Negative symptoms and EEG alpha activity in schizophrenic patients

Schizophrenia Research, 8 (1992)1 I-20 (0 1992 Elsevier Science Publishers B.V. All rights reserved

SCHIZO

11 0920-9964/92/$05.00

00248

Negative symptoms and EEG alpha activity in schizophrenic patients Edward L. Merrin and Thomas C. Floyd The Ps_vchiarry Service, San Francisco

(Received

VA Medicul

Center and lhe Department Francisco. CA, USA

29 July 1991; revision

received

13 April

of Ps.vchiatr_v, University

1992; accepted

20 April

of California

at San

1992)

Quantitative analyses of electroencephalographic (EEG) recordings in schizophrenic patients have often demonstrated reduced alpha band (S-13 Hz) activity. However, this finding is not universal and there is some evidence that subgroups of schizophrenics may differ in overall or lateralized levels of EEG alpha activity. To investigate this issue, the authors examined relationships between clinical ratings performed at the time of EEG recording and resting alpha power and coherence in 14 medication free schizophrenic patients. Nine channels of previously recorded resting (eyes open) EEG were transformed to average reference prior to spectral analysis and transformed to source derivation prior to calculation of interelectrode coherences. Patients were rated with the Brief Psychiatric Rating Scale (BPRS), from which subscales corresponding to negative symptoms, positive symptoms, paranoia, and anxiety/depression were derived. Ratings and EEG measures were also obtained on 10 of the schizophrenic patients after neuroleptic treatment. Multiple regression with repeated measures was used to examine the influence of the subscale scores on bilateral log alpha power and both within- and between-hemisphere alpha coherence at selected locations. Prior to treatment, negative symptoms varied inversely with alpha power (p< 0.02) withinhemisphere alpha coherence (p < 0.03), and between-hemisphere coherence (p = 0.053). The effect of negative symptoms on alpha power was bilateral, but the effect on within-hemisphere coherence tended (p= 0.053) to be right-sided. After treatment these relationships were no longer present. The possible implications of and the effects of drug treatment on an association between negative symptoms and reduced alpha activity are discussed. Key words; EEG; Negative

symptoms;

Alpha;

(Schizophrenia)

INTRODUCTION

Many investigators have described abnormalities in the alpha frequency band (approximately 8-13 Hz) of the electroencephalogram (EEG) in schizophrenic patients (Itil, 1979). These usually consist of reductions in the absolute amplitude of alpha activity or its relative proportion of total EEG activity. Since alpha activity is reduced under Correspondence to: EL. Merrin, VA Medical Center, 4150 Clement 94121, USA.

Psychiatry Service (116N), Street, San Francisco, CA

conditions of cortical arousal or activation (Ray and Cole, 1985), these and other related findings in the EEGs of schizophrenics have been interpreted as evidence of a state of sustained hyperarousal (Shagass et al., 1982). However, a closer scrutiny of the literature reveals that the relationship between alpha activity and schizophrenia is not entirely straightforward. For example, some studies have reported normal or increased resting alpha in schizophrenics (Lifshitz and Gradijan, 1974; Salamon and Post, 1965; Volavka et al., 1966). In addition, a number of strategies used to study reactivity of alpha to changes in exogenous conditions have yielded inconsistent results. These

12

strategies have included changing from prone to sitting positions (Goldstein et al., 1965) eye opening (Blum, 1957; Koukkou, 1980) comparing resting and task conditions (Colombo et al., 1989) and exposure to orienting or verbal stimuli (Bernstein et al., 1981; Koukkou, 1980; Krooth, 1971; Salamon and Post, 1965). Findings have included decreased (Blum, 1957; Colombo et al., 1989; Koukkou, 1980; Salamon and Post, 1965) increased (Davis et al., 1981; Krooth, 1971), or normal (Bernstein et al., 1981) levels of alpha reactivity in schizophrenics. Although methodological differences might explain some of these contradictions, we considered the possibility that alpha activity might vary with specific clinical features of schizophrenia. Various conflicting findings have been previously reported here also. Increased chronicity of illness has been associated with both higher (Abenson, 1970) and lower (Fenton et al.. 1980) levels of alpha activity in schizophrenics. Hebephrenic patients have been found to have more alpha than paranoid schizophrenics (Etevenon et al., 1979) but alpha reactivity was reported to be reduced in undifferentiated or disorganized patients (Colombo et al., 1989). Various relationships between clinical variables and alpha lateralization have also been reported (Etevenon, 1984; Gaebel and Ulrich. 1988). To pursue this question further we examined the relationships between clinical ratings performed at the time of EEG recording and measures of resting alpha power and coherence in a sample of medication-free hospitalized schizophrenic patients.

METHODS Subjects The subjects were 14 newly admitted male veterans who agreed to participate in a previously reported study of EEG lateralization in schizophrenics, affective disorders, and normal controls (Merrin and Floyd, 1991). All had been unmedicated at least two weeks and free of significant neurological or medical illness or recent clinically significant substance abuse. All met DSM-III criteria for schizophrenia or chronic schizoaffective disorder after interview with the SADS (Spitzer and Endicott, 1978) and a supplemental record review by

the primary author (E.L.M.). The resulting sample, aged 25-67 yr (43 I 14) included disorganized (2), paranoid (5) undifferentiated (6) DSM-III subtypes and I schizoaffective, manic. Nine of these patients had never married and none had obtained college degrees. Prior to admission, 2 had never been treated with neuroleptics, one had been medication-free for 2 yr, two for 1 yr, two for 2 months, and four for at least 1 month. The remaining five had been medication-free for 222.5 wk. One patient (subject 8 in Table 3) who had been medicationfree for 1 month received two 5 mg oral doses of haloperidol 72 h before his initial EEG session; all other subjects were studied prior to any psychotropic drug administration. A subset (n= 10) of these patients was restudied after 9-35 (24f8.4) days of treatment with the equivalent (Davis, 1976) of 533.7k523.7 mg of chlorpromazine. A variety of neuroleptic agents were employed, including haloperidol (4) chlorpromazine (2) perphenazine (1) loxapine (1) thioridizine, mesoridizine (1) and thiothixene (1). Most patients were also treated with anticholinergic antiparkinsonian agents but no other psychotropic drugs were used. The protocol was approved through the institutional review boards of both the University of California, San Francisco and VA Medical Center, San Francisco. Written informed consent was obtained prior to any experimental procedures. Clinical ratings Within 24 hours of each recording session, patients were interviewed and rated with the Brief Psychiatric Rating Scale (BPRS). Four subscales (Overall. 1974) were constructed from these ratings: (1) Positive symptoms (hallucinatory behavior, unusual thought content, conceptual disorganization); (2) Negative symptoms (motor retardation, blunted affect, emotional withdrawal); (3) Paranoia (hostility, suspiciousness, uncooperativeness); and (4) Anxiety/Depression (somatic concern, anxiety, guilt feelings, depressive mood). In addition, 10 subjects received a second clinical rating with the BPRS within 24 h of a posttreatment EEG session. EEG recording and analysis Complete recording

details of EEG recording methods and conditions were described earlier (Merrin

13

and Floyd, 1991); only eyes open resting EEG was used for this analysis. EEG data were recorded from 9 channels (including bilateral frontal, central, temporal, and parietal leads, and Cz or Fz) referenced to Fz (n= 12) or Cz (n = 2), with leads at Fpl and the right outer canthus for monitoring eye movements with the electrooculogram (EOG). Signals were filtered at 1 and 35 Hz and digitized at 64 samples/s. Artifact screening involved a two-step process. To eliminate gross artifacts, I-s epochs with EOG signals greater than 50 uV peak to peak were automatically discarded during data acquisition, which continued until 60 s (except for 30 s in the case of one subject) of EEG were obtained. The digitized EEG was later displayed second by second for each lead and screened for EOG, movement, muscle, and other artifacts that had passed through the automatic screening. The rater (T.C.F.) was blind to the identity and diagnosis of the subject, recording condition, and medication status. An epoch rejected from any lead disqualified that entire second of EEG record from further analysis. These procedures resulted in 39.419.7 s (range 22-55) of artifact-free EEG selected for analysis per subject. The final posttreatment EEG dataset consisted of 41.4+ 10.0 s (range 29-57) per subject. To minimize the influence of the reference site on spectral topography and coherence measures, raw EEG was algebraically transformed from its original common reference. For analysis of alpha power, the data were transformed to average reference (Walter et al., 1984), yielding an identical set of 10 channels of EEG data for all 14 subjects. A different procedure was followed for analysis of EEG coherence. Since EEG coherence indexes covariance by frequency between two channels over time (Shaw, 1981) a shared active reference, including an average reference, may inflate calculated coherence values because the contribution of activity from the reference is common to both signals (Fein et al., 1988). Although bipolar EEG recordings (Ford et al., 1986; Merrin et al., 1989) avoid this problem, their neurophysiological significance is less clear. An alternate solution is to calculate coherence after transforming the EEG to source derivation, which approximates current amplitude entering the scalp perpendicularly at each electrode site while minimizing contributions from neighboring scalp regions (Nunez, 198 1). We performed source derivation after the method

described by Hjorth (Hjorth, 1980), which essentially refers each electrode to a weighted (by distance) average of all surrounding electrodes. Although the accuracy of the source derivation estimates of amplitude we obtained are probably limited due to the small size of the recording montage, for the purpose of calculating coherence they provided spectral estimates relatively free of shared reference activity. EEG transformed to average reference or source derivation was subjected to Fast Fourier Transform for 0.5 Hz bins from O-32 Hz as described elsewhere (Merrin and Floyd, 199 1). Average reference EEG amplitudes between 7.5 and 12.5 Hz were summed into an alpha band and converted to log power. Mean coherence over the same frequency range was computed from source derivation EEG between selected electrode pairs. These included 12 within-hemisphere pairs (six corresponding right and left hemisphere homologous pairs) as well as 16 possible pairings across hemispheres (Fz and Cz were omitted). In order to reduce the degrees of freedom the analysis of between-hemisphere coherence was limited to four representing corresponding homologous pairs electrodes. The coherence pairs studied are listed in Table 1. Raw coherence values were transformed to Fisher’s z prior to analysis to permit the use of parametric statistics. Data analysis

Multivariate Analysis COVA) with repeated TABLE Listing

of Covariance (MANmeasures was used to test

I OJ coherence pairs Between

Wilhin

Wi/hin

hemisphere

hemisphere

left

right

F3C3

F4C4 F4T4 C4T4 F4P4 C4P4 T4P4

F3T3 C3T3 F3P3 C3P3 T3P3 F3 = Left C3 = Left T3 = Left P3 = Left

Frontal Central Temporal Parietal

F4 = C4= T4 = P4 =

F3F4 c3c4 T3T4 P3P4

Right Right Right Right

Frontal Central Temporal Parietal

hemisphere

14

for effects of the BPRS subscale measures on log alpha power, within-hemisphere alpha coherence, and between-hemisphere alpha coherence. A separate MANCOVA was performed on each independent variable with the four subscales entered as covariables. Within-subjects factors included: (1) for log alpha power, leadpair (4 levels: frontal, central, temporal, parietal) and side (2 levels: left, right), (2) for within-hemisphere coherence, electrode pair (6 levels) and side (2 levels: left, right), and (3) for between-hemisphere coherence, electrode pair (4 levels). When univariate and multivariate results differed, the Wilks’ Lambda F-statistic was used. Since an initial analysis suggested that several interactions between covariables approached significance, separate MANOVAs were also performed with each subscale entered as a single covariable. The results were essentially identical and for simplicity only the results obtained from using the four covariables together are described below.

RESULTS

The main effects of each of the 4 BPRS subscales on alpha power and coherence are summarized in Table 2. There were significant main effects of negative symptoms on log alpha power (F(l,9)= 8.75, p = 0.016) and within-hemisphere alpha coherence (F(1,9) = 7.18, p= 0.025) and a trend for a similar effect on between-hemisphere coherence (F( lq9) = 4.95, p = 0.053). The nature of these effects is demonstrated in Figs. 1-3, where negative symptoms are plotted against log alpha power and coherence in representative leads or coherence respectively; negative symptoms varied pairs,

TABLE

BPRS

inversely with each of the 3 measures of alpha activity. There were no main effects of any of the other three subscales on alpha activity. Closer examination of these analyses revealed a significant interaction between negative symptoms and leadpair (F(3,7) = 5.52, p = 0.02) on log alpha power. To identify the nature of this interaction we conducted separate multivariate analyses for each leadpair with log alpha power as the dependent variable, the four subscales as covariables, and side as the within-subjects factor. The effect of negative symptoms on log alpha power was stronger in central (F(1,9)= 12.01, p=O.O07) and parietal (F(1,9) = 11.57, p = 0.008) leadpairs than in frontal (F(1,9)=6.17, ~=0.035) or temporal (F( 1,9) = 4.41, p = 0.065) leadpairs. A marginally significant interaction (negative in symptoms x electrode pair x side) withinhemisphere alpha coherence (F(5,5) = 4.90, p = 0.053) suggested the presence of a lateralized effect. To identify the source of this interaction we performed separate linear regressions of negative symptoms on alpha coherence between each of the 12 within-hemisphere coherence electrode pairs. The only significant findings were all on the right side; in F4P4 (F(1,12)=4.83, p=O.O48) and C4P4 (F(1,12)= 6.17, p=O.O29), with a trend in F4T4 (p = 0.051). Tn addition, F values were larger on the right side than on the left in all electrode pairs except C3T3/C4T4 and T3P3/T4P4. Since age may be related to changes in EEG spectral power (Matousek et al., 1967) we investigated the relationships between age, negative symptoms, and alpha activity. Prior to treatment, age was not correlated with negative symptoms (r= 0.19;p = 0.7), left parietal log alpha power (r= 0.23; p = 0.4) or C4P4 alpha coherence (r= - 0.15; p = 0.6). To further evaluate the possible effects of

2

subscale

Negative symptoms Positive symptoms Paranoia Anxiety/Depr.

Log dpha po~vr

Wiihin-henzisphrre Alpha coherencc~

Be/~i’rm-hrmisphf,r~ Alpha coherence

F( I.91

P

F(1.9)

P

F( 1-9)

P

8.75

0.02 0.34 0.17 0.24

7.18 0.19 I .58 0.63

0.03 0.68 0.24 0.45

4.95 0.54 0.37 0.85

0.053 0.48 0.56 0.38

I .oo 2.23 1.55

15

. p=O.Ol

4

01 0

1

5

I

10

15

I

I 20

-I

o.40~

20

Negative Symptoms Negative Symptoms

Fig. 3. Figs. 1-3. Log alpha power (Fig. I) and both within- (Fig. 2) and between- (Fig. 3) hemisphere alpha coherence from representative locations recorded during an eyes open resting condition are plotted against the negative symptom score obtained from the Brief Psychiatric Rating Scale in 14 medication free schizophrenic inpatients. These figures demonstrate inverse relationships between negative symptoms and left parietal alpha power (E(l,12)= 9.41, p=O.Ol), coherence between alpha in C4 (right central) and P4 (right parietal) leads (F(1,12)=6.17, p
Fig. 1.

‘.2r--’

0.2 1 0

, 5

, IO

/ 15

20

Negative symptoms Fig. 2.

age on the relationship between negative symptoms and alpha power, MANOVAs similar to those above were performed on pretreatment log alpha power and within-hemispheric alpha coherence as dependent variables, with age and negative symptoms as the covariables. There was a weak trend (PC 0.2) for a main effect of age on log alpha power, but the main effect of negative symptoms was still significant (F(l,l l)= 10.91, p=O.O07). Similarly, age had no effect on within-hemispheric coherence (p =0.6), but a significant main effect

of negative symptoms (F( 1, I 1) = 8.36, p < 0.02) and a significant negative symptoms x side x pair interaction (F(5,7) = 7.03, p < 0.02) were again present. Thus, age does not appear to have made a significant contribution to these effects. We considered the possibility that the assessment of pretreatment negative symptoms may have been confounded by lingering extrapyramidal symptoms (EPS) such as masked facies and bradykinesia. Although the patients had been free of medication, drug-induced parkinsonism can persist after drug discontinuation. Unfortunately, quantitative assessments of EPS were not available on these subjects; however, it was possible to investigate the relationship between the initial medication-free interval (see Table 3) and negative symptoms. For this purpose a rank (Spearman) correlation was performed between the days off medication prior to study and the pretreatment negative symptom score. The medication-free periods of the two drugnaive subjects were set at 1000 days and that of

of medication

14 days I8 days

60 days 2 years

30 days

48 64

42 41

27

10 I1

I2 13

14

5

I5 I6

3 3

IO 5

8

9

12 3

7

3 8 10 9 _ _

Neuroleptic

21 _ 13

338 354

+ Posttreatment)

9 35

300 215

* 100.

chlorpromazine

60.0

~ 40.0

300.0 0.0

~ 30.0

~ 16.7 - 40.0 166.7 -9.1 12.5

symptoms

in negative

54 Change

changes in negative

loxepine thioridizine, mesoridizine chlorpromazine

haloperidol _

_

29 _

_

500

perphenazine haloperidol haloperidol thiothixene haloperidol

33

used

28 20 29 23

of

and associated

500 450 400 2000 _ _

(days)

treatmenl

Duration

treatment

280

Percent change was computed from (Pretreatment-Posttreatment/Pretreatment *Chlorpromazine equivalents in milligrams [Davis, 1976 # 12201.

30 days I8 days

59 35

8 9

5 3 11 8 15 9

10

Score

score

I2

dose*

symp f om

symptom

38 days drug naive I4 days 60 days I4 days 30 days drug naive

neuroleptic

negatiw

Pretx

negative

mean dail)

und dctuils

Medication P0Sit.X

subjects

free period

28 33 67 25 46 51

33

Age

1 2 3 4 5 6 7

Suhjeci

in schizophrenic

negative symptoms,

period,

Age, length of medication-free

power, C4P4 alpha coherence)

TABLE 3

-2.5 9.9 - 12.6 - 23.8 _ 14.7

22.0 30.0 303.7 _ 30.8

_

226.9 _

_

4.1 0.1 0.5 - 14.0 - 15.4 _ _

(C4P4j

coherence

in alpha

295.0 69.4 65.0 3.3 74.6

iP3)

power

in alpha

% Change

und alpha activity

% Chunge

symptoms

(kft

purietal

alpha

17

the subject with a 2-year interval was set at 700 days. If negative symptom scores reflected the presence of lingering EPS the correlation should be negative. Instead, the correlation was significant in the other direction (r= +0.59; ~~0.03). The source of this correlation is apparent from Table 3, where subjects with the higher negative symptom scores tended to be those with longer (rather than shorter) medication-free periods. Although the reason of this correlation, if real, is not apparent, the results make it unlikely that pretreatment negative symptom ratings were contaminated by lingering medication side effects such as EPS. There were no other findings suggesting relationships between BPRS subscales and location or side of power or coherence measures. Post treatment data After treatment, BPRS ratings in 10 patients were significantly decreased (paired two-tailed t-tests; 47.9+6.6vs. 32.824.3; t=5.724,df=9,p<0.001). Nine of these 10 patients had at least 20% reductions in total BPRS scores. Of the four subscales, the positive symptom score was significantly reduced (15.6k3.5 vs. 8.413.1; t= 10.885, df=9, p
clear from the table that the lack of a significant group change in negative symptoms for the total patient sample was partly a function of considerable heterogeneity in the sample. Negative symptoms were relatively unchanged in some patients, decreased in others, and considerably worse in several others after treatment. In contrast, a consistent pattern of increased left parietal log alpha power across subjects resulted in a significant difference between treatment conditions (paired two-tailed t-test, t = - 4.492, df = 9, p = 0.002). The percent change in negative symptoms was not correlated with the dose of medication used (Pearson r = - 0.12, p = 0.8) although it was negatively but nonsignificantly correlated with changes in alpha power (Y= - 0.45, p = 0.2). However, there was a significant negative correlation between percent change in negative symptoms and the duration of medication treatment (r= - 0.74, ~~0.02). In fact, the three patients with the largest increase in negative symptoms were treated for the shortest time periods. Thus, it is possible that the negative symptom ratings in those 3 patients were influenced by other factors such as drug-induced sedation or bradykinesia which the other subjects had accommodated to after longer treatment periods. There was no consistent pattern of change in C4P4 alpha coherence between the two recording sessions (t= -0.56, df=9, p=O.6) and the changes in coherence were not correlated significantly with changes in negative symptoms (r = 0.52, p = 0.12).

DISCUSSION

In this sample of medication-free schizophrenic patients, resting (eyes open) alpha power and coherence varied inversely with the ratings of a three item subscale of the BPRS used as a measure of negative symptoms of schizophrenia. This relationship was unique to negative symptoms, as neither measure of alpha activity varied with any of the other clinical measures derived from the BPRS, including positive symptoms, suspiciousness (paranoia), and anxiety/depression. Analysis of similar data obtained from ten of the patients after a period of exposure to neuroleptic medication indicated that similar relationships between these clinical ratings and alpha activity were no

18

longer present. A closer examination of the data suggested that changes in alpha activity and negative symptom ratings after treatment varied independently both within and across subjects. Although the findings were not consistent with any specific relationships between clinical ratings and localized or lateralized patterns of alpha power, they did suggest that negative symptoms were associated with reduced alpha coherence over the right hemisphere. Evidence presented elsewhere is compatible with an association between negative symptoms and right-sided dysfunction and between positive symptoms and left-sided dysfunction (Green and Walker, 198.5; Gruzelier, 1984; McCarley et al., 1989). Although it is possible that these data reflect this same pattern, it is still uncertain what specific brain events variations in EEG coherence represent. Coherence is independent of actual amplitude (Shaw, 1981) and may either increase or decrease, depending on the study, with cortical activation. In our laboratory coherence values have declined across frequency bands with task activity, whereas reductions in spectral power have been primarily limited to alpha (Merrin et al., 1989). Although this would immediately suggest an association between negative symptoms and right-sided activation in this subject sample, it is not supported by the lack of a corresponding pattern of right-sided alpha power reduction. An alternative hypothesis is that lower right-sided coherence in patients with negative symptoms resulted from a more diffuse organizational pattern (hence lower coherence) in their right cerebral hemispheres. This hypothesis would be supported by the presence of a similar relationship during eyes-closed recordings, which were not performed on these subjects. Our findings with alpha power conflict with those reported by at least two other laboratories. Guenther et al. (1988) reported higher alpha in Type II (predominantly negative symptom) schizophrenics than in normal controls, but not in Type I (positive symptom) patients. However, the schizophrenic patients in their study were already medicated. Itil et al. (1975) found that patients with the apparent equivalent of negative symptoms had higher alpha power than other schizophrenics prior to treatment as well as a less satisfactory treatment response. Their study procedures differed in several important ways from ours; specifically, they

recorded eyes-closed rather than eyes-open EEG and processed their data with period analysis rather than FFT. Period analysis measures the proportion of time spent in a given frequency in the EEG record rather than the absolute amplitude of signal in that frequency range. Although these measures may be correlated, they do reflect separate phenomena. Finally, the Itil et al. study was conducted prior to the use of DSM-III diagnostic criteria and may have included patients with other diagnoses. Whether these or other factors are responsible for the differing results is indeterminable at present. Reductions in alpha activity in schizophrenia have often been interpreted in terms of increased arousal or tension, but the lack of a relationship in these data between alpha activity and ratings of frank psychotic symptoms or anxiety seems to argue against this unless negative symptoms are accompanied by a form of hyperarousal less accessible to direct observation. Alpha activity may also be a physical trait, reflecting specific aspects of neurological organization in a given individual. For example, some EEG spectral characteristics, particularly in the alpha band, are almost identical between members of monozygotic but not dizygotic twin pairs (Lykken et al., 1974). More recently, Karson et al. (1988) suggested that lower alpha frequency in schizophrenic patients may be associated with ventricular atrophy. The possibility exists, then, that anomalies of various measures of alpha activity associated with negative symptoms in schizophrenic patients represent persistent (when not altered by drug therapy) central nervous system traits that predate the onset of illness. Like enlarged ventricles or neuropsychological deficits, they may provide further evidence that the negative symptom cluster signals the presence of a profound neurointegrative disorder. A degree of caution is called for when considering the results of this study. The findings are essentially post facto, since no specific hypothesis regarding negative symptoms and EEG activity was entertained during data collection or prior to data analysis. Although the major focus was on activity in the alpha frequency band, the conditions of EEG recording (eyes open) are not ideal for that purpose. An alternative, although not necessarily very compelling, explanation for the results

19

is that patients with more negative symptoms block more alpha when they open their eyes. An examination of the relationship between negative symptoms and the alpha component from eyes closed EEG would rule this out. Towards this end, a repeat study using eyes-closed EEG with a new sample of schizophrenic patients is currently underway.

ACKNOWLEDGEMENT

This research was supported by the Department of Veterans Affairs Medical Research Program.

REFERENCES

Abenson, M.H. (1970) EEGs in chronic schizophrenia. Brit. J. Psychiat. 116, 421-425. Bernstein, A.S., Taylor, K.W., Starkey, P., et al. (1981) Bilateral skin conductance, finger pulse volume, and EEG orienting response to tones of differing intensities in chronic schizophrenics and controls. J. Nervous Mental Dis 169, 5133528. Blum, R.H. (1957) Alpha-rhythm responsiveness in normal, schizophrenic, and brain-damaged persons. Science 126, 7499750. Colombo, C., Gambini. O., Macciardi, F., et al. (1989) Alpha reactivity in schizophrenia and in schizophrenia spectrum disorders: demographic, clinical and hemispheric assessment. Int. J. Psychophysiol. 7, 47-54. Davis, J.M. (1976) Comparative doses and costs of antipsychotic medication. Arch. Gen. Psych. 33, 858861. Davis, J.R., Glaros, A.G. and Davidson, G.S. (1981) Visual information processing and alpha blocking in schizophrenics and normals. J. Psychiat. Res. 16, 955102. Etevenon, P. (1984) Intra and inter-hemispheric changes in alpha intensities in EEGs of schizophrenic patients versus matched controls. Biol. Psychol. 19, 247-256. Etevenon, P., Pidoux, P., Rioux, P., et al. (1979) Intra- and interhemispheric EEG differences quantified by spectral analysis: comparative study of two groups of schizophrenics and a control group. Acta Psychiatr. Stand. 60, 57-68. Fein, G., Raz. J., Brown, F., et al. (1988) Common reference coherence data is confounded by power and phase effects. Electroencephalog. Clin. Neurophysiol. 69, 581-584. Fenton, G.W., Fenwick, P.B.C., Dollimore, J., et al. (1980) EEG spectral analysis in schizophrenia. Brit. J. Psychiat. 136, 4455455. Ford, M.R., Goethe, J.W. and Dekker, D.K. (1986) EEG coherence and power in the discrimination of psychiatric disorders and medication effects. Biol. Psychiat. 21, 11751188.

Gaebel, W. and Ulrich, G. (1988) Topographical distribution of absolute alpha-power in the EEG and psychopathology in schizophrenic outpatients. Acta Psychiatr. &and. 77, 390-397. Goldstein, L., Sugerman, A.A., Stolberg, H., et al. (1965) Electra-cerebral activity in schizophrenics and non-psychotic quantitative EEG amplitude analysis. subjects: Electroencephalog. Clin. Neurophysiol. 19, 350-361. Green, M. and Walker, E. (1985) Neuropsychological performance and positive and negative symptoms in schizophrenia. J. Abnormal Psycho]. 94, 460-469. Gruzelier, J.H. (1984) Hemispheric imbalances in schizophrenia. Int. J. Psychophysiol. 1, 227-224. Guenther, W., Davous, P., Godet. J.-L., et al. (1988) Bilateral brain dysfunction during motor activation in type II schizophrenia measured by EEG mapping. Biol. Psychiat. 23, 2955311. Hjorth, B. (1980) Source derivation simplifies topographical EEG interpretation. Am. J. EEG Technol. 20, 121-132. Itil, T.M. (1979) Computer analyzed EEG findings in adult schizophrenics, psychotic children and ‘high risk’ children for schizophrenia. In: Obiols, J., Ballus, C., Gonzalez Monclus, E. et al. (Eds), Biological Psychiatry Today, ElsevieriNorth Holland Biomedical Press, Amsterdam, pp. 123991308. Itil, T.M., Marasa, J., Saletu, B., et al. (1975) Computerized EEG: Predictor of outcome in schizophrenia. J. Nervous Mental Dis. 160, 188-203. Karson, C.N., Coppola, R. and Daniel, D.G. (1988) Alpha frequency in schizophrenia: An association with enlarged cerebral ventricles. Am. J. Psychiat. 145, 861-864. Koukkou, M. (1980) EEG correlates of information processing in acute schizophrenics. Adv. Biol. Psychiat. 4, 102-l 10. Krooth, D.M. (1971) Cortical arousal in paranoid schizophrenics as a function of a word association task. J. Nervous Mental Dis. 153, 366-376. Lifshitz, K. and Gradijan, J. (1974) Spectral evaluation of the electroencephalogram: Power and variability in chronic schizophrenics and control subjects. Psychophysiology 11, 4799490. Lykken, D.T., Tellegen, A. and Thorkelson, K. (1974) Genetic determination of EEG frequency spectra. Biol. Psychol. 1, 2455259. Matousek, M., Volavka, J., Roubicek, J., et al. (1967) EEG frequency analysis related to age in normal adults. Electroencephalog. Clin. Neurophysiol. 23, 162-167. McCarley, R.W., Faux, S.F., Shenton, M., et al. (1989) CT abnormalities in schizophrenia: A preliminary study of their correlations with P3OO/P200 electrophysiological features and positive/negative symptoms. Arch. Gen. Psychiat. 46, 698-708. Merrin, E.L. and Floyd, T.C. (1991) Average reference EEG lateralization in schizophrenic patients. J. Neuropsych. Clin. Neurosci. 3, 307-314. Merrin, E.L., Floyd, T.C. and Fein, G. (1989) EEG coherence in unmedicated schizophrenic patients. Biol. Psychiat. 25, 60-66. Nunez, P. (198 1) Electrical Fields of the Brain: The Neurophysits of EEG, Oxford University Press, New York. Overall, J.E. (1974) The Brief Psychiatric Rating Scale in

20 psychopharmacology research. Mod. Probl. Pharmacopsychiat. 7, 67-78. Ray, W.J. and Cole, H.W. (1985) EEG alpha activity reflects attentional demands, and beta activity reflects emotional and cognitive processes. Science 228, 750-752. Salamon, I. and Post, J. (1965) Alpha blocking and schizophrenia. Arch. Gen. Psychiat. 13, 3677374. Shagass, C., Roemer. R.A. and Straumanis, J.J. (1982) Relationships between psychiatric diagnosis and some quantitative EEG variables. Arch. Gen. Psychiat. 39, 142331435. Shaw, J.C. (1981) An introduction to the coherence function

and its use in EEG signal analysis. J. Med. Eng. Technol. 5, 2799288. Spitzer, R.L. and Endicott. J. (1978) Schedule for Affective Disorders and Schizophrenia, Biometrics Research, New York State Psychiatric Institute, New York. Volavka, J., Matousek, M. and Roubicek, J. (1966) EEG frequency analysis in schizophrenia: An attempt to reconsider the role of age. Acta Psychiatr. Stand. 42, 237-245. Walter, D.O., Etevenon, P., Pidoux, B., et al. (1984) Computerized topo-EEG spectral maps: Difficulties and perspectives. Neuropsychobiology 1I, 264-272.