Social and economie exchanges often occur between strangers who cannot rely on past behavior or the prospect of future interactions to establish mutual trust. Game theorists formalize this problem in several “one-shot” game – such as the trust game - predicting noncooperation – since the investor is not expecting trustee to reciprocate it is not rationally to invest. Bohnet and Zeckhauser (2004) suggest that, due to betrayal aversion, people seek to avoid situation in which one could be betrayed. We argue that this behavior could emerge also due to regret aversion.

The past decade has witnessed an unprecedented growth in our understanding of the brain basis of economic decision-making. In particular, research is uncovering not only the location of brain regions where certain processes are taking place, but also the nature of the (economically meaningful) latent variables that are represented, as well as how they relate to behavior. This transition from understanding where to how economic decisions are being made in the brain has been integral to relating neural processes to economic models of behavior. This progress, however, has been notably uneven. Neu-roeconomic studies of individual decision-making, such as those involve risk and time preferences, have the benefit of drawing on decades of work from neuroscientific studies of animal behavior. Critically, many of these findings are based on quantitative, computational approach that lends well to economic experimentation. In contrast, our understanding of the neural systems underlying social behavior is much less specific. A large measure of the current challenge in fact arises from the empirical shortcomings of standard game theoretic predictions of behavior, which are largely equilibrium-based. Using our own study as an example, we show how one can directly search for the latent variables implied by current economic models of strategic learning, and attempt to localize them in the brain. Specifically, we show that the neural systems underlying strategic learning build directly on top of those involved in simple trial-and-error learning, but incorporate additional computations that capture belief-based learning. Finally, we discuss how our approach can be extended to address fundamental problems in economics.

Cet article décrit une approche de la modélisation d'un système d'acteurs, particulièrement adaptée à la modélisation desentreprises, fondée sur la théorie des jeux [11] et sur l'optimisation par apprentissage du comportement de ces acteurs. Cette méthode repose sur la combinaison de trois techniques : la simulation par échantillonnage (Monte-Carlo), la théorie des jeux pour ce qui concerne la recherche d'équilibre entre les stratégies, et les méthodes heuristiques d'optimisation locale, en particulier les algorithmes génétiques. Cette combinaison n'est pas originale en soi, même si elle est rarement utilisée avec toute la puissance d'expression conjointe de ces techniques. La contribution de cet article est double : d'une part nous proposons un modèle qui permet de structurer de façon systématique cette collaboration entre différentes techniques et, d'autre part, nous utilisons la technique des algorithmes génétiques pour enrichir la recherche des équilibres de Nash sous forme de points fixes. Il s'agit d'une méthode de simulation, qui n'est pas destinée à la résolution de problèmes, mais à la validation et l'étude des propriétés d'un modèle associé à un problème particulier.

A cooperative game is defined as a set of players and a cost function. The distribution of the whole cost between theplayers can be done using the core concept, that is the set of all undominated cost allocations which prevent playersfrom grouping. In this paper we study a game whose cost function comes from the optimal solution of a linear integercovering problem. We give necessary and sufficient conditions for the core to be nonempty and characterize itsallocations using linear programming duality. We also discuss a special allocation, called the nucleolus. Wecharacterize that allocation and show that it can be computed in polynomial time using a column generation method.