EEG, RACE, SEX, AND MATURATION

John P. Ertl

(Internal Report #97-Neural Models Limited

Maturational changes in EEG frequen­cy, hemispheric synchronization, alpha percent and the interrelationships between these parameters is demon­strated in a sample of 5760 Kinder­garten, Grade I and Grade II children. In this sample, all EEG parameters and their interrelationships which are age dependent indicate a maturational lag for black children and a slight developmental acceleration for females as compared to males. No valid inferences, deductions or extrapola­tions can be made concerning IQ or intelligence from this data.

The main purpose of this large scale study was to evaluate the Educational utility of brain wave analysis for the early detection of learning disabilities and to develop a language free measure of learning potential. These aspects are reported elsewhere (1), together with a detailed technical description of the apparatus and procedures used. During data analysis some unexpected and unusual findings emerged concerning maturational changes and racial differ­ences in certain EEG parameters and their interrelationships.

The sample is described in Table 1. Approximately 55% of the entire Kindergarten, Grade I and Grade II population in Caddo Parish, Shreve-port, Louisiana, were tested between December 1st. 1975 and May 30th. 1976. Re-tests were done on 870 subjects, with an average interval of one month between test and re-test (2).

Sample statistics.

The EEG was derived from electrodes placed symmetrically over the left and right hemispheres (approximately C3 and C4 according to the international 10-20 system). The monopolar technique, with the left ear lobe as reference and the right ear lobe as ground, was used. Testing was done either in the classroom or in a motor home. No environmental controls of any kind were used (3). The EEG was band limited (6db down at 3 and 40 Hz) converted to pulses by comparators with 2 microvolt hysteresis. All further computations were made using these pulses.

Three EEG parameters were measured simultaneously, using the Neural Models Ltd. Brain Wave Analyzer, Model BWA-04 (4). 1. The average frequency of the EEG (F) derived from the left hemisphere, based on counting comparator output pulses over a 10 second period. 2. The frequency of occurrence of alpha events (A) over ten seconds, where an alpha event is defined as a time interval between comparator output pulses which is greater than 83 milliseconds and less than 143 milli­seconds (i.e. between 7 and 12 Hz). A reading of say 4.5 for alpha indicates that EEG frequency was between 7 and 12 Hz 45% of the time during the 10 second measurement period. 3.

The absolute value of the phase leads and lags between the two hemispheres averaged over ten seconds. The BWA-04 is so arranged that a reading of 1000 corresponds to a 180 phase shift, a reading of 500 to 90, etc. A reading of 500 also corresponds to a cross correlation coefficient of zero and a reading of zero to a cross correlation of one (5). This score is designated hemispheric synchroniza­tion (HS). The measurements were repeated four times in succession and their average used for analysis.

I am not aware of neurophysiological or other theories which would explain any of the observed matura­tional changes, the racial differences or the maturational changes in the relationship between the EEG para­meters. I therefore postulate specula­tive hypotheses which could explain some of these findings. Some of the speculations in Postulate III are not relevant to the problem at hand, but do indicate my approach to the interpretation of the EEG.

Postulate I: The efficiency of an information processing system is relat­ed to the operating frequency used or required by the system (6), where efficiency is defined in the context of information theory and not in terms of energetics.

Postulate II (a): If two or more information processing systems are physically interconnected, the cross-correlation between the measurable outputs of the operating systems is inversely related to the autonomy of the systems.

(b): A cross correlation of zero indicates that the systems are unconnected. A cross correlation of one indicates that there is only one system, not two, involved.

Postulate III: An intelligent system which learns and develops as a function of time tends to change its functions from simple to complex, from highly organized to a lower level of organization, from synchronization to desynchronization among its parts, from a high internal signal to noise ratio to a low signal to noise ratio, from precision to error, from large amplitude oscillations to smaller am­plitude oscillations, etc.

The maturational changes in EEG average frequency, hemispheric synchronization and the alpha percent are shown in Table 2 by grade levels for the entire sample. Thus, from age 5.5 (average age of Kindergarten) to 7.5 (average age of Grade II) There is an 8.9% (11.2 - 12.2 Hz) increase in the average frequency of the EEG and an 8.1% (235 - 254 Hs scores) increase in desynchronization of the hemispheres.

TABLE 1

TABLE 2.

The percent alpha scores did not change significantly with age within the age range of the sample. Correlation between EEG parameters: frequency (F), hemispheric synchroni­zation (HS) and alpha (A) by grade levels. All correlations are significant p » .01. The differences in average frequency and HS scores between each grade level are also significant p » .01. The differences in alpha percent are not significant.

Maturational changes are also evi­dent in the relationship between F and A and F and HS. Due to the large sample size, all age differences (except for alpha) and all intercorrelations (7) and the trend as a function of age are statistically significant at very high (p » .001) levels of confidence.

The magnitude of the correlation coeffi­cients is small and is influenced by at least two other factors, race and sex, as shown in Table 3. Correlations between EEG parameters by race and sex. A correlation of approximately .082 with a sample size of 2600 is significant p » .001.

The maturational change in the average frequency of the EEG has been observed by others (1), the interpretation that this relates to information processing efficiency, ac­cording to Postulate I, is the subject of another report. (1).

The maturational changes in the synchronization between the two hemispheres is interpreted according to Postulate II. Thus, as functional specialization develops in the two hemispheres, the process of electrical synchronization of multiple events in large aggregates of neurons becomes technically more difficult and possibly also counteracts the- development of autonomy of the various brain regions.

The cross correlation coefficient bet­ween the electrical activity from the two hemispheres in normal children is approximately 0.5 + .1 (corresponding to an HS reading of 250 + 50). Children with severe learning disabili­ties and in mentally retarded children (1) the cross correlation is either very high, .7 plus, or very low, .2 or less, indicating that within certain limits a degree of asymmetry is necessary for proper functioning of the brain.

The absence of a significant matura­tional change in the alpha scores in this age range simply indicates that the percent of frequencies in the 7-12 Hz band did not change significantly with age. It will be recalled that the alpha designation refers to a band of frequencies and NOT to the convent­ional definition of alpha. The age related changes in frequency could have occurred by an increase in the average frequency within the alpha band or outside the alpha band or a combination of both. The equation describing these relationships is as follows:

F = NE (1 - A x Ap) + A where

F = average frequency of the EEG

NE = average frequency of the EEG outside the alpha band A = frequency of occurrence of alpha events per unit time Ap = average period of alpha events in seconds

At present the measurements made were inadequate to resolve this ques­tion, because Ap was not measured, however, based on recent observations, I think it is unlikely that the age related frequency changes are due to an increase in the average frequency of the alpha band.

The maturational changes in the correlations between the three EEG parameters are extremely complex because all variables are interrelated and most of the correlations and the age trends in the correlations are small, but very robust statistically. Postulates I or II or III may help in integrating?

TABLE 3

The Frequency-alpha correlation: It is clear from the equation and from consideration of the EEG frequency spectrum that in order to get a high average frequency the alpha percent must be small. An inverse correlation between alpha and frequency is therefore expected.

This correlation will be larger as the average frequency increases. The magnitude and the maturational trend in this relationship are thus partly explained by the change in average frequency. Similarly, the substantial (r = .18, r = .26) racial difference is partly a function of the difference in average frequency. The statistical analyses used were inade­quate to determine if a residual relationship is present and significant.

There are a number of factors operating to account for the relation­ship between frequency and HS; the most obvious is that it is harder to synchronize low frequency signals than high frequency signals. Frequen­cy increases with age and it is remotely possible that this frequency change could account for some of the maturational changes in the correla­tion between frequency and HS.

A more plausible hypothesis is based on Postulate II; as the autonomy of the two hemispheres develops the various frequencies of EEG signals produced at different times are less dependent on what is happening in the other hemi-spheres at a given moment. The strong trend towards desynchronization as a function of age tends to lend support to this view.

It appears from the near zero (r = .04, Table 3) correlation between F and HS in white children that the developmental changes in this relationship are com­plete at this age level. In black children the developmental process is still continuing, since there is a significant correlation of .2 remaining at this age level.

There are no significant age changes in the relationship between A and HS in the sample as a whole (Table 2). This is probably due to the masking effect of several variables: the F - HS correlations and its age trend, the racial SEPTEMBER-OCTOBER 1978 difference in the F - HS correlation and changes in developmental rates.

There is a very significant difference in the A - HS relationship between black and white children. For white children the desynchronization of the two hemispheres is increased substantially (r = .4, Table 3) in the presence of alpha band activity. For black children the increase in desynchronization is moderate (r = .18). I am unable, at present, to explain these findings.

The near zero correlation between F and HS for white children, together with the strong (.4) correlation bet­ween the percent of alpha band activity and desynchronization of the hemispheres, is extremely difficult to explain, unless one postulates that alpha band frequencies are somehow “different” than other frequencies and furthermore that this difference is related to race and possibly matura­tion.

The slight frequency difference between the races is insufficient to account for the A - HS corre­lations in terms of the hypothesis that low frequency signals are harder to synchronize. The slight difference in alpha percent (46% vs 45%) is offset by a smaller standard deviation (.7 vs .8) for the white children; a restricted range tends to lower correlation between variables. Analysis of the data by different statistical methods at a future date may help to resolve some of these problems.

The results, with respect to sex differences, indicate a general tenden­cy for accelerated maturation of females compared to males at this age level. The average frequency difference was 0.8%, the HS difference was 3.3% in “favor” of females. The alpha percent was not significantly different. The differences in the relationships between variables in terms of sex were all minor.

In general, maturational changes in EEG frequency, hemispheric synchro­nization and the interrelationships between these parameters has been demonstrated. In this sample, all EEG parameters and their interrelationships which are age dependent, indicate a maturational lag for the black children and a slight devel-opmental accelera­tion for the females as compared to males.

No valid inferences, deductions or extrapolations can be made con­cerning IQ or intelligence from this data. A maturational lag in the development of certain EEG para­meters does not imply lower intelligen­ce or IQ. It is simply a maturational lag.

There is strong evidence (1,8) that none of the EEG measurements in this study correlates significantly with IQ scores. Furthermore, there is no evidence to indicate that a maturation­al lag at a particular age cannot be compensated by a maturational accel­eration at a later time.

REFERENCES AND NOTES

1. J.P. Ertl, The Louisiana Study of Learning Potential by Brain Wave Analysis. The Louisiana State Department of Education, in print September 1976.

2. Reliability coefficients for alpha and frequency were not comput­ed, but are estimated to be not less than .7. The reliability of the NE scores, which are a composite of the A and F scores, was .91. The reliability of the HS scores was .7.

3. The original purpose of the study was to evaluate “utility”. I did not think that a controlled labora­tory environment was either ne­cessary or practical. It is often better to tolerate random errors than to introduce unknown con­stant errors by controls whose effects are unknown.

4. U.S. Patent No. 3893450, U.S. Patent Pending, 1976.

5. An experiment was performed in which two independent band limit­ed (3 - 50 Hz) noise sources were used as an input to the two chan­nels of the BWA-04. The phase readings were 510 + 10 for repeated measurements. The 10 count error is due to the fact that one channel is set to lead the other by 2 degrees.

6. The conventionally defined alpha rhythm is not considered an “operating frequency.” Therefore, efficiency of the system is re­lated to the frequency of non-alpha activity.

7. None of the distributions of the EEG measures were Gaussian, but they were not so badly skewed that the use of the correlation measure would be invalid.

8. In the Louisiana Study (Note 1)