Or from a somewhat different perspective, if you provide information on preprocessing and if you label your figures correctly, reporting coordinates, then the reader can still compare activation patterns with an appropriate template. But he/she does not know how the actual data looked like. As long as everything worked well it's not dramatic, but still, it would be interesting to know e.g. whether there was no effect or whether the brain region was not covered at all at the scanner or was not part of the analysis due to low intensity. Accordingly, Poldrack et al. (2008, Neuroimage, https://dx.doi.org/10.1016/j.neuroimage.2007.11.048 ) suggested to make available the voxel mask of the analysis. Masks would indeed be interesting as one can expect many studies to include distorted data (failed coreg, failed normalisation in some subjects, impacting group results). ;-)

Best

Helmut