Does Mary Learn a Fact, or Only a Way of Knowing One?

Bu yazıyı site için hazırladım, bir ders ödevi değil. Bölümdeki metinlerin nereden geldiğini girişte ayrı ayrı yazdım.

---

Does Mary Learn a Fact, or Only a Way of Knowing One?

1. The Argument and Its Shape

Jackson's (1982) thought experiment is famous enough to state briefly. Mary is a brilliant scientist confined to a black-and-white room who learns everything physical there is to know about colour vision. On release, she sees a red tomato. Does she learn something? The intuition is that she does. If she does, then there was a truth she did not know despite knowing all the physical truths. Therefore not all truths are physical truths, and physicalism is false.

What makes the argument interesting is not the intuition but its logical form. It moves from an epistemic premise — Mary did not know something — to a metaphysical conclusion about what there is. Every serious response to the argument attacks that transition, and the various responses can be sorted by where exactly they think it fails.

My view is that the ability hypothesis fails, that the phenomenal concept strategy succeeds in blocking the argument, and that it succeeds at a price that its defenders sometimes understate: it commits physicalism to an explanatory gap it can describe but not close.

2. Denying That Mary Learns Anything

The most direct response is to deny the intuition. Dennett (1991) argues that we cannot actually imagine what it is to know all the physical facts, and that our confidence about what Mary would experience is an artefact of failing to take the stipulation seriously. A Mary who really knew everything physical might well recognise red immediately and be unsurprised.

This is less obviously wrong than it is usually taken to be. The thought experiment asks us to evaluate a scenario that no one has ever been in, on the basis of intuitions formed in ordinary epistemic conditions. The reply, however, proves less than it needs to. Even if Mary could predict her own reaction perfectly, the question is whether her prior knowledge included what it is like. Prediction of a response is not obviously the same as possession of the phenomenal knowledge, and simply insisting that it is begs the question against the argument.

Churchland (1985) offers a subtler version: the argument equivocates between knowledge-by-description and knowledge-by-acquaintance, which are different kinds of knowledge and not different facts. This is the right neighbourhood, and it leads directly to the strongest reply.

3. The Ability Hypothesis

Lewis (1988) and Nemirow (1990) concede that Mary gains something but deny that what she gains is propositional. She acquires abilities: to imagine red, to recognise red, to remember red. Know-how is not knowledge-that, so no new fact is learned, and the metaphysical conclusion does not follow.

This is elegant and, I think, ultimately unsuccessful, for two reasons.

The first is that the abilities seem to be consequences of the new knowledge rather than constituents of it. Mary can now recognise red because she now knows what red looks like. Reversing that explanatory order is possible only if we already accept the hypothesis.

The second is the case of an unrepeatable experience. Suppose Mary sees a colour once, in conditions she can never reproduce, and afterwards lacks the ability to imagine or recognise it. It still seems that during the experience she knew something she had not known before. If the knowledge can be present when the abilities are absent, the knowledge is not the abilities.

I take the second objection to be decisive. The ability hypothesis identifies something real that happens to Mary but does not exhaust what happens.

4. The Phenomenal Concept Strategy

The most promising physicalist response grants that Mary learns a fact and denies that it is a new fact. She comes to know an old fact under a new mode of presentation.

The machinery is familiar from elsewhere. Someone can know that Hesperus is visible in the evening without knowing that Phosphorus is, although there is one planet and one fact. Concepts are individuated more finely than the properties they pick out. Phenomenal concepts — the concepts we deploy when we think about experiences as we undergo them — are, on this view, distinctive in their mode of presentation but refer to physical properties (Loar, 1997; Stoljar, 2005).

Jackson (1986) anticipated a version of this and objected that Mary learns something about other people too: she learns what their experiences are like, which looks like new propositional knowledge about states of affairs beyond herself. The phenomenal concept theorist answers that she learns of those states under a concept she could not previously deploy, which is still a change in concepts rather than in the inventory of facts.

Notably, Jackson (2003) later abandoned the argument, on grounds close to this: he came to think the intuition can be explained by representationalist accounts of experience without abandoning physicalism.

5. The Cost of the Strategy

Here is where I want to press, because the strategy is usually presented as a clean victory and it is not.

To block the argument, the physicalist must claim that phenomenal concepts are special in a specific way: they must be conceptually isolated from physical concepts, so that no amount of physical knowledge yields them a priori. But they must also refer to exactly the same properties physical concepts refer to.

That combination is coherent. It is not obviously explicable. What is it about a physical property that makes it graspable under two conceptual modes which are, in principle, inferentially insulated from each other? Chalmers (1996) presses this as a dilemma: either the special features of phenomenal concepts are themselves physically explicable — in which case the explanation should let us derive the phenomenal from the physical, which is what was denied — or they are not, in which case something non-physical has been admitted after all.

I do not think the dilemma is decisive, but I think it shows what the physicalist has bought. The phenomenal concept strategy converts a metaphysical problem into a psychological one: why do creatures like us have concepts of this peculiar kind? That is a legitimate move, and it is a promissory note.

6. A Structural Observation

Step back from the specific replies and a pattern appears that I find more instructive than any of them.

The knowledge argument, the conceivability argument for zombies, and the explanatory gap are three presentations of one difficulty: our physical descriptions and our experiential descriptions do not entail one another, in either direction, and no amount of detail in the first seems to close the distance to the second.

The physicalist responses that work all have the same shape. They locate the failure of entailment in the concepts rather than in the world. The anti-physicalist responses all have the same shape too: they treat the failure of entailment as evidence about the world.

Which shape one prefers depends on a prior commitment about how much weight conceivability and a priori derivability carry as guides to metaphysics. That is why the debate has proved so stable: the disagreement is not about Mary but about method.

7. Conclusion

So: does Mary learn a fact, or only a way of knowing one?

My answer is that she learns a way of knowing a fact, and that the ability hypothesis mislocates what that way consists in. She does not merely acquire abilities; she acquires a concept that she could not have acquired from the outside, and deploying it she comes to know, in a new manner, something she already knew in another.

That answer saves physicalism from the knowledge argument. It does not save physicalism from the underlying problem, which is that we have no account of why a physical world should contain properties accessible under two such radically different modes. The argument fails as a refutation and succeeds as a demand for explanation — which, for a thought experiment three pages long, is not a bad result.