Frank Jackson on What Mary Didn’t Know

i) Introduction

ii) Knowledge?

iii) Knowledge By Description & Knowledge by Acquaintance

iv) Mary's New Abilities

v) Knowing How & Knowing That

vi) Conclusion

************************

Mary “doesn't know what it's like to see red”. This argument has nothing to do with imagination (or Mary’s inability to imagine red). As Frank Jackson puts it: “Powers of imagination are not to the point”. This is about Mary’s knowledge (or lack thereof), not her imagination. More precisely, “she would not know” what it's like to experience red. More to the point,

“if physicalism is true, she would know; and no great powers of imagination would be called for”.

Knowledge?

The first response to this is to ask what Jackson meant by the word “knowledge” (or by the words “knowledge of red”). This seems like an odd use of the word “know”. How would Mary (or anyone else for that matter) know what red is like? What is the epistemology of knowing red? Even if Mary could sense red: could she also know red (or what red is)? Could she (or anyone else) be wrong about what is or isn't red (i.e., without inter-communal responses)?

Is Jackson’s conclusion correct? That is:

i) If physicalism is true

ii) then Mary would know what red is.

Firstly, Jackson argues that given Mary’s “fantastic grasp of neurophysiology and everything else physical” she couldn't thereby work out the last phenomenal part of red “by making some more purely logical inferences”. Mary can't, then, infer the phenomenal from the purely physical - no matter how complete and exact her physical knowledge is.

Knowledge By Description & Knowledge By Acquaintance

Jackson then makes a distinction which has often been made in various areas of philosophy: the distinction between “knowledge by description” and “knowledge by acquaintance”. Presumably Mary had knowledge by description before she was let out of her black-and-white room. After she was let out, then she had knowledge by acquaintance. In other words, her previous complete descriptions of red weren't enough. On freeing, she also became acquainted with red (not only with red’s physical “supervenience base”, as Jaegwon Kim puts it).

Does this also mean that phenomenal red literally can't be described? This is a position which has been long accepted by many philosophers. That's why colours (or “colour words”) are taught (so the argument goes) purely by ostension – by the teacher pointing to something red and then saying to the student, “This is red.” However, if red is (in effect) purely phenomenal, and taught by ostension, then how can Mary (or anyone else) have knowledge of red?

Mary Acquired New Abilities, Not New Knowledge

I've questioned Jackson’s use of the notion of knowledge. So too does David Lewis and Laurence Nemirow. They do so by distinguishing knowing (or learning) that something is red from acquiring “a certain representational or imaginative ability”. How can the sudden new experience of red (outside the black-and-white room) be a knowledge of red? How does Mary learn something new? She experiences something new; though she doesn't learn something new or acquire new knowledge.

However, something new does happen to Mary. As stated, she acquires a certain representational or imaginative ability. Presumably that ability is to recognise red on further occasions (or to distinguish red from other colours). Though how would she know that it's red unless someone else tells her that's the case? This would be especially relevant if Mary's new experience of red was sudden and had no direct connection to her previous examinations of red’s physical micro-structure. Outside the room she would see something new; though how would she know that it's red? Indeed how would she have known that it was a colour of any description?

Knowing How & Knowing That

Earlier I commented on Jackson’s use of the knowledge by description and knowledge by acquaintance distinction. Now he brings in the “knowing how” and the “knowing that” distinction (as used by David Lewis).

Mary now knows how to recognise red. She has that “ability”. Does she know that it's red? Not if, as argued, she has no new knowledge of red at all. And even with such knowledge, she would still require third-person help (as it were) to tell her that the red wall outside is indeed red.

We can join up the two oppositions here:

knowledge by acquaintance - knowledge how

knowledge by description - knowledge that

However, according to this discussion, neither knowledge by acquaintance nor knowing how are, in fact, examples of knowledge (strictly speaking). Only knowledge by description and knowing that are true examples of knowledge. Alternatively, perhaps we can't have one without the other. That is:

i) We can't have knowledge how without knowledge that. Or,ii) We can't have knowledge that without knowledge how.

Mary needs knowledge that red is red before she can learn how to distinguish red from other colours. Alternatively, she must know something in order to know that it's a colour or that she has had a new experience outside her room. To know that red is red, she must know what red looks like. How does this work for the other distinction? Thus:

i) We can't have knowledge by acquaintance without knowledge by description. Or,

ii) We can't have knowledge by description without knowledge by acquaintance.

This works in a similar way to the above - and for similar reasons. How do we know we're acquainted with something (an x) without the help of some form of description? How do we know we're now acquainted with red without some kind of ostensive definition (or some other kind of help)? We may know that we're acquainted with something new; though not that it's red or even that it's a colour of any kind. Alternatively, red can't be described to us in order to give us knowledge that without our also being acquainted in some way with phenomenal red. Without being acquainted with something, we wouldn’t know what it is that's being described.

Yet what if both distinctions are false disjunctures between ostensibly two alternatives? Perhaps this is like Ned Block’s distinction between “phenomenal consciousness” and “access consciousness”. The difference here is that Block admits that in this case you may not be able to have the one without the other. Nevertheless, this shouldn't stop us making the distinction because it's still, after all, an acceptable distinction.

Conclusion

The important conclusion to all this is that

“a physicalist can admit that Mary acquires something very significant of a knowledge kind – which can hardly be denied – without admitting that this shows that her earlier factual knowledge is defective”.

Physicalists aren't, then, denying that extra little something. They only deny the increase in Mary’s knowledge. Thus it's strange that Jackson (at least at this point in his career) still insisted in using the word “knowledge” - suitably reduced to his “of a knowledge kind”. What did he mean by that? A kind of knowledge is still an example of knowledge, isn’t it?

Thus perhaps all this depends on what Jackson and physicalists mean by the word “knowledge”!

References

Block, Ned, 'On a Confusion About a Function of Consciousness' (1995)

Jackson, Frank, 'What Mary Didn't Know' (1986)

Lewis, David, 'What Experience Teaches' (1990).

A Wittgensteinian Fragment on Solipsism and the Self

According to Wittgenstein, the solipsist is contradicting himself when he says, "Only what I see exists." That's because the I/self itself isn't seen in experience. Therefore, like the rest of the things he doesn't see, he has no right to talk about the I/self at all.

As Quine also said, if we have no “criterion of identity” for an object, then we should deny that object any (official) existence. Thus the I/self has no criterion of identity and therefore it should be thrown overboard.

The solipsist doesn't exist by his own standards because the I isn't seen, only the things that the I sees are seen. Thus the subject is indeed “a vanishing point” (Tractatus). There's nothing to see, hear, or smell; but that which the I itself sees, hears and smells. Thus perhaps we should write the I/self in quotation marks because it can be taken as a word without a referent or designation.

What is a person? 

Is it something over and above (to use David Pears' words) “the subject which is living this mental life” or “the subject which is having these visual impressions”? Surely some thing needs to have a mental life or have visual impressions. Or do we just intuitively presume that there needs to be a subject of a mental life or a subject of visual impressions?

Instead of the word 'person', we now meet the word 'subject'; which doesn't seem identical. A subject is a subject of something whereas a person needn't be the subject of something. (At least this is the case on certain readings of the concept [person].)

What is a subject? 

Subjects in subject-predicate expressions are the subjects of predication. Can we predicate anything of the subject? We can predicate experiences or cognitive operations of the subject. For example, when I say "The rose is red", I'm predicating an attribute (or property) of the rose. However, experiences or cognitive operations aren't the same kind of predicate as “red”. These are things the subject does or carries out. The rose, on the other hand, doesn't do red, carry out red or experience red. (This is like saying “John works on the farm” rather than “John is tall”.) Do we have predicates of the latter kind which can be predicated of the subject? We can say what the subject does or what the subject experiences; though can we attribute properties to the subject? It's these kinds of criteria of identity that are missing from the subject. No; the subject has no intrinsic identity-conditions. However, if the subject has vanished, then how can it be the subject of experiences or of cognitive operations? Something must surely experience things and carry out cognitive operations.

Self-Reflexivity and Self-Identity

The notion of reflexivity (i.e., to self-referentiality) or, the more closely related notion, self-reflexivity.

We can start off by saying that a reflexive relation is something that certain things can have to themselves. Or, alternatively, literally everything has the reflexive relation of self-identity .(If this is a genuine relation, or, indeed, a genuine property.) Stones are seen, by some, to be self-identical; though they are seen by no one as being self-reflexive. The self-identity reflexive relation (though not the self-reflexive relation) is something everything has to itself.

The self-identity reflexive relation or property, for example, is sometimes called a “strongly reflexive” relation. A “weakly reflective” relation is, on the other hand, sometimes stated as “everything has it to itself that has it to anything”.

We can have beliefs about our beliefs, thoughts about our thoughts, and so on.

A thought about a thought is not self-referential or self-reflexive. It's not the thought itself that refers to or thinks about itself (or even the content of the thought that does so): it's the person who has the thought who thinks about his own thought. The thinker himself is self-reflexive, not the thought itself. The thought is not self-referentially related to itself either. It’s just that the thought itself is thought about. Again, it's the owner of the thought (so to speak) who is self-reflexive (about his thought). The thought (or thoughts) about the thought can therefore be called second-order thoughts. Again, the base thought (as it were) is neither self-referential nor self-reflexive. A thought could, of course, be self-referential; or at least its content could be. However, I don’t think that the thought (or a thought itself) can ever be self-reflexive. The possible self-referentiality of a thought’s content would not also be an example of self-reflexivity.

There is an example of self-reflexivity which is more closely related to examples of self-referentiality.

Once example is what takes place when the self/person analyses the (same) self/person. That is, when the subject and object of self-analysis can be said to be one and the same thing – i.e., the self that analyses is the (same) self that is analysed. Thus, in the light of what we've said about self-referential statements, we can now ask:

Can the self/person both have thoughts (about himself) and also be the object/subject of these thoughts?

This wouldn’t be like the putative property self-identity which was referred to in the last paragraph. (In which it was said that all objects are said, by some, to have the property of being identical to themselves.) There is no such strict identity when it comes to the self/person and the self’s thoughts about itself. On a traditional reading, thoughts about the self come and go; whereas the self’s substance stays put. Thus, on this reading at least, thoughts about the self are not identical to the self itself. And even if we don’t believe in selves as bare particulars, the thoughts about the self would still not be identical to the self the thoughts are about. This would even be the case if part of the self (at a particular point in time) included those thoughts about the self (that has such thoughts). In this instance, particular thoughts about the self can’t be identical to the self taken as a totality at any given time. There may, however, be a partial identity (at least if we accept the bundle theory of the self); though not complete identity (which is the case if we accept the property - or relation – self-identity).

To give a taster of self-reflexivity or possible self-referentiality, here are a few points that put obstacles in the way of such states.

We can ask:

How can a human subject take himself to be the subject of study; as Descartes famously claimed he did when sitting next to his fire?

Perhaps we can never actually get to the essence (or totality or complete truth) of ourselves because it's precisely the same self that's trying to get to the essence, etc. of itself. It is, therefore, attempting to be both subject and object. (In a Cartesian manner, if we can’t discover our own “true essence of mind”, how can we even begin to discover the true nature of external reality?) According to many philosophers, it's a Cartesian myth that we can objectively “perceive” our own minds or states of consciousness. It's not a fact, then, that the mind (or the items within it) are “transparent” to the self-analysing mind itself. Perhaps we can never get to every item or working within our own minds. Perhaps we can't, specifically, recognise that it has been “conditioned” to, say, believe certain things or see particular things in certain ways. To put it bluntly, if it's the self that is analysing the (same) self (or the self that's analysing itself), then the “unseen” conditionings, workings, non-transparent items, etc., of the self (as object) would also be the conditionings, workings, non-transparent items, etc. of the self that's doing the (self) analysis. It could therefore only be a self which is split into two (quite literally) which could be both subject and object at one and the same time. Though then we can simply say that after such a split we will no longer be talking about two selves, encamped within the same mind-brain: not one self that's split in two. And if that were the case, then there would be no authentic self-reflexivity and certainly no self-referentiality.

Proper Names and Individuation (3)

“… the role of the general term is to identify the referents – not to identify the ‘kind of identity’ asserted.” (John Perry, 1970)

The first kind of identification referred to above is simply a matter of ostension: of pointing (which may include a Kripkean ‘naming-baptism’); whereas the other kind of identification is dependent on the “kind of identity”. Michael Dummett expresses the first kind of identification thus:

“A bare knowledge of the reference of the name a will consist…in knowing, of some object, that a refers to it, where this is a complete characterisation of this particular piece of knowledge.” [Dummett, 1991]

Dummett argues against this distinction. He says:

“…there cannot be a proper name whose whole sense consists in its having a certain object as referent, without the sense determining that object as referent in some particular way.” [Dummett, 1973]

Dummett must think that the name, say, ‘Tony Blair’ is a de facto rigid designator: a designator that denotes via the mediation of some concept or description. A de jure designation, on the other hand, is an ‘unmediated’ reference. That is, a rigid de jure designator denotes what it denotes without the mediation by some concept or description.

Can we have basic naming (or the “identification of referents”) without the latter kind of identification? What is at the ostended point? Doesn’t individuation (therefore sometimes the identification of kinds) come before the basic kind of Kripkean naming and identification? Wouldn't this be an epistemological question, rather than an ontological one? It would be a question about our knowledge of an object, rather than (the reality) of the object. It may still be the case (in conformity with Kripke) that we can only know an object to be the object it is through its ontological essence. Here again we see a philosopher making the same distinction:

“Kripke did make a feasible distinction between a description giving the content of a name and merely fixing its referent.” [Jason Stanley, 1997]

According to Marcus,

“the causal theory of names elaborated by Kripke and others. Proper names, unlike definite descriptions, can be used by speakers to refer to objects without mediation of ‘concepts’ or descriptive clusters. They can be used to capture and institutionalise an act of ostension.” [Marcus, 1978]

My argument is that without individuation we can't have ostension. We need to bring along with us the concepts that will help us individuate a given spatial region or physical mass. The concepts don't come from that spatial region or physical mass itself, otherwise we would have concepts that wouldn't necessarily belong to the object within that spatial region or physical mass. Therefore an act of ostension does require conceptual mediation. How would we know when one object ends and another object begins? How would we distinguish an object as an object at all?

The traditional problem was that if referents depended on concepts (or ‘descriptive clusters’), then each person would bring along his own concepts (or descriptive clusters). It was supposed to follow from this that a reference couldn't rely on a single concept or a single descriptive cluster. The referent would therefore be lost amongst rival - or simply different - conceptual contents of names.

Why was it thought that if the reference couldn't rely on specific concepts that it could escape all concepts? Everyone would still be having causal contact with the same referent [see Donald Davidson, 1989]; though that causal contact alone wouldn't tell us which concepts or descriptions were the correct ones (if there are correct ones). If everyone were still having the same causal contact with the same object (or the same spatial region or physical mass), then that wouldn't matter. Different or even contradictory concepts would still be related to the same object or spatial region/physical mass.

Thus, in an indirect sense admittedly, acts of ostension could still be ‘institutionalised’ [Marcus, 1978]; thought we couldn't guarantee uniformity of conceptual choices. The object wouldn't be lost; though the guaranteed uniformity of concept choices would be. And, of course, if different people brought the same conceptual baggage with them, then that in itself would/could generate a certain degree of conceptual uniformity. However, that wouldn't be guaranteed. And it could be possible to generate agreement between different or contradictory concepts if the concepts of the object or spatial region referred back to the object as it is “standardly recognised” [Paul Moser, 1993]. Of course what is standardly recognised would still be a contingent matter. Even the standard conceptual description of the object would still not be the way the object describes itself. Therefore there's no reason to suppose that what is standard today will be standard next week.

Bertrand Russell's Ontic Structural Realism? (2)

Mathematical Structuralism

Bertrand Russell puts the case for for what may be deemed a mathematical structural realism when he wrote the following:

“Physics is mathematical, not because we know so much about the physical world, but because we know so little: it is only its mathematical properties that we can discover.” (163)

That appears to be an admission that mathematics is used in physics because we don't know everything that is to be known about the object, condition, event, etc. under description or scrutiny. Mathematical equations and values are the mere bones of the physics. Mathematics only deals with structures; not with what have been called “intrinsic properties”. This position seems to be at odds, then, with the realism of James Ladyman, Don Ross and other contemporary ontic structural realists.

It's here that there are two options available. One, to become a kind of thing-eliminativist and say, “Every thing must go." Or two, one can adopt a quasi-Kantian line and say that there are things-in-themselves (or intrinsic properties); though the problem is that we don't know anything about them. Indeed, as with option one, we can eliminate "substances" or things-in-themselves from our ontology.

Russell then elaborates on his structural and mathematical ontic realism. He states the obvious point that “[p]hysics is mathematical”. And then says that when it comes to the “physical world” it's “only its mathematical properties” that we have access to. Such properties are all “we can discover” (163).

The Maths of Point-instants

In one place you find Russell speaking about “point-instants” which seems to be a perfect example of ontic structural realism. Firstly he tell us that “we can define a point-instant in space-time as a group of events”. Then he he says that

“the 'points' (or point-instants) that the mathematician needs are not simple, but are structures composed of events, made up for the convenience of the mathematician”.

This is an articulation of structural realism, not a realism about or towards things. Not only that: when Russell tells us that these structures are “made up for the convenience of the mathematician”, this too sounds like some kind of constructivism – ontic structural constructivism! Then again, the word "structural" (or "structuralism") is, in itself, a tacit commitment to some kind of constructivism.

It is clear that Russell has taken the idea of a “centre” omitting radiations from Werner Heisenberg. Russell tells us that

“Heisenberg regards a piece of matter as a centre from which radiations travel outward”.

These radiations constitute the realism of this picture in that “radiations are supposed really to occur”. However, the irrealism (as it were) about things (or about centres) is put when Russell argues that the matter at their centre is reduced to a mere mathematical fiction.

Russell's Kantianism?

Russell puts what can be called the Kantian stance on these issues and problems. Basically, Russell believes that all we have are what he calls the “effects of a thing-in-itself”. Thus Russell comes to the conclusion (as contemporary ontic structural realists have done) that if we only have access to effects (or to external properties), then why not factor out that distinction between thing-in-itself (or intrinsic properties) and its effects (i.e., external properties).

We can now ask about those effects. What is left of the things-in-themselves? In terms of the science of this issue, Russell writes:

“We find that energy in various forms spreads outwards from various centres; we find also that such centres have a certain degree of persistence, though this persistence is not absolute...”

Prima facie, this seems like a simple substitution of the word "thing" (or "piece of matter") with the word "center". Instead of a thing “having a certain degree of persistence”, it's simply these physical centres which do so. Nonetheless, it can still be said that these centres are, effectively, things (or pieces of matter). In other words, are we simply debating the correct usage of words here?

Russell continues by expressing the basic science of an electron or proton and why it is that we take them to be things. Thus, in point of fact,

“the modern physicist faces cheerfully the possibility than an electron and a proton may mutually annihilate each other, and even suggests that this may be the main source of the radiant energy of the stars, because when it happens it makes an explosion”.

Thus firstly we discover that the electron and proton have some kind of mutual relation with one another. That alone will raise questions as to their ontological reality as separate entities. In other words, if two things - a and b - always have a necessary and “mutual” relation to one another, then what right have we to see them as distinct entities in the first place?

That later possibility is scientifically elaborated upon when Russell tells us “[w]hat can be asserted” about these matters. He writes:

“When energy radiates from a center, we can describe the laws of its radiation conveniently by imagining something in the centre, which we will call an electron or a proton according to circumstances, and for certain purposes it is convenient to regard this centre as persisting, i.e. as not a single point is spacetime but a series of such points, separated from each other by time-like intervals. All this, however, is only a convenient way of describing what happens elsewhere, namely the radiation of energy away from the centre. As to what goes on in the centre itself, if anything, physics is silent.” (165)

Whereas earlier I said that the word "centre" was being used as a substitute for the word "thing", now Russell speaks of '"spacetime points" instead. Thus even though Russell says that “what goes on in the centre itself, if anything, physics is silent”, he still feels comfortable talking about "spacetime points".

Reference

Russell, Bertrand. (1927, 1970) An Outline of Philosophy.