Interaction with a chair seems like a useful thing to study, as people do spend a lot of time at computers and sitting in a chair.  The specific interface presented is a little problematic, since people aren't likely to turn their chair and keep their head facing the screen.  Other interactions, such as leaning left, right, forward, or backward might be more natural.  These could map to, for example, a person leaning to peek around a corner.  The suggestion given in class of doing "wizard of oz" prototyping to see what types of interaction work best without investing a lot of time in implementing a working system is a good one.