The evolution of cooperation isn’t so puzzling

There’s a bit of a vibe in evolutionary anthropology/biology/theory that the evolution of cooperation is puzzling. A recent book — The Moral Brain — says “cooperation between unrelated individuals poses a puzzle from both the perspective of natural selection and that of rational self-interest”. Still there are loads (and I mean *loads*) of theoretical models showing a variety of ways for cooperation to evolve. In a moment of heightened procrastination, I decided to delve into the literature and find out just how puzzling scientists find cooperation (warning: this is not even remotely a scholarly piece of work).

First, I searched a bunch of academic databases for all papers about the evolution of cooperation (using the search term “evolution AND cooperation”) as well as those containing variations on the word puzzle (search term: “evolution AND cooperation AND puzzl*”). Here’s the state of the literature:

Puzzle papers All cooperation papers %
Europe PMC 2,190 43,568 5.0%
PubMed 63 2,495 2.5%
Web of Science 144 18,004 0.8%
JSTOR 14,567 131,650 11.1%
Google Scholar 73,200 2,450,000 3.0% (approx.)

Overall, only 3.4% of cooperation papers use the word puzzle.

Next, I mined all the cooperation papers in Europe PubMed Central with a little help from The Content Mine’s getpapers program and a couple of tools I wrote to summarise and visualise commonalities between papers. I’ve put together a pretty, interactive graph where you can look at the most frequent words in common across 125 open access cooperation papers. Click the image to play with the full graph.

Screenshot of the interactive graph of cooperation papers

Then I did some very basic analyses with all 43,568 cooperation papers listed in Europe PMC. PNAS has the highest numbers of cooperation papers at 2,536 ever published, 164 of which contain the word ‘puzzle’. This bar chart shows the ten journals who publish the most evolution of cooperation research along with the number of ‘puzzle’ papers in each:

The number of cooperation/puzzle papers published has increased massively over time, but only a small proportion of them contain puzzlement:

Cooperation was least puzzling in 1976 with only one paper out of 111 published using the word ‘puzzle’ — Evelyn Witkin’s “Ultraviolet mutagenesis and inducible DNA repair in Escherichia coli (which, to be fair, isn’t even about the evolution of cooperation and Witkin doesn’t specify whether or not she finds the topic confusing).

The increase in cooperation papers (puzzling or otherwise) most likely just reflects the increasing number of academic articles published each year. Looking at the proportion of cooperation papers that are puzzling, we see a decline in puzzlement over time:

Of course, some/many of the articles I looked at might refer to something other than cooperation as a puzzle. A better analysis would look for the presence of “evolution” and “cooperation” (and their variants) in the same sentence as “puzzle” and its synonyms. I’ll leave this up to somebody with more flair for procrastination than me.

I’ve put the code to create the graphs into a Gist and you can download the raw data here. The graphs are drawn with colours from Wes Anderson films using Karthik Ram’s colour palette for R.