Image from Flickr by Torley

I attended a captivating talk last night by Jeremy Bailenson, director of Stanford’s Virtual Human Interaction Lab. Their research focuses on manipulating social interactions in virtual environments, thereby testing subjects’ reactions to situations that could not occur in physical reality. For example, in virtual reality, my avatar could be made to maintain eye contact with yours at all times, even though in physical reality, I am avoiding eye contact. Creepy, right?

I highly recommend watching the lab tour video, but if you don’t have time, here are some of their most interesting findings he presented:

• Although constant eye contact makes subjects extremely uncomfortable, the result is greater engagement and increased learning. This has significant implications for teaching via VR environments. Imagine if each student perceived that the teacher’s avatar is looking at them.

• When study participants’ avatars were made to be 10 inches taller than their co-participants, they felt more confident and were always more able to convince the “shorter” person of a certain point of view. Same goes for more attractive vs. less attractive avatars. Also, the confidence that appeared in users with taller or more attractive avatars was maintained outside of the virtual environment for at least an hour. (No longitudinal studies have been done yet, but maybe repeated exposures to tall, attractive, virtual selves could increase feelings of self-confidence. VR is already being used to treat various phobias, like the fear of flying, but perhaps it could be extended for more general treatments of mental health.)

• There are almost no obese avatars in Second Life, where prejudice against obesity is as prevalent as in reality. Subjects who spent 10 hours a week using Second Life with obese avatars did not enjoy it, and their decreased self-esteem actually extended beyond the game and into their real lives to a degree unexpected (and undesired) by the researchers.

• Similarly, players in World of Warcraft advance more quickly in the game if their avatars are tall and “attractive,” even with all character skills being equal. Short, “ugly” characters progress more slowly.

• The lab has developed software that can track facial expressions using a regular web cam and relay that information to remote participants. Imagine video chat, but with a 3D on-screen avatar whose face mimics your actual facial expressions. (Watch the video. It’s eerily well-done.)

Side note: The talk was free, and presented by Café Scientifique at SRI. There are lots of local “Café Sci” groups all over the world. Check out their website to find a local group and tons of interesting speakers. The SF group, for example, has an upcoming talk on “How Computers Look at Art.”

Also, if you’re in the Bay Area, note that SRI is also presenting a sort of Doug Engelbart retrospective in December.

Here’s a video demo of my iTunes Library Visualization project. I recommend watching it in HD on Vimeo—click the outward-pointing arrows to make it full-screen and then click “HD” to get the best quality image. (You can also play it above, at low-quality.)

Each track is represented by a disc. Longer tracks are larger discs. The tracks can be organized in space by length and frequency of playback (i.e. most listened-to tracks fly toward the front, least listened-to recede). Grouping by genre adds color and clusters all tracks of the same genre around one point. Once the tracks are colorized, they can be reordered while maintaining the color (so, for example, you could see if you listen to jazz more often than hip hop).

Future enhancements will add text to label groups and track names, and better physics to handle collisions and spatial overlapping.

Built using Processing and the ProXML library.

The latest ad from the Alliance for Climate Protection contains some rotating, 3D type reminiscent of David Small’s Talmud Project. Watch the whole thing, or just jump to 15 seconds in.

If only I had taken my early anticipation project, extended it into the third dimension, and thought to make it more game-like, I might have ended up with something like Coign of Vantage.

While working on my own 3D visualization of my iTunes library, I came across this visualization of Netflix data. It seems interesting, although not very dynamic. It includes several essentially static 3D plots of the data.