In the physical world, every user exists at one and only one place, but in a collaborative virtual environment (CVE), other paradigms are achievable such as a user existing at more than one place at a time. In a collaborative environment, a user is typically engaged with one primary activity at a time. As the number of collaborative activities increases, users are unable to maintain focal attention on all activities, and must offload some cognitive effort to a peripheral attention sphere. This delegation of attention-the movement of primary activities as focal attention to secondary activities as non-focal attention-requires that users remember certain parameters of context switching, such as what the secondary activities are, and more importantly, when to switch their focal attention to these activities. Keeping track of these context -switching parameters is itself a cognitive load that often degenerates focal attention on primary activities. Our goal is to augment users' cognition by delegating the work of remembering context -switching parameters to other entities in a collaborative environment. We call these entities avatar proxies because our implementation is in a CVE where users are iconified as avatars, but the techniques and results are general to a broader range of collaborative environments. Avatar proxies notify users when they are required to switch their focal attention to secondary activities.