Also in yesterday's issue of Nature there is an interesting hypothesis regarding the causes of autism. While several genetic factors have been identified that cause a small proportion of the cases of autism, the large majority is still unaccounted for.
Badcock and Crespi discuss here (Nature vol 454, p 1054) their hypothesis (originally published in 2006 but I missed it) that autism might be due to a problem in imprinting, a process which ensures that of certain genes only the paternal copy is used, and of others only the maternal one. As has been discussed before, this can create a battle of the sexes, whereby paternal genes favour a big, demanding fetus (and child) at the expense of the mother's fitness, while maternal genes would tend to minimise this trend. Imprinting affects at least 63 human genes, possibly several hundred.
Badcock and Crespi suggest that autism is due to an imbalance in favour of paternal genes. They go further and say that neurotic disorders (schizophrenia etc.), which in some respects are the exact opposite of autism may be symmetrically caused by an imbalance favouring maternal genes. (Although the authors seem to be unaware of it, the symmetry between symptoms of autism and neurosis have already been discussed in the autism literature. Essentially, if you're autistic you don't care what other people think, and if you're neurotic you care way too much.)
Linking a whole catalogue of mental disorders to imprinting is a bold hypothesis, but it is clearly testable, so within a year or two somebody should be able to work out whether it holds up in practice. If it does, it also suggests treatment options that may become available in the long term.