As information technology has become more widespread, it is now increasingly possible for decision makers, such as CEOs and legislators, to obtain advice from a variety of experts with di!ering areas of expertise. Conventional wisdom suggests that, as a result, the decision maker will reach better decisions. This, however, ignores the fact that experts may withhold or distort information in the hope of influencing the decision in a direction favorable to them. Thus, a decision maker needs to consider how the structure of communication -- the conversation --affects how informative is the advice received from experts. This project investigates how the structure of communication a!ects the quality of decisions. The first part of the project concerns conversations between an uninformed deci- sion maker and an informed expert. When the expert sends a written report, unac- companied by any dialogue, this results in substantial information loss. A face-to-face meeting between a decision maker and an expert, however, leads to an improvement in the informativeness of expert advice--even though the decision maker is completely uninformed. Both the decision maker and the expert benefit from the conversation. This leads us to examine the following questions: How much information can be cred- ibly transmitted in a face-to-face meeting? Is it possible to induce the expert to fully disclose his information? In many situations, the decision maker himself may be knowledgeable about some aspects of the decision. As an example, an entrepreneur may have detailed knowl- edge about the growth prospects of his product, but still require financial expertise in taking his company public. The second part of the project concerns communication in such situations--where both the decision maker and the expert have some rele- vant information. Once again, the structure of communication can critically affect the amount of information exchanged. For instance, there are many circumstances where back-and-forth communication does not lead to information gains--even if the decision maker is informed--and we identify these circumstances. This leads us to ask the following questions: Is the situation improved by adding more rounds of com- munication? Is a face-to-face meeting superior to a back-and-forth exchange, such as via E-mail? The third part concerns situations where decisions are multi-faceted and informa- tion is dispersed among many "specialist" experts. For instance, a piece of legislation may impact both employment and the environment. Different lobbying groups are likely to possess expertise in each of these areas. We study whether providing experts with an opportunity to communicate with one another improves the informativeness of their advice. Is the situation improved if the decision maker also participates? Is the situation improved if there is overlap in the experts' specialized knowledge?