I disagree with that interpretation. They said average because they're almost certainly referring to the average over the 32 patient cohort. What likely happened is some patients responded well in the the P300 and CogState but not the majority and so they cited the average, which is the most flattering light. Or they were careless in their wording. I hope it was the latter but not counting on it.
And was the improvement (in the non-majority that did improve) small? Big? Huge? Again, we should find out Saturday. That stuff matters. Big improvements in many of the folks who were on the donezepil would be great news, and entirely consistent with the abstract. Though why the abstract didn't point that out would be a boner on the company's part.