To me, fMRI is the equivalent of measuring economic activity across the world by looking at GDP amounts and growth, and trade balances across countries, regions, or continents. Nobody can claim this or that person or city was specifically responsible for the changes, and that's part of what's built into these coarse yet presumably useful economic indicators: the particulars don't matter. (To the point that many measures are averaged that shouldn't be, hiding massive inequalities in opportunity, outcomes, family budgets, education, and more.)
When an fMRI paper claims to look at neural activity patterns, benevolently I presume the authors are speaking at the analogue level of precision of GDP, economic growth, and trade balances. Perhaps here it is useful to distinguish between "neural" and "neuronal".
Any conclusions on mechanisms responsible for the observed activity patterns must be ignored for there isn't any basis whatsoever on the data. Thankfully these are written in the discussion section of the papers: the opinions of the authors about their own results. Since only the methods and results can be read from these papers, often there isn't much or anything to learn from them: all the claims are in the form of (over)interpretations listed in the discussion section.
