The mismatch negativity as an index of temporal processing in audition

Clinical Neurophysiology 112 (2001) 1712±1719

www.elsevier.com/locate/clinph

The mismatch negativity as an index of temporal processing in audition T. Kujala a,*, J. Kallio a, M. Tervaniemi a, R. NaÈaÈtaÈnen a,b a

Cognitive Brain Research Unit, Department of Psychology, P.O. Box 13, FIN-00014, University of Helsinki, Helsinki, Finland b BioMag Laboratory, P.O. Box 340, 00029 HUS, Helsinki, Finland Accepted 15 June 2001

Abstract Objectives: The relation of the mismatch negativity (MMN) elicitation with behavioral stimulus discrimination as well as the replicability of the MMN was evaluated for intervals between paired tones. Methods: The MMN, obtained in a passive oddball paradigm in two sessions separated by 4±21 days and behavioral responses (button presses to target stimuli) in a separate session were recorded from 10 adult healthy subjects. The standard stimulus …P ˆ 0:79† was a tone pair separated by a 120 ms silent inter-stimulus interval (ISI) and the deviant stimuli were tone pairs with an ISI of 100, 60, and 20 ms (P ˆ 0:07 for each). Results: The 20 and 60 ms ISI deviant tone pairs elicited a signi®cant MMN during both recording sessions and they were also behaviorally discriminated, whereas neither did the 100 ms ISI deviant pair elicit signi®cant MMN nor was it behaviorally discriminated. Furthermore, there was a signi®cant correlation between the MMN and reaction times to the 20 and 60 ms ISI deviant pairs. The 20 ms ISI deviant stimulus elicited highly replicable MMNs …r ˆ 0:75†, whereas the less well discriminated 60 ms ISI deviant pair did not …r ˆ 0:60†. Conclusions: The MMN re¯ects discrimination accuracy of temporal sound intervals. Furthermore, when the physical difference between the standard and deviant tone pair in the temporal domain is large, it is elicited with high reliability. q 2001 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Auditory event-related potentials; Auditory sensory memory; Mismatch negativity; Temporal processing

1. Introduction The temporal aspects of sounds carry important information for various auditory functions such as speech and music perception. It has been proposed that the perception of speech signal is primarily based on temporal cues (Rosen, 1992; Shannon et al., 1995). Studies evaluating speech± sound intelligibility on the basis of temporal vs. spectral cues have indicated that when the temporal cues are preserved, almost all spectral cues can be removed from the speech signal and, yet, the speech content can be recognized (Rosen, 1992; Shannon et al., 1995). Moreover, single-channel cochlear-implant users are able to understand speech although their hearing aid principally transmits temporal sound information (Tyler, 1993). A de®ciency in discriminating temporal sound features could cause severe disturbances in the linguistic domain. Numerous studies showed that de®ciency in speech perception, caused by developmental factors or brain damage at some stage of * Corresponding author. Tel.: 1358-9-19123760; fax: 19122924. E-mail address: teija.m.kujala@helsinki.® (T. Kujala).

1358-9-

life, is associated with a dif®culty in discriminating temporal sound features (see, for example, Tallal and Piercy, 1973; Tallal, 1980; Robin et al., 1990; Farmer and Klein, 1995; Merzenich et al., 1996; Tallal et al., 1996; Kujala et al., 2000). The present work addresses the applicability of the mismatch negativity (MMN; NaÈaÈtaÈnen et al., 1978) in measuring the temporal-discrimination accuracy of the central auditory system. The MMN re¯ects sensorymemory functioning and can be used to study discrimination accuracy in the auditory system (NaÈaÈtaÈnen, 1992, 1995; Tiitinen et al., 1994; Amenedo and Escera, 2000). The MMN is obtained by presenting a repetitive auditory stimulus (`standard') which is occasionally replaced with a deviant stimulus differing from the standard stimulus in some respect. The MMN elicitation is based on a memory trace formed by the auditory system for the standard stimulus to which each incoming stimulus is compared, with a deviant stimulus eliciting an MMN (NaÈaÈtaÈnen, 1990). The MMN has been obtained for various auditory-stimulus differences, such as those in pitch, intensity, and duration (NaÈaÈtaÈnen, 1992). It has also been obtained for more complex stimulus features, such as phonetic (Aaltonen et

1388-2457/01/$ - see front matter q 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S13 88-2457(01)0062 5-3

CLINPH 2000818

T. Kujala et al. / Clinical Neurophysiology 112 (2001) 1712±1719

al., 1987; Sams et al., 1990; NaÈaÈtaÈnen et al., 1997) or soundpattern changes or even changes in abstract features of auditory stimuli (NaÈaÈtaÈnen et al., 1993; Tervaniemi et al., 1994; Paavilainen et al., 1998, 1999; for a review, see NaÈaÈtaÈnen et al., 2001). The MMN re¯ects sound-discrimination accuracy since it is elicited even by just-perceptible differences between sounds (Kraus et al., 1995; Tremblay et al., 1998). The MMN can be obtained without the subject's attention; therefore, it is a suitable tool for studying perceptual functions in individuals who cannot concentrate on task performance (NaÈaÈtaÈnen, 1995). The MMN has already been successfully used in studying auditory perception in unconscious patients (Kane et al., 1993) and in those having dif®culties in understanding or following instructions, like patients with receptive aphasia (Aaltonen et al., 1993; Alain et al., 1998; Ilvonen et al., 2001) or children and infants (Alho et al., 1990; Kraus et al., 1996; Cheour et al., 1998). Currently, there is an urgent need to improve the reliability of the MMN measurement paradigms, since the main problem in using the MMN at the single-subject level is its fairly poor signal-to-noise ratio. This problem is of course pronounced when the discrimination of ®ne auditory differences is studied. The reliability of the MMN as an index of discrimination has been studied for simple sound features, such as pitch, intensity, and duration (Pekkonen et al., 1995; Lang et al., 1995; Escera and Grau, 1996; Frodl-Bauch et al., 1997; Joutsiniemi et al., 1998; SchroÈger, 1998; Tervaniemi et al., 1999; Kathmann et al., 1999). In these studies, MMN test±retest reliability varied between 0.3 and 0.78. 2. Methods 2.1. Subjects Ten healthy subjects (5 males) of 21±27 years (mean 24 years) were included in the present study. Nine of them were right-handed. 2.2. Stimuli and procedure The stimuli were tone pairs with different silent interstimulus intervals (ISIs). The tones were 500 Hz in frequency and 30 ms in duration (including 5 ms rise and fall times). Their intensity was 50 dB above the individual hearing level. In the standard-tone pair, the offset-to-onset ISI between the two tones was 120 ms, and in the deviant pairs 20, 60, and 100 ms. The probability of the standard pair was 0.79, and that of each deviant pair 0.07. The stimuli were presented in a pseudo-randomized order (each deviant pair was preceded by at least 3 standard pairs) in 7 blocks of 1260 stimulus pairs. The stimulus-onset asynchrony (SOA) of the stimulus pairs was 500 ms. In addition, subjects were presented with two blocks of stimulus pairs in which all deviant stimulus pairs were equiprobably presented (0.33). In this way, we wished to obtain

1713

responses that approximate the obligatory responses elicited by each deviant stimulus pair. 1 The stimuli were presented via headphones to the subject who was sitting in a reclining chair in a sound-proof and electrically shielded chamber. During the electrophysiological recordings, the subject was watching a self-selected silent movie. In order to determine the replicability of the MMN, the data were collected in two sessions separated by 4±21 days. The relationship between the MMN and the behavioral discrimination was studied in a separate session administered always after the second electrophysiological recording session in order to avoid carry-over effects of attention. The subject was presented with a stimulus sequence containing 700 stimulus pairs with the same parameters as in the electrophysiological recording blocks. The subject's task was to press a response key with the thumb of the preferred hand as quickly as possible when she/he heard a deviant stimulus. 2.3. Electrophysiological data recording and analysis The nose-referenced electroencephalogram (EEG) (0.1± 100 Hz, sampling rate 250 Hz) was recorded with a 32channel electrode cap covering the frontal, central, temporal, and parietal scalp areas. Event-related potentials (ERPs) were obtained by averaging, separately for each stimulus-pair type, EEG epochs digitally ®ltered with a bandpass of 2±20 Hz (24 dB/octave). The analysis epoch began 50 ms before and terminated 500 ms after the onset of the ®rst tone of the pair. Voltage changes caused by horizontal eye movements were monitored with electrodes attached to the outer canthi and those caused by vertical eye movements with the 3 forehead electrodes of the electrode cap. Epochs contaminated by eye movements or other extra-cerebral artifacts producing voltage variation exceeding 100 mV at any electrode and responses to the 4 ®rst stimulus pairs of each block were omitted from averaging. 2.4. The ERP analysis The 0 mV baseline began 50 ms before the sound-pair onset and terminated 50 ms after it (note that all stimuli, and therefore responses, were similar up to this point of time). The data were re-referenced to linked mastoids in order to maximize the MMN amplitude at the fronto-central electrodes. The MMN-amplitude comparisons were performed by using responses recorded with an electrode located at Fz in the electrode cap. The signi®cance of the MMN was studied by comparing the ERP elicited by each deviant pair with that elicited by the same pair when presented equiprobably with the other deviant pairs. This comparison was 1 The advantage of this approach is to save recording time. A better estimate of the obligatory responses elicited by the pairs might be obtained by using each deviant pair as the standard in separate blocks and subtracting them from the corresponding pairs when presented as deviants. However, such an approach would extensively prolong the recording time.

1714

T. Kujala et al. / Clinical Neurophysiology 112 (2001) 1712±1719

performed for 12 consecutive time windows of 25 ms starting at stimulus-change onset (i.e. the second-tone onset) and continuing for 300 ms. The peak latencies and amplitudes of MMNs elicited by different deviant pairs were measured from the difference waves obtained by subtracting from the ERP elicited by each deviant stimulus pair the ERP to this stimulus pair in the equiprobable stimulus-pair sequence. The peak amplitudes were measured as the mean of a 25 ms window centered around the individual ERP peaks, determined at 100±175 ms from the secondsound onset of the deviant pair. The replicability of the responses was evaluated by calculating Pearson's test±retest correlations between the ®rst and second sessions. In order to evaluate the possible ISI-related effects on the amplitude and latency of the N1 elicited by standard stimulus pairs, as well as by stimulus pairs in the equiprobable stimulus-pair sequence, the N1 peak amplitude and latency, obtained at 50±200 ms from the second-sound onset, and their replicability were studied similarly as for the MMN. 2.5. The analysis of the behavioral responses Button presses occurring 100±1000 ms after stimuluspair change onset were classi®ed as hits. The hit rate (HR) (%) was calculated as the number of the hits divided by the number of targets. The average hit reaction time (RT) was calculated for each subject. Since all 3 types of deviant stimulus pairs were presented in the same block and the subjects had to press the same button to all deviant pairs, the false-alarm rate was not analyzed. 2.6. The correlation between ERPs and behavioral responses Spearman's rank±order correlation coef®cients were

calculated to evaluate the relationship between the N1 and MMN amplitudes and latencies obtained in the second session and behavioral responses (HR and RT). Since only 20 and 60 ms ISI tone pairs elicited the MMN as well as were behaviorally detected, combined data from these conditions were included in the analysis. 3. Results 3.1. Electrophysiological results 3.1.1. ERP to the auditory stimuli After artifact rejection, there were on the average 529± 536 sweeps in the single-subject averages for the different deviant pairs when presented among the standard pairs (standard error of mean, SEM, 19±24). The standard tone pair as well as the 20, 60, and 100 ms ISI equiprobable tone pairs elicited positive and negative responses that were largely similar in amplitude (Fig. 1). However, as Fig. 1 shows, the onset of the negative de¯ection was temporally displaced for the different types of tone pairs (in relation to the onset of the ®rst stimulus). The N1 mean amplitudes for these tone pairs were unaffected by the ISIs in both sessions (F…3; 36† ˆ 0:65±1:06, P . 0:3±0:5). Furthermore, the N1 mean amplitudes were stable across the sessions. Pearson's test±retest correlation for the N1 amplitude was 0.66±0.93 (F…1; 8† ˆ 6:06±59:93, P , 0:05±0:001). The N1 peaked at 101±133 ms from the onset of the second tone. The N1 mean latency was affected by the ISI in the ®rst (F…3; 36† ˆ 5:49, P , 0:01), but not in the second session (F…3; 36† ˆ 1:29, P . 0:2). Subsequent analyses revealed that the N1 mean latency was signi®cantly shorter for the 20 ms ISI tone pair than for the 100 and 120 ms ISI tone pairs (F…1; 8† ˆ 24:87±33:98, P , 0:001). The

Fig. 1. Grand-mean ERPs, at the Fz scalp location, elicited by the different tone pairs used in the present study. The responses elicited by 20, 60, and 100 ms ISI tone pairs were obtained in a condition in which these stimuli were equiprobably presented (P ˆ 0:33 for each pair). The response elicited by the 120 ms ISI tone pair was obtained in the main experiment in which this tone pair was the standard stimulus …P ˆ 0:79†.

T. Kujala et al. / Clinical Neurophysiology 112 (2001) 1712±1719

1715

and 3). There was no signi®cant difference in the MMN mean amplitude between the ®rst and second sessions (Pearson's test±retest correlation: 0.75 (F…1; 8† ˆ 9:98, P , 0:02; Fig. 4) or in the MMN mean latency between the ®rst and second sessions (Pearson's test±retest correlation: 0.42 (F…1; 8† ˆ 1:76, P . 0:2).

Fig. 2. Grand-mean ERPs (averages from the ®rst and second recording sessions), at the Fz scalp location, to the 120 ms ISI tone pair which was the standard pair in the main experiment and to the 20, 60, and 100 ms ISI tone pairs when presented as deviants (P ˆ 0:07 for each pair) and when presented equiprobably (P ˆ 0:33 for each pair) in a separate control condition.

N1 mean latency was stable across the sessions for the 20, 100, and 120 ms ISI pairs (Pearson's test±retest correlations 0.76±0.90, F…1; 8† ˆ 10:96±36:17, P , 0:02±0:001), but not for the 60 ms ISI pair (Pearson's test±retest correlation 0.14, F…1; 8† ˆ 0:16, P . 0:60). 3.1.2. The MMN to the deviant pair with the 20 ms ISI The response to the 20 ms ISI deviant tone pair was signi®cantly negatively displaced at 100±175 ms in the ®rst and second sessions (F…1; 9† ˆ 17:5±116:7, P , 0:01; time windows calculated from second-tone onset; Figs. 2

Fig. 3. Grand-mean difference waves, at the Fz scalp location, separately for the ®rst and second sessions, for the 20, 60, and 100 ms ISI tone pairs obtained by subtracting the ERP elicited by a stimulus pair in the equiprobable condition from that elicited by the same pair when it was a deviant in the main experiment.

1716

T. Kujala et al. / Clinical Neurophysiology 112 (2001) 1712±1719

Fig. 4. Difference waves, at the Fz location, of each individual subject obtained similarly as in Fig. 3 for the ®rst and second sessions.

3.1.3. The MMN to the deviant pair with the 60 ms ISI The response to the 60 ms ISI deviant tone pair was signi®cantly negatively displaced at 100±150 ms in the ®rst and second sessions (F…1; 9† ˆ 7:2±15, P , 0:05; Figs. 2 and 3). There was a signi®cant MMN-amplitude

difference between the two sessions at 100±125 ms (F…1; 9† ˆ 8:3, P , 0:02). Pearson's test±retest correlation between the MMN peak amplitudes measured in the ®rst and second sessions was 0.60 (F…1; 8† ˆ 4:47, P , 0:06). There was no signi®cant MMN latency difference between

T. Kujala et al. / Clinical Neurophysiology 112 (2001) 1712±1719

the ®rst and second sessions (Pearson's test±retest correlation: 0.32 (F…1; 8† ˆ 0:93, P . 0:3; Fig. 4). 3.1.4. The MMN to the deviant pair with the 100 ms ISI This deviant pair, deviating least from the standard tone pair, elicited no signi®cant MMN in the ®rst or second recording session (Figs. 2 and 3). There was no signi®cant MMN-amplitude difference between the responses obtained in the ®rst and second sessions. 3.1.5. Comparison of the MMNs to different deviant pairs The different deviant pairs elicited signi®cantly different MMN amplitudes, whereas no signi®cant peak-latency differences (measured from second-tone onset) were obtained. The 20 ms ISI tone pair elicited a signi®cantly larger MMN than did the 60 ms ISI tone pair in the ®rst session at 100±175 ms (F…1; 9† ˆ 7:0±34:6, P , 0:03). In the second session, the MMN amplitude for the 20 ms ISI differed from that for the 60 ms ISI nearly signi®cantly at 100±125 and 150±175 ms (F…1; 9† ˆ 4:8±5:0, P , 0:06) and signi®cantly at 125±150 ms (F…1; 9† ˆ 11:3, P , 0:01). The 60 ms ISI tone pair elicited a signi®cantly larger MMN than did the 100 ms ISI tone pair in the ®rst session at 125±150 ms (F…1; 9† ˆ 8:2, P , 0:02) and nearly signi®cantly larger one at 150±175 ms (F…1; 9† ˆ 4:7, P , 0:06). In the second session, the MMN amplitudes elicited by the 60 and 100 ms ISI tone pairs signi®cantly differed from each other at 100±150 ms (F…1; 9† ˆ 26±7:2, P , 0:03). 3.1.6. Behavioral discrimination The subjects detected on the average 92% of the 20 ms deviant-pair targets, and the RT was 363 ms (Fig. 5). For the 60 ms ISI deviant-pair targets, the HR was 65% and the RT 490 ms. For the 100 ms ISI deviant-pair targets, the HR was 1%. This deviant pair was indiscriminable for all subjects.

1717

One subject had 3 and another 6 correctly detected targets, whereas the rest of them had 0 detections. The 20 ms ISI deviant tone-pair targets were signi®cantly more accurately detected than the 60 ms (F…1; 9† ˆ 16:5, P , 0:003) and 100 ms (F…1; 9† ˆ 759, P , 0:001) targets. There was also a signi®cant difference in the detection accuracy between the 60 and 100 ms ISI deviant-pair targets (F…1; 9† ˆ 87, P , 0:001). The 20 ms ISI deviant-pair targets were detected signi®cantly faster than the 60 ms ISI deviant-pair targets (F…1; 9† ˆ 34:2, P , 0:002). Spearman's rank±order correlation revealed that larger MMN amplitudes were associated with shorter RTs, the correlation being 20.52 (t…18† ˆ 22:591, P , 0:02). No signi®cant correlations were found for the MMN amplitude and HR, the MMN latency and HR, or the MMN latency and RT (the correlation coef®cients: 20.05 to 0.32 (t…18† ˆ 20:224 to 1:445, P . 0:2±0:8). No signi®cant correlation was found between the N1 amplitude or latency and HR or RT to target stimuli (the correlation coef®cients: 20.23 to 0.36 (t…18† ˆ 21:007 to 1:624, P . 0:12±0:32).

4. Discussion Since there is an urgent demand for methods to objectively evaluate temporal-discrimination abilities applicable at the individual level, the present study aimed at determining how reliably the MMN could re¯ect the discrimination accuracy of sound intervals. In both measurement sessions, a signi®cant MMN was elicited by the 20 and 60 ms ISI deviant tone pairs, whereas the 100 ms ISI deviant tone pair elicited no MMN in either of the sessions. In addition, the MMN amplitude was larger for the larger (20 ms ISI) than smaller (60 ms ISI) standard-deviant separation, which is consistent with previous ®ndings (e.g. Tiitinen et al., 1994;

Fig. 5. Behavioral target-detection responses for each deviant stimulus pair (with SEM).

1718

T. Kujala et al. / Clinical Neurophysiology 112 (2001) 1712±1719

Jaramillo et al., 2000) showing a clear relation between the MMN amplitude and deviant-standard stimulus difference. The MMN elicitation correlated with the behavioral discrimination results. The deviant tone pairs of 20 and 60 ms ISIs which elicited signi®cant MMNs in both sessions were also behaviorally detected (HR 92 and 65%, respectively). Furthermore, no signi®cant MMN was elicited by the 100 ms ISI deviant tone pair and, consistently with this, the corresponding HR was only 1%. Moreover, the MMN amplitude and the RT were signi®cantly correlated. The larger the MMN was in amplitude, the faster the subjects responded to the respective deviant stimuli. In contrast, the N1 elicitation (by the sound pairs when presented equiprobably) did not correlate with the HRs or RTs for the 20 and 60 ms ISI deviant pairs. The present ®ndings are in good agreement with several previous reports suggesting that the MMN is elicited by discriminable stimulus differences (NaÈaÈtaÈnen et al., 1993; Kraus et al., 1995). According to Tremblay et al. (1998), the MMN can sometimes be seen even when the subject does not behaviorally discriminate the stimulus difference. Also in the single-subject data of the present study, there are indications of small MMNs in some subjects (Fig. 4) for the deviant pair with the 100 ms ISI although none of the subjects was able to discriminate this deviant pair behaviorally. A high test±retest replicability of the MMN was found in the present study for the deviant tone pair that differed the most from the standard one. A considerably smaller replicability was found for the deviant pair with the second largest interval deviation for which the signal-to-noise ratio was naturally much lower. The test±retest reliabilities of the MMNs elicited by the two deviant pairs differing most widely from the standard pair were comparable to the MMN test±retest reliabilities previously found for simple stimulus attributes, which varied between 0.3 and 0.78 (Lang et al., 1995; Pekkonen et al., 1995; Escera and Grau, 1996; Frodl-Bauch et al., 1997; Joutsiniemi et al., 1998; SchroÈger, 1998; Kathmann et al., 1999; Tervaniemi et al., 1999). Interestingly, the MMN amplitude was larger in the second than in the ®rst session for the deviant tone pair with the 60 ms ISI at 100±125 ms from stimulus-change onset, which might re¯ect a discrimination-learning effect. Previous studies have shown that active discrimination training is needed in order to produce learning-related plastic changes in the nervous system (Ahissar et al., 1992; Recanzone et al., 1992, 1993; NaÈaÈtaÈnen et al., 1993). It is possible that the subjects in the present study were not able to fully ignore the sound stimuli, which might explain the possible learning effect re¯ected in the MMN enhancement from the ®rst to the second session. In the present study, the MMN was obtained by including all deviant pairs in the same stimulus blocks which approach has the advantage of saving recording time and making a large amount of deviant-trial presentation possible (Ritter et al., 1995; Tervaniemi et al., 1999). It should be noted,

however, that the presentation of all deviant pairs in the same blocks might favor the grouping of 20 and 60 ms ISI pairs as deviants and the 100 and 120 ms ISI stimuli as standards. In a future study, it should be determined whether the 100 ms ISI deviant pairs would be better discriminated when presented alone as deviant pairs among 120 ms ISI standard pairs. The second tone of the pair (Fig. 2) always elicited an N1 which was not signi®cantly different in amplitude for the different tone pairs. In the ®rst measurement session, the N1 to the second tone peaked later for the 100 and 120 ms ISI than for the 20 ms ISI pair. However, these ®ndings were not replicated in the second ERP measurement, where there were no signi®cant differences in the N1 mean latencies for the different tone pairs. According to studies using magnetoencephalography, the N1 evoked by the second tone of a stimulus pair is enhanced when the ISI falls within 20±250 ms compared with longer (320±500 ms) ISIs (Loveless et al., 1989, 1996). In contrast, amplitude differences were found neither between the N1s elicited by the different short ISIs (20±250 ms) nor between the N1s elicited by the different long ISIs (320±500 ms). In the present study, the ISI had no effect on the N1 amplitude, which is in good agreement with the ®ndings of Loveless et al. (1989, 1996), since all tone pairs were presented with 20±120 ms ISIs in the present study. To summarize the present data, we found a clear consistency between behavioral stimulus discriminability and the MMN amplitude for temporal sound attributes. In addition, a good MMN replicability was found for the largest deviantstandard tone pair ISI separation. The MMNs to the largest and second-largest deviant-standard separation had a replicability that was comparable to those previously found for MMNs elicited by simple stimulus attributes. Furthermore, in the present study, the MMNs were elicited in the individual subjects by both the deviant with a 20 and 60 ms silent interval. Moreover, a signi®cant correlation was found between the MMN amplitude and the RTs to the 20 and 60 ms ISI deviant tone pairs. These results are encouraging in terms of using the MMN in studying auditory temporaldiscrimination de®cits. Acknowledgements This study was supported by the Academy of Finland (grant number 73038). References Aaltonen O, Niemi P, Nyrke T, Tuhkanen JM. Event-related brain potentials and the perception of a phonetic continuum. Biol Psychol 1987;24:197±207. Aaltonen O, Tuomainen J, Laine M, Niemi P. Cortical differences in tonal versus vowel processing as revealed by an ERP component called mismatch negativity (MMN). Brain Lang 1993;44:139±152. Ahissar E, Vaadia E, Ahissar M, Bergman H, Harieli A, Abeles M. Depen-

T. Kujala et al. / Clinical Neurophysiology 112 (2001) 1712±1719 dence of cortical plasticity on correlated activity of single neurons and on behavioral context. Science 1992;257:1412±1415. Alain C, Woods DL, Knight RT. A distributed cortical network for auditory sensory memory in humans. Brain Res 1998;812:23±37. Alho K, Sajaniemi N, Niittyvuopio T, Sainio K, NaÈaÈtaÈnen R. ERPs to an auditory stimulus change in pre-term and full-term infants. In: Brunia CHM, Gaillard AWK, Kok A, editors. Psychophysiological brain research, 2, Tilburg: Tilburg University Press, 1990. pp. 139±142. Amenedo E, Escera C. The accuracy of sound duration representation in the human brain determines the accuracy of behavioural perception. Eur J Neurosci 2000;12:2570±2574. Cheour M, Ceponiene R, Lehtokoski A, Luuk A, Allik J, Alho K, NaÈaÈtaÈnen R. Development of language-speci®c phoneme representations in the infant brain. Nat Neurosci 1998;1:351±353. Escera C, Grau C. Short-term replicability of the mismatch negativity. Electroenceph clin Neurophysiol 1996;100:549±554. Farmer ME, Klein RM. The evidence for a temporal processing de®cit linked to dyslexia: a review. Psychon Bull Rev 1995;2:460±493. Frodl-Bauch T, Kathmann N, MoÈller HJ, Hegerl U. Dipole localization and test±retest reliability of frequency and duration mismatch negativity. Brain Topogr 1997;10:3±8. Ilvonen T, Kujala T, Salonen O, NaÈaÈtaÈnen R, Pekkonen E. Mismatch negativity (MMN) indicates de®cits in pre-attentive auditory discrimination after left hemisphere stroke. Psychophysiology 2001;38:622±628. Jaramillo M, Paavilainen P, NaÈaÈtaÈnen R. Mismatch negativity and behavioural discrimination in humans as a function of the magnitude of change in sound duration. Neurosci Lett 2000;290:101±104. Joutsiniemi SL, Ilvonen T, Sinkkonen J, Huotilainen M, Tervaniemi M, Lehtokoski A, Rinne T, NaÈaÈtaÈnen R. The mismatch negativity for duration decrement of auditory stimuli in healthy subjects. Electroenceph clin Neurophysiol 1998;108:154±159. Kane NM, Curry SH, Butler SR, Cummins BH. Electrophysiological indicator of awakening from coma. Lancet 1993;341:688. Kathmann N, Frodl-Bauch T, Hegerl U. Stability of the mismatch negativity under different stimulus and attention conditions. Clin Neurophysiol 1999;110:317±323. Kraus N, McGee TJ, Carrell TD, King C, Tremblay K. Central auditory system plasticity associated with speech discrimination learning. J Cogn Neurosci 1995;7:27±34. Kraus N, McGee TJ, Carrell TD, Zecker SG, Nicol TG, Koch DB. Auditory neurophysiologic responses and discrimination de®cits in children with learning problems. Science 1996;273:971±973. Kujala T, Myllyviita K, Tervaniemi M, Alho K, Kallio J, NaÈaÈtaÈnen R. Basic auditory dysfunction in dyslexia as demonstrated by brain-activity measurements. Psychophysiology 2000;37:262±266. Lang AH, Eerola O, Korpilahti P, Holopainen I, Salo S, Aaltonen O. Practical issues in the clinical application of mismatch negativity. Ear Hear 1995;16:118±130. Loveless NE, Hari R, HaÈmaÈlaÈinen M, Tiihonen J. Evoked responses of human auditory cortex may be enhanced by preceding stimuli. Electroenceph clin Neurophysiol 1989;74:217±227. Loveless NE, LevaÈnen S, JousimaÈki M, Sams M, Hari R. Temporal integration in auditory sensory memory: neuromagnetic evidence. Electroenceph clin Neurophysiol 1996;100:220±228. Merzenich MM, Jenkins WM, Johnston P, Schreiner C, Miller SL, Tallal P. Temporal processing de®cits of language-learning impaired children ameliorated by training. Science 1996;271:77±81. NaÈaÈtaÈnen R. The role of attention in auditory information processing as revealed by event-related potentials and other measurements of cognitive functions. Behav Brain Sci 1990;13:201±288. NaÈaÈtaÈnen R. Attention and brain function, . Hillsdale, NJ: Erlbaum, 1992. NaÈaÈtaÈnen R. The mismatch negativity: a powerful tool for cognitive neuroscience. Ear Hear 1995;16:6±18. NaÈaÈtaÈnen R, Gaillard AWK, MaÈntysalo S. Early selective-attention effect on evoked potential reinterpreted. Acta Psychol 1978;42:313±329.

1719

NaÈaÈtaÈnen R, SchroÈger E, Karakas S, Tervaniemi M, Paavilainen P. Development of a memory trace for a complex sound in the human brain. NeuroReport 1993;4:503±506. NaÈaÈtaÈnen R, Lehtokoski A, Lennes M, Cheour M, Huotilainen M, Iivonen A, Vainio M, Alku P, Ilmoniemi RJ, Luuk A, Allik J, Sinkkonen J, Alho K. Language-speci®c phoneme representations revealed by electric and magnetic brain responses. Nature 1997;385:432±434. NaÈaÈtaÈnen R, Tervaniemi M, Sussman E, Paavilainen P, Winkler I. `Primitive Intelligence' in the auditory cortex. Trends Neurosci 2001;24:283± 288. Paavilainen P, Jaramillo M, NaÈaÈtaÈnen R. Binaural information can converge in abstract memory traces. Psychophysiology 1998;35:483± 487. Paavilainen P, Jaramillo M, NaÈaÈtaÈnen R, Winkler I. Neuronal populations in the human brain extracting invariant relationships from acoustic variance. Neurosci Lett 1999;265:179±182. Pekkonen E, Rinne T, NaÈaÈtaÈnen R. Variability and replicability of the mismatch negativity. Electroenceph clin Neurophysiol 1995;96:546± 554. Recanzone GH, Merzenich MM, Schreiner CE. Changes in the distributed temporal response properties of SI cortical neurons re¯ect improvements in performance on a temporally based tactile discrimination task. J Neurophysiol 1992;67:1071±1091. Recanzone GH, Schreiner CE, Merzenich MM. Plasticity in the frequency representation of primary auditory cortex following discrimination training in adult owl monkeys. J Neurosci 1993;13:87±103. Ritter W, Deacon D, Gomes H, Javitt DC, Vaughan Jr HG. The mismatch negativity of event-related potentials as a probe of transient auditory memory: a review. Ear Hear 1995;16:52±62. Robin DA, Tranel D, Damasio H. Auditory perception of temporal and spectral events in patients with focal left and right cerebral lesions. Brain Lang 1990;39:539±555. Rosen S. Temporal information in speech: acoustic, auditory and linguistic aspects. Philos Trans R Soc Lond Biol Sci 1992;336(1278):367±373. Sams M, Aulanko R, Aaltonen O, NaÈaÈtaÈnen R. Event-related potentials to infrequent changes in synthesized phonetic stimuli. J Cogn Neurosci 1990;2:344±357. SchroÈger E. Measurement and interpretation of the mismatch negativity. Behav Res Methods Instrum Comput 1998;30:131±145. Shannon RV, Zeng F-G, Kamath V, Wygonski J, Ekelid M. Speech recognition with primarily temporal cues. Science 1995;270:303±304. Tallal P. Auditory temporal perception, phonics, and reading disabilities in children. Brain Lang 1980;9:182±198. Tallal P, Piercy M. Developmental aphasia: impaired rate of non-verbal processing as a function of sensory modality. Neuropsychologia 1973;11:389±398. Tallal P, Miller SL, Bedi G, Byma G, Wang X, Nagarajan SS, Schreiner C, Jenkins WM, Merzenich MM. Language comprehension in languagelearning impaired children improved with acoustically modi®ed speech. Science 1996;271:81±84. Tervaniemi M, Maury S, NaÈaÈtaÈnen R. Neural representations of abstract stimulus features in the human brain as re¯ected by the mismatch negativity. NeuroReport 1994;5:844±846. Tervaniemi M, Lehtokoski A, Sinkkonen J, Virtanen J, Ilmoniemi RJ, NaÈaÈtaÈnen R. Test±retest reliability of mismatch negativity for duration, frequency and intensity changes. Clin Neurophysiol 1999;110:1388± 1393. Tiitinen H, May P, Reinikainen K, NaÈaÈtaÈnen R. Attentive novelty detection in humans is governed by pre-attentive sensory memory. Nature 1994;327:90±92. Tremblay K, Kraus N, McGee T. The time-course of auditory perceptual learning: which comes ®rst, the chicken or the egg. NeuroReport 1998;9:3557±3560. Tyler RS. Cochlear implants, audiological foundations, San Diego, CA: Singular Publishing, 1993.