Mediating human interaction with video: the fallacy of the talking heads

Given that communication network bandwidth remains at a premium, the design of multiparty mediated systems is a problem of conveying the least redundant information first. We investigated the redundancy of gaze direction in relation to typical full-motion visual cues in performing a triadic abstract office communication task. The results of the experiment indicate that the use of still-video does not necessarily have a negative impact on task performance. Gaze directional cues may provide extra regulative information, but with regard to task performance in small, well-organized groups this information seems to be largely redundant.