 2004
NEUROREPORT
G.NEULOHAND G.CURIO
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Page 5

familiarity typically leads to a clearly enhanced MMN
amplitude [11]. Likewise, an amplitude ceiling effect is
unlikely, because the MMN peak amplitudes of around 2mV
in the present data were not exceedingly high. It is also
possible that some roughness or dissonance in the atonal
chords might have enhanced the MMN amplitude in the
unfamiliar chord sequences [24,25], which would provide
an explanation for the (insignificantly) higher amplitude of
the MMN in the atonal block of experiment 2. However, a
dominant roughness effect, which might have even para-
doxically masked some amplitude facilitation by sound
familiarity, is unlikely since it should lead to an enhanced
primary cortical response (N1), which was not observed
(Fig. 1); likewise, it should be paralleled by facilitation of the
subsequent cognitive components and improved discrimi-
nation scores, which is contrary to what was found in the
present study.
CONCLUSION
The results of this study do not show a facilitation of
preattentive sensory discrimination of familiar music chords
as reflected by the MMN; only later AEP components
related to novelty-detection were facilitated, corresponding
to an improved behavioral discrimination performance.
This contrasts with the previously described facilitation of
the sensory discrimination response for familiar speech
sounds, and emphasizes a leading role for speech with
regard to long-term plasticity in preattentive sensory
processing. The undiminished MMN even for undetectable
atonal chord deviances indicates that sensory discrimination
is fully effective below the level of conscious analytical
categorization of complex music sounds, which was shown
to reflect long-term familiarity in the present study.
REFERENCES
1. Naatanen
attention effect on evoked potential reinterpreted. Acta Psychol 1978;
42:313–329.
2. Alho K and Sinervo N. Preattentive processing of complex sounds in the
human brain. Neurosci Lett 1997; 233:33–36.
3. Naatanen R, Lehtokoski A, Lennes M, Cheour M, Huotilainen M, Iivonen
A et al. Language-specific phoneme representations revealed by electric
and magnetic brain responses. Nature 1997; 385:432–434.
4. Tervaniemi M, Schroger E, Saher M and Naatanen R. Effects of spectral
complexity and sound duration on automatic complex-sound pitch
processing in humans – a mismatch negativity study. Neurosci Lett 2000;
290:66–70.
5. Tervaniemi M and Huotilainen M. The promises of change-related brain
potentials in cognitive neuroscience of music. Ann NY Acad Sci 2003;
999:29–39.
R, GaillardAW and MantysaloS. Early selective-
6. Paavilainen P, Simola J, Jaramillo M, Naatanen R and Winkler I.
Preattentive extraction of abstract feature conjunctions from auditory
stimulation as reflectedbythe
Psychophysiology 2001; 38:359–365.
7. Fujioka T, Trainor LJ, Ross B, Kakigi R and Pantev C. Musical training
enhances automatic encoding of melodic contour and interval structure.
J Cogn Neurosci 2004; 16:1010–1021.
8. van Zuijen TL, Sussman E, Winkler I, Naatanen R and Tervaniemi M.
Grouping of sequential sounds – an event-related potential study
comparingmusicians andnonmusicians.
16:331–338.
9. Naatanen R. Attention and Brain Function. Hillsdale: Lawrence Erlbaum;
1992.
10. Friedman D, Cycowicz YM and Gaeta H. The novelty P3: an event-related
brain potential (ERP) sign of the brain’s evaluation of novelty. Neurosci
Biobehav Rev 2001; 25:355–373.
11. Naatanen R. The perception of speech sounds by the human brain as
reflected by the mismatch negativity (MMN) and its magnetic equivalent
(MMNm). Psychophysiology 2001; 38:1–21.
12. Pantev C, Ross B, Fujioka T, Trainor LJ, Schulte M and Schulze M. Music
and learning-induced cortical plasticity. Ann NY Acad Sci 2003; 999:
438–450.
13. Besson M and Schon D. Comparison between language and music. Ann
NY Acad Sci 2001; 930:232–258.
14. Munte TF, Nager W, Beiss T, Schroeder C and Altenmuller E.
Specialization of the specialized: electrophysiological investigations in
professional musicians. Ann NY Acad Sci 2003; 999:131–139.
15. Nager W, Kohlmetz C, Altenmuller E, Rodriguez-Fornells A and Munte
TF. The fate of sounds in conductors’ brains: an ERP study. Brain Res Cogn
Brain Res 2003; 17:83–93.
16. Hashimoto T, Hirata Y and Kuriki S. Auditory cortex responds in 100 ms
to incongruity of melody. Neuroreport 2000; 11:2799–2801.
17. Grier JB. Nonparametric indexes for sensitivity and bias: computing
formulas. Psychol Bull 1971; 75:424–429.
18. Kohlmetz C, Altenmuller E, Schuppert M, Wieringa BM and Munte TF.
Deficit in automatic sound-change detection may underlie some music
perception deficits after acute hemispheric stroke. Neuropsychologia 2001;
39:1121–1124.
19. Tervaniemi M, Ilvonen T, Karma K, Alho K and Naatanen R. The musical
brain: brain waves reveal the neurophysiological basis of musicality in
human subjects. Neurosci Lett 1997; 226:1–4.
20. Cheour M, Martynova O, Naatanen R, Erkkola R, Sillanpaa M, Kero P
et al. Speech sounds learned by sleeping newborns. Nature 2002; 415:
599–600.
21. Novitski N, Tervaniemi M, Huotilainen M and Naatanen R. Frequency
discriminationat differentfrequency
electrophysiological and behavioral measures. Brain Res Cogn Brain Res
2004; 20:26–36.
22. Lonka E, Kujala T, Lehtokoski A, Johansson R, Rimmanen S, Alho K and
Naatanen R. Mismatch negativity brain response as an index of speech
perception recovery in cochlear-implant recipients. Audiol Neurootol 2004;
9:160–162.
23. Koelsch S, Schroger E and Tervaniemi M. Superior pre-attentive auditory
processing in musicians. Neuroreport 1999; 10:1309–1313.
24. De Baene W, Vandierendonck A, Leman M, Widmann A and Tervaniemi
M. Exploration of roughness by means of the mismatch negativity
paradigm. Ann NY Acad Sci 2003; 999:170–172.
25. Itoh K, Suwazono S and Nakada T. Cortical processing of musical
consonance: an evoked potential study. Neuroreport 2003; 14:2303–2306.
mismatch negativity(MMN).
J Cogn Neurosci
2004;
levelsas indexedby
Acknowledgements:The authors thank Professor Dr. A.Friederici, headof the Max Planck Institute ofCognitive Neurosciencein
Leipzig/Germany, andhergroup, for supportduringdata acquisition andevaluation, andfor numeroushelpfuldiscussions.
Vol 15 No 16 15 November 2004
2475
DISCRIMINATIONOFFAMILIARMUSIC SOUNDS
NEUROREPORT
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.