Should virtual humans be thought of as actors playing a role in a virtual play or real participants living in a virtual world? Although the question is of philosophical interest, it also has implications in how the virtual human’s knowledge, goals, communication and sensing are implemented. The question is also one aspect of the more general tension between realism and believability in virtual environments. In most cases, components of a virtual world (such as its physics) are made believable as a short cut when fully realistic components are more desirable but prohibitively expensive in development time and/or processing power. In the actor vs. participant case, however, an argument can be made that actors who behave believably, but have additional unrealistic knowledge (more knowledge than their characters would realistically know) and capabilities, are actually more useful than realistic participants. We propose the 2002 Spring Symposium on AI and Interactive Entertainment include a discussion on this subject. This proposal consists of an introduction to the topic and two one-page position papers to get the conversation started.