Yes, the issue is very much if introspection counts as a form of "detection" by itself, or if it just some illusory or dressed-up version of a previously established "valid detection mode", perhaps requiring consideration of behaviors or social contexts, etc. You and I appear to agree on the former view, regardless of whether or not it is possible to connect that detection to something that shows up in a PET scan, or perhaps more to the point, regardless of what is the footprint of that detection that shows up on a PET scan. I think of scientific learning as a process of taking projections, projections of every kind we can think of. The ancient Greeks were suspicious of some projections, only they couldn't agree on which projections were spurious (abstract ideals or empirical data). Modern science apparently recognizes that one uses a combination of these, because it's not the projection you're using that counts, it is the methodology you apply to the process of making that projection. Science is thus a set of functions, that map from "reality" (which we have no direct access to) onto various "image spaces" that we choose via these projections (this is what we have direct access to). When we apply scientific methodology to find the functions that explain the results of those projections, we are doing science-- there is no need to argue "which projection is the valid one" because scientific methodology doesn't include that question. The question we ask, as apparently understood by Ramachandran, is, "what projections are effective?" The choice of the projection is often the crucial step in scientific discovery, and adjudicating which types of projections are philosophically unacceptable is counterproductive.