In their standard formulations, stochastic games and Markov decision processes assume a rational opponent or a stationary environment. Online learning algorithms can adapt to arbitrary opponents and non-stationary environments, but do not incorporate the dynamic structure of stochastic games or Markov decision processes. We survey recent approaches that apply online learning to dynamic environm...

This paper considers a multi-unit ascending auction with two players and common values. A large set of equilibria in this model is not robust to a small reputational perturbation. In particular, if there is a positive probability that there is a type who always demands many units, regardless of price, then the model has a unique equilibrium payo¤ pro…le. If this uncertainty is only on one side,...

Competition has been studied as something to be averted, yet rarely has it been asked what processes may be involved in successful competition. We tested whether more accurate modeling of an adversary can assist competitive success. Pairs played a zero-sum game with no specific skill component over 40 trials. We measured the relative accuracy of the players’ second-order model (i.e., what I thi...

The prediction of the future states in MultiAgent Systems has been a challenging problem since the begining of MAS. Robotic soccer is a MAS environment in which the predictions of the opponents strategy, or opponent modeling, plays an important role. In this paper, a novel case-based architecture is applied in the soccer coach that learns and predicts opponent movements. Case-Based Reasoning(CB...

This study examined whether blushing after a sociomoral transgression remediates trustworthiness in an interdependent context. Participants (N = 196) played a computerized prisoner's dilemma game with a virtual opponent who defected in the second round of the game. After the defection, a photograph of the opponent was shown, displaying a blushing or a nonblushing face. In a subsequent Trust Tas...

The current state-of-the-art in intelligent game design using Artificial Intelligence (AI) techniques is mainly focused on generating human-like and intelligent characters. Even though complex opponent behaviors emerge through various machine learning techniques, there is generally no further analysis of whether these behaviors contribute to the satisfaction of the player . The implicit hypothe...

The field of multiagent decision making is extending its tools from classical game theory by embracing reinforcement learning, statistical analysis, and opponent modeling. For example, behavioral economists conclude from experimental results that people act according to levels of reasoning that form a “cognitive hierarchy” of strategies, rather than merely following the hyper-rational Nash equi...

Optimization under uncertainty is central to a number of fields, and it is by now well-known that one can learn to play a repeated game without a priori knowledge of the game, as long as one observes the payoffs (even if one does not see the opponent’s play). We consider the complementary scenario in which one does not observe the payoffs but does observe the opponent’s play. Curiously, we show...

The video game industry is an area of high and increasing profitability, with $US 6 billion spent by U.S. consumers alone in the year 2001 [1]. In order to attract buyers, there is increasing demand to simulate human opponents artificially within games. As of present, artificial opponents commonly have their game-playing strategies coded in advance, often resulting in overly simplistic and pred...