Faculty Proceedings, Presentations, Speeches and Lectures

QEEG Asymmetry Ration is Related to Intelligence

Event Name/Location

National Academy of Neuropsychology

Event Location / Date(s)

Orlando, FL

Presentation Date

11-2000

Document Type

Conference Proceeding

Description

QEEG asymmetry ratio is related to intelligence

Patton DE, Selevan-Eisenstein E, Burns WJ, Montgomery D, Simco ER, Edmayer S.

This study evaluated different methodological approaches to quantifying the relationship between quantitative electroencephalography (QEEG) alpha activity and intelligence in high-risk children.

Sixteen right-handed children (mean age = 89.3 1 months, SD = 4.67) who were hospitalized at birth due to per&natal complications served as participants. QEEG was recorded as the children quietly listened to a story. The relationship of WISC-III Full Scale IQ to QEEG variables was evaluated in a 3-level analysis. Level I correlations were calculated between FSIQ and alpha activity at individual electrode sites. Level 2 correlations were calculated between FSIQ and asymmetry simple difference (DIFF) scores wherein left hemispheric alpha is subtracted from right hemispheric alpha (i.e., R - L).

Level 3 correlations were calculated between FSIQ and asymmetry ratio (AR) scores wherein left hemispheric alpha is subtracted from right hemispheric alpha and divided by their sum [i.e., (R - L)/ (R+L)]. Results of Level 1 analyses revealed that 3 out of 24 (12.5%) correlations for the left-sided electrodes were significant, 0 out of 24 (0%) were significant for the right-sided electrodes, and 0 out of 15 (0%) were significant for centerline electrodes. Results of Level 2 analyses revealed that 7 out of 24 (29.17%) of the correlations between FSIQ and DIFF scores were significant, representing a notable increase in information explained compared with Level 1. Analyses at Level 3 revealed 10 out of 24 (42.67%) of the correlations between FSIQ and AR scores were significant, representing a further increase in information explained compared with either Level 1 or Level 2. The same pattern of results across analysis levels was observed when regression formulas were generated. Specifically, analyses at Level I produced a best-fit R2 = 0.424, which failed to cross-validate (R2rnnss = 0.000). At Level 2 a somewhat improved best-fit R2= 0.79 was produced, however this value also underwent a large amount of shrinkage (R2ranss =0.140). Finally, Level 3 yielded a maximum R2 = 0.844 which achieved an acceptable moderate amount of shrinkage upon cross-validation (R2raEss = 0.525) given the small sample size of the study. Given that 52.5% of the variance of FSIQ was accounted for by AR values, this brief exploratory study concerning the relationship of QEEG to intelligence shows that the elucidation of brain-behavior relationships via QEEG requires complex analysis.

Abstracts /Archives of Clinical Neuropsychology 15 (2000) 653-850

This document is currently not available here.

Share

COinS