Entirely agreed. Even if you more often than not get the same answers from fMRI and surveys, the fMRI externalizes the judgment of whether or not someone is empathizing/emotional/cognitive stating with regards to something else.

One might argue that we probably have a decent understanding of how well people's verbal statements line up with different facts, but where this diverges from the neurological reality is interesting enough to be spending money on the chance of finding the discrepancies. If we don't find them, that's also fascinating, and is worth knowing about.

Taking for granted that what people say about themselves is accurate, but externalized measurement is also worthwhile for it's own sake.