Back    All discussions

2009-01-28
On an argument for Russellian representationalism
We can understand (simple) Russellian representationalism as the view that the phenomenal character of a color experience supervenes on the color that is represented in the experience (together with the lighting conditions represented; let's leave that qualification tacit).

I discuss (1) an argument for Russellian representationalism about phenomenal character (2) some consequences of the argument (3) a reply to the argument (4) a moral in opposition to "phenomenological specimenism" (maybe not in Hinton's sense).

1. I think that something like the following simple argument is in the back of the mind of many Russellian representationalists. 
Suppose that you and Alva are looking at the same red strawberry under white light, and conditions are normal for both of you. 

To you, the strawberry looks red. Putting the point another way: thinking about the color you are visually aware of when you look at the strawberry, which one is it? Answer, red.

Now suppose that Alva is long-term spectrally inverted with respect to you. You know how to imagine a spectrally inverted subject, and how to determine what it's like for one of them to see something with this or that color. So reflect on what seeing the strawberry is like for Alva. When you do so, don't you find that the strawberry just *looks green* to Alva? Putting it another way: projecting yourself into Alva's experience, and thinking about the color he is visually aware of when he looks at the strawberry, which one is it? Answer, green.

If you need some prompting here, look over at a lime. To you, the lime looks green. Thinking about the color you are visually aware of when you look at the lime, which one is it? Answer, green. Reflecting on Alva's experience, now, isn't the color he is visually aware of  just the very same color you are visually aware of when you look at the lime? Answer, yes. So Alva is visually aware of green when he looks at the strawberry.

Now plausibly, for a strawberry to be visually represented as green is just for it to look green, or for its subject to be visually aware of green when looking at the strawberry. Accordingly, you represent the strawberry as red and Alva represents it as green. In this case, the difference in represented color tracks the phenomenal difference. Generalizing, it always does: in line with Russellian representationalism. 

(Thoughts along these lines show up in Byrne and Hilbert 1997's powerful case of Fred (Colors and reflectances, 270, 2d last para); also in Chalmers's case for "edenic content" (Perception and the fall from Eden, 262, after the Shoemaker quote); also early on in Jeff Speaks's forthcoming PPR paper. There's an obvious association with various things people have said about "transparency", but I won't belabor this point.)

2. If Russellian representationalism is true, then (short of endorsing relativism about the manifest image) one can't accept both of these doctrines: (A) neither Alva nor I make any perceptual mistake about the strawberry (B) colors are monadic properties which stand in familiar unity relations to one another (e.g. nothing is both red and green). 

(The literature contains alternatives to both (A) and (B). Concerning A: the friend of Edenic content (Chalmers) thinks both of us make an error in our phenomenal content, mis-ascribing edenic colors (though this could be combined with a view on which Fregean content correctly ascribes physical colors in both cases). The hardline Russellian physicalist (Byrne, Tye) thinks at least one of us makes a perceptual mistake, mis-ascribing a physical color. Concerning B: a color relationist (Cohen?) thinks that we're both correct, in that the strawberry is red-for-me and green-for-Alva. A friend of "centering features" (Egan) thinks that we're both correct-relative-to-ourselves, in that relative to me, the strawberry instantiates the centering feature redness, while relative to Alva, it instantiates the centering feature greenness. A color pluralist (Kalderon?) thinks that we're both correct, since the strawberry is both red and green. 

It also contains alternatives to our Russellian view. A "complex Russellian" view would assert that it's not the *color* represented that determines phenomenal character, but some other property (such as a disposition to cause a response, or a higher-order property of colors, or a primitive non-color property: Shoemaker, Thau). A representationalist alternative to Russellianism, of course, is Fregeanism, on which the phenomenal character supervenes on the sense of the experience (B. Thompson). Finally, perhaps nonrepresentational properties make a difference: an "impure" representationalist could assert that phenomenal character supervenes on qualified representational properties, so that you Rly represent redness while Alva Gly represents redness; while a qualia theorist could assert that phenomenal character does not supervene on representational properties.)

If you like (A) and (B) and dislike relativism, you'll have to defeat the argument for Russellianism. 

3. The key manoeuvre in the argument is the assessment of Alva's experience as one in which the strawberry "looks green" (or the assessment of the color that Alva sees the strawberry having as = green). 

Let us examine the reasoning underlying this manoeuvre more closely.

When I reflect on Alva's experience, I imagine myself having an experience with the phenomenal character Alva's is alleged to have -- let's call it the G character.

When I then go on to *describe* Alva's experience, I use the same conceptual faculties I use when reacting to my own experiences. If I imagine it as having a certain character, I then describe it using the same concepts I deploy when I myself have an experience with that character. Ordinarily, when I have experiences with the G character, I judge that the things I thereby see are green, and that the color I thereby see = green. So I thereby judge Alva's experience as being one in which something looks green/in which the seen color = green.

I might instead have used my conceptual faculties in a different way. I might have "inverted" my conceptual dispositions, rather than using them "straight". That is, when I imagine it as having the G character,  instead of describing it using the concepts I deploy when I have experiences with the G character,  I might have used the concepts I deploy when I have experiences with the R character. Ordinarily, when I have experiences with the R character, I judge that the things I thereby see are red, and that the color I thereby see = red. Had I inverted my conceptual dispositions, I would have judged Alva's experience as being one in which something looks red/in which the seen color = red. 

Using inverted conceptual dispositions, the argument for Russellianism would fail. The argument only goes through if it is reasonable to use my conceptual dispositions "straight", rather than inverted, in this context. What then justifies using one or the other of the straight or inverted conceptual dispositions? 

One might think that, by the oft-mentioned link between perceptual and belief content, I should not be using my conceptual dispositions, but rather Alva's. After all, otherwise there would be no guarantee that the contents I ascribe to his experience will appropriately match the contents of his beliefs. 

If I do that, an argument can be made that I should be using the inverted conceptual dispositions. Plausibly, Alva's beliefs about colors are mostly true. If so, Alva judges that the strawberry is red. But he does so on the basis of an experience with the G character. Accordingly, Alva's conceptual dispositions are the inverted ones. So since I should be using Alva's conceptual dispositions, I should be using the inverted dispositions.

But if so, the key manoeuvre in the case for simple Russellian representationalism fails. 

4. There's a more general moral to be drawn here about phenomenology. It can seem as though, when taking the pseudo-first person perspective on the experiences of others, I can treat them as specimens: "simply examine" them, and classify them on the basis of how they seem in the light of this examination. That doesn't seem right. A more reasonable approach is to project myself into them, and classify them from the inside. 

I think the case for Russellian representationalism results from this mistaken phenomenological specimenism. First, I need to recognize that this classification requires the use of color concepts as if in reaction to the experience. Second, this classification requires the use not of *my* color concepts, but of those of the other.


2009-01-28
On an argument for Russellian representationalism
Reply to Benj Hellie
I'm a bit unclear about the details of the argument for the key claim that the strawberry looks red to Alva. On a first pass it seems to go like this:

1. The correct conceptual dispositions to use are Alva's
2. If the correct conceptual dispositions to use are Alva's, the strawberry looks red to Alva

There are two different things conceptual dispositions could be:

a. Dispositions to associate certain sounds with certain meanings--what one means when one produces various sounds, e.g. "red" and "green".
b. One's set of beliefs about the world, including dispositions to form certain judgments in certain circumstances.

On either interpretation, we can agree that Alva's conceptual dispositions are acceptable. We can use terms the same way he does and adopt his beliefs. His language is the same as ours, and his relevant beliefs may be stipulated to be true (or I'm willing to grant that they are), so no problem adopting his conceptual dispositions on either reading--no problem with premise (1).

I think (2) looks a little less good, however. On interpretation (a) it's hard to see how speaking Alva's language when describing his experience (i.e. our own language) should incline us to say that the strawberry looks red to him. I don't see anything in your post supporting such a move, anyway.

My guess is that you're going with interpretation (b), and that the intended argument for (2) goes something like this:

i. If the correct beliefs to have about how things are and look in the situation (i.e. the correct conceptual dispositions on reading (b)) are Alva's and he thinks that the strawberry is red, then the strawberry looks red to him.
ii. Alva thinks the strawberry is red.

So premise (2) above is true.

Here I think a proponent of the kind of representationalism at hand would question premise (i). You seem to be moving without argument from the antecedent to the consequent ("Had I inverted my conceptual dispositions, I would have judged Alva's experience as being one in which something looks red/in which the seen color = red"). But one might object that we should distinguish between epistemic and non-epistemic senses of look. In the epistemic sense, maybe believing that something is P requires it's looking P to one. But representationalists can say that the kinds of "looks" which are relevant to the representational content of experience are not the epistemic ones. Everyone agrees that there are epistemic and non-epistemic looks, so that leaves a big gap in your argument.

There is of course the challenge of spelling out what the relevant non-epistemic notion of look is. Here I agree there is an issue for representationalists of the type your have in mind. Personally, I don't think the representational contents of experience which make representationalism true have much to do with how things look in experience except in something like Frank Jackson's phenomenal sense of "look", and obviously the pro-representationalist argument you outlined doesn't work as stated on this view. So I agree with your conclusion, but I think that a more direct line of argument here would be simply:

1) Alva believes the strawberry to be red (as shown by his reports, and the fact that he speaks English)
2) If (a) [[I meant (1) here]], then the strawberry looks red to him in the epistemic sense (or at least it doesn't look green in the epistemic sense)
3) If the strawberry looks red (or does not look green) to him in the epistemic sense, then there is no clear sense of "look" such that a) it looks green to him in that sense and b) it's so looking to him implies that his experience represents greenness.

I would say more about (3), but I've got a busy server to attend to right now.. :)

2009-01-28
On an argument for Russellian representationalism
Reply to David Bourget
Hi David,
Thanks for the reply.

I make a negative claim and two positive claims. The negative claim is that the argument for Russellianism fails. The weaker positive claim is that a certain alleged mode of support for Russellianism actually undermines it. The stronger positive claim is that Russellianism is false.


Concerning the negative claim: if it's unclear how to decide claims about how things "look" in any "nonepistemic" sense -- in particular, to the extent that there is no clear meaning to the intuitive claims about how things look on which the arguments of Byrne and Hilbert, Chalmers, and Speaks rely -- then the negative claim goes through. 

You seem to agree with me on this. (Autobiographically, I'm happy to rest here: I think nonedenic Russellianism is quite out of the question, and am not especially happy about the edenic view.) But you also seem to go further: you seem to agree with the antecedent. I'm not so sure about that.

(Note though that the familiar alleged distinction among senses of 'look' is not obviously germane to my argument: I express the case not just in terms of how things look, but also in terms of which color is "presented in experience". I'm not sure whether there is more than one coherent sense of this expression. I'm also not sure whether there is more than zero. One might reasonably press that point.)

Concerning the weaker positive claim, I describe a specific procedure for assessing how things look in the relevant sense that I think the Russellians have in mind: project oneself into the experience, and then "apply" color concepts on the basis of the phenomenal character of the projected experience. Your "semantic" or (b)-type reading of the discussion is the correct one here. The thought then is that the referents of those color concepts are the colors presented in the experiences.


That's the "alleged mode of support" for Russellianism that I claim to undermine it. The stronger positive claim, that Russellianism is false, follows if (I) this mode of support does indeed undermine Russellianism and (II) the procedure just outlined for assessing claims about which color is presented in an experience is legitimate. Byrne and Hilbert et al I take it agree with (II); perhaps you disagree with (II), since you claim I "move without argument" in advancing (II) -- the argument, of course, appears in the text above, namely the stuff diagnosing what the Russellians are up to.

The question for assessing (I) is: "how shall we "apply" the color concepts?" I claim that if we should apply them straight (in accord with the  rule: having an R-experience, deploy a concept of red), the Russellian wins, and if we should apply them "inverted" (in accord with the  rule: having an R-experience, deploy a concept of green), the Russellian loses. (If it's not obvious what the answer is here, then the negative claim goes through.)

Now I also claim that we should apply them inverted. The argument goes (a) Alva applies them inverted, (b) we should apply them as Alva does. The argument for (a) is that otherwise Alva's beliefs about which colors things have are massively wrong. The argument for (b) in the initial post appealed to a familiar line on what the content of a given experience is, namely (roughly) that belief tracks it ceteris paribus (cf Byrne and Hilbert, 263--4). (This is of course distinct from the "epistemic use" of  'look' as you characterize it, as requiring actual belief.) Another case occurs to me just now: otherwise, when Alva reasons: 'lo, something red; therefore this color presented in my experience = red', the argument is, absurdly, invalid.

Obviously some Russellians accept the first line. Those who do must grant (I) and abandon Russellianism. I'm inclined to think that any Russellian should accept the second line, since it seems to be in line with the rhetoric of transparency. Those who reject both need to say something about why we should apply the concepts straight rather than inverted; otherwise it's not clear what the answer is and the negative claim goes through.

Finally, concerning your argument: that's a nice argument; obviously (1) and (2) are correct, I'm not sure about (3). I'm inclined to doubt it will have much dialectical force against Russellians, since most of them will reject (3) (especially the ones who advance the argument I discuss).


2009-01-31
On an argument for Russellian representationalism
Reply to Benj Hellie
Hi Benj,

I'm mainly taking issue with (II), the claim that the procedure to ascertain what is presented in experience which you described is correct. First, I'm not clear on the procedure. What kind of concept applications are relevant? Surely not all concept tokenings correctly, rightly triggered by a G experience are relevant to what is represented in a G experience. For example, one is right to judge that something doesn't look yellow when one has a G experience of it, and one is also right to judge that it doesn't look purple. So one rightly has conceptual dispositions to token the concepts PURPLE and YELLOW when having G experiences. Obviously, it doesn't follow that G experiences represent yellow and purple (and this is clearly not a peculiarity of negative judgments). While I agree that Alva ought to have a disposition to token the concept RED as a result of having G experiences--namely, a disposition to judge that the cause of his experience is red--I don't think this fact indicates that his G experiences present things as red to him. Or at least I would like to know why you think it does.

I was also suggesting that the procedure you describe is not endorsed by representationalists, but this question is probably best left aside until the procedure has been clarified.



2009-02-02
On an argument for Russellian representationalism
Reply to David Bourget
Hi David,
Looks like you've got two questions here:

1) "What kind of concept applications are relevant? Surely not all concept tokenings correctly, rightly triggered by a G experience are relevant to what is represented in a G experience. For example, one is right to judge that something doesn't look yellow when one has a G experience of it, and one is also right to judge that it doesn't look purple. So one rightly has conceptual dispositions to token the concepts PURPLE and YELLOW when having G experiences. Obviously, it doesn't follow that G experiences represent yellow and purple (and this is clearly not a peculiarity of negative judgments)"


Answer, the concept applications I have in mind are those involved in giving a positive specification of the color the object has.


(I wonder if you're on to some subtlety here that I'm missing.)

2) "While I agree that Alva ought to have a disposition to token the concept RED as a result of having G experiences--namely, a disposition to judge that the cause of his experience is red--I don't think this fact indicates that his G experiences present things as red to him. Or at least I would like to know why you think it does."


There's more to the argument that his G experiences present red than just that Alva has the inverted dispositions. The other premiss (to reiterate) is that when we're assessing which color Alva's G experiences present, we should use his dispositions -- I give two arguments for that claim in the above discussion. If you want more detail, go back and read my two previous posts -- there's not much to be gained by my repeating them here.

2009-03-10
On an argument for Russellian representationalism
Reply to Benj Hellie
Hi Benj, finally got the time to follow up here. I think the procedure you describe is not relevant to the contents of experience because one's (correct) dispositions to judge (based on perceptual experience) that an object is a certain color involve one's background beliefs. Suppose for example that I know I'm looking at a white wall under a green light. I would rightly be disposed to judge that the wall is white, so my experience would represent whiteness if your procedure for determining its content were right. I'm pretty sure that's not what representationalists intend (there is certainly no indication that David C takes edenic contents to be determinable that way, anyway). Examples like this can be multiplied at will. I don't even have to see an object to be rightly disposed to judge that it is green based on my experience. Suppose for example that I see the shadow of a tree. I might rightly be disposed to judge that the tree is green.

2009-03-13
On an argument for Russellian representationalism
Reply to David Bourget
... ceteris paribus ...

2009-04-04
On an argument for Russellian representationalism
Reply to Benj Hellie
Benj is right, in my opinion, but I want to come at his point from the opposite direction. 

Here's the case: I view a strawberry in good light and have an experience with colour-phenomenal character R, while Alva, performing the same operation, has experience with colour-phenomenal character G, where G is different from R.  The question is: does Alva's experience tell him the same thing about the strawberry as mine tells me, or does it tell him something different?  

Observe first that the different phenomenal character of his experience does not imply that he gets a different message, because his colour-phenomenal-character to colour-of-object mapping might be different from mine.  After all, different sounds convey the same meaning: you say to-may-to, I say to-mah-to, but we mean the same thing thereby.  The strawberry case might be like this.  The argument in Benj's point 1 is widely accepted, but I for one don't know why.

The fact that both his experience and my experience are evoked by the same strawberry tells in favour of the message being the same.  After all the light is good, and the object-colour is the same.  (Compare:  how does the gavagai man learn the word for 'strawberry red' in a language he is studying?)  And we could make the evidence even stronger by seeing what happens when we present Alva and me with objects that a spectrometer certifies as being exactly the same as the strawberry in spectral reflectance.  Suppose it turns out that in such cases, Alva and I just about always have experiences of character G and R respectively.

At this point, I am inclined to close the case.  Alva's G "means" the colour of that strawberry, as does my R.  (Since G and R are different, it folllows that Russelian Representationism is false.)  I am inclined to sum this up as follows: to both Alva and me, the strawberry looks the same colour, though he and I have colour-experiences with different phenomenal character.

Now, some are inclined to ask: How can we ever know in such a case that Alva's experience has a different phenomenal character than mine?

Well: there are cases in which it seems that it must be so.  Suppose that Alex and David are looking at the same strawberry, and Alex says "That is a saturated red", while David says, "Really?  It looks slightly yellowish to me".  Suppose once again that this difference of opinion once again persists whenever things that spectrally match the strawberry are presented.  Here it seems clear, as it did before, that the two experiences denote -- aren't I getting bold? -- the same object colour.  But it also seems that when Alex describes the red as saturated and David as yellowish, they evince a difference in colour-phenomenal character between their experiences.

For more details, see my "Truly Blue", here: http://analysis.oxfordjournals.org/cgi/reprint/69/1/48

 

2009-04-25
On an argument for Russellian representationalism
Reply to Mohan Matthen
Ok, there's a challenge there, but I don't think it's exactly what you suggest it is. As I understand it (I'm not sure I fully understand it), what characterizes the Russellian position is the idea that the contents of experiences are akin to states of affairs, as opposed to abstract entities (Fregean senses, epistemic intensions, etc). This does not tell us much about what representing consists in, and here there are many ways to go. Many have been explored by representationalists, and all are consistent with the Russellian position. One could say that the content of an experience is: what one is inclined to believe when having it, its accuracy condition, what the experience is "about", how the world "seems" when one has the experience, what it has the function of indicating, etc. Not all of these clearly tie the contents of experiences to what they "tell" their subjects--most don't. In particular, if the content of an experience is just what it is about, how do you know that what an experience is about is what it tells its subject? The book I'm reading right now is about some fictional characters, but it told me a lot about the author, indirectly. How do you know Alva's learning that the strawberry is red from his experience is not a matter of interpreting the "face value" of his experience in this kind of way?

As I was telling Benj initially, I do agree that there is a challenge, but I disagree about its nature. My objection so far is that the strawberry argument against the Russellian doesn't go through because Russellian representationalism doesn't clearly commit one to the kind of connection between experiences' contents and what they "tell" their subjects which the argument presupposes. So far so good for the Russellian. Here's where I think there's a challenge for him: ultimately, the reason the view doesn't come with the commitments you ascribe it is that it is utterly vague.  I'm not objecting to Benj or you specifically, but to the whole debate. As it is, it seems sufficiently unclear to me what "content" is supposed to be that I don't think either arguments for or against representationalism in the neighborhood of the strawberry case can succeed.

Of course, I have a view about how to fix the problem. What the representationalist really wants to say is this, I think: there is a non-factive relation R such that experiences are states of standing in R to proposition-like entities (maybe Russellian ones, maybe others). "R" isn't characterized directly, it's a theoretical posit (I got this trick from Adam Pautz). I think if you reflect on the dialectical links between rep and the other views, you will see that this is just the claim that distinguishes representationalists as a whole from their opponents, and just the claim which plays the theoretical role representationalists want it to play. I have a dissertation in the making in which I argue for this claim and defend the view...





2009-04-25
On an argument for Russellian representationalism
Reply to David Bourget
You certainly have a good point about the difficulties of distinguishing between content and what a state "tells" you. It may be, as you say, that what Alva's experience of the strawberry tells him may go beyond the content of that experience. 

But I thought myself to be addressing the question: what colour does the strawberry look to Alva -- if Alva's quale is different from mine, does the strawberry look a different colour to him just in virtue of this fact?  It doesn't seem to me that this question turns on the complex questions you raise, such as the difference between Fregean senses and states of affairs.  Have I missed something?

I am not sure how the R-trick goes.  Do you mean: a representationalist is one who holds that every mental state consists in some relation between a thinker/perceiver and a proposition-like entity?  I can go along with that, and, I think I would like to embrace representationalism thus articulated.  But though the question about Alva has implications concerning which proposition-like entity he stands in R to, it doesn't seem to me that representationalism tells us anything about what colour the strawberry looks to him.  Does this strike you as right?

2009-04-26
On an argument for Russellian representationalism
Reply to Mohan Matthen
Yes, the Russellian / Fregean distinction is not relevant--that was part of my point: you and Benj seem to think there is a special problem for Russellians, I say there isn't because the Russellian view (and also the Fregean view, but I didn't emphasize that) leaves a lot of understandings of what representing amounts to open, many of which neutralize the argument against rep.

The point of the R-trick is to free us from having to spell out the metaphor of aboutness. This comes at the cost of leaving us without a clue as to how to determine what contents what experiences have (what one stands in R to in these experiences). But that can and must be provided by a separate theory. For this reason, I disagree when you say that "the question about Alva has implications concerning which proposition-like entity he stands in R". Prior to having a theory of R which goes beyond the generic representationalist claim, we can't know this. If you assume that R is a something like "x learns that P", "x is inclined to believe that P", or "x gets the message that P", you will conclude that Alva stands in R to a red-involving proposition. But R could be something quite different. For example, it could be a nonfactive version of the relation that sense-datum theorists think we can stand in to sense-data. It doesn't have to have any epistemic dimension.