In Lu Hong and Scott Page's 2001 paper in the Journal of Economic Theory, "Problem Solving by Heterogeneous Agents,"1 they construct a model of agents who strive to solve complicated problems described by bit strings. The bit string formulation of thorny problems faced by agents is a modeling approach common in complex systems, often used in agent-based models (game-theoretic computer simulations), but rarely found in pure theoretical treatments in economics, such as those typically found in the Journal of Economic Theory. Hong and Page's model contains adaptive agents who use imperfect information and strive to perform the best they can with their limited resources, vision, and knowledge. A reader of economic theory can see this complex, adaptive, evolutionary approach fully described in the analytical framework of pure theory. This reader can see how these ideas can meaningfully advance their own understanding of this topic. Hong and Page place these ideas -- on one hand, complex and evolutionary, on the other hand, rigorous, mathematical, and game-theoretic -- in dialogue.

This is placing research into the cooperative frame.

Classical and neoclassical economics is a collection of theories and tools that give certain kinds of insights and provide certain kinds of results. The cooperative frame suggests that as evolutionary and complex-systems-oriented economists we are to be in dialogue with those theories and tools. We are to convince them that our theories and tools are not substitutes for theirs, but rather complements. I argue that the cooperative frame is a useful approach for the aspiring evonomist.

My paper with Ken Kurtz "Evaluating Case-based Decision Theory: Predicting Empirical Patterns of Human Classification Learning" in the journal Games and Economic Behavior2 takes this approach. We take the analytical model of case-based decision theory and write a computer program that implements it, allowing us to leverage the structure of CBDT using a computer to find implications of this analytical model that would take quite a long time to work out on a blackboard.  (Computational models of individual choice to guide adaptive agents in a complicated environment is a hallmark of agent-based modeling and complex systems.)

We have the computer generate simulated choice data and we can statistically compare these data to human choice data from the laboratory and accept or reject this or that conclusion. The analytical model of CBDT can be brought to new questions, previously unanswerable. The analytical approach and the agent-based model are complements. They cooperate.

Lamoureux mentions behavioral economics. I define behavioral economics as the intersection of psychology and economics. In that spirit, I place Pape/Kurtz squarely in behavioral economics. Is it a "typical" behavioral economics paper? No: the typical behavioral economics paper imports a concept from psychology through a mathematical/statistical representation and places it in an economic context and/or runs it against economic data. We don't do that, but it doesn't matter. The cooperative frame says I should talk about how we are connected, not how we are distinct. In that spirit, this paper contributes to, and is squarely in, behavioral economics. You see?

The evolutionary approach tells us we can increase the chances of our group's survival through cooperation. The cooperative frame is a kind of code-switching for mainstream economics. One that can also make our work more rigorous, more careful, more precise, more useful to the field, deeper, and persuasive to a larger audience.

I teach a class, Agent-based Policy Modeling, which I consider a kind of class in evonomics. There are two main texts: Miller and Page's Complex Adaptive Systems3 and Elinor Ostrom's Governing the Commons: The Evolution of Institutions for Collective Action.4 Miller and Page's Complex Adaptive Systems is the book to read for an economist with these interests. Emergence is a concept central to complex systems and this perspective: the idea that agents on one level may, through their micro-interactions, create macro phenomena that are unexpected or unintended. Their first example? The invisible hand of Adam Smith. This book is a master class in the cooperative frame.

Chapter 10 of Complex Adaptive Systems is "Evolving Automata" and describes the genetic algorithm.5 For those who don't know, the genetic algorithm is an agent-based and evolutionary approach to problem-solving, in which there are virtual genes "agents" which are candidate solutions to a problem. The genes are subject to evolution: they are allowed to mutate and sexually reproduce, mingling their content, and then the best are selected, and generation after generation this repeats. An aspiring economist should aspire to write their own from scratch in their programming language of choice. Build it. Play around with it. Throw it at some problems. Have it in your pocket when you want to have a community of adaptive computational agents who have individual and collective challenges.

Elinor Ostrom provides a framework for understanding where markets and states come from, as institutions that people constructed to solve cooperation problems, to manage common-pool resources. Like all human institutions, they serve a purpose and they are a current evolutionary solution to some set of problems. The cooperative frame dictates that we think of markets as one of the human institutions we developed that serve purposes, but like all evolutionary solutions, may not be the best in the long run or as the environment changes and might just be the best institutions we have so far.

Arrow and Debreu wrote a complicated and nuanced market model about how a perfectly functional market can trade away risk and the assets can capture the probabilities of good and bad outcomes.6, 7 It is a marvelous, intricate example of neoclassical economics and rigorous mathematics that contains insights about prices and how we would expect them to move. And Ken Arrow also wrote the famous (Im)possibility Theorem,8 in which he shows that reasonably, fairly, and appropriately aggregating preferences in a society to find a solution to cooperation problems can be very, very hard or, as they say, impossible. We are not going to find tidy, analytical solutions to these problems, Arrow tells us. We must muddle through, therefore using adaptive, complex models that are our best attempts so far, revise them, evolve them, and try again.

Take an existing game-theoretic, mainstream, neoclassical analytical model and build an adaptive, complex, evolutionary model version of it. Design your model to have parameters or dials you can set to zero, so the complex model matches up with the analytical model. Then say: but now when we try all these other aspects, we can learn new things and explore in a new direction, but rest assured, dear neoclassical reader, that it is connected to what you already know. That's called docking your model, or agentifying a model.9 This is the advice of the cooperative frame: figure out how to dock your work with mainstream economics. Find the points of collection to bridge your work with theirs, then take them gently by the hand over that bridge.


[1] Hong, Lu, and Scott E. Page. "Problem-solving by heterogeneous agents." Journal of Economic Theory 97.1 (2001): 123-163.

[2] Pape, Andreas Duus, and Kenneth J. Kurtz. "Evaluating case-based decision theory: Predicting empirical patterns of human classification learning." Games and Economic Behavior 82 (2013): 52-65.

[3] Miller, John H., and Scott E. Page. Complex Adaptive Systems: An Introduction to Computational Models of Social Life. Princeton University Press, 2009.

[4] Ostrom, Elinor. Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge university press, 1990.

[5] Holland, John H. "Genetic algorithms." Scientific American 267.1 (1992): 66-73.

[6] Arrow, Kenneth J., and Gerard Debreu. "Existence of an equilibrium for a competitive economy." Econometrica: Journal of the Econometric Society (1954): 265-290.

[7] Debreu, Gerard. Theory of Value: An Axiomatic Analysis of Economic Equilibrium. No. 17. Yale University Press, 1959.

[8] Arrow, Kenneth J. "A difficulty in the concept of social welfare." Journal of Political Economy 58.4 (1950): 328-346.

[9] Guerrero, Omar A., and Robert L. Axtell. "Using agentization for exploring firm and labor dynamics." Emergent Results of Artificial Economics. Springer, Berlin, Heidelberg, 2011. 139-150.