Figure 5.

Temporal and spatial profiles of difference-CSDs (CSD-transformed ERPs in the scrambled condition minus those in the comprehension condition) that predict individual differences in reading comprehension performance. (A) Proportion of successful classification of participants into good and poor comprehenders based on difference-CSDs from eight overlapping 100-ms time bins. Solid green line shows the mean of 1000 classification results using the original data, and dotted gray line shows the mean of 1000 control classification results using the same data but with randomized labeling of good and poor comprehenders (see Methods). Error bars represent ±1 SD (standard deviation). It is apparent that the difference-CSDs during 400–500 ms following the onset of the target word are crucial for distinguishing good from poor comprehenders. (B) Topographic map showing the recording sites that were imporant for classifying good vs. poor comprehenders (warmer colors indicate greater importance). AFz is clearly the uniquely important scalp site. (C) Correlation between comprehension test scores and difference-CSDs at AFz during 400–500 ms following target word onset. The rectangles indicate the poor (left) and good (right) reading comprehenders, the ellipse indicates the boundary of the 95% confidence ellipse, and the line indicates the regression line without the three outliers.