My son sends along a link to Neuroskeptic casting doubt on some, and perhaps even most fMRI data. It seems that the subjects' regular breathing may be confounding the experiments. This would be pretty devastating, as functional MRI's are the sexiest thing going at present in neuropsych.
Neuroskeptic's blog is in general quite good. Other articles include: his decision that the suspicious data of German social psychologist Jens Förster's can only be explained by data fabrication (Well that's not good); an improper use of sockpuppets by an astrophysicist to shut down criticism; and a Swedish study about a physical technique to predict suicide not being anything clearly definable statistically.
Which reminds me, I have some links about predicting violence and sexual-offense recidivism that I keep meaning to fashion into an understandable post without bogging everyone down in jargon. Perhaps this will spur me to do it.