The correlation doesn't even show a link. Run GPT on other inputs, run fMRIs on other inputs: call that dataset, (G, F). Now consider all possible subsets, (g

The correlation doesn't even show a link. Run GPT on other inputs, run fMRIs on other inputs: call that dataset, (G, F). Now consider all possible subsets, (g_i, f_i) in (G, F) ... How many show correlation in their choice of NN property, and their choice of brain property (ie., blood flow)? My guess: *many*. This is trivial to refute if you have any sense of the scientific method. Construct the converse hypothesis and test it. They didnt.