Event-related potentials in newborns with and without familial risk for dyslexia: principal component analysis reveals differences between the groups.

Differences revealed by factor scores extracted by principal component analysis (PCA) from event-related potential (ERP) data of newborns with and without familial risk for dyslexia were examined and compared to results obtained by using original averaged ERPs. ERPs to consonant-vowel syllables (synthetic /ba/, /da/, /ga/; and natural /paa/, /taa/, /kaa/) were recorded from 26 at-risk and 23 control 1-7 day-old infants. The stimuli were presented equiprobably and with interstimulus intervals varying at random from 3,910 to 7,285 ms. Statistically significant between-group differences were found to be relatively similar irrespective of the methods of analysis (original ERPs vs. factor scores from PCA). Responses to /ga/ differed from those to /ba/ and /da/ between the groups in the right hemisphere at the latencies of 50-170 ms (Factor 4) and 540-630 ms (Factor 3). The groups differed also in their responses to /da/ in the posterior electrode sites at 740-940 ms (Factor 2). There were no group differences in the natural stimulus set. These results demonstrate that brain activation differences may be implicated in risk for dyslexia immediately after birth. The results also show that the PCA-ANOVA procedure is an effective way of identifying the group-related variance in the ERP-data when the component structure, such as those of infants, is not well-known in advance.