What are the most advanced theoretical models about "learning" such as learning-by-doing or learning from others?

Overall, learning in economics spans different areas. Here there is a largely incomplete list:

  • One of the pioneers of "learning" in economics is Arrow with his 1962 paper. In his paper, Arrow discusses the economics implications of theories about learning in social psychology. The main focus here is about firms "learning" over time. (For a recent survey of models within this literature: Thompson, 2010)

  • In the same year (1962), G.S. Becker adds another milestone. Becker's paper is a must read. From the point of view of individual agents, learning is an investment: it involves immediate costs and gives future gains. It raises the question of why firms should pay money for training their workforce in competitive markets (i.e., wages equal marginal costs). It also makes an important distinction between "general" (reusable) and "specific" knowledge.

  • Following up on Becker's work, a large literature in labor economics (e.g., Rosen, 1972) pushes the same question a little bit further. This literature culminates in Acemoglu and Pishke (1999) paper showing that market frictions are a possible explanation for firms paying for training.

  • A more recent literature has started to look at "learning" from a wholly different point of view: how people learn to play games. This literature originates from the results of many experimental works showing a number of anomalies in behavior compared to theory. A standard reference here is the book "Learning in games" by Fudenberg and Levine. (The book is somewhat old and there is much of research going on in this area. So I am not sure what would be an updated reference.)

Can you help me create a better list? I am looking for either area of work I have overlooked (e.g., is there something in behavioral literature?) or more advanced references for each of the listed topics.

  • $\begingroup$ This is quite a broad question- one with which we won't be able to provide a detailed answer. I would suggest looking at fields such as artificial intelligence, machine learning, and evolutionary game theory. $\endgroup$
    – ml0105
    Commented Jul 1, 2016 at 18:19
  • $\begingroup$ Thanks. I know. Perhaps I should modify the question adding a few references I've found. However, do you have any specific author in mind? $\endgroup$
    – mrb
    Commented Jul 1, 2016 at 19:36
  • $\begingroup$ To clarify: I am not much interested in "learning in games" but more in learning in the sense of acquiring new skills from experience, problem-solving, collaboration, observation. I understand that "learning in games" is also all of that but it seems to me a different issue. $\endgroup$
    – mrb
    Commented Jul 1, 2016 at 20:00
  • 1
    $\begingroup$ That's artificial intelligence and machine learning. Russell and Norvig is the classic AI text. The techniques in evolutionary game theory are closely related to machine learning techniques. $\endgroup$
    – ml0105
    Commented Jul 1, 2016 at 20:24
  • $\begingroup$ @ml0105 Provide a link and post this as an answer? $\endgroup$
    – Giskard
    Commented Jul 1, 2016 at 22:11

1 Answer 1


If you aren't interested in learning from a game and decision theoretic perspective, I'm not exactly sure why you're posting on an economics forum. Because that's what learning really is, from an economics perspective. AI really breaks down into three subfields: logical, behavioral, and machine learning. The logical subfield deals with logical reasoning, automated proof-writing, automated conjecture making, and such. Behavioral AI is really more in line with theoretical economics- game theory, decision theory, mechanism design, etc.

Machine learning is really what constitutes mainstream AI research nowadays. This is really due to the fact that we now have the resources and tools to better deal with big data, rather than a philosophical revelation or technical breakthrough. Hot topics in computer science are very industry and money driven (as a theorist, I would argue too much so). Machine Learning really boils down to applying a stochastic process and refining a model. There isn't so much learning going on, just enhanced stupidity that works well sometimes. That's not to dismiss Machine Learning, but we should call it what it is.

I would definitely recommend AI, Machine Learning, and Evolutionary Game Theory, though. Russell and Norvig is the established classic AI text, and it has sections on Machine Learning as well. Link: http://aima.cs.berkeley.edu/

Evolutionary Game Theory uses similar techniques as Machine Learning. It relaxes the assumptions of Neoclassical Game Theory. Agents need not be perfectly rational. They can have imperfect information and adjust their strategies myopically. The game is played repeatedly. We then see how "mutation" (such as human error or experimentation) in enough agents drives changes in equilibrium. So we have a dynamical process to select equilibria in games. We can apply dynamics such as the imitation dynamic, the best response dynamic, etc. Weibull (https://mitpress.mit.edu/books/evolutionary-game-theory) and H. Peyton Young are both good introductions (https://www.amazon.com/Individual-Strategy-Social-Structure-Evolutionary/dp/0691086877/).

I took a seminar in Economics this past spring, the first half of which was Evolutionary Game Theory. We went through a bunch of papers as well; a subset of the following. One paper I enjoyed was Competitive Behavior in Market Games: Evidence and Theory. It showed that as the number of agents was made sufficiently large, that evolutionary forces drove the Nash equilibrium to the Walrasian equilibrium (under certain assumptions). This provides a behavioral motivation for the study of r-fold replica economies in the general equilibrium setting.

Mark E. Schaffer (1989): “Are Profit-Maximisers the Best Survivors?” Journal of Economic Behavior and Organization, 12, 29-45.

Mark E. Schaffer (1988): “Evolutionarily Stable Strategies for a Finite Population and a Variable Contest Size,” Journal of Theoretical Biology, Vol. 132, No. 4, 469-478.

Hehenkamp, B., W. Leininger, and A. Possajennikov (2004): “Evolutionary equilibrium in Tullock contests: spite and overdissipation,” European Journal of Political Economy, Vol. 20, 1045–1057.

Duffy, J., A. Matros, and T. Temzelides (2011): “Competitive Behavior in Market Games: Evidence and Theory,” Journal of Economic Theory 146, 1437-1463.

Kandori, M., G. Mailath, and R. Rob (1993): “Learning, Mutation and Long Run Equilibria in Games,” Econometrica, 61, 29-56.

P. Rhode, and M. Stegeman (1996): “A Comment on ‘Learning, Mutation, and Long-Run Equilibria in Games’,” Econometrica, 64, 443-449.

W. Sandholm (1998): “Simple and clever decision rules for a model of evolution,” Economics Letters, 61, 165-170.

Young, P. (1993): “The evolution of conventions,” Econometrica, 61, 57-84.

Young, P. (1993): “An Evolutionary Model of Bargaining,” Journal of Economic Theory 59, 145-168.

M. Saez-Marti and J. Weibull. (1999): “Clever agents in Young's evolutionary bargaining model,” Journal of Economic Theory 86, 268-279.

Matros, A. (2003): “Clever Agents in Adaptive Learning”, Journal of Economic Theory, 111, 110-124.

Ellison, G. (1993): “Learning, local interaction and coordination,” Econometrica, 61, 1047-1072.

Glenn Ellison (2000): “Basins of Attraction, Long Run Equilibria, and the Speed of Step-by-Step Evolution,” Review of Economic Studies, 67 (1), 17-45.

Bergstrom, T. and Stark, O. “How Altruism Can Prevail in an Evolutionary Environment.”

American Economic Review, May 1993 (Papers and Proceedings), 83(2), 149-55. Eshel I., L. Samuelson and A. Shaked (1998): “Altruists, Egoists, and Hooligans in a Local Interaction Model,” The American Economic Review 88(1), 157-179.

Matros, A. (2012): “Altruistic Versus Egoistic Behavior in a Public Good Game,” Journal of Economic Dynamics and Control 36, 642-656.

Robson, A., and F. Vega-Redondo (1996): “Efficient Equilibrium Selection in Evolutionary Games with Random Matching,” Journal of Economic Theory, 70, 65-92.

Josephson, J., and A. Matros (2004): “Stochastic Imitation in Finite Games,” Games and Economic Behavior, Volume 49, Issue 2, 244-259.

  • $\begingroup$ Hi, I am not sure that "evolutionary game theory" is about learning. I understand it is about the creation of norms for the selection of particular equilibria in games. But this does not necessarily involve "learning" in its most natural meaning. For example, suppose there is a game with multiple equilibria: one is good for society and one is bad. People create norms and follow these norms to eventually hit the good state. How can this behavior be considered "learning" as opposed to "coordination" or even pure "imitation"? $\endgroup$
    – mrb
    Commented Jul 4, 2016 at 17:09
  • 1
    $\begingroup$ @mrb Much of learning is imitation. Imitating a movement, a tought pattern, etc. If you do not include this in "learning" then perhaps you should give a very precise definition. Evolutionary game theory is a model of bounded rationality where in equilibrium all actors play optimally given their environment. Evolution provided the way the solution is reached, but this is actually also a method for an algorithm to learn the optimal strategy. $\endgroup$
    – Giskard
    Commented Jul 4, 2016 at 17:54
  • $\begingroup$ This is why modern artificial intelligence is considered enhanced stupidity, which happens to get reasonable results in certain cases. In machine learning, it's really a "refine the model" approach. Evolutionary game theory also allows for actors to make mistakes. If most folks are playing a Pareto dominated equilibrium, it does not make sense for a small fraction of players to deviate to a Pareto Optimal norm and evolutionary forces will drive them back to the societal norm. Hence, a peer pressure model, which is realistically a common way folks learn. $\endgroup$
    – ml0105
    Commented Jul 4, 2016 at 18:29
  • $\begingroup$ The book The Theory of Learning in Games is by Fudenberg and Levine, not Tirole. I own a copy of this book. It is definitely an evolutionary game theory text. Your other papers study motivation for and investment in higher education, rather than how folks physically learn these skills. $\endgroup$
    – ml0105
    Commented Jul 4, 2016 at 18:34
  • $\begingroup$ @ml0105 thanks for spotting the wrong author. The reference is correct now. $\endgroup$
    – mrb
    Commented Jul 4, 2016 at 20:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.