B worldwide social and health problem. It has been

Transcription

1 /99/ $03.00/0 ALCOHOLISM: CLINICAL AND EXPERIMENTAL RESEARCH Vol. 23, No. 9 September 1999 Suppression of Mismatch Negativity by Backward Masking Predicts Impaired Working-Memory Performance in Jyrki Ahveninen, liro P. Jaaskelainen, Eero Pekkonen, Anja Hallberg, Marja Hietanen, Rauno Makela, Risto Naatanen, and Pekka Sillanaukee Background: Pronounced disruption of memory traces by subsequent distractors may result in impaired behavioral memory performance in alcoholics. Methods: This hypothesis was investigated with an electrophysiological index of auditory sensorymemory traces, mismatch negativity, a preattentive event-related potential component elicited by a deviant tone within a train of standard tones. Results: Inserting a masking stimulus after these tones abolished mismatch negativity in alcoholics (DSM-IV) but not in social-drinker controls. This effect predicted working-memory impairment in alcoholics, and correlated significantly with self-reported alcohol consumption of the subjects. Furthermore, the backward-masking mismatch negativity paradigm detected sensory-memory impairment in 9 of 20 alcoholics (sensitivity 45%), whereas all 20 social drinkers were unimpaired (specificity 100%). Conclusions: Vulnerability to memory trace disruption by shortly following sounds may be one of the factors contributing to behavioral memory dysfunction in alcoholics. The present result may provide an objective neurophysiological tool for investigation of alcohol-induced and other degenerative brain disorders. Key Words: Alcoholism, Auditory Sensoy Memory, Event-Related Potentials, Mismatch Negativity, Neuropsychological Tests. Y ANY MEASURE, alcohol abuse constitutes a B worldwide social and health problem. It has been estimated that in the United States alone, there are approximately 13 million individuals who need treatment for alcohol use disorders (Institute of Medicine, 1990). The presence of alcohol-induced structural brain lesions and neurocognitive disorders is well established (Bergman et al., 1980; Brewer and Perrett, 1971; Ishii, 1983; Jernigan et al., 1991; Knight and Longmore, 1994). Despite the high clinical need and profusion of intensive research, satisfactory clinical applications for early detection and monitoring From the Cognitive Brain Research Unit, Department of Psychology (J.A., I.P.J., E.P., R.N.), and Department of Neurology (E.P., M.H.), University of Helsinki, Helsinki, Finland; the BioMag Laboratory (E.P., R.N.), Helsinki University Central Hospital, Helsinki, Finland; the Jarvenpaa Social Hospital (A.H.), A-Clinic Foundation, Jarvenpaii, Finland; the A-Clinic Foundation (R. M.), Helsinki, Finland; Alcohol Related Diseases (P.S.), Pharmacia Upjohn AB Diagnostics, Uppsala, Sweden; the Medical School (P.S.), University of Tampere, and the Department of Clinical Chemistry (P.S.), Tampere University Hospital, Tampere, Finland; and the Department of Clinical Neursciences (P.S.), Karolinshz Instituter, Stockholm, Sweden. Received for publication March 5, 1999; accepted June 7, This study was supported by the Finnish Foundation for Alcohol Studies, the Finnish Cultural Foundation, and the Academy of Finland. Reprint requests: Jyrki Ahveninen, Cognitive Brain Research Unit, Department of Psychology, PO Box 13, FIN University of Helsinki, Helsinki, Finland; Fax: ; Copyright by the Research Society on Alcoholism. Alcohol Clin Exp Res, Vol23, No 9, 1999: pp of alcohol-related disturbances of brain function have not yet been developed. According to brain imaging and pathological studies, chronic alcoholism has been associated with brain atrophy (Bergman et al., 1980; Brewer and Perrett, 1971; Ishii, 1983; Jernigan et al., 1991) and loss or shrinkage of cortical neurons (Harper et al., 1987; Kril and Harper, 1989). Profound alcohol-related changes in cerebral blood flow and metabolism have also been observed (Gilman et al., 1990; Nicolas et al., 1993; Sachs et al., 1987). The studies on auditory event-related potentials (ERPs), electroencephalographic (EEG) changes time-locked to the presentation of auditory stimuli, have shown marked neurocognitive abnormalities in alcoholics (Nixon and Phillips, 1999; Porjesz and Begleiter, 1985). The most consistent result associated with alcoholism is the amplitude reduction of P3b (Patterson et al., 1987; Pfefferbaum et al., 1991; Porjesz et al., 1988; Realmuto et al., 1993), a positive ERP component elicited by attended, infrequently occurring task-relevant stimuli, with a 300- to 500-msec peak latency. Delayed auditory P3b peak latencies have also been frequently reported in chronic alcoholics (Cadaveira et al., 1991; Pfefferbaum et al., 1979, 1991; Steinhauer et al., 1987). Previous ERP studies have also demonstrated the postwithdrawal enhancement of auditory evoked electric or magnetic responses (J. Ahveninen et al., 1999; Pekkonen et 1507

2 1508 AHVENINEN ET AL. al., 1998; Porjesz and Begleiter, 1985). It has been proposed that this enhancement is caused by brain hyperexcitability (J. Ahveninen et al., 1999; Pekkonen et al., 1998; Porjesz and Begleiter, 1985), resulting from neuroadaptation from chronic alcohol abuse, and subsequently reduced inhibitory and increased excitatory synaptic transmission after alcohol withdrawal (Buck and Harris, 1991; Tsai et al., 1995). Recent benzodiazepine binding results suggest that these synaptic changes can be detected 3 months after detoxification (Lingford-Hughes et al., 1998). The most profound neuropsychological deficits, such as retro- and anterograde amnesia, are usually observed only in alcoholics suffering from Korsakoff s syndrome (Butters and Cermak, 1980), caused by thiamine deficiency and subsequent Wernicke s disease (Victor et al., 1989). Conversely, many of the initial studies failed to show any significant cognitive decline in neurologically intact alcohol abusers (Butters et al., 1977; Parson and Prigatano, 1977). More recently, subtle but significant verbal short- and longterm memory deficits have also been observed in alcoholics without a history of Wernicke s disease (Acker et al., 1987; Brandt et al., 1983; Knight and Longmore, 1994; Ryan et al., 1980). Although the short-term memory span is rarely impaired in alcoholics (even in those with Korsakoffs syndrome) (Knight and Longmore, 1994), their performance clearly deteriorates when active rehearsal of the information is blocked with a distractor task (Brandt et al., 1983; Knight and Longmore, 1994; Ryan et al., 1980). This vulnerability to interference correlates with alcohol consumption even in social drinkers, which suggests that this type of deficit is one of the first ones to develop with alcohol abuse (MacVane et al., 1982). There is some evidence that alcoholic memory impairment correlates with widening of cerebral sulci and ventricles (Acker et al., 1987; Carlen et al., 1981), and with overall reduction of brain glucose use in alcoholics (Wang et al., 1993). Despite these findings, the exact neurophysiological factors underlying alcohol-related memory dysfunction have remained ambiguous. One factor possibly underlying the above-mentioned alcohol-related auditory working-memory deficits, namely, the accelerated forgetting during distraction, may be pronounced vulnerability of earlier neural memory representations to interference by later ones (Cowan, 199.5). An analog of this hypothesis was used to study alcohol-related abnormalities in transient auditory sensory memory that precedes further processing in the auditory working memory and long-term storage of information (Baddeley, 1986). The rationale of studying auditory sensory-memory traces is that they can be easily and objectively indexed by mismatch negativity (MMN), a preattentive brain electric response evoked when a predictable series of homogenous standard stimuli is interrupted by a deviant stimulus (Naatanen, 1992,1995; Naatanen et al., 1978, 1997; Ritter et al., 1995; Tiitinen et al., 1994). A memory trace is made of the repeated standards, and attention-independent recognition of difference between physical features of the deviant and the memory trace of the standard leads to evocation of MMN (Naatanen, 1992, 1995; Naatanen et al., 1978). This memory-trace formation can be interfered with backwardmasking stimuli that shortly follow each stimulus of the MMN paradigm (Cowan, 1984; Winkler and Naatanen, 1995; Winkler et al., 1993). It was hypothesized that the neural abnormalities possibly affecting auditory sensory representations may accumulate at the higher levels of memory processing. Namely, the memory deficits have been shown to correlate with global, rather that focal, functional changes in the alcoholic brain (Wang et al., 1993; George et al., 1999). Furthermore, alcohol consumption may have a cortically widespread impact on buildup of neural representations. The alcohol-sensitive (Tsai et al., 1995) NMDA receptors, proposed to be central in MMN generation (Ritter et al., 1995), may govern experiencedependent synaptic plasticity in hippocampus and throughout the neocortex (Kirkwood et al., 1993). It has been shown that the MMN is completely abolished in healthy subjects by masks interposed at 0 to 50-msec after stimulus offset (Cowan, 1984; Winkler and Naatanen, 1995; Winkler et al., 1993). By widening the gap between the masks and the tones, the masking effect gradually decreases, and vanishes within 1.50 to 300 msec (Cowan, 1984; Hawkins and Presson, 1986; Winkler and Naatanen, 1995; Winkler et al., 1993). Here, the masks were interposed 100 msec after standards and deviants, to avoid the possible ceiling effect in the control subjects and to reduce the time required for the measurement. In addition to the investigation of auditory sensory-memory impairment in alcoholics, the effect of backward masking on MMN generation was correlated with behavioral working-memory performance. To date, this is the first clinical study that capitalizes on backward masking of MMN generation. Subjects METHODS Twenty consecutive male alcoholics (age, years; mean age, 40 years; SD, 10.6 years), from the Jarvenpaa Social Hospital, who met DSM-IV criteria for alcohol dependence and 20 age- and educationmatched healthy male social drinkers (age, years; mean age, 37 years; SD, 11.7 years) were studied. The alcoholics, who had between 1 and 35 years (mean, 11 years; SD, 9.2 years) of chronic drinking (selfreported ethanol consumption, ghveek; mean, 1236 ghveek; SD, g/week), had been abstinent for 7 to 45 days (mean, 20 days; SD, 10.4 days) and were without acute withdrawal symptoms [according to Clinical Institute Withdrawal Assessment for Alcohol (Sullivan et al., 1989)l. The social drinkers self-reported consumption was 21 to 216 ghveek (mean, 97 glweek; SD, 73.4 ghveek). The Alcohol Use Disorders Identification Test (Seppa et al., 1995) was used. The study was approved by the Ethics Committee of the A-Clinic Foundation, Helsinki, Finland, according to the Declarations of Helsinki. An informed consent was obtained after the procedures had been fully explained to the subjects. An experienced clinician (A.H.) recruited the alcoholics from a routine treatment program and examined their medical records. Subjects with hearing loss, neurological disorders, severe psychiatric disorders other than alcoholism, or other severe diseases were excluded. The social-drinker subjects abstained from alcohol and other drugs for at least 48 hr before the measurement. Six

3 BACKWARD-MASKING MMN IN ALCOHOLICS 1509 alcoholics used antidepressants or other central nervous system (CNS) medication (two used fluoxetine, one used mianserin, one used mianserin and promazine, one used doxepin, and one used promazine in the evenings). No statistically significant differences were obsewed between the results of the 14 unmedicated and the 6 medicated alcoholics; moreover, the results of this study were significant even after exclusion of these 6 alcoholics from the analysis. ERF Measurement ERPs were measured in an acoustically and electrically shielded room by using a 32-channel EEG. Mastoid-referenced EEG (sampling rate, 250 Hz) was averaged and filtered digitally (pass band, 2-12 Hz for the MMN and 1-30 Hz for the N1; prestimulus baseline, -100 msec). Epochs with artifacts at >75 pv at any electrode or at an electro-oculogram were discarded. Subjects reading self-selected material and ignoring the stimulation were binaurally presented with trains of 600-Hz standard tones (p = 0. 80) that were randomly replaced by 670-Hz deviant tones. Both the standards and the deviants (total, 1294 stimuli per block) were sinusoidal tones of 25-msec duration with 2.5-msec rise and fall times. The stimuli were presented 60 db over subjective hearing threshold. Stimulus onset asynchrony was 300 msec. In one of these conditions, we interposed randomly varying masking tones (300,400, 900, or 1000 Hz; duration, 25 msec; 2.5 msec rise and fall time) starting 100 msec after each stimulus offset to interfere with memory-trace formation for the standards (the backward-masking condition) (Cowan, 1984; Winkler and Naatanen, 1995; Winkler et al., 1993). During the other condition (the baseline condition), the standard and deviant tones were presented without masking tones. The MMN amplitude was determined from subtraction curves (the deviant-stimulus ERP minus standard-stimulus ERP) by using a signal-space projection (Uusitalo and Ilmoniemi, 1997). Specifically, the spatial distribution of the MMN was estimated from the within-group grand-averaged ERP wave forms at the MMN peak latency. This estimate was used to dissociate the MMN signal from the noise at the individual level, which resulted in an improved signal-to-noise ratio. The signal-space projection of MMN was calculated from signals obtained at seven electrode positions centered around the frontal electrode (FJ, where the MMN is usually largest in amplitude (Naatanen, 1992). Neuropsychological Testing To measure the auditory-verbal working memory, we used the Wechsler Memory Scale digit span (Lezak, 1995; Wechsler, 1945) and the Brown-Peterson procedure termed, auditory consonant trigrams (ACT), in which a distractor task is interposed between encoding and recall of three consonants (Lezak, 1995; Peterson and Peterson, 1959). In the ACT, the subject was presented with three consonants, and after hearing them, during either 3, 9, or 18 sec of distraction, he was to count backward by 3s from a given number, until signaled to report the consonants. A total number of correct consonants of 20 trials, including an additional baseline without distraction, was scored. In addition, two neuropsychological tests were used to assess attention and executive dysfunction (here, termed frontal tests), Trail-Making Test part A and part B (TMT; originally in Army Individual Test Battery, 1944) (Lezak, 1995) and Stroop Color and Word Naming Test (SCWT) color patch naming (part I) and false color word naming (part 11) (Lezak, 1995; Stroop, 1935). The frontal test performance was defined as a sum score of the following standardized difference and error scores: TMT B-A, time; TMT-B, errors; SCWT 11-1, time; SCWT-11, errors. Statistical Analysis Group and treatment differences in the ERP were statistically analyzed by two-way analyses of variance (two groups by two MMN conditions) with two a priori contrasts (using the error term of the overall effect) to determine the group MMN differences within the MMN conditions (Statistica 4.1 software, Stat Soft Inc., Tulsa, OK). Neuropsychological data were analyzed statistically by using multivariate analysis of variance (years.- L Y In m E c 3 0 Social Drinkers r, 5 0 ms > 2 pvj - Standard response - Deviant resdonse Fig. 1. Grand-average responses to standard and deviant tones from electrode F, in alcoholic subjects and control subjects. The gray area indicates mismatch negativity (MMN). of formal education entered as a covariate) with contrasts to determine the group differences in the distinct neuropsychological tests. The neuropsychological and ERP data were correlated by using Pearson s correlation coefficient, and to analyze the factors affecting working-memory performance of the alcoholics, a multiple regression analysis was used. RESULTS MMN With and Without Backward Masking The cortical auditory-sensory memory function was measured under two separate ERP conditions (with and without backward masking). Figure 1 shows the deviant and the standard ERPs under these conditions, and the grandaverage scalp distributions of MMN and subtraction curves are presented in the Fig. 2. Under the baseline condition without backward masking, the deviant tones elicited MMN in both groups (Figs. 1 and 2; Table 1). This baseline MMN was larger [F(1,38) = 6.00, p < and earlier [F(1,38) = 5.66,~ < in the alcoholics, compared with that in the social drinkers. In addition, the N1 deflection to the standard stimuli was larger in the alcoholics [F(1,38) = 7.62,~ < Under the backward-masking condition, the alcoholics MMN was abolished by the masking tones, interposed after each stimulus offset to interfere with sensory-memory function, but the social drinkers showed no such effect. There was a significant group-by-stimulus condition interaction in the analysis of variance [F(1,38) = 10.48, p < , with the masking condition MMN amplitude being larger in the social drinkers than in the alcoholics [F(1,38) = 4.53, p < When the cutoff point was 0.80 FV (mean MMN reduction by masking plus 1.7 SDs), 9 of 20 of the alcoholics (sensitivity, 45%), but no social drinkers (specificity, loo%), showed impairment of the auditory sensory memory (Fig. 3). Furthermore, the MMN reduction by masking correlated significantly with self-reported weekly alcohol consumption (Pearson s r = 0.41; p < 0.01) in the whole study group). 1 J 1

4 1510 AHVENINEN ET AL. ACT score (Pearson s r = -0.56; p < 0.001) correlated negatively with the self-reported weekly alcohol consumption in the whole study group. The backward digit span of the alcoholics was significantly reduced [F(1,37) = 10.4; p < , but no between-group differences were found in the forward digit span (Table 1) A I W I -!klcw&wem 1 - moms i -AleohdEg Fig. 2. (a) Surface potential maps of mismatch negativity (MMN) (from grandaverage subtraction curves, deviant minus standard event-related potentials, at the MMN peak latency) of social drinkers and alcoholics, both with (right) and without (left) backward masking projected on a model head. (b) Deviant minus standard subtraction curves from electrode F,. Maps and curves demonstrate between-condition double dissociation of MMN amplitude (indicated by shading in Fig. 1) in alcoholics and social drinkers. Neuropsychological Tests The function of auditory-verbal working memory was studied by using the ACT, in which a distractor task was interposed between encoding and recall of three consonants (Lezak, 1995; Peterson and Peterson, 1959). Figure 4 and Table 1 show that, as expected, the alcoholics performed significantly worse in the ACT than the social drinkers [F(1,37) = 14.5; p < In addition, two neuropsychological tests, TMT and SCWT, were used to assess attentional and executive dysfunction (here, termed frontal dysfunction). Figure 4 and Table 1 indicate that the alcoholics showed marked frontal dysfunction (measured with a sum score of TMT and SCWT) compared with the social drinkers [F(1,37) = 17.8;~ < Both the frontal test performance (Pearson s r = -0.48;~ < 0.002) and the Correlation Between MMN and Neuropsychological Test Pegormance In the whole study group, the MMN reduction by backward masking correlated negatively with ACT performance (Pearson s r = -0.61; p < 0.001) and with the frontal test performance (Pearson s r = -0.33; p < 0.039). The ACT score also correlated with frontal test performance (Pear- son s I = 0.65;~ < 0.001). Figure 5 indicates the association between pronounced distractibility of sensory-memory and working-memory dysfunction in the alcoholics. In the alcoholics, the MMN reduction correlated with impaired ACT performance (Pearson s r = -0.56; p < 0.012), and the ACT score correlated with frontal test performance [Pearson s r = 0.49; t(16) = 2.43; p < The MMN reduction by masking did not, however, significantly correlate with frontal test performance in the alcoholics. In the social drinkers, no significant correlations emerged between the MMN masking effect (this MMN effect being, notably, statistically insignificant in the social drinkers) and the neuropsychological test scores. When both the alcoholics and the social drinkers were considered, a linear regression analysis showed a significant multiple correlation [adjusted R2 = 0.62; F(3,36) = 21.9; p < between the ACT performance and three independent variables, i.e., (1) the MMN amplitude reduction by backward masking [partial r = -0.52; t(36) = -3.66; p < , (2) the frontal test performance [partial r = 0.56; t(36) = 4.08; p < 0.001], and (3) the Wechsler Mem- ory Scale digit span score [partial r = 0.35; t(36) = 2.21;~ < In the alcoholics, correspondingly, a significant multiple correlation emerged between the ACT performance and MMN masking effect, frontal test performance, and Wechsler Memory Scale digit span score [adjusted R2 = 0.42; F(3,16) = 5.47; p < Significant partial correlations were detected between ACT and MMN reduction by masking [partial r = -0.55; t(16) = -2.62; p < and between ACT and frontal test performance [partial r = 0.50; t(16) = 2.33;~ < DISCUSSION In conclusion, backward masking abolished the MMN in the abstinent alcoholics whereas only a weak trend was observed in the social drinkers. The decrement of MMN amplitude by backward masking is believed to reflect disruption of memory trace formation at the auditory cortices (Cowan, 1984; Winkler and Naatanen, 1995; Winkler et al.,

5 BACKWARD-MASKING MMN IN ALCOHOLICS 1511 Table 1. Results of MMN Measurement and Neuropsychological Testing in and Social Drinkers Social drinkers CI CI CI CI Mean SE (-0.95) (+0.95) Mean SE (-0.95) (+0.95) MMN, without masking Peak latency (msec) ' Amplitude (pv) ' MMN, backward masking Peak latency (msec) Amplitude (pv) ' MMN reduction by masking " N1 to standard tones Peak latency (msec) Amplitude (pv) " Neuropsychological tests ACT (r score units) "' Frontal tests (I score units) *** Digit span (forward) Digit span (backward) MMN, mismatch negativity; ACT, auditory consonant trigrams (test). Data are mean, standard error of mean (SE), and confidence interval (CI) values. ' p < 0.05; "p < 0.01: "'p < MMN reduction by mask a A MMN amplitude (Pv) n Soclaldrinkers B - MMN peak latency (ms) C Neuropsychology (z-score),, Nt * " Base- Mask Hne -1' ACT Frontal tests L 1L Fig. 4. (A) Group average mismatch negativity (MMN) amplitudes. (B) Group average MMN peak latencies, without (Baseline) and with backward masking (Mask). (C) Group average2 scores of auditory consonant trigrams (ACT) test and frontal test performance p < 0.05: ** p < 0.01; *** p < 0.001, error bars are SEM values). -1 Social drinkers Fig. 3. Reduction of mismatch negativity (MMN) amplitude by backward masking. Nine of 20 alcoholics but no social drinkers showed sensory-memory deficit (cutoff mean SD). Boxes represent SEM values: whiskers are SDs. 1993). As a result of chronic alcohol abuse, the formation of cortical auditory-sensory memory traces appears to be abnormally vulnerable to interference by closely succeeding auditory stimuli. The masking effect is normally most evident during the 0- to 50-msec poststimulus period, after which it gradually decreases, and vanishes within 150 to 300 msec (Hawkins and Presson, 1986). During the critical poststimulus period, the temporal window of integration (normally, msec), successive stimuli are encoded as an integrated unit (represented by an integrated memory trace) and a masking stimulus falling within this temporal window dominates this integrated unit to the extent that all other information is lost (Naatanen, 1992; Yabe et al., 1997). It appears that this temporal window may be widened in the alcoholic brain. The significant positive correlation between MMN reduction by masking and self-reported alcohol consumption supports the assumption that this disorder may indeed be related to severity of drinking history. Moreover, the correlation coefficient (r = 0.4) detected is at the same level as the ones obtained by using routine laboratory markers of alcohol abuse (Sillanaukee, 1996). This was supported by the observed 45% sensitivity with 100% specificity, which should be, however, confirmed later with larger patient populations. Although there is evidence that MMN decreases significantly even with low blood alcohol concentration (Jaaskelainen et al., 1995), a recent magnetoencephalographic study indicated that 2 to 6 weeks after withdrawal of chronic drinking, the MMN generation may even be somewhat enhanced (Pekkonen et al., 1998). Correspondingly, significant augmentation and acceleration of MMN without the masking, and augmentation of Nl elicited by the stan-

6 1512 AHVENINEN ET AL. PV Z-score in the ACT Fig. 5. Correlation plot of z score in auditory consonant trigrams (ACT) test and absolute reduction of mismatch negativity (MMN) amplitude by backward masking within the alcoholic group. Reduction of MMN by backward masking was a particularly valid predictor of working-memory dysfunction in alcoholics (Pearson s r = -0.55; p < 0.011; confidence limitsp = 0.95). dards, was observed in the alcoholics. This may reflect reduced neural inhibition in the alcoholic brain, because reduced benzodiazepine binding has been evidenced even in alcoholics abstinent for 3 months (Lingford-Hughes et al., 1998). However, in contrast to the recent finding of augmented middle-latency auditory evoked potentials in alcoholics (J. Ahveninen et al., 1999), neither the MMN nor N1 enhancement correlated with abstinence duration in the alcoholics in the present study. Hence, whether the MMN and N1 enhancement reflects CNS hyperexcitability after alcohol withdrawal (Buck and Harris, 1991; Tsai et al., 1995), or results from more permanent or premorbid CNS changes, must be confirmed by later studies. Nevertheless, it appears that the MMN enhancement without masking probably reflects sensitized response to the detected difference between the standard tone trace and deviant tone in alcoholics, rather than strengthened sensory-memory traces. The pronounced MMN masking effect and the subsequent correlation with impaired working-memory performance supports this assumption. One may question whether the overlapping components elicited by the tones and the masks could have affected the results. For instance, the negative N1 deflection to masks, peaking -100 msec after mask onset, and, thus, 225 msec after the standard or deviant onset, could have overlapped with the positive P2 to the standard or deviant tones (peaking -200 msec after tone onset). Given that the varying masks followed both the standards and the deviants with similar probability, the components elicited by the masks, and their possible effects on standard and deviant responses, should, however, be canceled out in the subtraction of deviant and standard ERPs. In addition, the masks were substantially different from the standards and the deviants in tone frequency to activate different frequencyspecific neuron populations (Naatanen, 1992), respectively, which minimizes the possibility that group differences in the refractoriness of these frequency-specific neurons would have confounded the results. The present results demonstrated that the abolition of MMN by backward masking correlated significantly with the impaired behavioral working-memory performance in the alcoholics. This implicates that the accelerated forgetting at higher cognitive stages of the memory system could, in part, be attributed to the initially impaired formation of preattentive sensory-memory traces. As previously observed in alcoholic Korsakoff patients (Leng and Parkin, 1989), the accelerated forgetting during the distraction task correlated also with frontal dysfunction in alcoholics. In the alcoholic group, however, no significant correlation emerged between the frontal dysfunction and MMN reduction by masking, which suggests that these factors may predict distinct aspects of working memory. It has been proposed that in Korsakoff patients, frontal deficits may impair the ability to switch from the counting task to retrieval of the target information (Leng and Parkin, 1989). It can thus be speculated that frontally induced retrieval problems may also contribute to working-memory impairment in non-korsakoff alcoholics. The abolition of MMN by masking, reflecting impaired buildup of cortical memory traces, may, in turn, be a predictor of impaired neural representations, also at the higher levels of memory system in alcoholics. Given that the working-memory impairment may be one of the first neurocognitive deficits caused by alcohol abuse (MacVane et al., 1982), and the exacerbating nature of alcohol-related changes in cognitive function in general (Lishman, 1990), early detection of these memory deficits is essential in the chain of treatment. Unlike most behavioral and ERP measures of neurocognitive disorders, the MMN is not modulated by the subject s higher order cognitive processes associated with expectancy, confidence, or attention (Naatanen, 1992, 1995; Ritter et al., 1995). The MMN paradigm applied in the present study may provide an objective method for research on the early phase changes in alcohol-induced CNS disorders. ACKNOWLEDGMENTS We thank Dr. T. Kujala for comments on the manuscript, J. Sinkkonen for constructive comments on statistics, and T. Rinne for technical support. REFERENCES Acker C, Jacobson RR, Lishman WA (1987) Memory and ventricular size in alcoholics. Psycho1 Med 17: Ahveninen J, Jaaskelainen IP, Pekkonen E, Hallberg A, Hietanen M, Naatanen R, Sillanaukee P (1999) Postwithdrawal changes in middlelatency auditory evoked potentials in abstinent human alcoholics. Neurosci Lett, 268: Baddeley A (1986) Working Memory. Clarendon Press, Oxford. Bergman H, Borg S, Hindmarsh T, Idestrom CM, Mutzell S (1980) Computed-tomography of the brain and neuropsychological assessment of alcoholic patients. Adv Exp Med Biol 126: Brandt J, Butters N, Ryan C, Bayog R (1983) Cognitive loss and recovely in long-term alcohol abusers. Arch Gen Psychiatry 40:

7 BACKWARD-MASKING MMN IN ALCOHOLICS 1513 Brewer C, Perrett L (1971) Brain damage due to alcohol consumption: An air-encephalographic psychometric and electroencephalographic study.,br J Addiction 66: Buck KJ, Harris RA (1991) Neuroadaptive responses to chronic ethanol. Alcohol Clin Exp Res 15: Butters N, Cermak LS (1980) Alcoholic Korsakoff Syndrome: An Znformation Processing Approach to Amnesia. Academic Press, New York. Butters N, Cermak LS, Montgomery K, Adinolfi A (1977) Some comparisons of the memory and visuoperceptive deficits of chronic alcoholics and patients with Korsakoff s disease. Alcohol Clin Exp Res 1: Cadaveira F, Grau C, Roso M, Sinchez-Turet M (1991) Multimodality exploration of event-related potentials in chronic alcoholics. Alcohol Clin Exp Res 15: Carlen PL, Wilkinson DA, Worzman G, Holgate R, Cordingley J, Lee MA, Huszar L, Moddel G, Singh R, Kiraly L, Rankin JB (1981) Cerebral atrophy and functional deficits in alcoholics without clinically apparent liver disease. Neurology 31: Cowan N (1984) On short and long auditory stores. Psychol Bull 96: Cowan N (1995) Attention and Memoy. Oxford, New York. George MS, Teneback CC, Bloomer CW, Homer MD, Anton RF (1999) Using neuroimaging to understand alcohol s brain effects. CNS Spectrums 4: Gilman S, Adams K, Koeppe RA, Berent S, Kluin KJ, Model1 JG, Kroll P, Brunberg JA (1990) Cerebellar and frontal hypometabolism in alcoholic cerebellar degeneration studied with positron emission tomography. Ann Neurol 28: Harper C, Kril J, Daly J (1987) Are we drinking our neurons away? BMJ 294~ Hawkins HL, Presson JC (1986) Auditory information processing, in Handbook of Perception and Human Perjormance (Boff KR, Kaufman L, Thomas JP eds) pp 1-64, Wiley, New York. Institute of Medicine (1990) Broadening the Base of Treatment for Alcohol Problems. National Academy Press, Washington, DC. Ishii T (1983) A comparison of cerebral atrophy in CT scan findings among alcoholic groups. Acta Psych Scand (Supp1.)309:7-30. Jaaskelainen IP, Lehtokoski A, Alho K, Kujala T, Pekkonen E, Sinclair JD, Naatanen R, Sillanaukee P (1995) Low dose ethanol suppresses mismatch negativity of auditory event-related potentials. Alcohol Clin Exp Res Jernigan TL, Butters N, DiTraglia G, Schafer K, Smith T, Irwin M, Grant I, Schuckit M, Cermak LS (1991) Reduced cerebral grey matter observed in alcoholics using magnetic resonance imaging. Alcohol Clin Exp Res 15: Kirkwood A, Dudek SM, Gold JT, Aizenman CD, Bear MF (1993) Common forms of synaptic plasticity in the hippocampus and neocortex in vitro. Science (Wash. DC) 260: Knight RC, Longmore BE (1994) Clinical Neuropsychology of Alcoholim. Lawrence Erlbaum, Howe, U.K. Kril JJ, Harper CG (1989) Neuronal counts of four cortical regions of alcoholic brains. Acta Neuropathol 79: Leng NRC, Parkin GI (1989) Aetiological variation in the amnesic syndrome: Comparisons using the Brown-Peterson task. Cortex 25: Lezak MD (1995) Neuropsychological Assessment. Oxford, New York. Lingford-Hughes AR, Acton PD, Gacinovid S, Suckling J, Busatto GF, Boddington SJA, Bullmore E, Woodruff PW, Costa DC, Pilowsky LS, Ell PJ, Marshall EJ, Kenvin RW (1998) Reduced levels of GABAbenzodiazepine receptor in alcohol dependency in the absence of gray matter atrophy. Br J Psychiatry 173: Lishman WA (1990) Alcohol and the brain. Br J Psychiatry 156: MacVane J, Butters N, Montgomery K, Farber J (1982) Cognitive functioning in men social drinkers: A replication study. J Stud Alcohol 43: Naatanen R (1992) Attention and Brain Function. Lawrence Erlbaum, Hillsdale, NJ. Naatanen R (1995) The mismatch negativity: A powerful tool for cognitive neuroscience. Ear Hear 16:6-11. Naatanen R, Gaillard AWK, Mantysalo S (1978) Early selective-attention effect on evoked potential reinterpreted. Acta Psychol 42: Naatanen R, Lehtokoski A, Lennes M, Cheour M, Huotilainen M, Ilvonen A, Vainio M, Alku P, Ilmoniemi R, Luuk A, Allik J, Sinkkonen J, Alho K (1997) Language-specific phoneme representations revealed by electric and magnetic brain responses. Nature (Lond.) 385: Nicolis JM, Catafau AM, Estruch R, Lomefia J, Salamero M, Herranz R, Monforte R, Cardenal C, Urbano-Marquez A (1993) Regional cerebral blood flow-spect in chronic alcoholism: Relation to neuropsychological testing. J Nucl Med Nixon SJ, Phillips JA (1999) Neurocognitive deficits and recovery in chronic alcohol abuse. CNS Spectrums 4: Parson OA, Prigatano GP (1977) Memory functioning in alcoholics, in Alcohol and Human Memoy (Birbaum IM, Parker ES eds) pp , Lawrence Erlbaum Associates, Hillsdale, NJ. Patterson BW, Williams HL, McLean GA, Smith LT, Schaefer KW (1987) Alcoholism and family history of alcoholism: Effects on visual and auditory event-related potentials. Alcohol 4: Pekkonen E, Ahveninen J, Jaaskelainen IP, Seppa K, Naatanen R, Sillanaukee P (1998) Selective acceleration of auditory processing in abstinent alcoholics. Alcohol Clin Exp Res Peterson LR, Peterson MJ (1959) Short-term retention of individual verbal items. J Exp Psychol 58: Pfefferbaum A, Horvarth TM, Roth WT, Kopell BS (1979) Event-related potential changes in chronic alcoholics. Electroencephalogr Clin Neurophysiol 47: Pfefferbaum A, Ford JM, White PM, Mathalon D (1991) Event-related potentials in alcoholic men: P3 amplitude reflects family history but not alcohol consumption. Alcohol Clin Exp Res 15: Pojesz B, Begleiter H (1985) Human brain electrophysiology and alcoholism, in Alcohol and the Brain (Tarter RE, Thiel DH eds) pp , Plenum Press, New York. Poqesz B, Begleiter H, Rawlings R, Eckardt M (1988) Auditory recovery function and P3 in abstinent alcoholics. Alcohol Alcohol 23:A41. Realmuto G, Begleiter H, Odencranz J, Porjesz B (1993) Event-related potential evidence of dysfunction in automatic processing in abstinent alcoholics. Biol Psychiatry Ritter W, Deacon D, Gomes H, Javitt DC, Vaughan HG Jr. (1995) The mismatch negativity of event-related potentials as a probe of transient auditory memory: A review. Ear Hear 16: Ryan C, Butters N, Montgomery K (1980) Memory deficits in chronic alcoholics: Continuities between the intact alcoholic and the alcoholic Korsakoff patient, in Biological Effects of Alcohol (Begleiter H ed) pp , Plenum Press, New York. Sachs H, Russel JAG, Christman DR, Cook B (1987) Alteration in regional cerebral glucose metabolic rate in nonkorsakoff chronic alcoholism. Arch Neurol 44: Seppa K, Makela R, Sillanaukee P (1995) Effectiveness of the alcohol use disorders identification test in occupational health screenings. Alcohol Clin Exp Res 19: Sillanaukee P (1996) Laboratory markers of alcohol abuse. Alcohol AIcoholism 31: Steinhauer SR, Hill SY, Zubin J (1987) Event-related potentials in alcoholics and their first-degree relatives. Alcohol Stroop JR (1935) Studies of interference in serial verbal reactions. J Exp Psycho1 18~ Sullivan JT, Sykora K, Schneiderman J, Naranjo CA, Sellers EM (1989) Assessment of alcohol withdrawal: The revised clinical institute withdrawal assessment for alcohol scale (CIWA-Ar). Br J Addict M Tiitinen H, May P, Reinikainen K, Naatanen R (1994) Attentive noveltydetection in humans is governed by preattentive sensory memory. Nature (Lond.) 372: Tsai G, Gastfriend DR, Coyle JT (1995) The glutamatergic basis of human alcoholism. Am J Psychiatry 152:

8 1514 AHVENINEN ET AL. Uusitalo M, Ilmoniemi R (1997) Signal-space projection method for separating MEG or EEG in to components. Med Biol Eng Comput 35: Victor M, Adam RA, Collins GH (1989) The Wemicke-Korsakofl Syndrome and Related Disorders Due to Alcoholism and Malnuhition. FA Davis, Philadelphia. Wang GJ, Volkow ND, Roque CT, Cestaro VL, Hitzemann RJ, Cantos EL, Levy AV, Dhawan AP (1993) Functional importance of ventricular enlargement and cortical atrophy in healthy subjects and alcoholics as assessed with PET, MR imaging, and neuropsychologic testing. Radiology Wechsler D (1945) A standardized memory scale for clinical use. J Psycho Winkler I, Naatanen R (1995) The effects of auditory backward masking on event-related brain potentials. Electroencephalogr Clin Neurophysiol 44(Suppl.): Winkler I, Reinikainen K, Naatinen R (1993) Event-related brain potentials reflect traces of echoic memory in humans. Perception Psychophys 53: Yabe H, Tervaniemi M, Reinikainen K, Naatanen R (1997) Temporal window of integration revealed by MMN to sound omission. Neuroreport 8: