1Institute of Behavioural Sciences, Semmelweis University, Budapest, Hungary

2Medical and Pharmaceutical University of Targu-Mures, Romania

3Department of Psychology, Károli Gáspár University, Budapest, Hungary

4Department of Neurology, National Institute of Psychiatry and Neurology, Budapest, Hungary

Keywords: intelligence, sleep spindles, slow oscillation

Brain plasticity, amount of early sensory experience and synaptic transmission efficacy are related to synchronized neural oscillations of NREM sleep. EEG measures of these oscillations correlating in a traitlike manner with certain cognitive performances are characterized by high internight reliability and stable, fingerprint-like topography. As intelligence related brain features may generally enhance neural network efficiency, we hypothesize that IQ correalates with measures of self-sustained corticothalamic resonance characteristic for NREM sleep. Fifteen healthy subjects slept two nights in the sleep laboratory and completed the Raven Progressive Matrices Test. FFT-based amplitude spectra of NREM sleep-EEG (21 channels, 10-20 system) was used for quantifying absolute and relative slow (0.5-1.25 HZ), delta (1.5-4.5 HZ), individual slow spindle (ISS) and individual fast spindle (IFS) amplitudes respectively, as well as automated ISS and IFS detection. Additionally we quantified ISS’s and IFS’s tendency to occur in the positive phase of the slow oscillation. No measures of the slow or delta activity correlated with intelligence. However, IFSs were independently related to intelligence, age and sleep efficiency. Absolute amplitude of IFSs in the frontocentral region correlated positively with age-corrected IQ scores and negatively with S4 time. Partialling out S4 time did not reduce the correlations (max. for F4; r = 0.65; P= 0.012). Similar results were found for relative IFS amplitudes. IFS density (spindles min -1) showed the most reliable relationship with intelligence: positive correlations with both IQ and raw test scores, and negative ones with age and S3 time were found. After removing the effects of age and S3 time, correlations not only remained significant but exceptionally high (F4; r = 0.82; P = 0.0006). The tendency for IFS-ing in the positive phase of the slow oscillation correlated with test’s raw scores (r = 0.52; P = 0.05 age corrected). In contrast to IFSs, ISSs increased with age and correlated negatively with test perfromance and S2 sleep time. After removing age and S2 effects correlations between raw scores and relative ISS amplitudes remained significant in the frontocentral and temporal region (F4; r = -0.64; P = 0.02). Breaking EEG data into sleep cycles resulted in almost identical relationships confirming the reliability of our findings. Results support the idea that the oscillatory behaviour of corticothalamic networks during NREM sleep expresses a state-dependent, easily measurable general network-efficiency possibly involved in reasoning and problem solving ability. Enhanced fast-type spindling is a positive predictor of general fluid intelligence in healthy, drug-free subjects.

Acknowledgement: This research was supported by the Hungarian Medical Research Council (ETT-162/2003).