Reconceptualising Scientific Understanding in Light of Mary's Room — Epoche C2
What the Mary case was designed to prove Mary is a neuroscientist raised in a monochrome room, who learns every physical fact about human colour vision from black-and-white monitors and textbooks, and who then walks outside and sees a ripe tomato. Frank Jackson introduced her in "Epiphenomenal Qualia" (1982) as the second of two cases — the first concerns Fred, who reliably sorts into two classes tomatoes that everyone else sees as uniformly red — and the argument he built on her is short enough to set out in full, because almost every subsequent dispute is a dispute about one of its four lines. Before release, Mary knows all the physical information there is to have about human colour vision: the spectral tuning of the cone photopigments, the transduction cascade, the opponent recombination of cone signals, the cortical processing, the behavioural and verbal dispositions that result. On release she learns something about colour vision: what it is like to see red. So there was information about colour vision that she lacked while she had all the physical information. So not all information is physical information — and physicalism, which Jackson in that paper defines as the thesis that all correct information is physical information, is false. Two features of the original are worth recovering, because they are usually dropped. First, the title is not decoration. Jackson accepted, and argued for, the consequence that the non-physical properties he was positing are causally inert with respect to behaviour: if they made a difference to what Mary says and does, then a complete physical account of her speech and action would already have captured them. He was prepared to pay this price and defended it against the objection that a causally idle property could not have been selected for. Second, the phrase "what it is like" is not Jackson's. It is Thomas Nagel's, from "What Is It Like to Be a Bat?" (1974), and Nagel's own conclusion is markedly weaker than the use commonly made of it. Nagel does not argue that physicalism is false; he argues that we have no idea how it could be true, because we possess no method of objective description that preserves the point of view from which an experience is had, and he says explicitly that physicalism may be true although we do not understand it. Citing Nagel as a dualist, as introductory treatments often do, gets him backwards. The third feature to recover is that Jackson no longer holds the argument. In "Mind and Illusion" (2003) he rejects it, on the ground that its intuitive force depends on treating experience as having intrinsic features over and above what it represents about the world; on the strong representationalism he came to hold, Mary acquires no new fact, and the persuasiveness of the second premise is diagnosed rather than endorsed. This matters for the shape of the present essay. The knowledge argument survives its author — David Chalmers, in The Conscious Mind (1996), presses a structurally similar case from conceivability to a modal conclusion — but nobody should now write "Jackson's challenge to physicalism" as though the challenger were still on the field. The version that must be answered is the impersonal one. Two questions, only one of which Mary settles The case is put to work on two questions that need separating before anything can be said about either. The ontological question is whether there are facts, properties or items of information that are not physical. The epistemic question is what understanding a phenomenon consists in — whether it is exhausted by holding true propositions about it, or requires something else. The literature runs these together because premise 2 is stated epistemically ("Mary learns") and premise 4 is stated ontologically ("not all information is physical"), and the inference between them is exactly the contested move. My contention is that the case is inconclusive on the ontological question and decisive on the epistemic one, in a direction opposite to the one usually taken. It shows that a conception of understanding as the possession of a complete inventory of true "that P" statements is wrong. But the correct replacement is not acquaintance. It is the capacity to answer questions about what would happen under intervention — and by that criterion Mary in her room already understood colour vision completely. What grasping requires, stage by stage To make that criterion do work rather than serve as a slogan, it needs the machinery of two accounts that are standard in the philosophy of explanation, stated before they are used. The first is James Woodward's interventionist account of causal explanation, in Making Things Happen (2003). On it, a variable X counts as a cause of a variable Y relative to a background of other variables just in case there is a possible intervention on X — an exogenous manipulation that sets X's value without acting on Y by any other route — which changes Y or the probability distribution over Y's values. Explanation, correspondingly, is the exhibiting of a pattern of dependence that answers a range of what-if-things-had-been-different questions: an explanation of a fact is good to the extent that it tells you how the fact would have differed had the conditions been otherwise. Understanding, on this account, is not a feeling that accompanies explanation; it is possession of the answers. The second is the account of mechanistic explanation given by Peter Machamer, Lindley Darden and Carl Craver in "Thinking about Mechanisms" (2000): a mechanism is a set of entities and activities organised such that they are productive of regular changes from set-up conditions to termination conditions. The point of the formulation is that activities are not reducible to the entities that engage in them, so describing a mechanism means saying what the parts do , not just listing them and their properties. With those in hand, the stages that the earlier version of this essay listed can be give