Early cognitive and language skills are linked to resting frontal gamma power across the first 3 years

Abstract

High-frequency cortical activity in humans and animals has been linked to a wide variety of higher cognitive processes. This research suggests that specific changes in neuronal synchrony occur during cognitive processing, distinguished by emergence of fast oscillations in the gamma frequency range. To determine whether the development of high-frequency brain oscillations can be related to the development of cognitive abilities, we studied the power spectra of resting EEG in children 16, 24 and 36 months of age. Individual differences in the distribution of frontal gamma power during rest were highly correlated with concurrent language and cognitive skills at all ages. Gamma power was also associated with attention measures; children who were observed as having better inhibitory control and more mature attention shifting abilities had higher gamma power density functions. We included a group of children with a family history of language impairment (FH+) and thus at higher risk for language disorders. FH+ children, as a group, showed consistently lower gamma over frontal regions than the well-matched FH- controls with no such family history (FH-). We suggest that the emergence of high-frequency neural synchrony may be critical for cognitive and linguistic development, and that children at risk for language impairments may lag in this process.

Figure 1

Electrical Geodesics Inc. 64 Channel Geodesic Sensor Net Map. The red-circled channels were used for our EEG analysis.

Figure 2A and 2B

(A) Frontal EEG power spectra (averaged across the frontal and prefrontal electrodes indicated in Fig. 1) for control children and children with a family history of language impairment. At 16 months, the two populations largely overlap; by 36 months a clear distinction is seen with the family history group showing reduced power at 31Hz and above. (B) Scalp distribution of mean gamma power for each group at 36 months. The difference between the two is primarily seen in the frontal lateral region.

Figures 3A and 3B

ANOVAs at 5-30Hz and 31-50Hz band power for 24 (A) and 36 months (B). The lower and upper lines of the “box” are the 25th and 75th percentiles of the sample. The “whiskers” show the extent of the rest of the sample. The plus sign is an indication of an outlier in the data. For 5-30Hz band, the null hypothesis is true, p = 0.6985 at 24 months and p = 0.9326 at 36 months. For 31-50Hz, the null hypothesis is rejected for both 24 and 36 months p = 0.0454 and p = 0.0088 and the two groups are shown to significantly differ in the amount of frontal power within the broad gamma range (31-50Hz). (See full-size graphs in supporting on-line materials.)

Figures 3A and 3B

ANOVAs at 5-30Hz and 31-50Hz band power for 24 (A) and 36 months (B). The lower and upper lines of the “box” are the 25th and 75th percentiles of the sample. The “whiskers” show the extent of the rest of the sample. The plus sign is an indication of an outlier in the data. For 5-30Hz band, the null hypothesis is true, p = 0.6985 at 24 months and p = 0.9326 at 36 months. For 31-50Hz, the null hypothesis is rejected for both 24 and 36 months p = 0.0454 and p = 0.0088 and the two groups are shown to significantly differ in the amount of frontal power within the broad gamma range (31-50Hz). (See full-size graphs in supporting on-line materials.)

Figures 4A, 4B and 4C

Correlations at 16 (A) and 24 (B) and 36 (C) months with mean gamma power. Legend: PLS expressive language score (PLS-EC); PLS Auditory Comprehension (PLS-AC); Bayley-Mental development index (BSID-MDI); Temperament (TBAQ): attention shifting (TBAQ-AS); inhibitory control (TBAQ-IC); Parent-Child interaction (PCIS): appropriateness (PCIS-Ap), quality (PCIS-Qu); Stanford-Binet (SB): - total composite IQ (SB-IQ), quantitative (SB-QR), verbal absurdities (SB-VA), verbal reasoning (SB-VR).

Figures 4A, 4B and 4C

Correlations at 16 (A) and 24 (B) and 36 (C) months with mean gamma power. Legend: PLS expressive language score (PLS-EC); PLS Auditory Comprehension (PLS-AC); Bayley-Mental development index (BSID-MDI); Temperament (TBAQ): attention shifting (TBAQ-AS); inhibitory control (TBAQ-IC); Parent-Child interaction (PCIS): appropriateness (PCIS-Ap), quality (PCIS-Qu); Stanford-Binet (SB): - total composite IQ (SB-IQ), quantitative (SB-QR), verbal absurdities (SB-VA), verbal reasoning (SB-VR).

Figures 4A, 4B and 4C

Correlations at 16 (A) and 24 (B) and 36 (C) months with mean gamma power. Legend: PLS expressive language score (PLS-EC); PLS Auditory Comprehension (PLS-AC); Bayley-Mental development index (BSID-MDI); Temperament (TBAQ): attention shifting (TBAQ-AS); inhibitory control (TBAQ-IC); Parent-Child interaction (PCIS): appropriateness (PCIS-Ap), quality (PCIS-Qu); Stanford-Binet (SB): - total composite IQ (SB-IQ), quantitative (SB-QR), verbal absurdities (SB-VA), verbal reasoning (SB-VR).

Figure 5A-C

Relationship of resting frontal gamma EEG power to 24 month expressive (A) and receptive (B) language ability as measured by the Preschool Language Score [PLS], (C) PLS receptive language ability in 36-month old children.

Figure 5A-C

Relationship of resting frontal gamma EEG power to 24 month expressive (A) and receptive (B) language ability as measured by the Preschool Language Score [PLS], (C) PLS receptive language ability in 36-month old children.

Figure 5A-C

Relationship of resting frontal gamma EEG power to 24 month expressive (A) and receptive (B) language ability as measured by the Preschool Language Score [PLS], (C) PLS receptive language ability in 36-month old children.

Figure 6 A-B

Bootstrap method is shown on two pairs of vectors, power-PLS-3 Expressive scores (A) and Power-Bayley-MDI (B), with 1000 times resampling.

Figure 7A-D

(A) Correlation topogram showing a map of the electrode-by-electrode regression slopes for PLS expressive language at 24 months. Correlations are broadly positive over most of the head, showing the strongest relationship frontally, with a lateral bias. (B) Correlation topogram showing a map of the electrode-by-electrode regression slopes for Bayley MDI at 24 months. Note again the left hemisphere bias in this cognitive measure with a heavy language loading. (C). Correlation topogram showing a map of the electrode-by-electrode regression slopes for PLS receptive language at 24 months. (D) Similar plot for PLS receptive language ability at 36 months. Again, highest correlations are seen in the frontal regions and there appears to be a left lateral bias.