There is a small possibility an extract of cocoa could help older people be better at certain memory tasks - but the trial done so far is not good enough to show this.
In the trial, a group of healthy people aged between 50 and 69 were split into four groups. Two groups spent three months on a diet that was high in cocoa, two on a low cocoa diet. At least, that's how the paper describes it - in reality the 'high cocoa' group took a supplement of 99mg of cocoa flavanols. To get this much naturally you would have to eat 25 individual chocolate bars (not recommended!) - I don't know how much that is in cups of cocoa, but I suspect it's a lot. Each group was also divided into half that were sedentary and half that took regular exercise.
The scientists then looked at two things - how a particular part of the brain responded in an fMRI scanner, and how well the test subjects did at two memory tests. What they found was that those on a high cocoa diet did significantly better at one of the memory tests - the equivalent, it was claimed, of being almost 30 years younger.
This is interesting, but it isn't enough yet to suggest we should all get out and start consuming lots of cocoa flavanols. The test groups were small with only 8 to 11 people in each. This doesn't mean that the results are meaningless, but it does suggest further tests are required. It has also been pointed out that the claim that result is statistically significant is doubtful. The value isn't what most scientists would consider significant - the results could be obtained in error with about 50 per cent probability, which isn't good enough to be considered useful.
What was claimed to be observed is that the cocoa increased blood flow to the dentate gyrus region of the hippocampus in the brain, which is thought to have a role in memory handling. In the trial, the high cocoa group did better at a memory test where they had to remember whether a shape they were shown was one of 40 they had just seen in a sequence. But they didn't do any better in a test where they had to recall words from a list, 60 minutes after three attempts to learn it.
Another oddity of the trial is that no improvement was found in those who performed exercise, even though in a previous trial by the same experimenters had found a benefit. This doesn't rule out the findings, but does emphasise the need to repeat the trial, several times and with bigger groups. Oh, and the authors declared no personal interest, but it wasn't strongly flagged up that the study was funded by the Mars chocolate company.
There seems to be some evidence here that this cocoa-sourced substance might help with the short-term recognition of shapes, which is something we get worse at as we get older. This can't be a bad thing if true. But it isn't a miracle cure for the way that ageing effect our memories, and taken on its own, this trial is not enough even to be sure of that.
You can see the full paper at Nature Neuroscience, though you would need a subscription to read more than a summary. It is Enhancing dentate gyrus function with dietary flavanols improves cognition in older adults - Adam M Brickman, Usman A Khan, Frank A Provenzano, Lok-Kin Yeung, Wendy Suzuki, Hagen Schroeter, Melanie Wall, Richard P Sloan & Scott A Small - Nature Neuroscience (2014) doi:10.1038/nn.3850