A computer analysis of EEG spectral signatures from normal and dyslexic children.

We have been able to differentiate between 12 dyslexic children and 13 normal age- and sex-matched children on the basis of spectral estimates of their electroencephalograms (EEG's). The children were monitored during various mental tasks and rest situations. Data dimensionality was reduced by ``banding'' various spectral components and eliminating others. The reduced spectral vectors were used as an input to a stepwise discriminant analysis program which, in effect, selected the variables most disparate between the two groups (dyslexic and normal).