A review of Naturalizing the Mind, by Fred Dretske
No contemporary philosopher has tried harder to demystify the mind than Fred Dretske. But how to demystify it without eviscerating it? Can consciousness be explained? Many philosophers think that no matter how detailed and systematic our knowledge becomes of how the brain works and how it subserves mental functions, there will always remain an “explanatory gap.” Call it a brute fact or call it a mystery, trying to explain consciousness, they think, is as futile as trying to explain why there is something rather than nothing. Dretske is not exercised by the explanatory gap-he’d rather exorcise it. He thinks we can get all the explanation we need by understanding what consciousness does. Consciousness is at bottom sensory experience and what it does, essentially, is to represent the world. Explaining consciousness, therefore, comes down to understanding the representational character of experience.
Dretske contends that “all mental facts are representational facts” (xiii) and that experiences do not have phenomenal features of their own. Rather, “the subjective quality of an experience, the phenomenal appearances, is the way the experience represents things to be” (22). His “Representational Thesis” is thus reminiscent of Brentano’s thesis that intentionality is the mark of the mental. Indeed, it echoes Moore’s famous remark on the diaphanousness of sensation, that to have a sensation is already to be outside the circle of one’s sensations. Dretske’s “phenomenal externalism” is externalist on two counts: the properties of a perceptual experience, insofar as it is mental, are determined entirely by the properties it represents things as having, and which properties it represents things as having is not determined by the perceiver’s physical state.
Dretske goes on to argue that introspective knowledge, at least of perceptual experience, is not a matter of looking inward, at distinctively experiential properties, but of conceptualizing properties already being experienced. Moreover, to explain perception and introspection just is to explain consciousness itself. He grants that a comprehensive account would have to include experiences of other sorts than those he considers, bodily sensation, emotion, and imagination, for example, but he is confident that perception is the central case and that the representationalist strategy should work for the rest. For Dretske engineering takes precedence over phenomenology: to understand the nature of mind is to understand what it does and how it works.
Representation and Function
Before describing what it’s like to be a representional system, Dretske explains what it is to be one. He sketches an externalist, teleological picture of perceptual representation, on which representation is more than indication-a system that merely indicates cannot be said to “misindicate,” but there is no representation without the possibility of misrepresentation. He makes the very strong claim that systems and their states have the power to represent only if they have the function of doing so, and that this requires being designed to do so. Representational systems have the function, by design, of indicating values of some range of properties. Dretske explains how this is so by means of his old favorites, gauges and instruments. Unfortunately, he offers little detail on how this account is supposed to extend to perceptual systems in people and animals. His biological examples generally illustrate sensitivity to just one property-a certain temperature, a certain type of animal-not to a range of properties. In explaining functions he often appeals to natural selection, but it is unclear how much his biological examples show. Even if natural selection explains why organisms having a system with a given function win out over rivals lacking it, natural selection would not seem to explain why the system has that function.
In any case, there are two serious disanalogies between Dretske’s illustrative examples of gauges and bona fide perceptual systems. First, whereas we gain information from pressure gauges and speedometers by perceiving (reading) them, we do not gain information from our senses by perceiving them. Dretske is of course aware of this disanalogy, but he does not fully come to grips with it. Also, whereas gauges and instruments indicate values of one specific range of properties, typically in a scale of one dimension, the senses are sensitive to many ranges of properties and these are not represented as points on a scale. When we compare a pair of colors, timbres, or flavors, we do not experience their differences solely as differences in magnitude. It may be that the processes underlying each sense are sensitive to variations in some small number of dimensions, but that is not how we experience things.
Dretske claims that experiences are natural representational states “whose function it is to supply information to a cognitive system for calibration and use in the control and regulation of behavior” (19). To the detriment of the phenomenal externalism he defends later, he never spells out in detail what sorts of properties this information concerns. Do they include relational properties, such as being slippery, prickly, or poisonous? What about secondary and tertiary qualities, which are essentially connected to how they are experienced? If the functions of our senses fix the sorts of properties we can perceive, what sorts of properties are excluded? Dretske claims not only that “experiences have their representational content fixed by the functions of the sensory systems of which they are states,” but that “the quality of a sensory state-how things look, sound, and feel at the most basic (phenomenal) level-is thus determined phylogenetically” (15). This makes it puzzling how we can develop our sensibilities and thereby acquire subtle powers of discrimination and complex recognitional abilities, as in learning to appreciate tastes of wines, smells of perfumes, and sounds of music.
Dretske relies here on a sharp but not entirely clear dichotomy between perception and conception. If representational properties at the “basic (phenomenal) level” are “systemic,” not “acquired,” but “the function of experience is to help in the identification and recognition of objects” (121), not to mention recurrent situations, then such identification requires acquiring concepts of those objects and situations. That may be so, but perhaps these recognitional concepts are experiential in character and modality-specific, maybe even task-specific in some cases. This would collapse Dretske’s dichotomy, which precludes perceptual learning and counts as conceptual the seemingly perceptual skills developed in the course of learning such things as gymnastics, French horn playing, or Grand Prix driving. If the “basic (phenomenal) level” really is “determined phylogenetically,” Dretske needs to explain how biologically-based perceptual capacities get extended beyond what they are selected for (are they somehow had for free?). Otherwise, he will have discovered a remarkable new source of error about perceptual experience: one could be utterly mistaken about the character of one’s experience simply because one’s perceptual apparatus does not have the function of representing what one thinks it represents.
Dretske uses the Representational Thesis to explain introspective knowledge of perceptual experience. Introspection is not, as the word suggests, a matter of looking inward. There are no “internal scanners monitoring the clockwork of the mind,” for “what one comes to know by introspection are … representational facts” (40). Introspection is not inner perception, as Locke and many others have thought, but a special case of what Dretske calls “displaced perception.”
Displaced perception is perception of a fact about one thing by way of perceiving something else. If you stand on a bathroom scale and look at the dial, you see that your weight is so many pounds. In so doing you see the dial, not yourself. Similarly, by hearing a barking dog you can hear that the postman has arrived. You can do this even though you don’t hear the postman-you have a “conceptual but no corresponding sensory representation” of him (41). Thus displaced perception “enlarges the number of facts one perceives without a corresponding enlargement in the number of objects one perceives” (42). But you must know the relevant correlations. Without this background knowledge you wouldn’t see that your weight is such-and-such by looking at the dial or hear that the postman has arrived by hearing the dog bark. Moreover, this background knowledge must apply in the given case. The scale must correctly indicate your weight and the dog’s barking must correctly indicate the presence of the postman.
Now if introspective knowledge were just like this, it would require knowledge and correct application of the relevant correlation. But unlike displaced perception in general, introspection does not require independent knowledge of the relevant correlation. This is not necessary because you “occupy the [very] state whose respresentational content is under investigation” (52). No prior “calibrational procedure” is required for you to know what representational state you are in, because how it represents things as being is the very information it carries about itself. In this way “self-knowledge [is] an instance of displaced perception-a process whereby a system gets information about itself … by perceiving not itself, but something else” (53).
Unlike other kinds of displaced perception, not only does introspection not require independent knowledge of the relevant correlation, it is also immediate. In contrast, seeing what one’s weight is by looking at the scale and hearing that the postman has arrived by hearing the dog bark are indirect. So, Dretske asks, how could introspection “have the immediacy we know it to have” and still be a kind of displaced perception? Here he identifes “two important differences between introspective knowledge and other forms of displaced perception” (60): the intermediate fact need not be represented veridically and the connecting belief is not defeasible. Whereas my perception of the dog barking might be an illusion and whereas I would merely seem to hear that the postman has arrived if he has not, in the case of introspection “the postman cannot fail to be there when this dog barks” (61). So “if this is inferential knowledge, it is a very unusual form of inference. The premises need not be true and the inference cannot fail” (61-62).
This is a big “if” with which to explain the immediacy of introspective knowledge. Inference which cannot fail even if its premises are untrue is indeed a “very unusual form of inference,” so unusual, in fact, that one might doubt that it is inference at all. It sounds like some kind of unmediated cognitive transition, in which case it is not really a form of displaced perception but something else instead. After all, whereas one hears that the postman has arrived by hearing the dog bark, one doesn’t hear (or see) that one is hearing the dog bark. In no way does one perceive that one is hearing the dog bark- one merely notices or realizes it. Dretske acknowledges that “there is no need for an experience of the experience. All one needs is a belief about it” (63). Then why regard this belief as a case of perception at all? If it were a case of displaced perception, it would be in a particular sense modality, but it is not.
Dretske need not rely on the model of displaced perception to formulate a conception of introspection of experience that is consistent with his view that the character of an experience consists in the properties it represents its objects as having. He could say simply this: one is aware of an experience by way of being aware of what its object are experienced as-and that they are being experienced as such.
Given his thesis that perception is representational and that introspective knowledge about perceptual states is a kind of displaced perception, Dretske thinks he has explained whatever needs to be explained about consciousness, at least insofar as a philosopher can explain it. “Consciousness is the genus; seeing, hearing, and smelling are species” (99). Perception is awareness, awareness is consciousness, and that is that (well not quite, but in this book Dretske is not concerned with bodily sensations, emotions, fantasy, etc). He thinks that the phrase ‘conscious awareness’ is redundant (98), hence that there is no need to explain, as higher-order theories try to do, why some states of awareness are conscious and others are not. A state of awareness is a state of consciousness. Dretske complains that higher-order theories characterize the difference between conscious and unconscious states as extrinsic to the state itself. They say that an experience is conscious just in case one has an experience of it or, alternatively, one (noninferentially) thinks that one is having the experience. Higher-order theorists fail to appreciate that experiences are conscious “but not because we are conscious of them” (103). A state of awareness, just by being the sort of state it is, is conscious even if one is not conscious of being in it.
I think Dretske is not talking about what higher-order theorists are talking about. Why else would he claim that the phrase ‘conscious awareness’ is redundant or, by implication, that ‘unconscious awareness’ is an oxymoron? By his lights, even dream states are conscious. Clearly Dretske is conflating states of consciousness with conscious states. In discussing the classic case of the drowsy or distracted driver who motors for miles allegedly without being consciously aware of the wheel, the traffic, or the road, Dretske claims that all “the driver lacks is introspective awareness” (105). Questioning the view that the thoughts and experiences of an animal or an infant are not conscious, he asks, “what is the point of insisting that because they know less about their experiences, their experiences are different?” (111). His point is that there is no inherent difference between conscious and allegedly non- or unconscious experiences.
The confusing multiplicity of uses of the terms ‘conscious’ and ‘consciousness’ may explain why Dretske and the higher-order theorists are talking at cross purposes. We describe sentient beings as conscious of things or simply as conscious. We speak of conscious states and states of consciousness. Sentient beings have to be conscious to be in conscious states but, crucially, they do not have to be conscious to be in states of consciousness-they can be asleep and dreaming. So not all states of consciousness are conscious states. This is not paradoxical. Conscious states are not conscious in the way sentient beings are conscious. Conscious states are no more conscious than a healthy diet is healthy or a happy face is happy. Dretske is right to claim that all perceptions of things are states of being conscious of them, in the sense of being aware of them, but wrong to deny that there is a sense in which some of these states are not conscious. They can fail to be conscious in the same way that intentions and motives can fail to be conscious. They can fail to be immediately accessible to awareness.
Qualia and Phenomenal Externalism
The most daring of Dretske’s claims is his “phenomenal externalism”: the phenomenal contents of perceptual experience do not supervene on the physical properties of the experiencer. Phenomenal externalism is supposed to follow from the Representational Thesis, but just how, and with the help of what other premises, is never made clear. Dretske seems to think that the qualitative character of experience loses its mystery if the properties of experience are, to put it crudely, “out there” rather “in here.” As he explains, “the Representational Thesis … identifies mental facts with representational facts, and though representations are in the head, the facts that make them representations-and therefore the facts that make them mental-are outside the head” (124). As a result, “although one has privileged information about the character of one’s experience, one does not look inward to get it” (149). Thus we can reject the “Internalist Intuition” that the character of an experience depends entirely on the physical state of the person having the experience. Once we recognize that the character of experience consists in the properties it represents external objects as having and get clear on the whereabouts of these properties, we can embrace phenomenal externalism. We can learn to stop worrying about the explanatory gap. Unfortunately, there are some obstacles in the way of this lesson.
1. The internal red herring. Dretske suggests that the Internalist Intuition “may be traceable to an implicit commitment to an act-object view of sense experience, a view of experience that regards different experiences as experiences of different internal objects” (129). His diagnosis is, in effect, that fans of qualia are unwitting victims of the invidious old sense-data thesis, which in its simplest form says that to experience an apparent F is to experience an F appearance. Otherwise, why would they think that we have direct access to the qualities of experience themselves? In Dretske’s view, “the access one has to the quality of one’s experiences (unlike the access one has to the qualities of the external objects the experience is an experience of) is only through the concepts one has for having thoughts about experience” (134). We do not experience the intrinsic qualities of experiences themselves, for “experiential qualities are not, as it were, on display in the shop window of the mind” (139). Whereas concepts are required for being aware of the qualities of one’s experience, concepts are not required for being aware of the qualities of their objects. Having the experience is what makes one aware of those qualities.
Granting all this, I wonder if Dretske’s diagnosis of qualiaphilia isn’t based on a false dilemma. He is right to argue that qualia, considered as intrinsic properties of experience as opposed to properties objects are experienced as having, are not epistemically accessible in the way some qualiaphiles allege. Yes, we are not aware of our experiences in the way we are aware of external objects-we do not experience experiences as we do tables and chairs. But why can’t we construe qualia as intrinsic properties of experience and also identify them as experienced properties of objects? Isn’t that just to acknowledge the intentionality of experience? If phenomenal externalism is correct, however, these properties are not intrinsic to experience.
2. Qualitatively identical twins. Dretske does not so much argue for phenomenal externalism as rely on thought experiments to show that if you are a conceptual externalist, you have no reason not to be a phenomenal externalist too. One such experiment involves Twin Earth, where, as you may have heard, the common, clear, thirst-quenching liquid is not water but twater. The question is, does twater look to Twin Fred as water looks to Fred? Dretske thinks that whereas water looks like water to Fred, twater looks like twater to Twin Fred, and that looking like twater is different from looking like water. He thinks this because, assuming concept externalism, there is a difference between what Fred and Twin Fred think, respectively, about what water and twater look like-and each is right about what that is. Hence Fred and Twin Fred’s experiences are phenomenally different. This contradicts the internalist intuition that looking like twater is the same as looking like water. In that case, however, either Fred or Twin Fred would be wrong about the qualitative character of their experiences. There is no reason to think that one is wrong and the other is not, for by symmetry both are wrong if either is. But if both are wrong, internalists would score but a Pyrrhic victory, for the qualitative character of experience would not have the sort of direct access to consciousness that they suppose it to have.
The natural objection to Dretske’s argument is that even though Fred and Twin Fred are both right about the looks of water or twater, looking like water is the same as looking like twater. Neither Fred nor Twin Fred can tell the difference between the two, certainly not by just looking, and Dretske never gives us any idea what the phenomenal difference might consist in. So why can’t Fred be right to think that water looks like water and Twin Fred be right that twater looks twater, but without there being any difference between how their respective substances look to them? The mere fact (given conceptual externalism) that Fred and Twin Fred have different concepts does not entail any phenomenal difference between their experiences. If that were so, then presumably this difference would persist even if each acquired the other’s concept in addition to his own. Suppose they had read their Putnam or Twin Putnam and then traded places. Water and twater would look to each as they would to the other. Water would look to Fred as it always had and twater would look to Twin Fred as it always had-clear and liquidy-and the fact that one is water and the other twater would make no difference to how they look. Water and twater look the same in character. Fred and Twin Fred’s experiences are the same in phenomenal character even if their beliefs are different in conceptual content.
3. The alternative modality problem. Phenomenal externalism rules out the following possibility. It seems that two creatures could have different sense modalities that are responsive in the same degrees to the same properties but that the experiences they have in their respective modalities are radically different.
Dretske’s response to this, the alternative modality problem, might be predicted from his stance on the inverted spectrum problem. That, he thinks, is a problem for functionalists and behaviorists but not for him: “In identifying qualia with experienced properties, experienced properties with properties represented, and the latter with those properties the senses have the natural function of providing information about, a representational approach to experience makes qualia as objectively determinable as the biological function of bodily organs” (72). But the alternative modality problem is that there may be radically different ways of gaining sensory information about the same properties. Because “the Representational Thesis identifies the qualities of experience-qualia-with the properties objects are represented as having” (65), it rules out the possibility that exactly the same properties could be experienced in subjectively different ways. It seems that the only sort of objective properties for which this possibility could be ruled out are relational ones, namely dispositions to cause (in experiencers like oneself) experiences of a certain sort. But these are properties of just the sort that undermines the externalist import of the Representational Thesis.
Dretske considers something like this objection when he writes, in regard to whether a blind person knows what it is like to visually experience motion, “there is more-much more-involved in seeing an object move than experiencing the object’s movement. One also experiences the object’s shape, size, color, direction of movement, and a host of other properties. This is why seeing and feeling movement are much different even though the same thing (movement) is represented in both modalities” (95). Even so, he excludes a priori the possibility that two different creatures could each have a modality that is sensitive to exactly the same range of exactly the same properties and yet experience them in different ways.
4. Objects and phenomenal properties. The most obvious objection to phenomenal externalism is that there are some phenomenal properties that really are attributable to experiences themselves. I am not referring to properties that objects are experienced as having but do not really have, on account of perceptual error or illusion (or even outright hallucination), though it would have been helpful if Dretske had explained the status of such properties. I mean, rather, properties that objects are not even experienced as having. For example, visual experiences can become blurry, as when one removes one’s glasses, without their objects appearing to have become fuzzy. Their objects look different, of course, but do not look to have changed. Also, qualitative character is not limited to the properties things appear to have, for people can and do have different quality spaces even if things appear to them to have the same properties. That is, they can differ in their judgments of relative similarity even if they agree on the properties they judge things to have. Two people might both judge a to look F, b to look G, and c to look H, and yet disagree on whether b looks more like a in this respect than like c.
Still, suppose we grant that the qualitative character of experience consists in the properties objects are experienced as having. We can agree with Dretske that these properties are not to be found by looking inward-experience is intentional-but the issue does not concern their whereabouts. It concerns their ontological status. If experiences are distinguished by which properties their objects are represented as having, the question arises of just what sorts of properties these are. For Dretske’s Representational Thesis to lead to phenomenal externalism, these properties must be objective and nonrelational. They must not be experience-dependent or experiencer-relative, like smells and tastes, let alone aesthetic properties, e.g., of sunsets, bolts of lightning, or leaves fluttering in the wind. This is not the place to take up the question of which properties belong to which category, but there do seem to be properties of ontologically different sorts which, nevertheless, are all experienced as properties of objects. If you experience the sun as round, yellow, and painfully bright (or the sound coming from next door as metallic, loud, and raucous), their difference in ontological type makes no phenomenological difference. You experience them jointly, as properties of the object of the experience. Experience is an equal-opportunity representer.
Intentionality cuts both ways. If some of the properties objects are represented as having are experience-dependent or at least experiencer-relative, there is a certain sense in which they are not where Dretske says they are. They are not “out there.” But they are not “in here” either. Even if some philosophers have mislocated qualia, the mystery of qualia cannot be dispelled just by putting them in their place. That perceptual properties are properties objects are experienced as having does not remove the mystery (if there is one) about the qualitative character of experience but merely identifies what the mystery is about. The qualitative character of experience consists primarily in the qualitative character of objects as experienced. That’s what it means for experience to be intentional (in Brentano’s sense) and diaphanous (in Moore’s sense). Dretske frames the debate about qualia as concerning their location, but qualia are properties, not particulars. They are neither “out there” nor “in the head.” His claim that the Representational Thesis explains “why conscious experiences have that peculiar diaphanous quality-the quality of always being present when, but never where, one looks to find them” (xiii) is highly misleading. It suggests that they are somewhere or other. They aren’t anywhere. Only the objects that have them-or merely appear to them-have locations.
KENT BACH, San Francisco State University