Functional MRI

The MRI is a very useful tool for providing anatomic information, and it's useful in so many diagnostic situations. Is it possible that we could redirect the use of the functional MRI in more productive directions? I mean, would it help, for example, if we changed from this pooling of data from several people pooling those together? Would it be better to start doing serial studies of individuals? Even there, the problem of reliability raises its ugly head.