What makes a Nobel Prize winner? There's several suggested factors: Perseverance? Good luck? Good mentors and students? Here's one possible factor that I would have never imagined in my wildest dreams; chocolate consumption. Chocolate consumption tracks well with the number of Nobel Laureates produced by a country.
At least that's what a paper published in the New England Journal of Medicine - one of the world's premier journals of medical research - claims. I have to say I found the study bizarre when I read it, and a few hours of strenuous, perplexed thought have done nothing to shake that feeling off. The study itself is amusing and rather brief and I think it makes for entertaining reading; what I am left contemplating is why this paper constitutes serious research and why it would have been published in a journal which over the years has presented some of the definitive medical findings of our time.
The paper starts by assuming - entirely reasonably - that winning a Nobel Prize must somehow be related to cognitive ability. It then goes on to describe a link between flavanols - organic molecules found among other foods in chocolate, green tea and red wine - and cognitive ability. Now I haven't read the literature on flavanols and cognitive ability, but I am sure that flavanols themselves couldn't possibly be responsible for improved cognitive effect, especially when they are part of a complex cocktail of dietary and environmental factors affecting brain function.
But let's say that's true; flavanols are indeed a strong indicator of cognitive function. From this idea the author basically jumps to the dubious and frankly bizarre question of whether chocolate consumption could possibly account for Nobel Prize winning ability. However, from a purely scientific standpoint the hypothesis is testable, so the author decides to simply plot the number of Nobel prize winners per 10 million people in different countries counted from 1900-2011 vs the chocolate consumption in those countries. The figures for chocolate consumption come from Caobisco and Chocosuisse and cover only four years, none before 2002. This fact itself makes any such comparison dubious to say the least; how can you compare two variables when they are sampled from such radically dissimilar sample spaces? And what about other compounds containing flavanols; why not also consider red wine or green tea?
In any case, a plot of chocolate consumption vs number of Nobel Prizes reveals a strong correlation of 0.79. Sweden is an anomaly (and the author thinks it could be a result of "patriotic bias" from the Nobel Committee); take it out and the correlation improves to 0.86. The graph in all its glory is illustrated above.
What does one make of this? Well, I have said before that if only three rules of scientific deduction were inscribed on the doors of every university and research organization in the world, one of them should be that "correlation does not mean causation". Conflating the two can lead you to believe, for instance, that storks deliver babies. Now the author recognizes this, but what I find absolutely baffling is that he makes no attempt to dissect other possible contributing factors. In fact at the end of the article he acknowledges the existence of such factors and then proceeds to dismiss them by saying that "differences in socioeconomic status from country to country and geographic and climatic factors may play some role, but they fall short of fully explaining the close correlation observed."
Some role? Well, I don't know what to say. Or actually, I do. The most likely explanation for that correlation is that it's caused by a third factor. We can be almost sure that for a variable like "cognitive function" which undoubtedly depends on thousands of environmental and genetic factors, there are going to be several co-dependent ones both correlated to it and causing it, and in fact causing each other. There's also going to be thousands correlated with it but not causing it. Chocolate consumption is going to be a small dot in this complex universe of potential factors, and it's certainly not the obvious one which would have occurred to me.
What other factor might possibly be related to the number of Nobel Laureates? I think the graph provides a strong suggestion and I am again surprised that the author missed speculating about it. The Scandinavian countries rank at the top of the graph on the right and they are known to rank high both on the Human Development Index (HDI) and in per capita income. Couldn't it just be possible that higher chocolate consumption simply means greater affluence and an improved lifestyle? Put simply, people who eat more chocolate are likely to be better off (and perhaps even happier?). Greater affluence means better higher education, research opportunities and perhaps Nobel Prizes. I still don't think this set of socioeconomic factors can be directly connected to Nobel Prizes, but I really think it's far more likely to track with Nobel Prizes than chocolates and their flavanols. Other questions besiege my embattled brain; was chocolate consumption in Germany very high during those miserable post-World War 1 years when the country produced all those physics Nobel Prize winners? Does the high percentage of Nobel Laureates in Switzerland reflect its neutral, relatively peaceful status during wars rather than the health benefits of Swiss chocolate?
I have to quite plainly admit that this is one of the strangest and most bizarre papers I have seen in a long time. As I said before, the hypothesis itself is not unscientific per se, but the methodology and conclusions are simplistic at best, the complexity of the system and other possible contributing factors are hardly acknowledged and the paper in general is just not something which I would have expected to appear in the NEJM. Is this really supposed to be serious research? Could it be possible that this is a spoof akin to the absolute zero paper from 1931? My head spins. I think it's time for me to munch on a few Toblerones to improve my cognitive function (the author himself reports a daily diet of Lindt's dark varieties).
Note: I thank Nick Terrett for pointing out the paper and for useful discussion.