This thesis is composed of three chapters, distinct in focus but unified by the basic question of how communication affects strategic behavior. The first chapter deals with forward induction, a common equilibrium selection principle in game theory which relies on a sort of implicit communication. The second studies communication in a face-to-face environment. The third chapter explores the effects of minimal, explicit communication in a repeated game with frequent actions and imperfect monitoring.Several general conclusions can be drawn from the work presented below. The first is that the sort of implicit, rational communication that economic theory describes does not seem to be widespread. Only about 30% of the subjects who participated in the experiments described in Chapter 1 demonstrated strong evidence of understanding the forward induction principle. The second is that face-to-face communication can have adverse effects on an individual's earnings that are not predicted by economic theory. Thus, it is shown in Chapter 2 that agreeable workers are paid less by their managers in a controlled bargaining environment. The third is that minimal, explicit communication which in theory has no bearing on behavior improves the earnings of individuals a repeated interaction with imperfect monitoring. On the other hand, information management institutions that in theory can allow players to sustain more cooperation can in practice have adverse effects.The subject of Chapter 1 is implicit communication in a strategic environment. Imagine that two players, Person 1 and Person 2, are engaged in a strategic interaction with two stages, and that Person 1 made a move in the first stage. Forward induction is the notion that from this first stage move Person 2 can infer what Person 1 believes will happen later. Theoretical models of forward induction take a prominent place in the literature, but whether or not people make forward induction-like inferences in the laboratory is an open question. In the experiments described in Chapter 1, detailed reports from participants playing a battle of the sexes game with an outside option. Approximately a third of these reports exhibited an excellent understanding of forward induction, and these reports were associated more strongly with forward induction-like behavior than reports consistent with first mover advantage and other reasoning processes. The experiments also provide some evidence that forward induction thinking can be learned through observation of other players' actions. In most laboratory experiments, subjects interact anonymously through a computerized interface. Communication in the field, however, often takes place in environments where agents interact face-to-face. In Chapter 2, I report the results of a study that allowed this kind of interaction to test the hypothesis that an individual's economic choices and outcomes are significantly affected by the personality of others, measured in terms of the ``Big Five." Specficially, the study introduced a bargaining experiment where output was produced by a worker, a manager determined how this output was divided between the worker and herself, and the two parties interacted face-to-face to determine their bargaining positions. The results of this study attribute a significant effect of the worker's personality on her bargaining power.Chapter 3 studies the effects of canonical information management institutions on cooperation in a game with imperfect monitoring. Delay of information is the first institutional manipulation this study considers; the second is the ability of players to communicate their strategies, and the third is bounded rationality in the form of constrains on reaction time. The results of this study show that subjects earn significantly more without delay of information, a result that cannot be explained by standard repeated games models, that communication always improves welfare, and that average payoffs in one of our treatments (with communication and no delay) are significantly greater than the upper bound on public Nash equilibrium payoffs. Exploring the possibility that this is driven by bounded rationality in the form of reaction lags, the study finds that slowing down the experiment has no significant effect on behavior.