Ž .International Journal of Psychophysiology 25 1997 227]235

A follow-up study of EEG coherences in children withdifferent pedagogical evaluations

Erzsebet MarosiU, Thalıa Harmony, Alfonso Reyes, Jorge Bernal, Thalıa´ ´ ´Fernandez, Vicente Guerrero, Mario Rodrıguez, Juan Silva,´ ´

Guillermina Yanez, Hector Rodrıguez´˜ ´Department of Neurosciences, National Uni®ersity of Mexico, ENEP Iztacala, Mexico, Mexico

Received 10 September 1996; revised 15 November 1996; accepted 15 November 1996

Abstract

A follow-up study on EEG coherence values was carried out with 46 school-agechildren divided into threeŽpedagogical groups considering their reading and writing abilities normal subject and children with mild and with

.serious reading]writing problems . EEGs were recorded at rest, with eyes closed in 15 referential derivations thatgave 105 possible coherence values. Coherence was calculated after the elimination of the average reference.Repeated Measure ANOVAs were calculated to evaluate the effects of time, sex, reading]writing problems and theirinteractions. The second EEG recordings were taken 2]3 years after the first. The results revealed significantincrease of coherence values during the time lapse studied, consisting of an increase of frontal coherences in allbands, with the exception of the theta band. The most numerous changes occurred in the alpha band. Very fewdifferences in coherence values were observed between the two EEG recordings by pedagogical groups and by sexsuggesting a similar but not igual maturation of coherences of boys and girls and of the three pedagogical groups, inthis age range studied. These findings suggests an abnormal coherence development for the reading disabled group,as well as different coherence maturation for boys and girls in this time lapse studied. Q 1997 Elsevier Science B.V.

Keywords: EEG coherences; Pedagogical evaluation; EEG frequency analysis; Sex differences

1. Introduction

Coherence is defined as a spectral correlationbetween two time series. A value of 1.0 in fre-

U Corresponding author. Apartado Postal 82 Atizapan de´Zaragoza Estado de Mexico C.P. 52971 Mexico´

quency f indicates that the power in frequency fof the first signal can be completely predicted bythe power of the second one. If signals of twoEEG channels are interrelated over a particularfrequency band, that fact can be detected by thepresence of coherences that are significantly dif-ferent from zero. The coherence between each ofthe channel pairs is an indicator of the linear

0167-8760r97r$17.00 Q 1997 Elsevier Science B.V. All rights reserved.Ž .P I I S 0 1 6 7 - 8 7 6 0 9 6 0 0 7 4 5 -3

( )E. Marosi et al. r International Journal of Psychophysiology 25 1997 227]235228

dependency or correlatedness between the corre-Ž .sponding time series Gersch, 1987 and is related

to structural and functional interrelation. Coher-ences have been used to study brain development

Ž .and cognitive processing Tucker et al., 1985 .Based on previous works by Thatcher and WalkerŽ . Ž .1985 and Gasser et al. 1987 , we know thatEEG coherences are inversely related to intelli-gence and increase with age.

In an earlier paper we demonstrated a directrelationship between coherence values and ageŽ .Marosi et al., 1992 with higher significance inthe delta and alpha bands. Age-related coherencechanges were found to be increase between pos-terior regions and vertex and decrease of thetacoherences between frontal areas, in the controlgroup. In groups with reading]writing problemsage effect on coherences was observed mainlybetween left temporal conections. We proposedqualitative differences in brain maturation thatmay be monitored more precisely with longitudi-nal EEG studies.

Reading disability is frequently related to brainlesions or other biological factors, although itsetiology has not been clearly understood. In aprevious study we proposed that children withborderline or poor reading]writing abilities mayhave a different brain organization, at least at theleft temporal cortical region and this fact can bedemonstrated by studying their EEG coherencesŽ .Marosi et al., 1995 . The suggestion that lefttemporal regions may play an important role inlanguage related disorders, sounds logical, iswidely accepted and supported not only by elec-trophysiological but also by neuropsychological

Ž .studies. McCarthy et al. 1990 suggested thatboth spelling and writing impairment are typicallyassociated with the posterior region of the lefthemisphere. But there exists other theories to

Ž .explain reading poblems: Harmony et al. 1995proposed a developmental lag in the maturationof the reading disabled group and Galaburda et

Ž .al. 1985 suggested emotional problems causedby cortical anomalies as an important factor inreading disability.

We suppose that our subjects also shall havesex-related differences in the coherence values.We reported sex dependent right hemispheric

differences in coherence values in an earlier studyŽ .Marosi et al., 1993 , so we consider that the agedetermined changes in the coherences values mustbe related to sex, too. Differences in EEG coher-ences by sex were also reported by Beaumont et

Ž . Ž .al. 1978 and Corsi-Cabrera et al. 1989 .The goals of this paper are multiple. In part it

seeks to determine the changes in electrical activ-ity of the brain during maturation through alongitudinal examination of EEG coherence val-ues. On the other hand, it has the purpose ofdemonstrate differences in brain maturationbetween the normal and the reading disabledchildren. We expect that the differences observedin brain organization through time may shed lighton the nature of reading disability suggestingwhether brain lesion or maturational problemsare involved in reading impairment. At last, thisstudy could reveal sex dependent differences inbrain maturation and show the role of sex inreading disability.

We propose to obtain the following findings:

v An increase of coherence values between thetwo recording periods at posterior regions.

v Higher increase of coherences for girls thanfor boys, at right hemispheric interconnec-tions.

v If reading problems are related to brain lesionor funtional disability of cortical languageareas, we expect to find pedagogical groupdifferences in the left temporal areas.

v If reading problems are due to a maturationallag, we expect a higher increase of coherencesbetween the two recording periods in the groupŽ .1 good readers , than in the other two groups.

v If emotional problems caused by corticalanomalies are responsible for the reading dis-ability then, the differences between groupsshould be observed at frontal areas.

2. Material and methods

2.1. Subjects

This study was carried out with 46 children intwo recording periods, the second performed2.5]3.1 years after the first. All subjects came

( )E. Marosi et al. r International Journal of Psychophysiology 25 1997 227]235 229

from a public school. A pedagogical test de-Ž .scribed by Harmony et al. 1990 applied to all

subjects at the time of the first and secondrecordings, considered reading]writing achieve-ment as compared to the requirements of theirschool grade. Based on test performance on vari-ous subitems on reading and writing } exploringoral reading, reading comprehension, copying,dictation and functional writing } subjects wereclassified into three different groups. Group 1consisted of children with good performance, ac-cording to their age and grade. Group 2 consistedof subjects having below-level performance withminor difficulties. The children in Groups 1 and 2had no history of academic failure and had neverrepeated school years. Group 3 was comprised ofchildren having below-level performance with se-vere difficulties. The subjects of this group hadrepeated more than 2 school years and wereattended in a behavioural clinic.

Our previous studies using the sameŽreading]writing test Harmony et al., 1990, 1995;

.Marosi et al., 1992, 1995 demonstrated that thistest achieved a good separation of subjects withdifferent reading]writing abilities. To be sure thatage did not affect in a different way the coher-ences of the three pedagogical groups, analyses ofvariance were performed. No significant differ-ences in age or in intervals between the two EEGrecordings were found. The distribution of thesubjects in the three performance groups andtheir ages are shown in Table 1.

All subjects were completely healthy and freeof medication with an IQ higherthan 85 as mea-

sured by the Wechsler Intelligence Scale for Chil-Ž .dren WISC . They had normal pediatric and

neurological evaluations. Children with neurolog-ical abnormalities or epileptic seizures wereeliminated from this study.

The anormalities observed in EEG recordingswere paroxistic focal spike activity, posteriorslowing and very low voltage. None of our sub-jects had generalized paroxistic bursts. We haveto stress that none of the abnormal findings indi-cated epileptic seizure activity and, other, thantheir EEGs, these children appeared neurologi-

Ž .cally healthy Table 2 .

2.2. EEG recording and analysis

The EEGs of all subjects were recorded in ashielded, dimly-lit room. The recordings weremade at rest, with eyes closed. The subjects wereawake in a sitting position with the head sup-ported by the back of a reclining chair. All EEGswere made without sedation. The recordings werevisually monitored and segments with artifactswere eliminated. Recordings were obtained from15 referential derivations with silver]silver chlo-ride electrodes located at F3, F4, C3, C4, P3, P4,O1, O2, F7, F8, T3, T4, T5, T6 and Cz in accor-dance with the International 10r20 System, withlinked earlobes as reference. All electrodeimpedances were less than 5000 V. EOG activitywas continuously monitored. The EEGs wererecorded by a multichannel MEDICID-3M com-puterized amplification system. The amplifiers hada bandwidth of 0.5]30 Hz. Twenty-four 2.56-s

Table 1Distribution of age values in pedagogical groups

Group n Sex 1st EEG 2nd EEG Mean diference

Mean SD Mean SD Mean SD

1 18 Bs11 9.68 1.5 12.14 1.28 2.46 0.76Gs7

2 17 Bs9 9.21 1.4 12.01 1.71 2.8 1.07Gs8

3 11 Bs7 9.12 1.1 12.26 0.74 3.14 0.75Gs8

( )E. Marosi et al. r International Journal of Psychophysiology 25 1997 227]235230

Table 2Results of conventional EEG examination of the sample

Results of EEG Group 1 Group 2 Group 3

ExaminationNormal EEGs 12 7 4Multifocal EGGS 1 1 1Focal spike act. 1 5 4Postetrior slowing 1 1 1Very low voltage 1 0 1

segments, consisting of a total of 61.44 s of arti-fact-free EEG, were obtained for each subject ata sampling rate of 100 Hz. As common activity atthe reference electrode causes correlated signalsŽ .Nunez, 1995 , the effect of the reference elec-trode upon coherence values was calculated and

Ž .eliminated Valdes et al., 1992 and cross-spectral´matrixes were then computed between all leadsfor each EEG segment. Average values were usedto compute coherences for the following fre-quency bands: delta 1.5]3.5 Hz; theta 3.6]7.5 Hz;

Žalpha 7.6]12.5 Hz; beta 12.6]30 Hz Alvarez et.al., 1987 . Coherence was computed for all possi-

ble pairwise combinations of the 15 leads, result-ing in 105 combinations for each frequency band.

The data obtained was Z-transformed by theFischer transformation in order to bring thecoherences closer to Gaussian distribution and touse properly the parametric statistical analysis.Z-transformed coherence data was examined tomonitor normality with the Shaphiro]Wilk

Ž .method SAS Procedure 6.03 .

2.3. Statistical analysis

Based on the Z-transformed values of coher-ence, the Repeated Measures Analyses of Vari-ance were conducted, taking into considerationthe effect of the time of EEG recording, thepedagogical category and sex, as well as theirinteractions: recording sessionrsex, recordingsessionrpedagogical category and recording ses-sionrsexrpedagogical category for each possiblecombination of coherences in the four broadbands. As we have a great number of variablesonly results with significance higher than 0.005were considered.

3. Results

The results of the Repeated Measure Analysesof Variance were obtained as follows:

Delta coherence values changed little betweenthe first and second EEG recordings. The changeswere observed at frontal interconnections, in-crease at F3-O2, F3-T6 and F4-C4 and decreaseat F3-F8.

Theta coherences decreased during this timelapse at F3-P4, F3-F7, F4-F7. We may observethat all significant changes occurred at frontalinterconnections.

Fig. 1 shows the changes of coherence valuesbetween the two EEG recordings in the alphaband, where coherences increased significantly atF3-F4, F3-O1, F3-O2, F3-Cz, F4-C4, F4-O1, F4-T5, F4-Cz, C4-F8, C4-T5, O1-Cz, O2-F7 and T5-Cz. Frontal alpha coherences reflected most ofthe recording session related changes in thecoherences values.

In the beta band, we obtained significant in-crease of coherences between the two EEGrecordings at F3-Cz, F4-C4, F4-T5, P3-T4, F8-T5,

Ž .T3-T5 and T5-Cz Fig. 2 . All these changes wereincreases and occurred at frontal and temporalderivations.

Fig. 3 shows the significant differences byrecording sessionrpedagogical category. Changesin the coherence values of the three pedagogicalgroups between the two recording periods wereobserved only at P4-O2 in the theta band and atC4-T3 in the alpha band. Theta coherences atP4-O2 decreased between the two recordings in

Ž .Group 1 normal readers and increased inŽGroups 2 and 3 children with reading]writing

.problems . In the alpha band at C4-T3 coher-ences of normal readers increased, while coher-ences of Group 2 decreased and of Group 3 didnot change during the time elapsed between thetwo EEG recordings.

Fig. 4. shows the changes in coherence valuesbetween the two recording sessions by sex andpedagogical groups. The upward bars representan increase in coherence values between the twoEEG recordings and downward bars show de-crease. The only derivation where significant

( )E. Marosi et al. r International Journal of Psychophysiology 25 1997 227]235 231

Ž . Ž .Fig. 1. Differences between the coherence values of the first diagonal lines and second gray EEG recordings in the alpha band.

Ž . Ž .Fig. 2. Differences between the coherence values of the first diagonal lines and second gray EEG recordings in the beta band.

( )E. Marosi et al. r International Journal of Psychophysiology 25 1997 227]235232

ŽFig. 3. Significant pedagogical group differences between the coherence values of the first and second EEG recordings Group 1:.diagonal lines; Group 2: black; Group 3: gray .

ŽFig. 4. Significant differences in coherence values by recording sessionrsexrpedagogical group in the beta band at C3-P4 Group 1:.diagonal lines; Group 2: black; Group 3: gray Ps0.0038.

( )E. Marosi et al. r International Journal of Psychophysiology 25 1997 227]235 233

Table 3Sex, pedagogical group and sexrpedagogical group differencesin the coherence values

Band Derivation Sex Pedagogical groups

Delta F3-F4 Boys)girls Boyss1)2)3Girls s3)2)1

Theta F4-T4 2)1)3Alpha F7-T4 Boys)girls

changes occurred was C3-P4 in the beta band. WeŽcan see that coherences of Group 1 normal

.readers increased in boys and girls alike, while inGroup 2 boys’ coherences increased and girls’coherences decreased. In Group 3 boys’ coher-ences decreased, while girls’ coherences increasedduring the time lapse observed.

The results obtained without significant sessioneffect were observed in Table 3.

Significant pedagogical grouprsex interactionwas observed in the delta band at F3-F4 whereboys had higher coherences than girls in bothgroups, and differences by pedagogical categoryshowed that boys had higher coherences in thegroup of normal readers and lowest ones in poor

Ž .readers 1)2)3 ; and girls had higher coher-ences in the group of poor readers and lowest

Ž .ones in the control group 3)2)1 . Pedagogicalgroup effect was observed at F4-T4 in the thetaband, where Group 2 had higher significantlycoherences than the other two groups and Group

Ž3 had the lowest coherences of the three 2)1).3 . Sex differences were observed at F7-T4 in the

alpha band, where boys had significantly highercoherences than girls.

4. Discussion

The correlational assemblies of the brain canbe defined, crudely, as groups of neurons, activeto the same extent in the same way at the same

Ž .time Gerstein et al., 1989 . High coherence val-ues show functional interaction between differentgroups of neurons. Increase of coherence valuesduring brain maturation was reported by various

Ž .authors Thatcher et al., 1987; Gasser et al., 1987that could consist of more numerous or moreeffective brain connections. Nevertheless, this ar-

gument is controversial, as higher coherences inmentally retarded children were observed by

Ž .Gasser et al. 1987 .Ž .In a previous paper Marosi et al., 1992 we

found a marked increase of coherences with age,especially in the delta and alpha bands. The re-cent follow-up study confirms these findingsshowing that EEG coherences mostly increasedŽ .delta, alpha and beta bands during the timeelapsed between the two recording periods. Asthe second EEG recordings were made only 2]3years later, the significant changes were not toonumerous. The greatest coherence increase wasobserved in the alpha band. The reason of thegreatest change in the alpha band may be due tothe fact that our subjects ranged in age between 7and 14 years old, when the maturation of alpha

Želectrical activity is still in process John et al.,.1980 . The coherence increase did not occur at

posterior regions, but mostly in frontal intercon-nections. This finding is supported by the results

Ž .of Goldman-Rakic 1984 who found that frontalcortex completes maturation at the second decadeof life.

In this study we also obtained a reduction infrontal theta coherences as it was observed in our

Ž .previous paper Marosi et al., 1995 that may bedue to a reduction of frontal synaptic density,though the reasons of this frontal theta coher-ence reduction are not clear.

In a previous paper we reported right hemi-spheric differences between girls and boys, having

Žgirls higher coherence values than boys Marosi.et al., 1993 . Our recent results do not reveal a

different maturation by sex groups. At least in thetime studied, boys and girls matured in the sameway. Differences without taking into considera-tion the session effect were observed at F7-T4 inthe alpha band, having boys higher coherencesthan girls. The failure to repeat the our previoussex related results may be due to various facts:our previous study was done with control chil-dren, our recent experimental group consisted ofboth normal, borderline and poor readers. Also,the level of significance was determined as p igualor higher than 0.05. In this paper the significancewas considered as igual or higher than 0.005.

( )E. Marosi et al. r International Journal of Psychophysiology 25 1997 227]235234

Moreover, we have eliminated the effect of thereference electrode, and the number of subjectsstudied was not balanced by sex, we had less girlsthan boys.

Differences considering the reading]writingability were observed at P4-O2 in theta and atC4-T3 in the alpha band. The coherences of thenormal readers decreased in the theta and in-creased in the alpha band. In the other twogroups with reading]writing impairment, coher-ences changed during the intervals between thetwo recordings in an opposite way. We obtainedan increase of theta coherences and a decrease ofalpha coherences. This finding supports a differ-ent cerebral organization and coherence matura-tion of children with reading impairment, morethan a simple maturational lag. The changes oc-cured between right parietal and occipital regionsin the theta and between right central and lefttemporal areas in the alpha band. If our logic iscorrect, it means different maturation of the lan-guage related cortical areas. When session effectwas not considered, pedagogical group differ-ences were observed between right frontal inter-connections in the delta and theta bands.

At the time of the second EEG recording, werepeated the evaluation of the reading- writingabilities. As expected, all subjects improved theirexecution on the test, surprisingly children ofGroup 3 improved more than children of theother two groups and were no more so impaired.This fact may be due to a higher motivationobtained due to the psychological assessment.

Girls and boys of the three different pedagogi-cal groups showed different coherence matura-tion at C3-P4 in the beta band. Dyslexia is foundto be strongly sex-linked. Language related prob-lems are three time more frequent in boys than in

Ž .girls McGlone, 1980 and the fact that girls learnto read earlier and with more ease than boys iswell known for every teacher. When the sessioneffect was not considered, we observed sex re-lated differences between the three pedagogicalgroups at bifrontal interconnections in the deltaband.The difficulty in the interpretation of thesedata shows how meager is our knowledge of thedistribution, variance and correlates of this signof dynamics of neural assemblies.

Acknowledgement

This project was supported in part by a schola-ship given by the Government of Estado deMexico and in part by grant number IN210193´from DGAPA, UNAM. The authors thank Sandor´John for the correction of the English version ofthe manuscript.

References

Ž .Alvarez, A., Valdes, P. and Pascual, R. 1987 EEG develop-´mental equations confirmed for Cuban schoolchildren. EEGClin. Neurophys., 67: 330]332.

Ž .Beaumont, J.G., Mayes, A.R. and Rugg, M.D. 1978 Asymme-try in EEG alpha coherences and power: effect of task andsex. EEG Clin. Neurophys., 45: 393]401.

Ž .Corsi-Cabrera, M., Herrera, P. and Malvido, M. 1989 Corre-lation between EEG and cognitive abilities: sex differences.Int. J. Neurosci., 45: 133]141.

Galaburda, A.M., Sherman, G.F., Rosen, G.D., Aboitiz, F. andŽ .Gescjwind, N. 1985 Developmental dyslexia: four consec-

utive patients with cortical anomalies. Ann. Neurol., 18:222]233.

Ž .Gasser, T., Jennen-Steinmetz, C. and Verleger, R. 1987 EEGcoherence at rest and during visual task in two groups ofchildren. EEG Clin. Neurophys., 67: 151]158.

Ž .Gersch, W. 1987 Non-stationary multichannel time seriesŽ .analysis. In: Gevins and Remond Eds. , Methods of Analy-

sis of Brain Electrical and Magnetical Signals, Elsevier.Ž .Gerstein, G.L. Bedenbough P. and Aertsen A. 1989 Neuronal

assemblies. IEEE Trans. Biomed. Eng. 36: 4]14.Ž .Goldman-Rakic P.S. 1984 Modular organization of pre-

Ž .frontal cortex. Trends Neurosci., 7 11 : 419]429.Harmony, T., Hinojosa, G., Marosi E., Becker, J., Rodrıguez,´

Ž .M., Fernandez, T., Reyes, A. and Rocha C. 1990 Correla-´tion between EEG spectral parameters and an educationalevaluation. Int. J. Neurosci., 54: 147]155.

Harmony, T., Marosi, E., Becker, J., Rodrıguez, M., Reyes, A.,´Ž .Fernandez, T., Silva, J. and Bernal, J. 1995 Longitudinal´

quantitative EEG study of children with different perfor-mances on a reading]writing test. EEG Clin. Neuro., 95:426]433.

John, E.R. Ahn, H., Prichep, L., Trepetin, M., Brown D. andŽ .Kaye H. 1980 Developmental equations for the electroen-

cephalogram. Science, 210: 1255]1258.Marosi, E., Harmony, T., Sanchez, L., Becker, J., Bernal, J.,´

Reyes, A., Dıaz de Leon, A.E., Rodrıguez, M. and´ ´ ´Ž .Fernandez, T. 1992 Maturation of the coherence of EEG´

activity in normal and learning disabled children. EEGClin. Neurophys., 83: 350]357.

Marosi, E., Harmony, T., Becker, J., Bernal, J., Reyes, A.,Ž .Rodrıguez, M. and Fernandez, T. 1993 Sex differences in´ ´

EEG coherence in normal children. Int. J. Neurosci., 72:115]121.

( )E. Marosi et al. r International Journal of Psychophysiology 25 1997 227]235 235

Marosi, E., Harmony, T., Becker, J., Reyes, A., Bernal, J.,Fernandez, T., Rodrıguez, M., Silva, J. and Guerrero, V.´ ´Ž .1995 Electroencephalographic coherences discriminatebetween children with different pedagogical evaluation Int.J. Psychophysiol., 19: 23]32.

Ž .McCarthy, R.A. and Warrington, B. 1990 Cognitive neu-ropsychology. Academic Press, p. 251.

Ž .McGlone, J. 1980 Sex differences in human brain asymme-try: a critical survey. Behav Brain Sci., 3: 215]263.

Ž .Nunez, P.L. 1995 Neocortical Dynamics and Human EEGŽ .Rhythms. In: Cutillo, Gevins, Inber, Lopes da Silva Eds. ,

Pilgrren Silberstein, Oxford University Press, p. 164.SAS Procedure Guide for Personal Computer. Version 6.03.

Ž .Thatcher, R.M. and Walker, R.A. 1985 EEG coherence andintelligence in children. EEG Clin. Neurophys., 61: S61.

Ž .Thatcher, R.W., Walker, R.A. and Guidice, S. 1987 Humancerebral hemispheric development at different rates andages. Science, 236: 1110]1113.

Ž .Tucker, D.M., Dawson, S.L. and Roth, D.L. 1985 Regionalchanges in EEG power and coherence during cognition:intensive study of two individuals. Behav. Neurosci., 99:564]577.

Valdes, P., Bosch, J., Grave, R., Hernandez, J., Riera, J.,´ ´Ž .Pascual, R. and Biscay, R. 1992 Frequency domain model

of the EEG. Brain Topogr., 1992. 4: 309]319.