We recently had a paper published in PLoS ONE:
A man goes alone to a party. There are not many people there, a few sitting around, and one lone woman who starts looking at him. Very soon, the lone woman walks towards him and starts a conversation. At first the conversation is mundane and she maintains a normal social distance from the man. But eventually she tells him that she thinks he looks very nice, moves much closer to him breaking normal social boundaries, and then asks him whether he is involved with anyone at this time.
How does the man react? Well for the experiment we recruited men who were either were quite socially confident in their relationships with women, or quite socially anxious. Our expectation was that those who were socially confident would simply enjoy the encounter, whereas the anxiety level of those who were socially anxious would go through the roof.
Oh - don't forget that all this happened in virtual reality, in the Cave system at UCL, London.
So what happened in fact? On the whole the men in both groups were more anxious than previously, when the woman first approached. Then for the confident group the anxiety levels returned to what they had been beforehand, but, and to our surprise, the anxiety levels of the anxious group actually seemed to go even lower than they had been beforehand.
I believe that what happened is that men who are socially anxious with women tend to avoid such encounters (because it makes them anxious). In particular they do not initiate encounters, and avoid situations where they might be required to approach the opposite sex - such as dances, parties etc (except perhaps unless they are fortified by alcohol). But here there was a 'woman' who was doing all the work, making all the small talk, and also interspersing a more mundane conversation (about living in London etc) with the more challenging topic of relationships. This allowed the anxious men to relax more than usual, certainly more than they would have expected than when the woman first approached.
I think that this points an interesting way forward in therapy for such social phobic conditions. In this case the virtual woman initiates the conversation, and reduces the 'threat' level, by keeping it at a mundane level for a while. Then more problematic issues can be raised. This interspersing of mundane and more difficult topics in the conversation, could, I think, be a key to therapy in this area.
We carried out this study quite some time ago, but only recently wrote and submitted the paper. Now we can do avatars and scenarios of much higher quality than we used then. Moreover, we now routinely animate virtual characters through motion capture, so overall the experience for any participant has a much greater level of realism. But the interesting point is that in spite of the low level of realism in this study, the men still tended to respond as if it were real. I'm not sure how much additional realism would actually change things.
Finally, social phobia is a hidden illness, with very severe consequences for the sufferers. Many years ago we were doing a case study with Prof. David M. Clark and the patient had a particular form of social phobia - fear of public speaking. This had affected his career - choosing a career where he could avoid speaking in public was a critical consideration for him. He told us he was worried about making a speech at his daughter's wedding. His daughter was, at the time, 3 years old.