At each time frame, the 15 estimated parameters for the Gaussians can be rendered for viewing as the stick figure in Figure 3.3. This is also the display provided to each user so that he may view the gestures of other human (or computer) players through his personal computer screen.
The output is kept simple to avoid confusing users into believing that more sophisticated perception is taking place. This is critical since users will immediately realize how to use this limited channel expressively. If a sophisticated rendering of a face and hands was used on top of the stick figure, humans might mistakingly believe that these features contain information that is being relayed to the learning system. Thus, we avoid misleading the users and hope that they will emphasize large head and hand gestures.
