Since we based all our neurobiological assumptions on reproduced experimental results from different authors they should be considered valid. Quite a lot of neurobiological experimental results don't reproduce, but that is not a phenomenon isolated to fmri but to all of science.