Bowles and Gintis’s A Cooperative Species


Jason Collins


March 26, 2012

Sam Bowles and Herb Gintis have an interesting reputation within the fields of economics and evolutionary biology. The recent paper of Nowak and colleagues has given Bowles and Gintis some competition as the most prominent advocates of a group selectionist approach (or multi-level selection as Nowak and colleagues would term it), but I still have not come across anyone in evolutionary biology who will argue for the importance group selection to the extent they will (as this video of Herb Gintis demonstrates).

A Cooperative Species: Human Reciprocity and Its Evolution continues Bowles and Gintis’s group selection approach to the evolution of cooperation, with many of their earlier papers forming the basis for chapters or models.

The opening chapters are excellent as they tour the findings of experiments into cooperation (particularly dictator games and the prisoner’s dilemma). They argue that these present strong evidence that people undertake actions that are costly to themselves, and they address most of the typical criticisms of the experiments and the interpretations. While I did not always agree, their discussion is thorough and builds a strong argument. In particular, their discussion of whether people could distinguish between experiments and real world situations is well covered, with Bowles and Gintis arguing that people are aware that most experiments are single shot games.

Once we move from the experimental to the theoretical, the remainder of the book is more an exploration of evolutionary game theory models involving cooperation than an attempt to describe the evolution of cooperation. In many ways, I found this approach frustrating. For example, when examining the feasibility of the evolution of cooperation through inclusive fitness, they developed a model in which slight error rates in recognising whether someone was cooperating or not prevented cooperative traits from evolving. They then continually referenced this model as showing that cooperation could not evolve without group selection. That a single theoretical model can form the basis of such a claim, without seriously considering the broad literature and range of models in which it is argued that cooperation can evolve, leaves the argument short. Similarly, they constructed some very complicated game theory models and suggested they were too complicated for mere mortals as there were infinite equilibria.

As a test for these arguments, you might ask whether a generally cooperative approach leads to higher benefits or costs in modern society. It would seem the benefits are huge, with more cooperative people (also having higher IQ and patience) obtaining significant dividends. Even if you make the odd mistake, cooperation appears optimal for purely selfish reasons. If my error rate is so costly and my ability to coordinate with others is so difficult, as Bowles and Gintis imply, why do I find cooperation so rewarding?

At times their points were important. For example, Bowles and Gintis argued that while signalling theory tells you that agents will engage in costly activities to signal, their group selection argument provides a basis for why the signals are “positive” in the sense that people engage in costly cooperation, rather than simply beating the weaker person up. But, by this stage of the book, their dismissal of reciprocal altruism and inclusive fitness has left the obvious answer unavailable.

Their models of group selection are interesting, although hard work. They use an agent based approach, which is useful, but such models are hard to assess without playing with the models and the parameters yourself. Each chapter feels more like an academic paper than book chapter, which had me skimming much of the maths with occasional stops at interesting points.

One of the pillars to their claim that group selection was vital to the evolution of cooperation is that intergroup conflict was frequent and that groups regularly wiped each other out. They survey the evidence for group conflict and suggest that the frequency and severity of war created the intergroup selection pressures necessary for group selection to operate. I tend to agree with the evidence they present concerning war, but the evidence presented by Napoleon Chagnon, among others, suggests that war also has significant individual benefits.

Ultimately, the book’s weakness is the lack of time that is spent addressing models of the evolution of cooperation that do not rely on group selection. If the focus given to defending the experimental evidence was also given to to alternative models of the evolution of cooperation, the book may have delivered some serious arguments. In their absence, I am not sure this book will convince anyone to change their mind.