Abstract of presentation at the
European International Society of Neuronal Regulation,
EEG and Intelligence: Univariate and Multivariate Comparisons Between EEG Coherence, EEG Phase Delay and
Power
Robert W. Thatcher, Ph.D.,1,2, Carl J. Biver,
Ph.D.1 and Duane North, M.S.1
NeuroImaging Laboratory, Bay Pines VA Medical
Center1 and
Department of Neurology,
Objectives: There are two inter-related categories of EEG measurement: 1- EEG currents or power and, 2- EEG network properties such as coherence and phase delays. The purpose of this study was to compare the ability of these two different categories of EEG measurement to predict performance on the Weschler Intelligence test (WISC-R).
Methods: The Weschler Intelligence test (i.e., mazes, coding, block design, digit span, picture completion, math, vocabulary, verbal I.Q., performance I.Q., full scale I.Q) and the EEG was recorded from 19 channels with a linked ears reference in the eyes closed condition from 329 normal control subjects, age 6 to 18 years of age. The EEG and the WISC-R were measured on the same day but at different times. Varimax factor analyses were performed on different categories of the EEG auto and cross-spectrum and the two highest loading EEG variables on each factor were selected for the correlation analysis of EEG vs. full scale I.Q. on all 171 combinations of electrodes. EEG variables that correlated at P < .01 were entered into a multivariate regression analysis (other WISC-R tests were also correlated but only full scale I.Q. is presented in this paper) in which full scale I.Q. was the dependent variable and EEG Phase delays, EEG coherence, EEG amplitude asymmetry and EEG power were the independent variables. The multivariate regression equations, referred to as the Brain Performance Index (BPI) also computed the + and – 95% confidence bands. Two cross-validation tests were then performed to test the regression equations using the BPI to predict I.Q. from two independent groups of subjects: 1- a group of adult subjects (N = 102) and, 2- a group of learning disabled children (N = 87).
Results: EEG phase delays had the most complex factor structure, followed by EEG coherence followed by EEG amplitude asymmetry and followed by EEG power. Multivariate regression of the two highest loading EEG variables on each factor ranged from 0.38 (coding) to 0.55 (full scale I.Q.) at P < .0000001. EEG phase was the strongest EEG measure and represented 66.25% of the variables that survived the multiple regression analysis, EEG coherence = 21.25%, EEG amplitude asymmetry = 10 % , relative power = 2.5% and power ratios = 0%. EEG phase delays were positively correlated with I.Q. in the frontal regions (i.e., shorter left frontal lobe phase delays predict higher I.Q.) while inverse correlations were present in short distance electrode combinations (i.e., longer phase delays predicted higher I.Q.). In addition, a decrease in short distance coherence (7cm to 14 cm inter-electrode distances) especially in temporal, parietal and occipital regions was positively related to I.Q.
Conclusions: The EEG correlations presented in this paper primarily concern the “G factor” of the Weschler intelligence test which is a measure of a general or integrative property of human intelligence. In the present study, the EEG network measures were more strongly correlated with the “G factor” in intelligence tests than was EEG power or power ratio measures. Peak frequency of alpha was not investigated in this study, however, a network hypothesis of short vs. long distance phase delay dynamics is not inconsistent with higher peak alpha frequency and positive correlations with intelligence.
Based on the results of this study it is hypothesized that general intelligence is positively correlated with increased system integration and faster processing times in long distance frontal connections as reflected by shorter phase delays correlated with increased intelligence. Simultaneously, intelligence is also positively related to increased differentiation in widespread local networks or local assemblies of cells as reflected by longer EEG phase delays in local posterior and local frontal-temporal distances which are correlated with increased intelligence. Higher frequency beta in the right frontal cortex reflects the requirement for high speed social integration, whereas the lower frequency delta reflects the requirement for longer integration times of analytical details. Emotional intelligence and analytical intelligence have different left vs. right hemisphere temporal dynamics.
|
|
|
Fig. 1- Scattergrams N = 329. Y-axis is measured full scale I.Q.
and X-axis is the EEG predicted full scale
I.Q. |
|
|
|
Fig. 2 – Significant
correlations between EEG phase delays and full scale I.Q. Top row (red color) are Negative
correlations (i.e., the shorter the phase delays the higher is I.Q.) and
the bottom row (green color) are Positive correlations (i.e., the longer
the phase delays then the higher is I.Q.). Left column is the delta frequency
band (1 to 3.5 Hz), the middle column is the
beta frequency band (12.5 – 25 Hz) and the right column is the direction
of correlation between phase delay and intelligence.
|
