The sleep EEG spectrum is a sexually dimorphic marker of general intelligence

Abstract

The shape of the EEG spectrum in sleep relies on genetic and anatomical factors and forms an individual "EEG fingerprint". Spectral components of EEG were shown to be connected to mental ability both in sleep and wakefulness. EEG sleep spindle correlates of intelligence, however, exhibit a sexual dimorphism, with a more pronounced association to intelligence in females than males. In a sample of 151 healthy individuals, we investigated how intelligence is related to spectral components of full-night sleep EEG, while controlling for the effects of age. A positive linear association between intelligence and REM anterior beta power was found in females but not males. Transient, spindle-like "REM beta tufts" are described in the EEG of healthy subjects, which may reflect the functioning of a recently described cingular-prefrontal emotion and motor regulation network. REM sleep frontal high delta power was a negative correlate of intelligence. NREM alpha and sigma spectral power correlations with intelligence did not unequivocally remain significant after multiple comparisons correction, but exhibited a similar sexual dimorphism. These results suggest that the neural oscillatory correlates of intelligence in sleep are sexually dimorphic, and they are not restricted to either sleep spindles or NREM sleep.

Figure 1

Spectro-correlograms of the age-corrected relationship between relative REM EEG power spectral density (by 0.25 Hz bins) and RES by electrode. Axis X represents frequency between 1 Hz and 40 Hz, while axis Y shows the partial Pearson correlation coefficient between RES and relative EEG power in the given bin, corrected for the effects of age. A horizontal line indicates the critical partial correlation coefficient (p = 0.05) if at least one significant correlation is present on the given electrode.

Figure 2

Spectro-correlograms of the age-corrected relationship between relative NREM EEG power spectral density (by 0.25 Hz bins) and RES by electrode. Axis X represents frequency between 1 Hz and 40 Hz, while axis Y shows the partial Pearson correlation coefficient between RES and relative EEG power in the given bin, corrected for the effects of age. A horizontal line indicates the critical partial correlation coefficient (p = 0.05) if at least one significant correlation is present on the given electrode.

Figure 3

Scatterplots illustrating the age-corrected relationship between RES and relative EEG power for each of the four maximal associations within Rüger areas of potential significance, at the frequency and derivation where the absolute value of the partial correlation coefficient was maximal, separated by sex. Data points indicate the unstandardized residuals of RES (axis X) and relative power (axis Y) after regressing for the effects of age, as the Pearson correlation of these values is equal to the age-corrected partial correlation coefficient of the original values.

Figure 4

The topographical distribution of association strength at the frequencies of maximal association illustrated on Fig. 3 in females (left column) and males (right column).

Figure 5

Top: Sample EEG in REM sleep, showing transient bouts of beta activity (“REM beta tufts”). Bottom: spectral composition of the highlighted signal (frequency in Hz, spectral power in arbitrary units).