For some reason fMRI terrifies me in the sense that it will be misused to predict human behavior. The idea that we can effectively read your brain leads to the assumption that we can understand the behaviors, and I think that is false.
http://en.wikipedia.org/wiki/Polygraph#Validity
As with many technologies, though, you can't be a Luddite and slow progress because the general population might misunderstand it.
Such judgement appears unfair, and so (to misapply the conclusions of the original paper) rational people should be uncomfortable with it.
Hm, that is unlucky I supposed fMRI is the modern lie detector, where the test administrator just makes up results that the authorities rely on.