Information, strategic behavior and learning in games : an experimental analysis

The current doctoral thesis is comprised of three distinct papers with a unifying theme of studying human behavior in strategic environments of incomplete information. Each paper considers a particular manifestation of information guiding, shaping or in some other way, affecting human behavior when the environment is not known to the fullest extent. The general contribution of this work goes to the literature that originated in the second half of the last century with the works of such prominent economists as George Akerlof, Kenneth Arrow, Michael Spence, George Stigler, and is now usually referred to as the economics of information. The first paper takes the popular perspective on information by considering it a stock variable, that is, knowledge that can be used by players as a tool to achieve certain goals. Specifically, it investigates whether knowledgeable players can be better off than their uninformed opponents in a number of repeated 2x2 games of incomplete information. The second paper considers information as a flow variable by studying the relation between the amount of payoff feedback the choice of action provides and speed with which human subjects learn to play repeated 2x2 games of incomplete information. The third paper examines information as a screening protocol. It takes the well-established notion of conditional cooperation in the public goods game framework to see how sensitive the concept itself as well as its corollaries are to the type of information used to define them in the first place. The main research method used in all three papers is that of an economic laboratory experiment. The underlying theoretical framework relies on the modeling concepts from game theory and behavioral economics. Both parametric, including regression analysis, and non-parametric methods of statistical analysis are used. The first paper also uses computer simulation techniques.