I'm not really qualified to comment and I didn't spend very much time analysing the paper. However, I'll comment a bit since reading nutrition papers used to b

I'm not really qualified to comment and I didn't spend very much time analysing the paper. However, I'll comment a bit since reading nutrition papers used to be a hobby of mine. First, the sample size is relatively good - nearly 1000 people, taken from a similar area and over a long period of time (30 years). What's missing is any obvious mention of how the participants were chosen, so it's hard to see if there is any selection bias. However, I didn't look at the links provided, so hopefully it goes into more detail there. I mention this because I have often seen papers published in this area with obviously flawed sample selection. This seems better than most. The main things that stand out for me is that they did not attempt to measure how much chocolate people consumed, nor the type of chocolate consumed. They simply measured the number of times a week "chocolate" was consumed (which could have been a large variety of things). The other main point is that they mention that there was a significant reduction in alcohol consumption in the groups that ate the most chocolate, however they did not try to correct for this. Just scanning through the data they present, I don't think it is possible to tell if the increased cognitive performance is related to chocolate intake, or a lack of alcohol intake. So, it is an interesting paper, but I wouldn't go out and start eating chocolate because of it.