Brain Response Correlates of Psychometric Intelligence

John P. Ertl

Edward W. P. Schafer

Previous attempts to correlate electrophysiological variables with behavioural indices of intelligence have been inconclusive. Recent findings suggest that the average evoked potential (AEP) recorded from the human scalp may reflect the neural correlates of higher mental activity or information processing by the brain.

The speed of this process, measured by the latency of sequential AEP components, could be the biological substrate of individual differences in behavioural intelligence. AEP latency data from cretinized rats3, hypothyroid patients4, aribo-flavinotic children5 and humans with differential intelli­gence6 provide preliminary support for this hypothesis.

We now report evidence of a relationship between the time delays of the brain's electrical responses and psycho­metric intelligence. The latencies of sequential visual evoked potential components bear highly significant inverse correlations with IQ scores on three commonly used intelligence tests in a random sample of 573 primary school pupils.

The visual AEP waveforms of high and low IQ subjects are characteristically distinct from each other.

The experimental sample comprised 317 male and 256 female pupils randomly selected from the population of 8,000 children attending grades 2, 3, 4, 5, 7 and 8 in the thirty-nine schools of the Ottawa separate school system.

The Wechsler intelligence scale for children (WISC), the primary mental abilities test (PMA) and the Otis quick-scoring mental ability tests were used. A visual AEP record was also obtained from each subject. The electroencephalograph (EEG) was re­corded from bipolar scalp contact electrodes 6 cm apart, parallel with the midline and astride C4 in the 10-20 system with ground to the right ear lobe (upward deflexion in reported data indicates negativity of the anterior electrode with respect to the posterior).

The raw EEG was amplified to the required voltage in a bandwidth 3 dB down at 3 and 50 Hz and recorded on magnetic tape along with trigger pulses corresponding to the onset of flash stimuli. Subjects sat with eyes open in a darkened shielded room fixating a spot on a reflecting screen. Bright photic stimuli of microsecond duration were delivered according to a uniform stimulus interval distribution which ranged from 0-8 to 1-8 s. AEPs in response to 400 stimuli in a 625 ms interval after stimu­lation were extracted from the EEG by two methods— amplitude summation and zero-crossing analysis—using the Enhancetron ND-801 digital computer. These two techniques in combination facilitated objective identifica­tion of sequential AEP components.

Initially, the evoked potentials of each subject were amplitude averaged in alternate response sets of 200 and read out by XY recorder. Short term reliability assess­ments of the AEP and preliminary identification of its sequential components were made by visual cross- correlation from these graphs. A zero-crossing analysis technique was then used to confirm statistically the presence of sequential AEP components7.

The EEG was converted to pulses corresponding to zero-crossings where the EEG waveforms passed from positive to negative voltage. A histogram of zero-crossing occurrences against time, following 400 stimuli, was made by the Enhancetron

using its multi-scaling mode across sixty-four channels (Fig. 1). The same EEG data were used to generate a control histogram by triggering the multi-scaler 400 times randomly with no time relation to experimental stimula­tion.

The mean channel counts of both stimulated and control histograms were not statistically different. The standard deviation of the control histogram was also computed. If the number of zero-crossing counts in any channel of the stimulated histogram exceeded two stand­ard deviations above its mean a statistically significant event was identified. The corresponding peaks of the summated AEP waveform were called sequentially El, E2, E3 and E4 and their latencies from the onset of stimulus were measured with an error of measurement estimated to be plus or minus 5 ms.

A statistically signifi­cant zero-crossing event was always accompanied by a visible component in the AEP record. Visible peaks in. the AEP waveform not supported by significant z«ro-crossing events were rejected as sequential components.

The latencies of the first four sequential AEP com­ponents detected in this manner were intercorrelated with the intelligence test scores using the Pearson r correlation coefficient across the entire sample.

Results, presented in Table 1 and illustrated in Fig. 2, are evidence of a relationship between the electrical responses of the human brain and psychometric intelli­gence. The latencies of sequential components of the visual AEP bear highly significant inverse correlations with IQ scores from the three intelligence tests used (Table 1). It would appear that AEP component latency is related to some common factor tapped by the highly intercorrelated intelligence tests.

IQ scores correlate higher with the late components (E3 and E4) of the visual AEP than with the early com­ponents (El and E2). This observation accords well with the generally accepted position that the late com­ponents of the AEP are more significant for conscious processes than the early primary components8.

Specimen AEPs from ten high and ten low IQ subjects are shown in Fig. 2. It is evident that these two sets of waveforms are internally consistent but distinct from Table 1. pearson r correlation coefficients and descriptive statistics

for psychometric and physiological variables each other.

The AEPs of the high IQ subjects are more complex, characterized by high frequency components in the first 100 ms which are not observed in the AEPs of the low IQ subjects. The ten high IQ subjects illustrated have a mean E3 (third sequential peak) latency of 88 ms. The ten low IQ subjects have a mean E3 latency of 194 ms. The mean AEP component latencies for our entire the first 100 ms which are not observed in the AEPs of the low IQ subjects. The ten high IQ subjects illustrated have a mean E3 (third sequential peak) latency of 88 ms. The ten low IQ subjects have a mean E3 latency of 194 ms. The mean AEP component latencies for our entire sample correspond closely with normative data from other laboratories9. Our range of response latencies, however, is much greater, possibly because of the uncontrolled influence of intelligence and the visual and therefore subjective method of AEP component identification used in these other studies.

Several variables are known to affect the waveform of the visual AEP10. The inter-subject AEP component latency differences noted in our sample, however, are much greater than any demonstrated latency changes attribut­able to potentially contaminating variables such as attention, arousal level, pupil diameter and so on. Furthermore, these potentially contaminating variables are not related to psychometric intelligence and are assumed to be randomly distributed in the population.

Reported intercorrelations between AEP component latency and IQ are not large but highly statistically significant. When attempting to relate variables in the psychological domain to variables in the physiological domain, it can be argued that even very small correlations, if statistically significant, could identify a fundamental relationship. Our findings suggest that evoked potentials, which reflect the time course of information processing by the brain, could be the key to understanding the biological substrate of individual differences in behavioural intelligence.

This work was carried out in accordance with a contract with the US Department of Health, Education and Welfare, Office of Education. We also acknowledge the contribution of the Ontario Mental Health Foundation, the Educational Records Bureau of New York and the Ottawa Separate School Board.

John P. Ertl

Edward W. P. Schafer

Center of Cybernetic Studies, University of Ottawa. Received January 21, 1969.

1 Vogel, W., and Broverman, D. M., Psychol. Bull., 62, 132 (1964).

2 Sutton, S., Tueting, P., Zubin, J., and John, E. R., Science, 155, 1436

(1967); John, E. R., Herrington, R. N., and Sutton, S., ibid., 155, 1439 (1967); Shevrin, H., and Fritzler, D. E., ibid., 161, 295 (1968). Cohen, J., and Walter, W. G., Psyche-physiology, 3, 187 (1966). Clynes, M., and Kohn, M., EEG Clin. Neurophysiol., suppl. 26, 82 (1967).

3 Bradley, P. B., Eayrs, J. T., and Richards, N. M., EEG Clin. Neurophysiol.,

17, 308 (1964).

4 Nishitani, H., and Kooi, K. A., EEG Clin. Neurophysiol., 24, 554 (1968).

5 Arakawa, T., Mizuno, T., Chiba, ¥., Sakai, K., Watanabe, S., Tamura, T.,

Tatsumi, S., and Coursin, D. B„ Tohoku J. Exp. Med., 94, 327 (1968).

6 Chalke, F. C. R., and Ertl, J., Life Sci., 4, 1319 (1965); Ertl, J., in Bio-

cybernetics of The Central Nervous System (edit, by Proctor, L. D.) (Little Brown, Boston, 1969); Whitaker, H. S., Osborne, R. T., and Nicora, B., Trans. Amer. Neurol. Assoc., 92, 194 (1967); Bennett, W. F., Nature, 220, 1147 (1968); Rhodes, L. E., Dustman, R. E., and Beck, E. C, EEG Clin. Neurophysiol., 26, 237 Q969); Engle, R., and Butler, B. V., EEG Clin. Neurophysiol., 26, 237 (1969).

7 Ertl, J., EEG Clin. Neurophysiol., 18, 630 (1965); Krekule, I., ibid., 25, 175

(1968).

8 Libet, B., Alberts, W. W., Wright, jun., E. W., and Feinstein, B., Science,

158, 1597 (1967).

9 Kooi, K. A., and Bagchi, B. K., Ann., NY Acad. Sci., 112, 254 (1964);

Gastaut, H., and Regis, H., NASA SP-72, 7 (1965). 10 Beinhocker