99

ATTENTION AND MENTAL PAINT1 Ned Block New York University Abstract Much of recent philosophy of perception is oriented towards accounting for the phenomenal character of perception—what it is like to perceive—in a non-mentalistic way—that is, without appealing to mental objects or mental qualities. In opposition to such views, I claim that the phenomenal character of perception of a red round object cannot be explained by or reduced to direct awareness of the object, its redness and roundness—or representation of such objects and qualities. Qualities of perception that are not captured by what one is directly aware of or by representational content are instances of what Gilbert Harman has called “mental paint” (Block, 1990; Harman, 1990). The claim of this paper is that empirical facts about attention point in the direction of mental paint. The argument starts with the claim (later modified) that when one moves one’s attention around a scene while keeping one’s eyes fixed, the phenomenology of perception can change in ways that do not reflect which qualities of objects one is directly aware of or the way the world is represented to be. These changes in the phenomenology of perception cannot be accounted for in terms of awareness of or representation of the focus of attention because they manifest themselves in experience as differences in apparent contrast, apparent color saturation, apparent size, apparent speed, apparent time of occurrence and other appearances. There is a way of coping with these phenomena in terms of vague contents, but vague contents cannot save direct realism or representationism because the kind of vagueness required clashes wth the phenomenology itself. 1. Introduction Are phenomenological characters of perception—e.g. what it is like to experience redness or roundness—philosophically reducible to the redness24 NedBlock or roundness of the objects one sees or to representation of redness or roundness? If there is no such reduction, then there can be said to be mental paint2. The mental paint issue cross-cuts the metaphysical dispute between physicalism and dualism. I am friendly to reductive physicalism in the sense of a scientific reduction of phenomenological qualities to biological properties of the brain, a reduction thesis comparable to the “micro-reduction” of water to H2O, light to electromagnetic waves, and heat to molecular kinetic energy (Nagel, 1961; Oppenheim & Putnam, 1958), where micro-reduction is a reduction of properties of a whole to conglomerations of properties of the parts of that whole. Though I am open to reductive physicalism in that sense, I reject any philosophical reduction of the phenomenological qualities of perception to properties of objects such as redness or roundness or representations of redness or roundness and in this I am in agreement with some dualists (Burge, 2005; Chalmers, 2004). Philosophers may disagree about what the difference is exactly between scientific reduction and philosophical reduction. Perhaps it is the difference between an a posteriori reduction and an a priori reduction. I will not be discussing this issue further here. One aim of this paper is to argue that philosophical reductions of phenomenological properties to direct awareness of properties of objects or representations of those properties are wrong. A second aim is to present evidence for some surprising phenomena involving attention against can be explained better with mental paint than without it. A third aim is to argue that certain kinds of indeterminacy in perception can be accommodated in terms of an indeterminacy in representational content without any corresponding indeterminacy in phenomenology. It has been widely recognized that shifts of attention can affect the phenomenology of perception (Block, 1995; Chalmers, 2004; James, 1890; Macpherson, 2006; Nickel, 2007; Peacocke, 1993). What is controversial is whether such attentional changes in the phenomenology of perception can be accounted for in terms of what properties one is directly aware of or what properties one’s phenomenological state represents. The dialectic of this paper starts with a preliminary claim that a difference in attention can produce a difference in phenomenology without any difference in what properties one is directly aware of or what properties one’s experience represents. More specifically, two non-illusory percepts of the same feature of an object can differ in respect of, for example, perceived size or contrast. Later, I will suggest that indeterminate contents or awareness of indeterminate properties can accommodate the phenomena, but that experience is not indeterminate in the way it would have to be to save direct realism or representationism. This may seem paradoxical: two percepts can differ in respect of say perceived size or perceived contrast, yet neither be illusory. How can that be? By way of understanding why there is no paradox, it might be useful to consider, briefly, loudness. Loudness—the perceived intensity of a sound3—isAttention and Mental Paint 25 a function of a number of variables, aside from actual intensity, namely frequency, bandwidth and the duration of the sound. Although loudness in some sense presents intensity and is experienced as presenting intensity, the same intensity can sound differentially loud depending on other variables. Analogously, although perceived size presents actual size, perceived size is a function not only of the actual size but of other variables notably the distribution of attention. Just as there can be two phenomenally different but non-illusory presentations of the same sound intensity, there can be two phenomenally different but non-illusory presentations of the same size. Despite this analogy between perceived size and loudness, there are are some disanalogies between the two cases that will be mentioned later. I will argue that empirical facts concerning attention point in a direction incompatible with direct realism, and then move to arguing that the same points conflict with forms of representationism like that advocated by Alex Byrne, Peter Carruthers, Fred Dretske, Gilbert Harman, Chris Hill, William Lycan, Adam Pautz, John Searle and Michael Tye4 according to which the phenomenal character of perception is or supervenes on its representational content (how it represents the world to be). Similar points apply to Colin McGinn’s “cluster of properties” view (1999, p. 319) and Mark Johnston’s (2004) “sensory profiles” account. In previous papers, I have argued that the phenomenal character of perception goes beyond its representational content. Here the argument is importantly different: that there is something about the representational content of perception (and direct awareness) that is actually incompatible with the phenomenal character of perception. The paper will end with a brief discussion of a respect in which the mental paint view argued for here is less radical than what I have argued for previously. But first, I will briefly characterize the direct realist and representationist opposition and the sense in which these views reject and I embrace mental paint. 2. Direct Realism and Representationism According to direct realism, the phenomenal character of perceptual experience is “object-involving” in the following sense: an experience of the redness of the tomato depends for its existence and individuation on the tomato and (on some versions) its color. Any experience, even a possible experience, that is of a different tomato or no tomato or of a tomato that is not red is not an experience with that particular phenomenal character. Thus (surprisingly, counterintuitively) even an experience of a perfect duplicate of the tomato is bound to be different in phenomenal character. This claim is often justified by appeal to introspection. C.D. Broad (1951) famously said: “In its purely phenomenological aspect seeing is ostensibly saltatory. It seems to leap the spatial gap between the percipient’s body and26 NedBlock a remote region of space. Then, again, it is ostensibly prehensive of the surfaces of distant bodies as coloured and extended…It is a natural, if paradoxical, way of speaking to say that seeing seems to “bring one into direct contact with remote objects” and to reveal their shapes and colours.” (Quoted in (Fish, 2009; Hellie, 2007)). This “openness to the world” seems to be expressed by Martin Heidegger (1977, p. 156): “Much closer to us than any sensations are the things themselves. We hear the door slam in the house and never hear acoustic sensations or mere sounds.” (Quoted in Smith, 2002). G.E. Moore’s (1903) idea of the diaphanousness of experience is taken to combine the positive claim of openness with the negative claim that one cannot be aware of the experiences themselves (Crane, 2006; Martin, 2002; Siewart, 2003; Stoljar, 2004). Moore says “…the moment we try to fix our attention upon consciousness and to see what, distinctly, it is, it seems to vanish: it seems as if we had before us a mere emptiness. When we try to introspect the sensation of blue, all we can see is the blue; the other element is as if it were diaphanous…”5. I accept the positive introspective claim of openness. However, I will be arguing that, ironically, when combined with facts of attention, it dooms the very direct realist and representationist perspectives that it has been used to support.6 The object-involving view of perceptual experience is often explained in terms of constitution. As Bill Brewer (2004) puts it, “the subjective qualities of experience…are constituted by the actual spatial distribution of the various displays as these are accessible to the subject.” John Campbell (2002, p. 116) describes a similar constitution view, according to which “the phenomenal character of your experience, as you look around the room, is constituted by the actual layout of the room itself: which particular objects are there, their intrinsic properties, such as colour and shape, and how they are arranged in relation to one another and to you.” The point of direct realism is to capture the phenomenological differences among different percepts in terms of the world, specifically the direct pickup of worldly objects, properties and relations rather than any mental ways of perceiving those items that go beyond the objects and properties that are perceived. Note that the difference between a percept as of green and a percept as of red can be acknowledged by all sides to be normally caused by the difference between red and green in the world. The controversial issue is whether this relation is constitutive rather than causal. The second of the two views opposed to mental paint is representationism. Representationism (also called ‘representationalism’ and ‘intentionalism’) assumes that perceptual experiences have accuracy conditions (Siegel, 2008a) and holds (in one form) that the phenomenal character of experiences can be identified with the representational contents defined by those conditions (Tye, 1995). Or alternatively, what it is for a subject to have a certain phenomenal character is to have a certain representationalAttention and Mental Paint 27 content (Byrne, 2001; Crane, 2007). Representationism is sometimes described as a supervenience doctrine: phenomenal character supervenes on representational content (no difference in phenomenal character without a difference in representational content). As has often been noted (Kim, 1991; McLaughlin, 1995), supervenience is too weak a relation to ground a form of reductionism. My case against representationism depends on a version of it in which phenomenal character is determined by or flows from representational content rather than merely supervening on it. On certain views of the content of perception (for example, Tye, 2009), representationism can be a version of direct realism–for example, if the contents of perception are just the objects and properties perceived rather than something more like a proposition. Confusingly, the term ‘representationalism’ is used for what I call representationism and also used for indirect or representational realism, the view that one is aware of the apple only by being directly aware of something mental, for example a sense datum.7 The mental paint view I will be defending here is opposed to representational realism, so it rejects representationalism in both senses of the term. If we hold that a veridical perceptual state is individuated by what it is a direct awareness of, it we may be led to the disjunctivist view (Fish, 2009; Hinton, 1973; Martin, 2002; McDowell, 1982; Putnam, 1999; Snowdon, 1979–80) that veridical, illusory and hallucinatory cases share “no positive mental characteristics other than their epistemological properties of not being knowably different from some veridical perception” (Martin, 2004, p. 82). This is an epistemic conception of hallucination (Siegel, 2008b) in that no further mental properties are supposed to underlie the indiscriminability. (See Susanna Siegel’s (2008b) critique.) Tyler Burge (2005) has refuted disjunctivism in a way that suggests that the direct realist intuition and the disjunctivist view that grows from it derives ultimately from a type/token error. Burge points out (see also Burge, 1991; Siegel, 2008a) that particular (datable) token perceptual states could be said to have object-involving contents (contents individuated in relation to actual existing things). For after all the veridicality condition of my percept of the whiteness of this page is that this page itself—not some other qualitatively similar page–is white. Since my perception could not have that content without the existence and presence of that particular page, the perceptual content could be described as object-involving. But as Burge notes, token perceptual states may be object-involving in this way even though there are non-object-involving perceptual types that apply both to the veridical perception of this page and equally to indistinguishable illusory perceptual states. These non-object involving types would constitute a mental similarity among indistinguishable veridical and non-veridical perceptions. I won’t be discussing disjunctivism explicitly in this paper except in a few side remarks such as that of the last paragraph.28 NedBlock Figure 1. Attending to the face rather than the house changes the phenomenal character of experience. From (Tong, Nakayama, Vaughan, & Kanwisher, 1998), where it was used for a different purpose. 3. Selective Attention I say that facts about attention point away from direct realism and representationism. But there are some attentional phenomena that direct realism and representationism are well equipped to accommodate. It is well known that differences in attention can make for differences in which aspects of the environment the subject is aware of. Many people have noticed what is often called the “cocktail party effect” (Cherry, 1953) in which people at a cocktail party focus on one conversation while at least partially losing conscious awareness of other conversations. Colin Cherry tested this idea using the “dichotic listening” paradigm in which subjects wearing headphones are instructed to listen to, say, the left channel and ignore the right. Such subjects were often unable to tell what language the speech in the unattended channel was in or even whether the speech was forwards or backwards. Later work showed that at least some information from the unattended channel was processed to a high level unconsciously, but still it was at least partially gated out of conscious awareness. The reader can experience an example of selective attention via Figure 1. Attending to the face makes the house recede, phenomenally speaking and conversely for attending to the house.Attention and Mental Paint 29 According to direct realism, one role of selective attention is to shift within a single scene from direct awareness of some properties to direct awareness of other properties (Campbell, 2002). So direct realism can easily accommodate the role of attention in selecting some objects, locations and properties rather than others. A similar point can be made using Christopher Peacocke’s (1992) adaptation of Ernst Mach’s (1959) famous example: one can see a 45◦ tilted square either as a tilted square or as an upright diamond. (See also the discussion in (Macpherson, 2006).) The direct realist can accommodate the phenomenal difference by noting that the actual layout instantiates both properties and one’s awareness of the layout is constituted by which of the actually instantiated properties are the ones that are selected by attention and are the properties the perceiver is directly aware of. What I am calling direct realism is often called na¨ ıve realism (Smith, 2002), but the more na¨ ıve form of the view is that the phenomenal character of your experience as you look around the room, is constituted by the actual objects and properties instantiated in the room, including their spatial locations and relations to you. This view leaves no room for two observers at the same location to have different experiences and so does not allow for the role just described for selective attention. Unfortunately, statements of direct realism often do not distinguish sharply between this genuinely na¨ ıve form of realism and what I am calling direct realism.8 David Chalmers (2004) notes that if one shifts attention back and forth between two pinpoint red lights, one’s phenomenology changes. He mentions however, that the change in phenomenology may be treatable as a representational change in which one’s experience represents with varying specificity or else represents salience.9 (I will be discussing salience later. For now, I will just mention that the direct realist may prefer not to allow that we are directly aware of salience, since salience is mental.) The world certainly contains properties of varying levels of abstractness, specificity or determinacy—for example, red in addition to vermilion and crimson– and changing attention can be a matter of changing which properties one is directly aware of. Alex Byrne (cited in Tye, 2005, 2006) has noted that a grid of 9 dots in a 3 by 3 matrix can be seen either as 3 rows or as 3 columns, and arguably these shifts are at least in part a matter of shifting attention. Michael Tye (2005, 2006) replies that the experiential change is a matter of change of which properties are represented in the experience. When we see the square/diamond as a square, we visually represent its symmetry about an axis that bisects two sides, whereas when we see it as a regular diamond, we represent its symmetry about an axis that connects two angles. In the case of Byrne’s matrix, we can represent 3 rows or 3 columns and that makes a phenomenal difference. (The difference may involve attention to one axis of symmetry rather than another.) Opinions differ as to the adequacy of these replies. (See Macpherson (2006); and see Nickel (2007) for a version of the Byrne example that may be less susceptible to Tye’s reply.) But I will assume30 NedBlock for the sake of argument that selection of the sort discussed can save direct realism (and representationism) from these problems. Both realism and representationism have an “anti-mentalistic” feel to them, and it is important to get a bit clearer about it. We should distinguish between two questions a theory of conscious experience may wish to answer: (1) What is the difference—constitutively rather than causally—between consciously perceiving red and green? (2) What is the difference between conscious perception (say, as of green) and unconscious perception (as of green)? The direct realist and the representationist are opposed to any appeal to conscious mental properties in the answer to the first question but not the second. The answer to the first question for the direct realist is: the difference between red and green, that is, the difference between the two colors in the world; and for the representationist it is: the difference between representing red and representing green. Direct awareness and conscious personal level mental representation come into the 2nd question, not the first—for we can perceive (and represent) red and green and other properties unconsciously and sub-personally.10 The representationist view of the difference between perception of red and green is the same for both conscious and unconscious perception, namely the difference between representing red and representing green. Representationists differ among themselves in answering the 2nd question. Tye (1995) says that unconscious perceptual representations differ from conscious representations in that the unconscious ones are not appropriately poised for use by the cognitive system. Lycan (1996) combines this sort of functionalism with an appeal to higher order states.11 On both views, representation comes in to the answer to the first question but not the second. As mentioned, the direct realist answer to the second question is: direct awareness; that is, the difference between conscious and unconscious perception of a face is the presence/absence of direct awareness of the face. Aquestion may have occurred to the reader: what is the direct realist account of unconscious perception? Suppose the direct realist says unconscious perception of red is a matter of perceptual representation of red without direct awareness of it. But that answer rather undermines the opposition of direct realists to conscious representation. I have never seen a direct realist discussion of this issue, even a brief one. It is one more sign of the profound disconnect between direct realism and the science of perception (Burge, 2005), especially since most conscious perception also involves unconscious perception (Debner & Jacoby, 1994; Jacoby & Whitehouse, 1989). The upshot then is that opposition to mental paint centers on the first rather than the second question. What the mental paint view denies is that the difference between conscious perception of red and of green is just the difference between red and green or representing red and representing green. This point will figure in what follows because the direct realist response to the points I will be making about attention may be to appeal to attentiveAttention and Mental Paint 31 vs inattentive direct awareness or more generally to degrees of attentive direct awareness. Similarly, the representationist may want to appeal to degrees of attentive representation. What I will be arguing is that these moves are not true to the phenomena because the effect of attention on phenomenology is to change perceived contrast, perceived size, perceived hue saturation and so on for certain other perceived properties. That is, the difference between one degree of attentive direct awareness and another is experienced by the subject as a difference in specific phenomenological properties such as perceived size or perceived hue saturation. I will argue that this puts the direct realist in a box—the direct realist has to say either that the subject is experiencing an illusion (an option I claim to be able to close off) or that the real properties of objects that the subject is directly aware of are changing—another option that I claim to close off. This point may be thought to be fatal to direct realism while at the same time rescuing representationism, since the representationist can respond by postulating a representational content that corresponds to the change in perceived contrast, perceived size, perceived hue and so on. However representationists cannot postulate contents at will; rather the contents they appeal to must be grounded in normal veridical perception. And I will argue that the move just canvassed fails that test. But before I get to these points, I will mention an attentional phenomenon that does not require mental paint. I regard selection as what happens when because of the joint effect of amplifying some representions and suppressing others, some things that could be seen are not seen. Other attentional phenomena are non-selectional.