What you seem to have plotted here is the "Canolty ERP": the signal filtered around 10Hz, and epoched/averaged around the peaks of the filtered data. This is would be therefore in the same units as the original signal, and your conditional A may simply have less alpha, or more "phase resets"...
Can you please post screen captures of the files in the database explorer, and of the figures you get when you double-click on the them?