Conference draft. Please do not quote.
Externalism and the varieties of self-awareness
Philosophy and Cognitive Science
Ottawa, Ontario, Canada
When externalism is considered in the context of self-awareness, theorists usually focus on consciousness of one=s own psychological states. I will consider this kind of consciousness, too, but first I want to look at consciousness of oneself, the subject of these states.(1)
Externalism is the view that some crucial element in the content of our representational states is outside of not just the states whose content they are but even the person who has those states. If so, the contents of such states (and, many hold, the states themselves) do not supervene on anything local to the person whose has them. There are a number of different candidates for what that element is: function (Dretske), causal connection (Putnam, Kripke, Fodor), and social context (Davidson). (Burge has foot in both the causal connection and the social context camps and Dennett fits in here somewhere, too.) This diversity will turn out to be important. The paper starts with Dretske but gets to other varieties of externalism eventually. Here is how we will proceed.(2)
After identifying a problem that consciousness of self purportedly creates for externalism and distinguishing two current views of consciousness, we lay out one approach that externalists take to the problem. They use the supposed transparency of one=s representational states to oneself to urge that what we call >introspection= of an internal state is really an inference from what appears in the state. We will consider Dretske=s displaced perception version.
First we will examine the displaced perception approach in the context of consciousness of oneself as the subject of one=s experiences; the latter will turn out to contain a problem for this approach. Then we will turn to consciousness of one=s own conscious states. Here again there are problems for the displaced perception approach. HOT and HOE (higher order thought, higher order experience) models are a leading alternative. They face problems, too. I will sketch what I call a SOE (same order experience) model as an alternative to both of them.
An important objection to HOT and HOE models is that they lead to a Aperception@ picture of introspective self-awareness. Does the SOE model face the same problem? I will argue that it does not. The appearance of a problem is an artefact of an atomistic approach to what it is to be a conscious state. A sketch of how conscious states fit into a cognitive system shows that it is not a real problem.
If displaced perception is not the solution, was there a real problem in the first place? As I will argue, the appearance of a problem here might also be an artefact of an overly atomistic approach to conscious states. An application of our sketch of how conscious states fit into a conscious cognitive system suggests that self-awareness is not a problem for most forms of externalism.
The shape of the supposed problem
There is an apparent problem for externalism in psychological states that is broader than self-awareness, that arises out of content in general. Some psychological phenomena seem to be entirely in us, in our Aminds@ (merely existing in the body is not good enough). Dretske sees this:
When I close my eyes, I cease to see [the world around me]. The world does not vanish but something ceases to exist when I close my eyes. And this something has to be in me. [1995, 36]
Other examples. When I move to a new location (certainly on this world, whatever may be true of twin earth), the vast proportion of my mental content B and, of course, any consciousness that I have of it B moves with me unchanged: my thoughts, imaginings, emotions, memories, and so on. Concerning those states of which I am conscious, I would still be conscious of what I was thinking and feeling, what my words mean, what my beliefs are, and of myself having these states, even if I were in a sensory deprivation chamber and cut off from all current causal contact with the world. And so on. How could all this be true if the content of those states is partly external to me? In this paper we will restrict ourselves to consciousness of these states. Here the leading question is, how could my consciousness of myself and my states be so secure if the content of those state is partly external?(3) Considerations like these point toward internalism.
On the other hand, adapting some comments of Dretske=s, staring at the face of a gauge is not the way to discover what information it provides (1995, 109) and observing a meaningful symbol is not the way to discover its meaning.(4) If so, why should we expect to learn what a psychological state represents by Astaring@ at it B even from the Ainside@ (Dennett, 1978a)? Since many internalist pictures see introspection as a kind of internal Astaring@, here externalism appears to have the better idea. And we have a problem.
Transparency and the representational theory of consciousness
What is a conscious state? Currently, there are two main approaches to this question, representationalism and what Block is currently and inelegantly calling phenomenism. On the representationalist view, a state being conscious is one way of it representing something. As Dretske puts it (1995, 116), conscious states are simply states that make you conscious of things. For the phenomenist, by contrast, there is something about conscious states that outruns representational content. Burge (1988), Rosenthal (1991), Dretske (1995), and Tye (1995) are prominent representationalists. McGinn (1991), Chalmers (1996) and Block (forthcoming and earlier papers) are prominent phenomenists.(5)
The representational approach has some prima facie advantages. Consciousness comes out as a relationship between the mind and something, not some mysterious Afelt quality@ in the states themselves.(6) Thus there is no chasm separating consciousness from cognitive functioning nor any serious risk of conscious epiphenomenalism.
We said that the leading question about externalism and self-awareness goes like this: how could my consciousness of myself and the content of my states be so secure if the content of those state is partly external? (Cranked up a bit, this security becomes Afirst person authority@.) One common representationalist response is to appeal to the supposed Adiaphanousness@ of one=s conscious states to oneself (the term is Moore=s ). This is the idea that my own conscious states are transparent to me, so that when I focus on my experience of seeing X, all I am conscious of is X B as it appears to me, of course. Versions of the idea can be found in Burge, Harman, Tye, Dretske, and others. In Dretske, the idea takes the form of what he calls >displaced perception=. If conscious states are transparent, then our consciousness of them must be Adisplaced@ from what is not transparent, namely, what the states are representing.
If this approach is right, consciousness of our own states would pose no threat either to representationalism or to externalism. It would be no threat to representationalism because there would not be anything nonrepresentational about consciousness. It would be no threat to externalism because consciousness of our own states would start from consciousness of the external element in content.(7)
When the supposed transparency of conscious states is used to defend externalism against the threat of self-awareness, usually, as I said, the only kind of self-awareness examined is consciousness of one=s states. There is another and equally important kind of self-awareness: consciousness of oneself, the subject of those states. This is where I am going to start. We will get to consciousness of one=s states eventually but consciousness of self has received little attention in the context of externalism. A new issue may yield a fresh perspective.
Consciousness of self
As has been demonstrated by a number of writers (Castañeda 1966; Shoemaker 1968; Perry 1979; Evans 1982), consciousness of self as subject has some unusual features. Consciousness of self is via what Shoemaker what calls self-reference without identification.
My use of the word >I= as the subject of [statements such as >I feel pain= or >I see a canary=] is not due to my having identified as myself something [otherwise recognized] of which I know, or believe, or wish to say, that the predicate of my statement applies to it. [1968, 558]
Note that in this form of self-awareness, I am not just conscious of myself, I=m conscious of myself as myself. I am conscious that it is myself of which I am conscious. Even more, as we said, with respect to a certain group of conscious states, I appear to myself as their common subject, a single thing that is having (or, in a case that includes past conscious states, had and is having) all of them.(8) And I am conscious of myself this way, as myself and as the single Acommon subject@ (Kant 1781, A3350) of these conscious states, without inferring this from any other features of myself.
Nagel (1965), Castañeda, Shoemaker and Perry all argue not only that I do make non-identifying references to myself but, if I am to know certain ascriptive things about myself, I must. Here is one standard argument (Castañeda 1966, Perry 1979). To know that I wrote a certain book a few years ago, it is not enough to know that someone over six feet tall wrote that book, or that someone who teaches philosophy at a particular university wrote that book, or ... or ... or ... , for I could know all these things without knowing that it was me who has these properties (and, I think, I could know that it was me who wrote that book and not know that any of these things are properties of me). Nor would it help to add details of a more identifying kind B the person whose office number is 123 in building ABC, the person who office phone number is ... . If I don=t know that that office is my office, that that phone number is my phone number, I could know all these things and still not know that it was me who wrote the book.(9) And vice-versa B through bizarre selective amnesia, I could cease knowing all such things about myself and yet continue to know that it was me who wrote the book. As Shoemaker puts it,
... no matter how detailed a token-reflexive-free description of a person is, ... it cannot possibly entail that I am that person. [1968, 560]
What a curious piece of knowledge B if it really is knowledge at all!
If I am and, for certain purposes, must be conscious of myself as myself without inferring this from anything else that I know about myself, my knowledge that it is myself of whom I am conscious is independent of knowing anything else about myself. I can be conscious of myself as myself without being conscious of myself as anything except B myself. (Being myself is not a property of me, i.e., something that I and other things could have in common.) In fact, since it is not just identifying properties that I need not know but any properties whatsoever. As Kant put it,
In attaching >I= to our thoughts, we designate the subject ... without noting in it any quality whatsoever B in fact, without knowing anything of it either directly or by inference. [1781, A355]
Thus I think the terms >non-ascriptive reference to self= captures what is special about this form of consciousness of self better than >self-reference without identification=.(10)
Non-ascriptive reference to self, displaced perception, and externalism
How does non-ascriptive consciousness of self fit with externalism? I will answer this question in two steps. First I will examine whether non-ascriptive consciousness of self could be a form of displaced perception (and, by extension, whether it is compatible with inference theories of introspection in general). Then I will examine where this leaves us vis-à-vis externalism.
The idea behind displaced perception is that all the experience, all the representing we need in order to be conscious of a conscious state is that the state represent something, anything. When we are conscious of one of our representational states, it is by the state representing something. What makes this possible is that,
you cannot represent something as F without, necessarily, occupying a state that carries the information that it is F (not G or H) that you are representing something as. [1995, 56]
By being conscious of those properties, I can become conscious that I am conscious of those properties. The former is phenomenal consciousness, consciousness of, but the latter is not. It is conceptual consciousness, consciousness that.
Here is how displaced perception yields this consciousness. In the same way that the ringing of the doorbell can tell me that the mail is here (and would not tell that to anyone who did not have the concept of mail or even just the link between the doorbell ringing at a certain time and mail), representing something can also tell me that I am representing it. If so and if there is not a problem for externalism in the original activity of representing something as F, there cannot be any problem for externalism in my consciousness of the representation. So goes the argument.
Now ask: In the statement, Ainformation that it is F (not G or H) that you are representing something as@ (my emphasis), where does the consciousness of this you come from? I don=t think that displaced perception can be the answer.
I can infer or otherwise know that information represented by me is me representing something only if I am already know of myself, and of myself as myself. Certainly this is true if the view of Castañeda et al. has anything going for it but it seems plausible in its own right. If so, consciousness of self is not displaced perception. Having information about something, simply having information about it, may be enough for me to infer that something has that information but will never be enough for me to infer that I have the information. That=s one argument. Here=s another. As we argued, consciousness of myself as myself is not via consciousness of properties of myself (or properties of any other kind). If so, this consciousness of self cannot be any form of displaced consciousness of properties.
What is the basis of it, then? That is a longish story, but the short answer goes something like this: The thing that is in fact me becomes conscious of itself as me by being conscious of (the thing that is in fact) itself from the point of view of having certain experiences, i.e., being the thing that has those experiences. Being conscious of representations by having them is sufficient to provide this point of view.(11) (I will enlarge on these dark comments later.)
If my story about the basis of consciousness of myself as myself is anywhere near right, then Dretske might be wrong about something else, too. Consciousness of myself as myself is not conceptual by his criteria (1995, 10-12); it is phenomenal. I am conscious of myself B as myself. I do not have a thought, judge, form a proposition, that it is me. (If so, consciousness of myself as myself is also a counterexample to HOT theories of consciousness.)
If consciousness of oneself as oneself is not displaced perception and is not conceptual, is it a problem for externalism? For Dretske=s brand of externalism, the answer would depend on what kind of evolutionary story we can tell. For Dretske, a state gets to be a conscious state by having the function of conveying a certain kind of information to an organism. It gets this function when information available in it confers survival or reproductive advantage. If it regularly confers such advantage, it eventually gets the function of conveying such information. This is the external element in content. What information a state has the function of conveying, i.e., what it represents, is determined in part by the evolutionary history of states of that kind, i.e., by something external.
Whether consciousness of oneself as oneself creates a problem of this brand of externalism would thus depend on the following: Can we make sense of a mechanism conferring a survival or reproductive advantage of a type such that it could gain the function of representing the organism to itself as itself? (Burge 1998 says some things relevant to the question of what advantages consciousness of self might confer.)
There is a more general reason for suspecting that non-ascriptive consciousness of self may be no threat to externalism. These acts of consciousness hardly have any content. Recall Kant=s claim: when one is conscious of oneself as oneself, one can be conscious of the self Awithout noting in it any quality whatsoever@ (1781, A355). If so, this consciousness of self may not have any external content. But it remains a counter-example to displaced perception theory.
Consciousness of one=s conscious states
Let us now turn to consciousness of one=s own conscious states. Studying consciousness of self may have been a useful prelude but consciousness of one=s own states is the main act; it is the kind of self-awareness that philosophers always have in mind when they set out to show that self-awareness raises difficulties for externalism. So: the states that make us conscious of the world B is consciousness of these states a problem for externalism?
Dretske developed his displaced perception story to argue that they are not. Whatever its success with consciousness of self, does it work here? First, distinguish two questions:
1. Are conscious states transparent, as Dretske and others urge?
2. Could we be conscious of them in the way that we are if they were transparent?
(1) appears to be straightforwardly empirical and so is a peculiar question for a philosopher to take a stand on. (2) is more philosophical. Here=s the question for transparency raised by (2). How much information is provided merely by representing a thing?
Recall Dretske=s claim:
You cannot represent something as F without, necessarily, occupying a state that carries the information that it is F (not G or H) that you are representing something as. [1995, 56]
This does not sound right. Merely representing something as F does not provide enough information for you to know that representing is going on, not unless you can know that representing is going on without having any idea as to what kind of representing. Information that something is F would not tell you whether you are perceiving or remembering or imagining or dreaming about the thing that is F. Nor would it tell you anything about what attitude you are taking the thing (apparently) being F. To know those things, you would need additional information. This information would have to come, so far as I can see, from the representation, not from how it is representing something. A fortiori, representing a thing as F will not tell you that it is you who is representing it, a point we argued in previous sections.
Nor is it concepts that you need. You do need concepts B as Dretske urges, in addition to phenomenal information, you need the power to give Aconceptual embodiment@ to that information (1995, 60). I agree; consciousness of our own conscious states involves concept-using judgment more than is commonly recognized. (Two philosophers who do recognize this are Kant 1781/7 and Dennett 1991). I am not talking about knowing that one is representing, however; I am talking about the conditions of even having information about what representing is going. For this, you need phenomenal information about the representing, not just about what is represented by it. Dretske says that all the phenomenal information that we need is phenomenal access to what is being represented.
If we know not only what we experience and think ..., but that we experience and think it ..., we do not know this by introspection. [1995, 57]
Perhaps not (it depends on what you mean by >introspection=) but you need some kind of phenomenal information about the representation.
With this, we also seem to have an answer to (1): conscious states are not transparent in the requisite way. And this seems right; they do seem to give us the information we need for our answer to (2). In the same way that it seems like something to represent the world, it feels like something when we are aware of our representation (Flanagan 1992, 67-8). We are aware, for example, of how we are representing the world (as any near-sighted person knows as soon as they take off their glasses). Indeed, such information seems to abound. Dretske himself worries over those peculiar emotions that do not seem to represent anything [1995, 103]). Then there is sensory modality and emotional toning. Equally, awareness of pain seems to be more than displaced perception from information indicated by the pain-indicating state (if pains represent anything; most of them seem not to). And what about states in which something seems the way it does because of an interaction between what is being represented and the information-indicating state (color) or other states in us (hue).(12) All the states mentioned in this paragraph seem to be counter-examples to the transparency thesis. In short, we seem to have all kinds of phenomenal access to the Avehicle@ of representation, not just to what it is representing. And the access in question is not inferential, not displaced. Perhaps Aqualia@ cannot be rendered safe for externalism quite so easily.
HOT, HOE, and SOE
One of the reasons for suspicion about the idea that we have phenomenal information not just about what a representation represents but also about the representation itself is that this latter information itself would need to be represented and we would seem to be off down the road to HOTs or HOEs. For Dretske, by contrast, it is the conscious states themselves that tell us about our conscious states, not thoughts or experiences of those states. This is attractive.
First, it is intuitively plausible. Second, we should avoid the extravagance of representations of representations if we can. For HOT and HOE theorists, a state is conscious when we are Anoninferentially conscious of that state@ (Rosenthal MS, 4). Depending on the theory, this could be by having a thought or an experience of it. (The thought or experience need not itself be conscious. For that, it would have to become the object of another thought or experience.) We should avoid the extra layer of representations if we can. Third, Dretske thinks that HOT and HOE models face serious difficulties. Against HOE models, Dretske urges that we are as unlikely to be able to Aread@ the nature of a representing state off its face as we are to determine what information an instrument in a cockpit indicates by looking at that instrument. Against HOT theories, he argues that there is a great deal in our conscious states that is not represented in any thought about them (1995, 108-16).
So here is our task. Can we overcome the problems facing displaced perception and other transparency theories without falling into a HOT or HOE model? To become conscious of a state, we may need information from the state itself B this was the difficulty for Dretske B but do we need some other representational state, too?
As an alternative to both displaced perception and HOT/HOE models, I want to offer what might be called a same-order experience (SOE) model. On this model, what gives us phenomenal information about our representational states is not information they convey about something else but it is not any higher order thought or experience of these states either. The states themselves provide the required information. A representation provides phenomenal information about something other than itself B but it also provides such information about itself.
To see a bit more clearly what I have in mind, think of one of Dretske=s favorite examples, a gauge. A gauge presents information about something other than itself B whatever it has the function of indicating information about. So, an altimeter presents information about height off the surface of the earth. (It need not present correct information but its function is to present this kind of information.) However, and this is what Dretske seems not to take into account, an altimeter also presents information about itself: that its dials are moving in certain ways, that it has a face with a metric for calibrating information on it, and so forth. No displaced perception here! However, it is the gauge that presents this information about the gauge, not some higher-order gauge pointed at it. To have information about an altimeter, all we need is the altimeter. Similarly with our conscious states. To have information about a conscious state, all we need is that conscious state. So no HOTs or HOEs either.(13)
Put the idea in information-theoretic terms. For A to provide information about B, A must co-vary with B (and be a representation). What could co-vary better with a representation than that representation?(14) In short, we can avoid the problems that face displaced perception and other transparency theories while keeping their metaphysical elegance.
Does SOE require a problematic form of introspection?
Or can we? It will be objected that we have succumbed to one of the weaknesses of the HOT/HOE approach, the idea that in introspection we Aread@ the nature of a representation off its Aface@. As Dretske notes, there is no more reason to think that we can discover how a representation is representing something by Astaring@ at it than that we can discover what kind of information a gauge indicates by staring at the gauge. This is an important challenge and we need to answer it. As we will see, the answer is useful for another reason: it opens the way to a good account of why self-awareness is no threat to externalism, most forms of externalism at any rate. But first the answer to the challenge.
The answer to the challenge lies in how conscious states sit in the cognitive system. Dretske says that a conscious state is one that makes you conscious of things (1995, 116). What is this Ayou@ and how must a state relate to you to make you conscious of something? You first. Consciousness states not only indicate something, they indicate it to someone. When they do, what are the properties of the thing to whom they indicate?
One important property of a cognitive system to whom conscious states indicate is a high degree of mental unity. Mental unity takes at least three different forms but we will consider only one of them here, namely, unity of consciousness.(15) The latter appears in at least four ways:
unified consciousness of items not themselves conscious
$ unified consciousness of one=s conscious states
$ unified consciousness of self; and,
$ unified focus.
The four might be instances of a single phenomenon. The core idea is unity of consciousness. Intuitively, this is the idea of being aware of many things at once. A better definition is:
The unity of consciousness =df. a representing in which a number of items as represented are combined in such a way that to be aware of any of these items is also to be aware of other represented items as connected to it and of the whole as a single complex Aobject@.
We need not consider the last two. Consciousness of self was discussed earlier; it would add nothing to note that it is unified. Unity of focus, the capacity to select and bring resources to bear on particular items, is not relevant to our current discussion.
Unified consciousness of items not themselves aspects of our consciousness
$ Unified consciousness of one=s conscious states
In the first, we bring together a variety of things of which we are conscious, and something our states of consciousness of them, in such a way that we can relate them to one another. In the second, one is aware of seeing things, having feelings, thinking things, remembering things, etc., as all linked together in one subject, namely, oneself.
Now, what is needed for a state to be conscious, i.e., make us conscious of something? Philosophers responding to this question often focus to atomistically on individual mental states and their contents. Dretske starts with this:
A state is a conscious experience of F ... if the state has the natural ... function of providing information about the F-ness of objects standing in the appropriate contextual relationship (C) to the system. [1995, 162]
So far this looks atomistic. But he also says that conscious states are Astates whose functions it is to supply information to a cognitive system for calibration and use in the control and regulation of behaviour@ (1995, 19) and cites with approval Evans= suggestion that the states in question Aserve as input to the concept-exercising and reasoning system@ (Evans 1982, 227). All this is probably meant to specify merely necessary conditions of a state being conscious, not sufficient conditions. Indeed, Block may be right when he suggests that sufficient conditions for a state being conscious cannot be non-circularly given (Block, 1996). Even as necessary conditions, however, we need to be more specific. It depends on the kind of input.
We all have a huge number of representational states that are input into some cognitive system in us (e.g., the autonomic nervous system) but are not conscious. Kant=s way of specifying the kind of input needed for a state to be conscious was to say that the state must be Asomething to me@.(16) By this he meant that the cognitive system can recognize the state, base judgments upon it, connect it to other similarly recognized states, focus attention on it B i.e., make it part of the group of states about which one has unified consciousness. Another way to put this point is to say that, for a state to be conscious, it must play a functional role in a cognitive system, whatever role gives the system the ability to recognize and integrate it.
Moreover, individual conscious states carry information about their functional role(s) and links to other states in the system. Suppose I wish that this paper was finished. This conscious state would not only carry information about its content, this paper, but also about its being a wish, about it being in conflict with another wish to go see a film, about a further wish that I ignore the second wish and concentrate on the task at hand, about the role of these wishes as reasons for some of my actions, ... and so on and so forth. In short, individual conscious states carry information about a good deal more than the items they are representing.
We can now address the challenge. In urging that we have phenomenal information not just about what a representation represents but also about the representation itself, we are not succumbing to the bad picture of introspection as a process of Areading things off@ individual conscious states. In the same way that a conscious state is linked, via its indicating function, to the items it represents, it is linked, via whatever function allows it to be something to a cognitive system, to the cognitive system that has it and it carries information about that link and its place in that system. So we can deny the transparency thesis without falling into a Aperception@ picture of introspection. We just need to stop thinking about conscious states atomistically.1(17)
Before we move on, let me return to something left hanging earlier: my dark sayings about the basic of consciousness of oneself as oneself. Its basis, I think, is the unified representation that results from tying individual representations and what they represent together. This unified representation provides information about itself, too. This information about itself is input to itself (where else?). When this happens (and the right concepts are there), the unified representation will, I think, represent itself as itself. If so, this sheds some light on the earlier dark comments, that the thing that is in fact me becomes aware of itself as me by being aware of itself from the point of view of being that thing.1(18)
Was externalism ever in danger?
We saw that displaced perception and the transparency approach in general cannot protect externalism from self-awareness. But was externalism ever in danger? I do not think so. As long as content is accessible (in the right way) to the system, I will argue, it doesn=t matter if some of the information that makes it up is external to, i.e., not carried by individual conscious states of that system. When we think of conscious states one by one, self-awareness seems like a problem for externalism. If we think of conscious states as part of a whole cognitive system, we can make the problem go away, at least for many kinds of externalism.
Dretske=s kind of externalism provides for such access. For him, representational function and thus one of the elements determining what information a representation carries is external to the representation. The representation may not carry information about its function but the cognitive system has this information. The system has to have this information to know what information the representation is supposed to carry.1(19) And we do. We know what a gauge has the function of indicating and we know what a perception has the function of representing. We may know nothing how they got that function in either case and we may even be unable to describe the function. But we know what their functions are. If so, self-awareness is no threat to at least Dretske=s brand of externalism. Note that this result is quite independent of the transparency thesis or the theory of self-awareness as displaced perception.1(20)
From one perspective, this result is not very surprising. Compare the roughly parallel claim that functional role plays an essential role in the individuation of conscious states. Now, functional role is external to individual conscious states. Yet no one has ever thought that the special Aauthority@ of self-awareness threatened the claims made for functional role. Why should the claims for externalism be any more threatened?1(21)
Is externalism entirely out of the woods with respect to consciousness of our own conscious states? It depends on the type of externalism. The qualification introduced just above is crucial: so long as content is accessible (in the right way) to the system. On Dretske=s version of externalism, it is, but that is not so on all versions of externalism. In particular, it is not true of versions of externalism derived from Putnam=s (1975) twin earth experiments or thought experiments like them (including some of Burge=s 1979 ones). Here is how these theories go. Suppose that Adam and his twin, Twadam, have beliefs about a certain clear liquid in front of them and both call it >water=. One liquid is H2O and one is XYZ. They have beliefs about different things and hence different beliefs. Here, however, the element of content external to them is not something that either of them knows about. For versions of externalism that have this feature, self-awareness may continue to be problem for anything that my analysis has shown.2(22)
Dretske=s brand of functionalism is not the only brand that allows for subjects= knowledge of the external element in content. The same is true of any version of the causal theory of reference that allows that the cognitive system can learn the relevant causal connections (or, in the case of theories that rely on nomic connections, the counterfactually supported causal laws). Kripke, 1972, and Fodor, 1994, are two examples.2(23) I think that Davidson=s and Dennett=s brand(s) of externalism are safe, too.
In short, even if conscious states do not carry information about the element of their content that is external to them, so long as this external element can be known by the system, consciousness of our own states will not be a problem .
Go back to the two requirements laid down at the beginning. One was that we account for the fact that the nature and contents of our conscious states change in only minor ways with change of environment or causal connections or eyelid position (or even, counter-factually, history). Even if both individual states and the system as a whole are hostage to something external, we can meet this requirement. So long as the system knows how the relevant external element enters content, it has the autonomy it needs to move around (or, counterfactually, have a different history) without its contents changing.
Yet this picture also meets the second requirement. That=s the requirement that our account not hold or imply that we read the nature and contents of conscious states off the state itself. On the model that we have developed, the information that conscious states carry about themselves is the result of a complex interaction among information carried by those states in isolation, information about their Alocation@ in the cognitive system, and other elements in the system (Dennett 1991 sketches one such model.)
To summarize: displaced perception theories and other appeals to transparency do not work but we can save externalism from all likely ensuing risks: the special Aauthority@ of self-awareness, HOT/HOE models, and theories of introspection as internal perception. All we need is a SOE model of self-awareness and a cognitive system that knows the external element in the contents of its representations.2(24)
Anscombe. G. E. M. 1957. Intention. Oxford: Blackwell Publishing Co.
Bilgrami. A. 1992. Can externalism be reconciled with self-knowledge? Philosophical Topics 20, 1: 233-67
Block, N. 1996. Consciousness in Anglo-American philosophy. In The Encyclopedia of Philosophy Supplement . London: Macmillan, pp. 96-100
Block, N. forthcoming. Mental ink
Brook, A. 1975. Imagination, possibility and personal identity. Amer. Phil. Quart. 12, 185-98
Brook, A. 1994. Kant and the Mind. New York: Cambridge University Press.
Brook, A. and Stainton, R. 1997. Fodor=s new theory of content and computation. Mind and Language 12, pp. 459-74
Burge, T. 1979. Individualism and the mental. Midwest Studies in Philosophy 4, 73-121
Burge, T. 1988. Individualism and self-knowledge. J. Phil. 85, 649-63.
Burge, T. 1998. Reason and the first person. In Knowing Our Own Minds, ed. C. Wright, C. MacDonald and B. Smith. Oxford: Oxford University Press
Castañeda, H.-N. 1966. >He=: A study in the logic of self-consciousness. Ratio 8, 130-57
Chalmers, D. 1996. The Conscious Mind Oxford: Oxford University Press
Davidson, D. 1987. Knowing one=s own mind. Proc. and Add. of the Amer. Phil. Assoc. 60, 441-58
Dennett, D. 1978a. Brain writing and mind reading. In Dennett, 1978c
Dennett, D. 1978b. A cure for the common code. In Dennett 1978c
Dennett, D. 1978c. Brainstorms. Bradford Books: Montgomery, VT
Dennett, D. 1987. Consciousness. In The Oxford Companion to the Mind, ed. R. Gregory. Oxford: Oxford University Press
Dennett, D. 1991. Consciousness Explained. Little, Brown
Dretske, F. 1995. Naturalizing the Mind. MIT Press
Evans, G. 1982. The Varieties of Reference. Oxford: Oxford University Press
Flanagan, O. 1992. Consciousness Reconsidered. Cambridge, MA: MIT Press
Fodor, J. 1994. The Elm and the Expert. MIT Press
Harman, G. 1990. The intrinsic quality of experience. In Philosophical Perspectives 4: Action Theory and Philosophy of Mind, ed. J. Tomberlin. Atascadero: Ridgeview Publishers
Kant, I. 1781 and 1787. Critique of Pure Reason. Trans N. Kemp Smith. London: Macmillan 1927
Kripke, S. 1972. Naming and necessity. In The Semantics of Natural Languages, ed. D. Davidson and G. Harman. Dordrecht: Reidel Publishers
McGinn, C.1991. The Problem of Consciousness. Oxford: Blackwell Publishing Co.
McKinsey, M. 1991. Anti-individualism and privileged access. Analysis 51, 9-16
Moore, G. E. 1922. Philosophical Studies. London: Routledge
Nagel, T. 1965. Physicalism. Phil. Rev. 74, 339-56
Perry, J. 1979. The problem of the essential indexical. Nous 13, 3-21
Parfit, D. 1970. Personal identity. Phil. Rev. 80, 3-27
Putnam, H. The meaning of >meaning=. In Mind, Language and Reality: Philosophical Essays, Vol. 2. Cambridge: Cambridge University Press
Rosenthal, D. 1991. The Nature of Mind. Oxford: Oxford University Press
Rosenthal, D. undated MS. State consciousness and higher-order thoughts
Shoemaker, S. 1968. Self-reference and self-awareness. J. Phil. 65, 555-67
Shoemaker, S. 1970. Persons and their pasts. Amer. Phil. Quart. 7, 369-85
Tye, M. 1995. Ten Problems of Consciousness. Cambridge, MA: MIT Press
1. 1. I will follow the literature and use >consciousness= to talk about the topic of this paper except that I will use the term >self-awareness= to talk about consciousness of self.>Self-consciousness= already has a nonphilosophical meaning, so . In general, I treat >consciousness= and >awareness= as synonyms.
2. 2. Dretske will be my main stalking horse in this paper. Not only does he tell a very interesting story about externalism and consciousness but he is here.
3. 3. McKinsey (1991) raised another problem about self-knowledge and externalism, one that I won=t consider, the problem that if self-knowledge is authoritative and if its contents are partly external, we should be able to infer, and infer a priori, that certain things outside us exist simply from what we are conscious of in ourselves. Given that knowledge of these external items may be a condition of the self-awareness in question, I am not sure how big this problem is but in any case I won=t examine it here.
4. 4. Dretske is actually after HOE (higher order experience) models of conscious experience in this comment but my modification is in the spirit of his broader analysis.
5. 5. Some consciousness theorists are less easy to classify, e.g., Davidson and Dennett.
6. 6. This way of distinguishing consciousness and self-consciousness may make it look like I am heading for a HOT (higher order thought) or HOE (higher order experience) model. To put it mildly, that is not the case. I see no reason why consciousness of a state that is itself conscious requires anything higher order. Why can=t the original conscious state itself do the job? I will urge later that it can.
7. 7. The transparency claim also provides a neat distinction between two kind of consciousness much in need of separating, namely, consciousness of oneself and one=s states, and, consciousness of things that are not themselves conscious states.
It is worth remarking that the terminology in consciousness studies is a mess. There is creature consciousness and state consciousness and phenomenal consciousness and access consciousness and monitoring consciousness and transitive consciousness and intransitive consciousness ... and goodness knows what else. One of the worst bits of chaos is this: for many theorists, what I am singling out as merely one form of consciousness, namely consciousness of oneself and one=s states, is taken to be what consciousness is. The consciousness that Freud contrasted with the unconscious, for example, is clearly consciousness of self. Dennett goes both ways in different works. In (1987), he carefully distinguishes the two. In (1991), he never does and sometimes it not clear which he is talking about. Consciousness of items not themselves conscious should never be conflated with consciousness of one=s states or oneself.
8. 8. Which group of conscious states can be specified this way? They are the states (in principle states of anybody) of which the experiential basis of my gaining conscious and/or cognitive access to them is having them (in the case of a certain kind of memory, having had them). There may well be other, more cognitive requirements for me to gain conscious or cognitive access to them but all the experience I need is the state itself (contrary to HOE models).
9. 9. Other things could bridge the gap, too; the point is, I have to know that something is true of me.
10. 10. I discussed non-ascriptive reference to self first in (1975, 188).
11. 11.It is also when, and only when, we are conscious of ourselves by having experiences that Shoemaker=s (1970, 269) immunity to error through misidentification of something as myself holds (Brook 1994, 89-90). Dretske makes room for the introspectibility of misrepresentations by urging that both when we are misrepresenting something (or even nothing) as F and when we are correctly representing something as F, we are in the same state of representing it as F. If so, we can have the same displaced perception of thus representing it. Does Shoemaker=s immunity to error rule out being the same situation as I would be in if I were conscious of myself as myself and yet not being conscious of myself? Strangely enough, the answer is no. When I remember some thought or feeling or action from the point of view of thinking or feeling or doing it, that it is me who thought or felt or did it is something that I will Aautomatically assume@ (Parfit 1970, 15). That is to say, I will be in the same situation as if it were me that I am conscious of. Yet, if traces of autobiographical memories (more exactly, autobiographical q-memories) can be transferred from person to person in certain ways, this will not be true. I (q-)remember seeing the scene from the top of Mt. Robson. I will >automatically assume= that is was me who saw that scene. However, memory-traces of Sally seeing that scene have been transferred to me, so it is Sally that I am actually remembering seeing it. I take her to be me because I remember the scene from the point of view of the person who saw it. (This is the only issue about what Shoemaker calls Aimmunity to error though misidentification@ of something as myself using I and cognates that I will consider.)
12. 12. Dretske mentions both of them (1995, 72, 89) and also our Asense@ of what it is like to be ourselves (1995, 94) but I don=t see how he can accommodate these cases.
13. 13. The core of this idea goes back as far as Anscombe=s (1957) notion of non-observational, non-inferential consciousness of one=s own states.
14. 14. Robert Stainton suggested this way of putting the point. That states carry information about themselves does not entail that they carry it better (or worse) than they carry other information.
15. 15. The other two kinds are cognitive unity and unity of behavior. Cognitive unity consists in our ability to bring an wide range of cognitive resources to bear in an integrated way on a cognitive task: what we want; what we believe; our attitudes to self, situation, and context; input from each sense; information of all kinds; language; memory of various kinds; bodily sensations; our skills; and so on.
Unity of behavior is the coordination of limbs, eyes, bodily attitude, etc., in ways the precision and complexity of which would be difficult to exaggerate. Think of a concert pianist performing a complicated concerto.
16. 16. In both versions of the Transcendental Deduction, Kant used the (German cognate of) the phrase Anothing to me@ to describe states not in the appropriate relationship to a cognitive system. Thus he must have had being something to me in mind as right relationship (1781, A120; 1787, B132).
17. 55. This analysis of how a conscious state is related to a cognitive system may itself pose a problem for HOT and HOE theories, at least as usually construed. The HOT or the HOE in these theories is usually thought of as a single thought or experience directed onto a single psychological state. On the view I have just sketched, consciousness of our own states is a matter of them becoming part of a unified consciousness involving a great many conscious states linked to other conscious states in a single representing. It is unlikely that there are single HOT=s or HOE=s here and there may be no HOT or HOE of any kind. Kant at least sometimes talked of the unified representation as a representation of its included representations but then what of the unified representation itself? It is certainly a conscious state but it seems strained to suggest that we have a thought or experience of it.
18. 66. One implication of the story I have just told may be that the unified representation and the subject who has it are one and the same. This will strike many as counter-intuitive. I myself do not find it counter-intuitive at all. Intuitions based on introspection have hardly any credibility at all but I do not experience myself and my unified representation (which includes a unified representation of my conscious states and of myself) as different from one another. Moreover, there are significant theoretical advantages to identifying the two. On my account, the Aexempt agent@ as Dennett calls it (1978b, 101), the unanalyzed homunculus that some theories feel constrained to posit, turns out itself to be a kind of representation B a rather exotic kind to be sure, but still a kind of representation. If so, a representational theory can in principle be a complete theory of consciousness and the cognitive system. We no longer need a separate theory for the thing that is conscious.
I present the theory behind these remarks in more detail in Brook 1994, where I claim that the theory is present in Kant, albeit rather obscurely. When I say a Acomplete theory@, I mean complete at a certain level B roughly, the procedure level of the triple of task, procedure, and implementation. We will still need neuroscience to tell us about implementation, for example. Moreover, even at the level of procedure, it is complete only in the sense that nothing has to be left out. It is not complete in the sense of providing all significant details.
19. 77. By >know= here I mean no more than >have reasonably secure beliefs about= or the like.
20. 88. If Dretske is right about function, the beliefs about function that the system has would themselves have a function, namely, to convey information about functions. If so, these beliefs would not be locally supervenient any more than representations of the world are. I don=t see any special problem here. There could even be multiple levels of belief about function. We could have beliefs about the function of our function-attributing beliefs, for example.
21. 99. A point that Martin Davies has emphasized.
22. 00. That may turn out to be a reason for rethinking twin earth and similar thought-experiments. Attempts have been made to argue that the ensuing varieties of externalism is compatible with self-awareness, of course. Davidson 1987, Burge 1988 and Bilgrami 1992 are among the best known examples. I myself suspect that Adam and Twadam do know everything they need to know about the content of their beliefs. I suspect that they both believe AThis [the substance in front of them] is water@ but not AThis is water-rather-than-twin-water@ (in their respective idiolects). If so, their beliefs have the same contents in the relevant respects and self-awareness is no problem. (Brook and Stainton 1997 offers a different approach but reaches the same conclusion.)
23. 11. Fodor also argues that consciousness of one=s states presents no problem for twin earth cases and twin earth cases are not a problem for psychological theory because there are no twin earth cases in any world nomically near to ours. And those are the only worlds that we need to worry about (Brook and Stainton 1997).
24. 22. My thanks to Robert Stainton and Martin Davies for very helpful comments on an earlier draft.