I think this is fascinating, although there seems to be some controversy around it. However, this is about socialisation for Asperger’s patients – rather than psychotherapy. So the patient learns interactions inworld and then is able to translate them into RL. Well, that’s the hypothesis.
BUT the article also notes that people with this condition are notoriously inflexible – so if someone in RL uses, say, sarcasm slightly differently to how the patient learnt to recognise it in SL, the idea doesn’t work. I’m looking forward to the results of the research and indeed the new platform they are developing which seems to be more catered towards this client group.