Representationalism and the Argument from Hallucination

Brad Thompson

Department of Philosophy

Southern Methodist University

bthompso@smu.edu

 

Introduction

           

            Representationalism is an increasingly popular theory of consciousness among philosophers of mind. Though representationalism can take both reductive and non-reductive forms, much of the attraction of representationalism comes from its prospects for offering a theory that takes the phenomenal character of experience seriously while being compatible with materialism. According to reductive representationalism, phenomenal character is nothing over and above the having of representational content of a certain sort. Paired with a naturalistic theory of representational content, the resulting view is fully materialist.

When it comes to perceptual experiences, reductive representationalism has a second apparent virtue. Intuitively, it seems to be possible to have hallucinations that are phenomenally just like veridical experiences. But if this is so, then it seems that what determines phenomenal character is something that both veridical and hallucinatory experiences have in common. The presence of a “common factor” between veridical and hallucinatory experience is incompatible with naïve realism, since naïve realism identifies phenomenal properties with properties of external objects (which are absent in the case of hallucination). Traditionally such considerations have been taken to support some form of indirect realism, such as a sense-datum theory. But representationalism makes possible a non-naïve direct realism that is compatible with the common factor assumption. Phenomenal character is determined by representational content, which both hallucinatory and veridical experiences can share. But in the case of veridical experience, unlike hallucination, the external objects of experience literally have the properties one is aware of in experience. The representationalist can accept the common factor assumption without having to introduce sensory intermediaries between the mind and the world, thus securing a form of direct realism.

It is an appealing picture. But as I will argue, much of the attractiveness of representationalism stems from a failure in the literature to distinguish between two distinct possible versions of representationalism, what I call “content-based representationalism” and “vehicle-based representationalism”. In what follows, I present a version of the argument from hallucination. Generically, representationalism appears to have a response to the argument from hallucination that avoids a commitment to qualia or sense-data. But I argue that, once the distinction between content-based representationalism and vehicle-based representationalism is recognized, this response to the argument from hallucination is inadequate. Content-based representationalism fails to adequately accommodate hallucination. The upshot is that if you want to be a representationalist, you have to be a vehicle-based representationalist. But vehicle-based representationalism lacks many of the virtues that are frequently advertised for representationalism more generally, and fails to avoid a kind of indirect realism.

 

The Argument from Hallucination

 

            A familiar and widely reviled argument in the philosophy of perception proceeds as follows. Suppose that Mary is dreaming or hallucinating that she is looking at a red fire hydrant. One might argue as follows:

1.     Mary perceives red.

2.     There is nothing red before her—it is a hallucination.

3.     The red that Mary perceives is something mental or internal.

4.     Mary’s hallucination has phenomenal character that is identical to a veridical perception of red.

5.     When Mary has a veridical perception of red, the immediate object of her perception is something mental or internal.

Philosophers have criticized the argument in more ways than I can enumerate here.[1] But one very significant difficulty with the argument as stated is that it is not entirely clear what it is an argument for. What exactly is an “object of perception”, and what does it mean to be an “immediate” object of perception? Similar questions arise for statements of the argument from hallucination that employ notions such as “direct perception” or “direct awareness”.

On at least one straightforward reading of the argument, it is primarily an argument about the epistemology of perception. It is an argument in favor of the view that our epistemic access via perception to the external world is mediated by mental or internal properties or objects to which we have more “direct” epistemic access. This reading of the argument takes the central issue to be the question of whether or not we directly perceive the external world or whether we only perceive external properties and objects indirectly, and that which we directly perceive or are directly aware of in perception is a mental or internal item such as a sense-datum. It is an argument against what we might call “epistemological direct realism”.

Unfortunately, these epistemic matters have often been entangled with metaphysical questions regarding perception. The epistemology and the metaphysics of perception are no doubt intimately related. But the difficulties that confront the argument from hallucination as an argument concerning the epistemology of perception can obscure the merits of a related, metaphysical version of the argument. On this alternative reading of the argument, it attempts to show that in addition to the properties possessed by external objects of perception, there are other properties and/or objects that must exist in order to account for the phenomenal character of hallucinatory experiences. It is an argument against the view that external objects or properties contribute directly to phenomenal character. This feature of the argument is especially apparent when one considers premise 4. Since hallucination can be phenomenally just like veridical perception, it is argued that the properties or objects that constitute phenomenal character are not external to the subject but are instead internal or mental items.

Corresponding to these two conclusions are two distinct theses that have occasionally gone under the common name “direct realism”. There is a metaphysical thesis of direct realism, related to naïve realism, according to which phenomenal properties are properties of external objects. The external world itself at least partially constitutes the phenomenal character of a perceptual experience. And then there is direct realism as an epistemological thesis. Epistemological direct realism is the thesis that our perceptual knowledge of external objects and properties is “direct” rather than mediated by our knowledge of some other sort of object or property.

Though they are not often distinguished carefully, both epistemological direct realism and metaphysical direct realism are familiar in the literature, as are the corresponding ways of understanding the argument from hallucination. Smith (2002) adopts a form of direct realism that only makes sense as a version of epistemological direct realism, seeing that in the very first chapter of the book he concludes that the only way to defend direct realism is “to construe sensory qualities as real qualities of experience itself” (p. 64). Smith takes the argument from illusion (construed broadly to include the argument from hallucination) to be a successful argument against metaphysical direct realism in my sense, and considers the task of the book to be that of finding a way of reconciling this with epistemological direct realism.

Epistemologists often discuss “direct realism” as a view that is clearly epistemological direct realism.[2] In such discussions, a central issue is whether or not perception is direct in the sense of providing non-inferential knowledge about the external world. The central concern regarding hallucination becomes the fact that such experiences can be indiscernible from veridical perceptions, and this indiscernability poses a potential threat to the idea that genuine perception has a special justificatory status given that the two types of experiences seem (prima facie) to place the perceiver in the very same epistemic situation. It also seems that Thomas Reid was a direct realist in the epistemological and not the metaphysical sense.[3] Sensations determine phenomenal character, but they “suggest to us” conceptions of their external causes.

Discussion of the argument from hallucination as an argument against metaphysical direct realism is also common in the literature. Valberg’s (1992) “problematic reasoning” is a puzzle about the “objects of visual experience”, understood as what is “present” in experience. Valberg is clear that although presence connotes some sort of direct availability or acquaintance, that epistemic feature is not definitive of the difficult to define notion of “presence”. Rather, it is presence to the mind that makes an object directly available. “Presence to the mind” thus seems to be something akin to directly contributing to how things are phenomenally within one’s experience, a metaphysical rather than epistemological notion.

Traditional discussions of the argument from hallucination or argument from illusion, as can be found in Price (1932), Moore (1957), and Broad (1923), often employed the troublesome verb “to appear”. In ordinary usage, to appear some way is, in part, to be represented as being that way. And such representations can, ordinarily, be accurate or inaccurate. Formulations of the argument from hallucination that utilize the language of appearing can thus lend themselves to being interpreted as arguments that are most centrally concerned with the epistemology of perception.

But as Robinson (1994) has pointed out, these advocates of the argument from hallucination (especially Price and Broad) are committed to what he calls the “phenomenal principle”—the claim that whenever something sensibly appears to have some quality, there is something that actually has that quality of which the subject is aware. The sense in which this principle is at all plausible, however, is only when “sensibly appearing” a certain way is understood as describing a fact about the (metaphysical) nature of the phenomenal character of one’s experience. It is unfortunate that the language of appearing can be used both to describe the metaphysical nature of one’s experience and to describe facts about the way things seem in an epistemic sense. But I think there is a charitable reading of these earlier advocates of the argument from hallucination, especially given their frequent claims that something like the phenomenal principle is simply undeniable, as offering an argument against what I have called “metaphysical” direct realism.[4]

Johnston (2004) and Martin (2004) both discuss the argument from hallucination in a way that suggests that its central challenge is to metaphysical rather than epistemological direct realism. Crane (2005) is quite explicit in treating the argument from hallucination, as well as the “problem of perception” more generally, as a metaphysical issue rather than primarily an epistemological one:

Note that the problem is expressed as one about the “nature” of an experience. This means its nature from the point of view of the subject having the experience—its phenomenological nature …. The nature of an experience from the subject’s point of view is also called the “phenomenal character” of the experience.

The two versions of the argument from hallucination are also mirrored in two distinct roles that sense-data might be asked to play. The metaphysical version of the argument is an argument for the existence of sense-data to play the role of constituting phenomenal character, and against the view that no such entities are needed to constitute perceptual experience. The epistemological version of the argument is an argument for the existence of such entities in order to play an epistemological role, as the things that we are most directly aware of.

The two views have not always been carefully distinguished, I would suggest, because sense-data have traditionally been understood to play both roles. And indeed, it is not implausible that the properties or objects that ground phenomenal character are also thereby the properties or objects that we are most directly aware of. But such a view is not obligatory, and the failure to carefully distinguish the two positions obscures the possibility of a variety of other views about the relationship between the phenomenal character of perceptual experience and our epistemic access to the external world. More importantly, the failure to keep the two positions separate can make certain positions appear more plausible than they are.

The primary target of the paper, representationalist theories of phenomenal character, are in the first instance metaphysical rather than epistemological theses. They are theories about what phenomenal character is. If there is a metaphysical version of the argument from hallucination that challenges representationalism, then this is an important issue independently of whether or not it is an argument that advocates of the argument from hallucination have traditionally had in mind. And representationalists such as Tye (2000) and Harman (1990) have in fact taken something like the metaphysical version of the argument from hallucination as potential objections to their views.

In what follows, I will evaluate the argument from hallucination as an argument concerning the metaphysics of perceptual experience. In particular, I will treat it as an argument for the following view:

The phenomenal properties associated with sensory experience are not physical properties of the external objects of perception (such as middle-sized physical objects like tables, trees, and fire hydrants).

            As I use the phrase, a “phenomenal property” is a property characterizing an aspect of the phenomenal character of an experience. It is thus a thoroughly metaphysical notion. The phenomenal character of an experience is the “what it is likeness” of an experience. To illustrate, consider an inverted spectrum scenario. Jack has visual experiences when looking at red things that are like the visual experiences Jill has when looking at green things. The phenomenal character of their experiences is the feature that we imagine to be reversed when we consider such possibilities.[5] Anyone who accepts that there is such a thing as phenomenal character can accept that there is such a thing as a phenomenal property.

            Traditional articulations of the argument from hallucination have as their conclusion that the objects of experience or perception are mental or internal. But one can also give the argument in terms of properties, and indeed, such an argument is more modest in its conclusion. As I will present the argument from hallucination, it is an argument concerning the nature of phenomenal properties, and it is neutral regarding the question of whether or not the phenomenal character of perceptual experience is constituted by a special sort of object that bears phenomenal properties, such as a sense-datum. That is, the argument is neutral between sense-data theories and more contemporary conceptions of conscious experience as grounded in sensory qualia (which need not be properties of a special sort of object distinct from minds or physical bodies).

Let us say that a subject “experiences” some property r if and only if r (at least partially) characterizes the phenomenal character of the subject’s experience. For a subject to experience some particular phenomenal property is thus a metaphysical fact about the subject and her experience, rather than an epistemic fact (such as being aware of or perceiving that property). Whether a subject’s experiencing r entails that the subject is aware of r is a further matter. The revised argument proceeds as follows:

 

1.     Mary experiences phenomenal property r.

2.     There is no physical thing with phenomenal property r before her.

3.     The property r that Mary experiences is not a physical property of an external object of perception.

4.     Mary’s experience has phenomenal character that is identical to a veridical perception of red.

5.     When Mary has a veridical perception, the phenomenal property r that she experiences is not a physical property of the external object of perception. Metaphysical direct realism is false.

 

It is compatible with the conclusion of this argument that phenomenal properties are physical properties of brains, properties of sense-data, or irreducibly mental properties. The argument is thus an argument against what we might call “metaphysical direct realism”, the view that the phenomenal characters of perceptual experiences are constituted by properties of the external objects of perception.

There are currently two popular responses to arguments of this sort. Disjunctivists can be understood as denying premise 4.[6] They deny that there is any positively characterizable feature in common between hallucination and veridical perception. But disjunctivism has features that make it unattractive to many. For one, in denying premise 4, it endorses a rather severe restriction on our ability to notice differences in the phenomenal character of our experiences. It certainly seems to us that hallucinations can be phenomenally just like veridical experiences. The disjunctivist claims that although hallucinations are indiscriminable from veridical perceptions, they are nonetheless phenomenally distinct. But one would think there is an explanation for the fact that a hallucination and veridical perception can be indiscriminable. And the most plausible explanation is that they have some feature in common, namely, their phenomenal character.[7]

Intentionalism rejects the entailment from premises 3 and 4 to the conclusion. Intentionalism, unlike disjunctivism, accepts that there is a common factor between veridical and hallucinatory experiences. But the presence of a common factor on this view does not entail that phenomenal properties are not properties of external objects. The common factor is a shared intentional content. In veridical perception, the external object of perception literally has the properties that are present in perceptual experience. In hallucinatory experience, one is not made aware of a sensory intermediary that has these properties. One is simply presented with the properties that one’s experience represents the external object as having. Such properties and objects are mere intentional objects.

 

The Intentionalist Response

 

The intentionalist response to the argument from hallucination begins by the observation that perceptual experiences have intentional content. Just as other intentional states, such as beliefs, can be about non-existent objects, so can perceptual experiences. The sense in which it is true to say that “Mary sees red” or “Mary perceives red” is an intentional sense; Mary has a visual experience that represents redness. But from this premise, one cannot conclude that there exists something red that Mary sees. As Harman (1990) points out by analogy, the fact that Ponce de Leon searched for the Fountain of Youth does not entail that there exists a Fountain of Youth that Ponce de Leon was searching for. And we certainly ought not conclude that Ponce de Leon was searching for something mental. Likewise, it is claimed that in hallucination we need not posit something mental to serve as the object of experience.

Early versions of the intentionalist response can be found in Anscombe (1965) and Hintikka (1967).[8] Those versions focus on the way we talk about perception and hallucinatory experiences, observing that verbs like “to see” or “to dream” are intentional. But the challenge for direct realism mounted by the argument from hallucination is not merely verbal, as the revised argument from hallucination attempts to make clear.

The intentionalist response has more recently been adopted by reductive representationalists about phenomenal character. Representationalism is the view that there can be no difference in phenomenal character without a difference in intentional content. That is, representationalism claims that phenomenal character supervenes on intentional content. Most advocates of representationalism have found it promising as a solution to the “hard problem” of consciousness. These reductive representationalists hold that phenomenal character can be reduced to purely physical intentional content. On this view, the phenomenal is nothing over-and-above the intentional.

Reductive representationalism offers an intentionalist response to the argument from hallucination that does not require that we treat the challenge as merely verbal. According to representationalism, perceptual experiences, just like propositional attitudes, are representational states. Hallucinations are misrepresentations—perceptual states that represent nonexistent things. And just as beliefs about non-existent things do not require the positing of existents as the objects of belief, perceptual experiences that represent non-existent things do not require the positing of existents as the objects of perception.

This response has two attractive aspects. First, it explains why hallucinations can have the same phenomenal character as veridical perception. According to representationalism, phenomenal character supervenes on intentional content. Hallucinations and veridical perceptions that share intentional content will thus share phenomenal character. Second, the representationalist response does not require the positing of sensory intermediaries as the “common factor”, thus allowing for some form of metaphysical direct realism. The phenomenal properties that we are aware of in experience really are, in the case of veridical perception, properties of the external objects of perception.

 

Two Versions of Reductive Representationalism

           

            The representationalist response to the argument from hallucination does the seemingly impossible, keeping the attractive part of indirect realism without the metaphysical baggage, and retaining a form of metaphysical direct realism without flying in the face of phenomenology. Alas, it really is too good to be true. The attractiveness of this response to the argument from hallucination is deceptive, relying as it does on a frequent ambiguity in discussions of representationalism.

            Before presenting the two versions of reductive representationalism, I should make explicit two assumptions about such views upon which my criticisms will rest. First, I assume that the views on offer, and their responses to the argument from hallucination, are realist about the phenomenal character of both veridical perception and hallucinatory experiences. Certainly one can be a representationalist and not be a realist about phenomenal character. Dennett’s (1991) view can arguably be understood as an anti-realist representationalist theory of phenomenal consciousness. And it may be that at the end of the day, the most charitable reading of some of the representationalists I will discuss here is a reading that takes them to be anti-realists about phenomenal character. But I want to assume here that the representationalist way of responding to the argument from hallucination is a way of reconciling realism about phenomenal character with both the common factor assumption and with some form of direct realism. And indeed, prominent representationalists such as Dretske and Tye do present themselves as being realists about the phenomenal character of both perception and hallucination. If the adoption of representationalism requires that one be an anti-realist about the phenomenal character of hallucinations and/or veridical perceptions, this will at least be a useful conclusion for determining whether one ought to adopt the view. For many, this will show that representationalism must be mistaken.

            The second assumption that I will be making in what follows is that reductive representationalism is to be paired with an ontologically conservative theory of intentionality. An ontologically conservative theory of intentionality is one that gives an account of intentionality solely in terms of properties, objects, or relations that are needed as part of one’s ontology independently of explaining intentionality. I take naturalistic theories of intentional content, such as causal-covariation theories or teleological theories, to be instances of ontological conservatism about intentionality.[9] I take Brentano’s (1874) theory, or Meinong’s (1904), to be ontologically liberal. I have no doubt that if one reduces phenomenal character to intentional content within a suitably liberal theory of intentionality, one can successfully accommodate hallucination. It is a little less clear whether such a view, such as one that might have been offered by Brentano, could qualify as a version of metaphysical direct realism. But if intentional properties are allowed to be sui generis, then I suspect that some such view is possible. Such a view would not, however, be the kind of view that has ordinarily been offered by those advocating representationalism. Most reductive representationalists adopt a naturalistic theory of intentional content. It would, again, be a significant conclusion if representationalism can only be made plausible by adopting an ontologically liberal theory of intentionality.

With those assumptions out of the way, I want to now focus on the fact that representationalism, as defined above, is underspecified. Reductive representationalism identifies phenomenal properties with intentional properties of a certain sort. But once we get more particular about what phenomenal character is according to reductive representationalism, we find two possible views: vehicle-based representationalism and content-based representationalism. According to content-based representationalism, phenomenal properties are represented properties. They are the properties that enter into the propositional content of the experience. For instance, phenomenal redness is, on this view, the property that the object of perception is perceived to have. Thus, if the experience is veridical then the phenomenal property that one is aware of is in fact a property possessed by the object of perception. According to vehicle-based representationalism, a phenomenal property is the property of representing p rather than the represented property p itself. On both views, phenomenal character supervenes on intentional content. But it does so for different reasons on each view. On content-based representationalism, phenomenal character just is the content. On vehicle-based representationalism, phenomenal character is a property of the vehicle of content—the property of having a particular content.

The dominant form of representationalism, though not always stated unequivocally nor recognized as such, is content-based representationalism. Advocates include Tye (1995, 2000), Dretske (1995), and Harman (1990). According to content-based representationalism, phenomenal properties are not properties of experiences. Rather, the phenomenal character of an experience is identified with at least some aspect of the experience’s intentional content. Phenomenal properties are the properties represented by experiences, rather than properties of the vehicles of representation.[10]

Here are two representative statements of content-based representationalism, by Michael Tye and Fred Dretske, respectively:

What, then, is visual phenomenal character? …. The best hypothesis, I suggest, is that visual phenomenal character is representational content of a certain sort—content into which certain external qualities enter (Tye 2000).

 

In accordance with the Representational Thesis, I continue to identify qualia with phenomenal properties—those properties that (according to the thesis) an object is sensuously represented (represented_s) as having (Dretske 1995).

 

The prominence of content-based representationalism among the advocates of representationalism has frequently gone unrecognized. For example, David Chalmers (2004) explicitly defines representationalism as the view that phenomenal properties are identical to representational properties, where the latter are defined in vehicle-based terms, as the property of representing a certain intentional content. To say otherwise, he suggests, is a category mistake. He notes that others have articulated the view in what I am calling content-based terms, but suggests that the difference is merely terminological. Likewise, Alex Byrne (2002), in a review of Tye’s book (2000), finds it necessary to re-interpret Tye’s view as a form of vehicle representationalism. Byrne finds it to be definitional that phenomenal character is a property of experiences, and confusing of Tye to talk as if it is a property of content.

But in his reply to Byrne, Tye (2002) says the following:

Byrne says that we should agree at the outset that the phenomenal character of a state is a property of that state. I don’t see why we should so agree. To be sure, we talk of states having phenomenal character, but nothing in ordinary usage or thought commits us to the view that phenomenal characters are properties. After all, we talk of beliefs as having content and of words as having meaning, but we don’t take belief contents or word meanings to be properties of beliefs and words respectively. Why do that from the outset for phenomenal character? Buy into the Cartesian view of experiences as inner ideas or pictures viewed by an inner eye and it may be natural to take the ‘feel’ of an experience as a property of the idea or picture. But that isn’t commonsense. It’s philosophical dogma — precisely the dogma which representationalism opposes.

 

            The matter of dispute here is not merely terminological. Part of the motivation of representationalism (or at least Tye’s version of it) is to, as Shoemaker (2003) has described it, solve the explanatory gap problem by “kicking the phenomenal character downstairs, into the external world” (p. 256). Content-based representationalism has a story to tell about why an experience that represents redness has the phenomenal character that it has: the quality that we are aware of is redness. There is, on this view, no inner surrogate for redness that mediates our perception of it. Content-based intentionalism thus salvages the epistemological virtues, and the appeal to common sense, that motivate direct realist theories of perception. And it avoids that mystery so well-posed by Wittgenstein, as a seemingly unbridgeable gulf between consciousness and brain processes. If there is still a gulf, it is a gulf between external properties understood in scientific terms versus their appearances in conscious experience. Whether this gulf is just as wide as the former gulf is a question I will leave open.[11]

            Chalmers seems to take the difference as being terminological because of a difference in how, in the case of Dretske, “qualia” is defined. But this difference is philosophically significant. Qualia can be understood as the properties that we are directly aware of in perception. They can also be understood as the properties that directly constitute phenomenal character.[12] Chalmers appears to assume that Dretske would find it unobjectionable to identify the latter with properties of experiences rather than as properties of external objects (in the case of veridical perception). But as the quotation from Tye suggests, many representationalists are explicitly motivated to reject this position. Instead, they hold that both what we are directly aware of in introspection and what constitutes phenomenal character are the represented properties, not the representing. Thus Chalmers’ and Byrne’s assumption, though perhaps ultimately the most plausible view, is far from innocent.

            Even among those philosophers who have, at times, been quite explicit in their adoption of content-based representationalism have articulated the view in a way that seems more compatible in some respects with vehicle-based representationalism. For example, Michael Tye’s PANIC theory of phenomenal consciousness appears to waffle between the two views. Vehicles of content, not intentional contents themselves, can be “poised”. But contents, not vehicles, can be “abstract”. And much confusion has arisen among philosophers about whether or not being “non-conceptual” is a feature of vehicles or of contents.[13] Given many of Tye’s remarks on the matter (as indicated above), I will take Tye to be a content-based representationalist. But there does seem to be legitimate grounds for confusion on the proper interpretation of representationalism more generally, and as I will argue, that confusion has made reductive representationalism (in either form) appear to be a more satisfactory account of perception than it actually is. Once the two possible views are distinguished carefully, one finds that there is no coherent and successful representationalist response to the argument from hallucination.

 

Content-based Representationalism

 

            How does content-based representationalism respond to the argument from hallucination? In the case of hallucinations, the phenomenal properties represented by the experience are not instantiated. They are mere intentional objects or properties of intentional objects. The red fire hydrant that Mary hallucinates is no more a real existing thing than the Fountain of Youth that Ponce de Leon was searching for. In both cases something is represented. In the case of Ponce de Leon, it is an object of belief or desire. In Mary’s case, it is an object of hallucination. But the object of hallucination is a mere intentional object, just as the Fountain of Youth is a mere intentional object of de Leon’s desire.

But this raises a problem. Content-based representationalism avoids sensory intermediaries, but cannot adequately account for the phenomenal character of hallucinations. The inadequacy here is metaphysical. In the case of hallucination, the content-based representationalist lacks the ontological tools to fully account for the fact that the subject has an experience with a certain phenomenal character. If the property that Mary represents is what constitutes the phenomenal character of her experience, then it seems that in order for her to have an experience with that phenomenal character, the property must exist and be instantiated.

This can be seen clearly when considering cases of phantom pain. Suppose that Mary has a phantom pain, which she feels as occurring in her nonexistent left leg. The phenomenal property associated with her pain clearly exists and is instantiated. Mary would be quite upset to be told by her doctor that she needn’t worry, that medication is unnecessary because the existence of her pain is illusory. There is an illusion here, but it is not illusory that the phenomenal property associated with pain exists and is genuinely instantiated. But in the case of phantom pain, the intentional object of Mary’s pain does not exist. And so any view that identifies the phenomenal properties of hallucinatory experiences with properties of mere intentional objects—intentional objects that do not exist—cannot possibly be correct. We can state the argument carefully as follows:

1.     When Mary has a phantom pain, the “painy” phenomenal property she experiences really exists and is instantiated.

2.     The intentional object of Mary’s experience does not exist.

3.     The phenomenal property that Mary experiences is not a mere intentional object (nor a property of a mere intentional object).

It is worth pausing over what representationalists have said about phantom pains, as it illustrates the way in which the frequent ambiguity in the literature between content-based representationalism and vehicle-based representationalism has obfuscated the problems posed here in giving a metaphysically adequate account of hallucinatory experience. Harman responds to this concern regarding phantom pains by stating that the word “pain” can sometimes refer to the intentional object and sometimes refer to the experience itself.[14] The referred pain exists in the sense that the experience (a representation) exists. The referred pain does not exist in the sense that the intentional object does not exist.

            But this response to the worry about referred pain is a case of Harman wanting to have his cake and eat it too, waffling between content-based and vehicle-based representationalism. The sense in which the pain surely exists is the sense in which the phenomenal quality we find so discomforting is instantiated. In discussing color experience, Harman discusses at great length that the qualities we are aware of in color experience—phenomenal color properties—are the properties represented by our experiences rather than properties of the experience. They are not even, according to Harman, properties of the experience in virtue of which the experience has the representational content that it has—what Harman calls “mental paint”. They are the colors that, if some of our experiences are veridical, belong to the external objects of perception. And in the case of illusion or hallucination, these properties are not instantiated. They are merely properties of the intentional object of the experience.

            This view is clearly false in the case of pain experience—in a way that Harman appears to want to concede. The phenomenal property associated with pain (the property most of us would simply be inclined to call “the pain”) clearly exists and is instantiated. But in the case of phantom pain, the intentional object does not exist. Harman is thus inclined to identify the pain with the representation. But that is to no longer identify phenomenal properties with intentional objects or that which is represented by an experience, but instead to identify phenomenal properties with the vehicle or properties of the vehicle of representation itself. Thus Harman’s response abandons content-based representationalism in favor of vehicle-based representationalism (to be discussed below).[15]

            This problem is not isolated to pain experiences. The sense in which the pain clearly exists even when it is a referred pain is precisely the sense in which there is a phenomenal color property instantiated even in the case of hallucination. The phenomenal color property cannot be, just as the pain cannot be, a nonexistent and merely intentional object or property. There is an argument against content-based representationalism regarding color hallucinations that is perfectly analogous to the one given above regarding phantom pains:

1.     When Mary hallucinates red, the red phenomenal property that she experiences really exists and is instantiated.

2.     The intentional object of Mary’s experience does not exist.

3.     The phenomenal property that Mary experiences is not a mere intentional object (nor a property of a mere intentional object).

One response to the argument is for the content-based representationalist to give an account of the intentional objects of experience that is equipped to metaphysically ground the phenomenal character of hallucinatory experience. For example, Lycan (1987, 1996) claims that “phenomenal individuals” are intentional inexistents. He then argues that intentional inexistents are best thought of as denizens of other possible worlds. They exist, just not here in the actual world.

Now this response clearly does not help. The painfulness of Mary’s phantom pain is actual, existing in the here and now! Likewise, when Mary has a hallucination of redness, there is a certain actually existing property that she is aware of, in virtue of which her experience has the phenomenal character that it has. There is perhaps something right about Lycan’s view. We might agree that when Mary has a red hallucination, her experience has an intentional object that does not exist. And a possible worlds account of intentional objects might be promising here. This view about the intentional content of experiences is not being called into question. The question is whether phenomenal properties are to be identified with these intentional contents. An adequate account of what Mary’s hallucination represents may not require positing something that exists—after all, it is a hallucination. But as I have been arguing, the phenomenal character of her experience surely does exist, and a proper account of this phenomenal character does require existent and actual properties or objects.

There is an alternative elaboration of content-based representationalism that identifies representational content with something actual. One might identify the properties that we are aware of in color experience (even in cases of hallucination) with color universals.[16] This view has some attraction independently of the argument from hallucination. Phenomenal content has a certain abstractness or lack of particularity. For example, suppose that I am looking at a red fire hydrant and am having a visual experience that is typical in those circumstances. There are an indefinite number of different scenarios that would make my experience veridical. My experience does not require that any particular fire hydrant be at that location, only that it have a certain shape and color. Arguably, there need not even be a fire hydrant at that location, but only something that looks a certain way. A red fire hydrant-shaped critter could make this experience veridical.[17] Similarly, this very instantiation of redness, the token that actually belongs to the fire hydrant, need not be at that location in order for my experience to be veridical. All that my experience requires is that there be some instantiation or other of this determinate shade of redness at a particular place before me. That is, my experience represents only that an instance of the universal redness is present.

Now all of this may be right, but just as before, we should be careful not to confuse at the outset an attractive view about the intentional content of experience with an attractive view about what metaphysically grounds the phenomenal character of an experience. The present suggestion on behalf of content-based representationalism identifies the phenomenal properties that we are aware of in both veridical and hallucinatory experience with a universal. On its face, this is quite a peculiar view. We might well wonder, how can universals do this metaphysical job? Or, how can we be acquainted with universals in the way that seems to be required in order to account for our experiences of redness?

Let’s distinguish two general views about universals. An Aristotelian view about universals holds that universals are nothing over and above their instances. A Platonic view about universals holds that universals are something over and above their instances.

If universals are taken to be Aristotelian, then this response does not resolve the challenge. There is no instance of the universal redness before Mary when she has a hallucination of redness. And surely what Mary experiences is not all of the instantiations of redness spread out across the universe, or some arbitrary instance of redness that is not before her.

One might object that such an Aristotelian view merely requires that the representationalist adopt phenomenal externalism, a doctrine that many representationalists have willingly accepted as a consequence of their views.[18] According to phenomenal externalism, the phenomenal character of an experience does not supervene narrowly on the subject’s physical state, but instead also supervenes in part on factors external to the subject. If intentional content is wide, and phenomenal character supervenes on this wide intentional content, it follows that phenomenal character is wide. This much is required by many representationalist views. And when one considers many of the familiar thought experiments in the literature—brains in vats, Twin Earth cases, and Swamp Man—this minimal version of phenomenal externalism is itself implausible to many.

But it should be noted that in adopting the Aristotelian view, content-based representationalism is forced to take on a more extreme version of externalism than might be thought. Criticisms of externalist representationalism and the phenomenal externalism that goes with it do not normally distinguish between vehicle-based and content-based representationalism. They instead focus on what the two views have in common—that since phenomenal character does not supervene on “what’s in the head”, microphysical duplicates situated in different environments can have experiences with distinct phenomenal characters. Once we look more carefully at what phenomenal properties are, according to these two forms of representationalism, the resulting phenomenal externalism in the case of content-based representationalism becomes even less plausible.

Vehicle-based representationalism identifies phenomenal character with the property of representing that p. According to most representationalists, this representational content p is a kind of wide content. It follows that these representational properties of experiences also must be individuated widely. The property of representing that p is a relational property of experiences. And since phenomenal character is identified with these representational properties, vehicle-based representationalism entails that phenomenal character is a relational property of an experience. According to vehicle-based representationalism, phenomenal character fails to supervene narrowly on a subject’s microphysical properties in the same way that the weight of an object fails to supervene on that object’s microphysical properties. An object has the weight that it does only in relation to the gravitational force of a large nearby body. But weight is still a property of the object. Likewise, the phenomenal character of an experience is still a property of the experience, just one that it has only in virtue of its relations to the external environment.

Content-based representationalism requires something far more radical. Content-based representationalism tells us that phenomenal properties are not properties of experiences at all, and so they are not even relational properties of experiences. The Aristotelian suggestion identifies these properties with universals, understood as nothing over and above their instances. On this view, the redness that Mary experiences while hallucinating is not the property of representing physical redness. It is not simply, like the weight of an object, a relational property of her experience that depends on relations to the external environment. The redness that is phenomenally manifest to Mary is entirely external, even in the case of hallucination. The redness that she experiences is the sum total of instances of redness scattered across the universe, or some arbitrary instance of redness. This, I think, is phenomenal externalism gone mad.

Besides being implausible, content-based representationalism paired with an Aristotelian view of universals also admits of a simple refutation. It is possible to have a hallucination of redness even in a world at a time in which there are no physically red things. Suppose, for instance, that everything in the universe were painted or modified so as not to be red.[19] One could, nonetheless, have a hallucination of redness. But since there is no actual present instance of redness, the phenomenal property that the subject experience’s cannot be an external instance of physical redness.

A different development of the Aristotelian universals view might identify the phenomenal property with an instance or set of instances that were part of the learning history in the formation of the creature’s representation of that property, along Dretske’s teleological lines. But this has the absurd consequence that the phenomenal character of one’s present experience can be constituted by property instances that are in the past and not in the present.

Perhaps all of this suggests that the content-based representationalist should adopt the view that phenomenal properties are uninstantiated properties. But if universals are taken to be Platonic, then the response is far outside the naturalistic spirit that motivates reductive representationalism. How could we be acquainted with such uninstantiated universals in the way that we are acquainted with phenomenal properties? Platonic universals presumably exist outside space-time. And Platonic universals lack causal powers. The problem is not that of making sense of how one could represent such things as Platonic universals. Rather, the puzzle is how a property could, despite being uninstantiated, constitute phenomenal character.

Conscious experiences, including hallucinations, have a particularity to them that seems to rule out the position that phenomenal properties are uninstantiated properties. Whatever phenomenal properties are, they need to be instantiated in order to do the metaphysical job of constituting phenomenal character. As mentioned earlier, Tye (1995) has pointed out that perceptual experiences have, in some sense, a general or abstract character.[20] Experiences are abstract in that different objects can look precisely the same, and hallucinations can be phenomenally just like veridical experiences. What such considerations show, however, is that phenomenal content is abstract. My experience of a red fire truck does not require, in order to be veridical, that the fire truck have any particular instance of the determinate redness that I represent. It is sufficient that it has some instance or other of that property. Indeed, that particular fire truck need not be present, but only a qualitatively identical truck (at least with respect to the portion of the truck that I am sensing), or perhaps only something with the appropriate shape, size, location, and color. The content of experience is plausibly abstract in this sense. But such considerations show that experiences do not represent particular property instances, not that phenomenal properties are not particular property instances. In fact, a metaphysically adequate account of the phenomenal character of hallucinations seems to require that there exist particular instances of the relevant phenomenal properties. The content of hallucination and veridical perception may be abstract, but the phenomenal characters of those experiences have particular, not abstract, natures.

Johnston (2004) has recently defended a view on which hallucination involves awareness of uninstantiated properties.[21] According to this view, the primary objects of perception are what Johnston calls “sensible profiles”, which are complexes of qualities and relations. In veridical perception, we are aware of instantiations of sensible profiles. When we hallucinate, we are aware of sensible profiles which are not in fact instantiated in the scene before the perceiver.

Johnston suggests that the idea of being aware of uninstantiated sensory profiles is no more problematic than being aware of manners of presentation (by which he appears to mean something like Fregean senses) that have no referent.[22] But even on Johnston’s own view there is an important difference. He says of sensible profiles that they are themselves objects of sensory awareness, and he takes this to account for what is distinctive about sensing as opposed to discursive thought.[23] If this is so, then an analogy with Fregean modes of presentation is of no help in clarifying how one could be aware of uninstantiated properties in hallucination. Most philosophers would not endorse the idea that Fregean modes of presentation can be objects of sensory awareness. Indeed, Fregean modes of presentation seem to be too abstract to play that sort of role. The problem is that uninstantiated properties also seem to be ontologically such that one could not sense them either.

For uninstantiated properties to play the role of grounding phenomenal character, as required by content-based representationalism, it seems that ontological conservatism about intentionality would have to be abandoned. Naturalistic theories of intentional content do not provide an explanation for how one could become related to an uninstantiated universal in such a way that the universal, such as phenomenal pain, could contribute to the phenomenal character of an experience. Such theories generally reduce intentional content to a complex set of ultimately physical relations (including causal and historical relations) between the organism and property instances. One could introduce an irreducible relation that can hold between minds and Platonic universals, in addition to the relation of exemplification. Short of that, uninstantiated properties do not seem to be eligible candidates for the phenomenal properties present in hallucination.

There is the further problem of causal efficacy on this view, which can again be illustrated by considering the experience of pain. Unless one is willing to adopt epiphenomenalism, one wants to say that pains can cause various behaviors, such as exclamations of “ouch” or the seeking of medication. In such cases, it is the phenomenal property associated with pain—the “painfulness” of the pain, that plays some role in effecting behavior. Uninstantiated properties simply cannot play that causal role.

It might seem that this problem is nothing more than the more general problem of wide intentional causation.[24] To see that the problem is greater here for the content-based representationalist, consider Dretske’s (1995) solution. There he clearly adopts content-based representationalism, including for pains and other bodily sensations.[25] In the final chapter he considers objections to his view on the grounds that his phenomenal externalism is incompatible with the causal efficacy of conscious experiences. He responds by invoking his distinction between structuring and triggering causes. He grants that mental states must exist in the here-and-now in order to serve as triggering causes of a person’s behavior. But Dretske denies that the mental must supervene on the here-and-now in order to be causally relevant as structuring causes.

It is not clear that phenomenal pain’s being a merely structuring cause is sufficient to account for the intuition of causal efficacy. The quality I am aware of, and which seems to cause my behavior, is in the here and now! The causal efficacy of the phenomenal may in this sense be importantly different from the causal efficacy of intentional content for propositional attitude states. If one is an externalist about belief content, it is not implausible to think that the causal relevance of content is merely at the level of structuring rather than triggering causes.[26] Twin Earth cases provide one motivation for such a view. One can make sense of the idea that the difference between believing that water is wet and believing that twin-water is wet is a difference that does not make a difference with regard to the causing of bodily motion. Here Dretske’s distinction becomes useful. Only when the effect is understood as a form of behavior, and questions arise about why a subject is configured such that being in some internal state has the effects that it has, does external mental content become causally relevant. By contrast, lying behind the intuition that pains and other conscious experiences are causally efficacious is the idea that such qualities can literally cause bodily motion. The kind of epiphenomenalism about consciousness that most philosophers seek to avoid is the view that denies this. Uninstantiated properties would seem to be, because they are not instantiated, incapable of this sort of causal efficacy.[27]

An objection to the argument from hallucination as it concerns content-based representationalism might still linger. Why think that the object of a hallucination must exist? After all, the object of a belief about a non-existent object need not exist. In both cases, to think that something must actually exist to account for the psychological state of misrepresenting is to commit an intensional fallacy.

In reply, it should be noted that the object of a hallucination, in at least one important sense of “object”, need not exist. A hallucination represents something as existing that in fact does not exist—and this thing that is represented is appropriately considered the object of the hallucination. But an adequate theory of phenomenal consciousness must have the metaphysical equipment to account for the phenomenal character of hallucinations. Beliefs do not have phenomenal character in the way that sensory experiences do, and thus beliefs do not pose the same problem as hallucinations.[28]

The dagger that Macbeth “saw” or “hallucinated” did not exist. But the phenomenal properties that he experienced while hallucinating, various phenomenal colors and shapes, surely did exist and were instantiated. Without them, his experience would lack phenomenal character. Contrast this with beliefs about nonexistent things. The pink elephant that Mary believes is in the next room does not exist. But it is not clear that there are any relevant phenomenal properties that she experiences as part of having that belief.[29] If that is right, then hallucinations pose a metaphysical problem that is simply not present in cases of belief about things that do not exist.

 

Vehicle-based Representationalism

 

            The preceding suggests that the only viable form of representationalism is vehicle-based representationalism. Vehicle-based representationalism identifies phenomenal character with the property of representing that p, rather than with the intentional content p. Prima facie, the property of representing that p is a property that is really instantiated during hallucination, and so vehicle-based representationalism appears to be immune to the criticisms above aimed at content-based representationalism.[30]

            On this view, phenomenal properties are properties of experiences, rather than properties of external objects. Rather than answering the argument from hallucination, vehicle-based representationalism concedes the conclusion of the argument. Recall that the conclusion of that argument was the following: the phenomenal properties associated with sensory experience are not physical properties of the external objects of perception (such as middle-sized physical objects like tables, trees, and fire hydrants).

            Vehicle-based representationalism no more vindicates metaphysical direct realism or challenges the argument from hallucination than did Russell’s (1927) view that sense-data were brain states. On both views, the properties that we are sensibly aware of in perception and hallucination are not properties of external objects but instead properties of minds or brains. One difference that might be appealed to is that on externalist theories of content, the property of having a particular representational content is a relational property of the vehicle that involves the external world. But a relational property is distinct from its relata, and it remains the case that the phenomenal properties that we are aware of in experience are not, even in veridical experience, the external properties of objects that such experiences represent. Representational properties, even if externalistically individuated, are properties of minds or brains. Vehicle-based representationalism therefore, like Russell’s view, fails to vindicate the naïve realist intuition that properties of the external world are made manifest to us or are present in perceptual experience. Furthermore, some sense-data theorists, such as Jackson (1977) and Moore (1957), held that sense-data were “out there” in physical space. A view qualifies as a version of metaphysical direct realism not simply by being physicalist (as was true of Russell’s theory) or by being externalist (as was true of Jackson and Moore’s sense-data). Metaphysical direct realism seems to also require, at a minimum, that the phenomenal properties are non-mental.

            Vehicle-based representationalism is also not supported by some of the usual arguments for representationalism, such as the transparency argument. According to one strong version of the transparency thesis, the phenomenal properties are properties that, in the case of color experience, we experience as belonging to the surfaces of external objects.[31] Under the assumption that visual experiences are by and large veridical, this thesis supports content-based representationalism. But it does not support vehicle-based representationalism, since vehicle-based representationalism claims that phenomenal properties are properties of the vehicles of representation rather than properties of the external objects of perception. Vehicle-based representationalism, plus the transparency thesis, yields a form of projectivism, convicting experience of massive error. Properties that really belong to experiences are systematically projected onto the external objects of those experiences. In short, if the phenomenal properties seem to be properties “out there”, but are in fact properties of the vehicle of representation, then perceptual experience is systematically delusive.

One might interpret the transparency thesis not as a thesis concerning phenomenal properties but as a thesis concerning only what we are “directly aware of”—that the properties we are directly aware of in perceptual experience are the external represented properties, but these are not the phenomenal properties (those properties that constitute phenomenal character). Given that discussions involving “qualia” or “phenomenal properties” are sometimes ambiguous (in the way discussed above) between their epistemological and metaphysical roles, there is a genuine question here about how one should interpret appeals to transparency in arguments for representationalism (which can also be ambiguous between content-based and vehicle-based versions). But if the transparency thesis is interpreted in this more modest way, then it fails to motivate representationalism (of either variety). It fails to motivate representationalism because it is compatible with a “natural sign” theory of the relationship between perceptual experience and representational content, according to which the having of a sensory impression or non-representational quale “brings to the mind” an associated representation of external properties. Such a view, like representationalism, can allow that we are not directly aware of the properties that constitute phenomenal character, but are instead in some sense directly aware of external properties. It is no accident then that representationalists who typically appeal to transparency, most notably Tye (2000), seem to be content-based representationalists rather than vehicle-based representationalists. The only interpretation of transparency that supports representationalism is one that supports content-based representationalism and not vehicle-based representationalism.

Unlike content-based representationalism, reductive vehicle-based representationalism also makes the connection between a particular phenomenal character and a particular intentional content mysterious. Why would representing that p feel like this, rather than like that? Why does visually representing something as being green feel the way it does, rather than the way representing something as red feels? Content-based representationalism is a little better off on this count, however implausible the view might be. It is due to the nature of redness, the property represented, that experiences of red feel the way that they do. Vehicle-based representationalism cannot offer this answer, since in general, the property of representing some property or object does not resemble that property or object. Beliefs that represent cows do not resemble cows, nor does the property of representing cows resemble cows. Vehicle-based representationalism thus fails to vindicate the naïve realist intuition that in having a perceptual experience, external properties or objects are present in our experience. And it fails to satisfy the perhaps weaker intuition that there is at least a qualitative resemblance between the phenomenal properties we experience and the external properties that those experiences represent.

            Vehicle-based representationalism is a form of representationalism that goes along with an idea Brian Loar (2003) expresses as the view that identifies qualia with “paint that points”—playing off Harman’s (1990) notion of “mental paint”. Such a view countenances the existence of phenomenal properties that determine intentional content but are distinct from that content. But phenomenal properties on this view are not like mental paint—they are not raw qualia that represent, if at all, in virtue of an interpretation or a resemblance to their objects (as in the manner of paint on a canvas). They are intrinsically intentional.

            A view like this does not salvage metaphysical direct realism. But it does allow for epistemological direct realism, barring some argument that the latter requires the former. For unlike sense-data theories of perception, vehicle-based representationalism does not hold that we represent external properties and objects indirectly, and only represent mental items directly. According to one way of developing vehicle-based representationalism, we do not at all, except when introspecting, represent our own mental states that are involved in perception. We represent external properties and object by having (and not necessarily representing) experiences with a certain phenomenal character. That the phenomenal character of an experience is not constituted by that representational content does not entail that experiences fail to directly represent that content. On such a view, the external world is not made manifest to us in the way that naïve realism wants, but our representation of it is not mediated by any other sort of representation. This is, I have argued, the only form of direct realism that reductive representationalism of any variety can allow.
References

 

Anscombe, G. E. M. 1965. The intentionality of sensation; A grammatical feature, in R. J. Butler, ed., Analytical Philosophy, second series. Oxford: Basil Blackwell.

Austin, J. 1962. Sense and Sensibilia. Oxford: Clarendon.

Block, N. 1990. Inverted earth. Philosophical Perspectives 4, J. Tomberlin, ed., Northridge: Ridgeview Publishing Company.

Bonjour, L. 2002. In search of direct realism. URL= <http://philosophy.rutgers.edu/EVENTS/EPIS2002/bonjour.pdf>

Brentano, F. 1874. Psychology from an Empirical Standpoint. London: Routledge and Kegan Paul.

Brewer, B. 1999. Perception and Reason. Oxford, UK: Oxford University Press.

Broad, C. D. 1923. Scientific Thought. London: Routledge and Kegan Paul.

Byrne, A. 2002. Don’t PANIC: Tye’s intentionalist theory of consciousness. URL=<http://host.uniroma3.it/progetti/kant/field/tyesymp_byrne.htm>

Chalmers, D. 2004. The representational character of experience. In B. Leiter (ed.), The Future of Philosophy. New York: Oxford University Press.

Copenhaver, R. 2004. A realism for Reid: mediated but direct. British Journal for the History of Philosophy 12.

Crane, T. 1992. The non-conceptual content of experience. In T. Crane, ed., The Contents of Experience. Cambridge, UK: Cambridge University Press.

Crane, T. 2005. The problem of perception. The Stanford Encyclopedia of Philosophy (Spring 2005 Edition), E. Zalta, ed., URL = <http://plato.stanford.edu/archives/spr2005/entries/perception-problem/>.

Cummins, R. 1990. Pappas on the role of sensation in Reid’s theory of perception. Philosophy and Phenomenological Research 50.

Dennett, D. 1991. Consciousness Explained. Little Brown.

Dretske, F. 1995. Naturalizing the Mind. Cambridge, MA: MIT Press.

Dretske, F. 2003. Experience as representation. Philosophical Issues 13, E. Sosa and E. Villanueva, eds., Oxford: Blackwell Publishing.

Gunther, Y. (ed.) 2003. Essays on Nonconceptual Content. Cambridge, MA: MIT Press.

Harman, G. 1990. The intrinsic quality of experience. Philosophical Perspectives 4, J. Tomberlin, ed., Northridge: Ridgeview Publishing Company.

Heil, J. and A. Mele. (eds.) 1993. Mental Causation. Oxford, UK: Oxford University Press.

Hilbert, D. 2004. Hallucination, sense-data, and direct realism. Philosophical Studies 120: 185-191.

Hintikka, J. 1969. The logic of perception. In Models for Modalities. New York: Humanities Press.

Hinton, J. 1967. Visual experiences, Mind 76: 217-27.

Huemer, M. 2001. Skepticism and the Veil of Perception. Lanham, MD: Rowman and Littlefield.

Horgan, T. and Tienson, J. 2002. The intentionality of phenomenology and the phenomenology of intentionality. In D. Chalmers, ed., Philosophy of Mind: Classical and Contemporary Readings. Oxford University Press.

Jackson, F. 1977. Perception: A Representative Theory. Cambridge, UK: Cambridge University Press.

Johnston, M. 2004. The obscure object of hallucination. Philosophical Studies 120: 113-183.

Loar, B. 2003. Phenomenal intentionality as the basis of mental content. In M. Hahn and B. Ramberg, eds., Reflections and Replies: Essays on the Philosophy of Tyler Burge. MIT Press.   

Lycan, W. 1996. Consciousness and Experience. Cambridge, MA: MIT Press.

Lycan, W. 2001. The case for phenomenal externalism. Philosophical Perspectives 15:17-35.

Martin, M. 2004. The limits of self-awareness. Philosophical Studies 120.

McDowell, J. 1982. Criteria, defeasibility, and knowledge. Proceedings of the British Academy.

Meinong, A. 1904. Äuber gegenstandtheorie. Engl. transl. “The theory of objects,” in Chisholm, R. M., ed., 1960. Realism and the Background of Phenomenology. Glencoe: The Free Press.

Moore, G. E. 1957. Visual sense-data. In Mace, ed., British Philosophy in Mid-Century, 203-211.

Pappas, G. 1990. Causation and perception in Reid. Philosophy and Phenomenological Research 50.

Pitt, D. 2004. The phenomenology of cognition, or, what is it like to think that P? Philosophy and Phenomenological Research 69.

Price, H. H. 1932. Perception. London: Methuen.

Reid, T. 1997. An Inquiry into the Human Mind on the Principles of Common Sense.          Edinburgh: Edinburgh University Press.

Reynolds, S. 2000. The argument from illusion. Nous 34: 604-21.

Robinson, H. 1994. Perception. London: Routledge.

Russell, B. 1927. The Analysis of Matter. New York: Harcourt, Brace.

Shoemaker, S. 2003. Content, color, and character. Philosophical Issues, 13:253-278.

Siegel, S. forthcoming. Which properties are represented in perception? In T. Gendler-Szabo and J. Hawthorne, eds., Perceptual Experience.

Siewart, C. 1998. The Significance of Consciousness. Princeton University Press.

Smith, A. D. 2002. The Problem of Perception. Cambridge, MA: Harvard University Press.

Snowdon, P. 1982. Perception, vision, and causation. Proceedings of the British Academy 68:455-479.

Stich, S. and T. Warfield, eds. 1994. Mental Representation: A Reader. Oxford, UK: Blackwell.

Strawson, G. 1994. Mental Reality. MIT Press.

Tye, M. 1995. Ten Problems of Consciousness. Cambridge, MA: MIT Press.

Tye, M. 2000. Color, Consciousness, and Content. Cambridge, MA: MIT Press.

Valberg, J. J. 1992. The Puzzle of Experience. Oxford: Oxford University Press.