Robert Axelrod's short book was not the first to contend that genuine cooperation can emerge both from seemingly non-cooperative and Neo-Darwinian interactions. But it was the first, however, to furnish this argument with its most cogent and clearly interdisciplinary account. In so doing, it has secured its status as a foundational work.  It has become textbook orthodoxy that by unraveling the chemical structure of DNA, Watson and Crick satisfactorily answered Schroedinger's famous question: 'what is life?'  And, similarly, almost as much could be prophesized for Axelrod's wielding of Hobbes' challenge: how is cooperation possible without a central, enforcing authority?  (Or — phrased alternatively — how can cooperation evolve in a world of "pure" egoists?)   Indeed, if the frequency of field-related scholarly citations is indicative of a seminal idea's success among peers qualified to assess it, Axelrod's book passes the test with flying colors. As is revealed in Richard Dawkins' preface for the revised edition, the Evolution of Cooperation's (EC) referencing ratio has been increasing quasi-exponentially since its publication date — elevating some of its themes from professional consensus to nearly established cannon.    

The deployment of a number of theoretical gymnastics could be predictably expected in a work with a scope of this magnitude; unjustified heuristics such as the overuse of specialized jargon, sweeping but unverifiable generalizations and impenetrable statistical or mathematical analysis. Lesser books, to be sure, have gotten away with more. But it is not the case with this one. On the contrary, the EC's greatness lies precisely in supplanting rhetorical and technical frill for commonsensical intelligibility. Nonetheless, various sections and appendices (some even featuring harder math!) have been added for the expert's relish in more complex formulations of cooperative phenomena, extending thereby the work's credit.  Axelrod's central tenet, consequently, is rendered with extraordinary — though somewhat deceptive — simplicity. That is, what the minimum conditions need to be in order for cooperative behavior to evolve without assuming in the process a priori cooperation. In other words, without positing or 'begging' the same notion of cooperation that is being put to trial, as other works on the subject seemed to have done.   Therefore the onus, and a particularly weighty one, is on the author to produce a convincing mechanism with which to demonstrate how cooperation emerges out of the participating agents' self-interest. This Axelrod does adroitly with the prisoner's dilemma.

As game theorists have for long noted, the context-based "currency" of punishments and rewards for differing interactions doled out in the prisoner's dilemma lends itself naturally for modeling the goals of agents engaged in fulfilling their self-interest. Moreover, it proves, with uncommon accuracy, how (given a certain projected time-frame) rational self-seeking behavior is in the long run best served by the operation of seemingly contradictory strategies involving genuine cooperation; hence the paradox embedded in the term dilemma. Note that, due to Axelrod's pragmatic use of the concept of strategies, "agents" can be anything from computer programs to living organisms to commercial firms or international decision-makers. And the measuring 'currencies' can range as well from information to surviving offspring to dollars or negotiating power — and it is in this ample margin of applicability, thus, that the EC's interdisciplinary luster shines through.  Accordingly, different chapters are devoted to exploring several factual applications through detailed case studies.  Chapter five on the biological evolution of altruism, for example, was co-written with William D. Hamilton — a founding figure of the so-called Neo-Darwinist turn in theoretical biology. Besides earning the authors the Newcomb Cleveland Prize of the American Association for the Advancement of Science, thereby becoming a little classic in its own right, this chapter formed the basis of an important addendum to Richard Dawkins' magnum opus, The Selfish Gene.

The book's most salient contribution, however, is its design to stage those minimal conditions necessary for cooperation to develop over a specified number of interactions (rounds) amongst self-interested agents.  Given the strong requirement for simplicity, the chief difficulty consisted in finding the most parsimonious way to pit self-maximizing, competing agents against each other. In addition, the agents' behavior had to be as straightforward so as to fit squarely within the model's targeted minimalism, implying no more foresight than that predetermined perhaps by a sort of algorithm or rule of thumb. Consequently, these desiderata made its enactment ideal in a series of computer simulations, which resulted in the book's two celebrated tournaments. Being no fun to play against oneself, Robert Axelrod called for interested participants, and the respondents were as oddly assembled as the strategies which they submitted (computer hobbyists, political scientists, mathematicians, economists, evolutionary psychologists and biologists signed up, among others.) To the shock of all, the same strategy, Anatol Rapoport's TIT FOR TAT, won both tournaments. More shockingly, its closest competitors were just slight variants of its uncomplicated design, defeating in the process much more elaborate programs. Mocking reciprocal altruism in biology, TIT FOR TAT always cooperated on the first move, and just responded after in kind to its antagonist's preceding action. If the latter was previously cooperative, for instance, TIT FOR TAT cooperated also — but if its opponent defected, TIT FOR TAT retaliated.   Besides being carefully tabulated and scrutinized, these results are properly bolstered and contextualized with the rest of the claims in the book.

Although its content remains descriptive for the most part, the EC's does commit to issue, on chapter 7, some normative recommendations. For it adopts there — somewhat unabashedly — the role of the "reformer" to show "how the strategic setting itself can be transformed in order to promote cooperation among the players." (p. 124) But it does so quite prudently. And so the book wraps up with five resulting guidelines, all well-anchored in its theoretical framework, but which are advised nevertheless to be implemented cautiously: 1) enlarging the shadow of the future; 2) changing the payoffs; 3) teaching agents to care about each other; 4) teaching reciprocity; 5) Improving recognition abilities.

© 2007 J. Elias Saiden-nunez

J. Elias Saiden-nunez has fulfilled a visiting stay at Lund University, Sweden. His current research interests lie in the philosophy of biology and complex systems; particularly in areas regarding emergence and agent-based consciousness. He has also published work on modern philosophy, metaphysics and epistemology.

Categories: Genetics, Psychology