Sounds interesting. Too bad the article is so weak on details. BTW, this sort of thing can already be done in computer neural networks using back-propogation or some of the other methods of neural training. Of course, living neurons are likely to have better training algorithms, and I want to know them!