Melting Faces

At Stanford, my graduate career was spent conducting experiments in immersive virtual reality. We were interested in presenting what seemed to be physical reality but then changing the rules in the background. We had subjects interact with an agent who shared 30% of their facial features. Or we had subjects interact with an agent who mimicked their head movements at a 4 second delay. Or we created worlds where other participants would appear to be looking attentively at you even when they were looking elsewhere.

In one particular well-publicized study, we morphed a nationally-representative sample of voting age citizens into either Bush or Kerry right before the 2004 election (at a 35% ratio). We found that people were more likely to vote for the candidate they had been morphed with, enough to have swung the election if everyone were morphed with Kerry. What was even more interesting was that in our sample of about 200 participants, not one detected the self morph.

This line of research intrigued me because it foregrounded the possibility of breaking the rules of interaction in virtual worlds in productive ways. Melting faces could lead to an avatar becoming more persuasive. I was intrigued because it reminded me that virtual worlds were good for doing what was impossible to do in the physical world rather than simply produce replicates of physical reality.