12
Naive Realism and Hallucinations
All visual experiences present,1 or at least purport to present,2 mind-independent objects and their features and so seem to relate us to those objects and their features. Suppose, for example, that you are looking at a bowl of fruit on the table in front of you. You can visually attend to those mind-independent objects—the fruit and the bowl—and note their features: their color, their shape, the way they are illuminated, and so on. You can also introspectively reflect on the visual experience that you have while looking at the bowl of fruit. In doing so, you might notice features of your experience—that only parts of some of the fruit are visible from this perspective, for example, or that from this close the three-dimensional shape of the fruit is easily visible. Although in introspection you are interested in the character of your experience of the bowl of fruit, your attention is still on the objects of your experience: the fruit and the bowl. Since attending to your experience involves attending to the mind-independent objects of your experience, your experience seems such that were the bowl of fruit not actually there, you could not be having this particular experience. In general, when we introspect a visual experience, our experience seems in this way to involve the mind-independent objects or features that the experience presents, and so our experience seems to be a relational; it seems to have mind-independent objects and features as constituents.3 We can call the property of having some mind-independent object or feature as a constituent the naive realist property of experiences. It is fairly widely accepted that visual experiences seem to have the naive realist property;4 naive realism is the view that some experiences—the veridical ones—actually do have it: it’s the view that veridical perceptual experiences have mind-independent objects and features as constituents.5
In a plausible conception of phenomenal character, the phenomenal character of a perceptual experience is those properties of the experience that explain the way it introspectively seems. We characterize the way an experience seems at least partly in terms of the objects and features it seems to involve. Naive realism is therefore the view that veridical perceptual experiences have a phenomenal character that consists of relations to mind-independent objects and features; it is the view that perceptual experiences have, as I shall say, a naive realist phenomenal character.
We have two reasons for thinking that naive realism about visual experiences is true, both of which appeal to the naive realist phenomenal character of experiences. The first is straightforwardly phenomenological: the naive realist claims that the best explanation of the fact that visual experiences introspectively seem to have the naive realist property is that veridical experiences actually do have it: their having a naive realist property explains why they seem to have it. Those who reject naive realism claim that although experiences introspectively seem to have the naive realist property, their seeming that way is not explained by the fact that they actually do have it. Representationalists about perceptual experience, for example, explain the fact that experiences seem to involve mind-independent objects by appeal to the fact that experiences have contents that represent mind-independent objects. According to such views, visual experiences have a phenomenal character that does not consist of relations to mind-independent objects and features; they have what I call a non-naive realist phenomenal character.
The second reason is epistemological but also appeals to the distinctive phenomenal character of veridical experiences. Visual experiences have epistemic authority over a subject’s beliefs in the following sense: in the absence of any countervailing or defeating reasons, it is rational to believe that things are as they are visually presented. Visual experience is therefore “nonneutral” about the presence and character of the objects of experience in a way that visual imagination, for example, is not. Furthermore, any experience that is phenomenologically indistinguishable from a veridical visual experience will have the same epistemic authority as the veridical experience, so there is a connection between phenomenology of experiences and their epistemic authority. The naive realist claims that this connection can best be explained on the assumption that naive realism is true. According to naive realism, a visual experience seems to present mind-independent objects and their features. That it seems to the subject that there are objects and features present gives the subject an authoritative reason to judge that there are. When the experience is veridical, the fact that it seems to the subject that there are objects present is explained by the fact that there actually are the objects present, so the reason is a good reason.6
If naive realism is true, then it follows that the kind of experience you have when veridically perceiving a bowl of fruit could not occur in the absence of the fruit and the bowl that are its constituents, and so could not occur were you merely hallucinating the bowl of fruit rather than seeing it. In general, naive realism entails that the kind of experience you have when veridically perceiving could not occur were you having a hallucination. Naive realism therefore entails a rejection of the common-kind assumption, that is, the assumption that whatever kind of experiential episode occurs when you perceive, the same kind of episode can occur when you hallucinate. It is by virtue of rejecting the common-kind assumption that naive realism is a form of disjunctivism about visual experiences. Rejecting the common-kind assumption does not, however, entail the further claim that there is nothing in common between veridical perceptual experiences and nonveridical and hallucinatory experiences, and for reasons discussed later, naive realists must reject that further claim.7
But what positive account can the naive realist give of hallucinatory experiences? Saying that the kind of experience that you have when veridically perceiving cannot occur when hallucinating tells us what hallucinatory experiences are not, but it doesn’t tell us what they are. In particular it doesn’t explain the phenomenology of hallucinatory experiences. The naive realist needs to say something more, but what more can she say? It is possible to show, given some plausible assumptions, that whatever is true in general of hallucinatory experiences must also be true of veridical experiences. This limits the characterization that the naive realist can give of hallucinatory experiences in a way that some writers have found unacceptable. In what follows, I try to show that the naive realist’s characterization of hallucinatory experiences does not have the unacceptable consequences that it is sometimes thought to have; to do so, I need first to set out the reasons for the limitation. The argument begins with the naive realist’s response to a familiar argument for the common-kind assumption.
Many people find the common-kind assumption compelling because they accept some version of the causal argument.8 The causal argument appeals to the idea that the character of hallucinatory experiences supervenes on the state of the subject’s brain. Suppose that it is possible to keep fixed the exact state of your brain when you are looking at the bowl of fruit while at the same time removing the bowl and the fruit. Were this to happen, there would be no change that you could detect in the character of your visual experience even though you would no longer see the bowl of fruit: you would be having a hallucinatory experience brought about by a state of your brain that matches (or is of the same kind as) the state that is involved in your seeing the vase. There is therefore a state of your brain that is sufficient to determine your experience, and that state can be brought about in the normal way by the bowl of fruit reflecting light that stimulates your retinas, produces activity in your visual cortex, and so on; or it can be brought about in some other, abnormal, way that doesn’t involve the bowl of fruit. According to the causal argument, since your brain is in the same kind of state in both the veridical and the hallucinatory situations, your experience in both situations must be of the same kind too; and since your experience in the hallucinatory case doesn’t have the bowl and the fruit as constituents, your experience in the veridical case doesn’t either. Therefore naive realism is false.
This argument appeals to the idea that two events are of the same kind if they are produced by the same kind of cause (the “same-cause, same-effect” principle):
An event e1 is of the same kind (K) as an event e2 if event e1 is produced by the same kind of proximate causal condition as e2.
Such a general causal principle is false if there are kinds of events that have noncausal necessary conditions for their occurrence (Martin, 2004, 56–58; 2006, 368–371). The naive realist thinks there are such events, as do externalists and singular-thought theorists in the philosophy of mind. A necessary condition for having a water thought is that there be water in my environment; a necessary condition for thinking about the cat sitting on my sofa is that the cat exists; thoughts brought about by the same proximate causes but in an environment that doesn’t contain water or in which the cat doesn’t exist would be different thoughts. That suggests we need to modify the same-cause, same-effect principle to take into account such noncausal necessary conditions. The naive realist accepts such a modified version of the same-cause, same-effect principle:
An event e1 is of the same kind (K) as an event e2 if event e1 is produced by the same kind of proximate causal condition as e2 in circumstances that do not differ in any noncausal conditions necessary for the occurrence of an event of kind (K).
According to naive realism, there are noncausal conditions necessary for the occurrence of a perceptual experience—the bowl and fruit that are its constituents—that are absent when you hallucinate, so the kind of event that occurs when you veridically perceive the bowl of fruit cannot occur when you hallucinate a bowl of fruit; but there are no noncausal conditions for the occurrence of the hallucinatory experience that cannot also obtain when you veridically perceive the bowl of fruit, so the kind of event that occurs when you hallucinate can occur when you veridically perceive. Therefore, according to the modified same-cause, same-effect principle, whatever kind of experience is produced when you have a causally matching hallucination of a bowl of fruit, the same kind of experience will be produced when you veridically perceive the bowl of fruit.
The naive realist is therefore forced to accept the conclusion that “whatever the most specific kind of mental event that is produced when having a causally matching hallucination, that same kind of event occurs when having a veridical perception” (Martin, 2006, 369); but that falls short of establishing the common-kind assumption because “no instance of the most specific kind of mental event that occurs when having a veridical perception occurs when having a (causally matching) hallucination” (Martin, 2006, 361).
Since there is nothing inconsistent in the idea that a single event (a veridical perceptual episode) can be an instance of two different kinds—veridical and hallucinatory—that conclusion is not inconsistent with naive realism. It does, however, threaten to undermine the reasons for accepting naive realism in the first place. The problem comes from the role of experiences in explanation. When we have the following kind of asymmetric relation between properties:9
for all e, if e has p, then e has h, and for some e, e has h but lacks p, and some p-event e occurs and produces an effect,
we can ask: did e have that effect because e had p, or because e had h? When we answer that question, it seems plausible to appeal to the following screening-off principle:
An event e being h screens off being p as an explanation of some effect of e if, had e occurred and been h but not p, then e would still have had that effect.10
For example, a cloth sample being red screens off its being scarlet as an explanation of its being picked out by a machine if, had the sample been red but not scarlet (by being crimson, say), it would still have been picked out. In that case, it would be false to say that it was picked out because it was scarlet, since it would have been picked out whether or not it was scarlet (Martin, 2004, 62).
Given the screening-off principle, anything that we appeal to the occurrence of a veridical perceptual experience to explain and which could also be explained by a hallucinatory experience will, even in the veridical case, be explained by the fact that the experience is of the same kind as a hallucination and not by the fact that the experience is a veridical perception. There are effects of veridical perceptual experiences that cannot be explained by hallucinatory experiences (see Martin, 2004a, 64; Peacocke, 1993), but the effects that concern us are those that the naive realist claims are explained by the phenomenal character of veridical perceptual experiences.
When you introspect a (causally matching) hallucinatory experience of a bowl of fruit, the experience introspectively seems to involve the bowl of fruit—and so seems to have the naive realist property—in just the same way as the veridical experience. For the purposes of argument, we can assume that there is some property of the hallucinatory experience—its phenomenal character—that explains its seeming that way. Since hallucinatory experiences lack the naive realist property—and so have a non-naive realist phenomenal character—the hallucinatory experience’s seeming to have the naive realist property is not explained by its actually having the naive realist property.
According to the conclusion of the modified causal argument, your veridical experience of the bowl of fruit is an event of the same kind as the matching hallucinatory experience, so we can ask: What explains the fact that the veridical experience seems to have the naive realist property? Does the veridical experience seem to have the naive realist property because it is an experiential episode with a naive realist phenomenal character, or because it is an experiential episode with a non-naive realist phenomenal character? According to the screening-off principle, the non-naive realist phenomenal character screens off the naive realist phenomenal character in explaining the fact that the veridical experience seems to have the naive realist property, since if the experience had occurred and had the non-naive realist phenomenal character but lacked the naive realist phenomenal character, it would still have seemed to have the naive realist property. Therefore, having a naive realist phenomenal character doesn’t explain the fact that veridical perceptual experiences introspectively seem to have the naive realist property. But if having the naive realist property doesn’t explain the fact that veridical perceptual experiences seem to have the naive realist property, then the claim that naive realism provides the best explanation of that fact is false, and the argument for naive realism is undermined. We could construct a similar argument for the conclusion that the naive realist phenomenal character of experience doesn’t explain the epistemic authority of visual experiences. So although the conclusion of the modified causal argument is not directly inconsistent with naive realism, when combined with the plausible assumption that hallucinatory experiences have a non-naive realist phenomenal character, it undermines the two reasons for thinking that naive realism is true.
The naive realist response to this objection is to question the assumption that causally matching hallucinatory experiences have a non-naive realist phenomenal character that could be in explanatory competition with the naive realist phenomenal character of veridical experiences (Martin, 2004, 65–81; 2006). Since the phenomenal character of an experience comprises whatever of its properties explain the way it introspectively seems, the naive realist is questioning the assumption that there are any properties common to all hallucinatory experiences sufficient to explain the way they introspectively seem. If hallucinatory experiences have a phenomenal character—if, that is, they have properties sufficient to explain the way they introspectively seem—then naive realism is undermined. So the naive realist must claim that the only thing such hallucinatory experiences have in common is that they cannot introspectively be told apart from the corresponding veridical experience; that, according to the naive realist, is all that they have in common: the “only positive mental characteristics [of hallucinations brought about through matching causal conditions of veridical perceptions] are negative epistemic ones—that they cannot be told apart by the subject from veridical perception” (Martin, 2004, 73–74). In other words, “there is no more to the phenomenal character [of a causally matching hallucination] than that of being [introspectively] indiscriminable from a corresponding visual perception” (Martin, 2006, 369).
What does it mean to say that a hallucination is indiscriminable through introspection from a veridical perception? If two things are discriminable, it is possible to know that they are not identical, so two things are indiscriminable if it is not possible to know that they are not identical: experience1 is indiscriminable from experience2 by a subject S at a time t if and only if S cannot know at t by introspection alone that experience1 is not experience2.11 The naive realist claims, therefore, that there is nothing more to the phenomenal character of a hallucinatory experience than having the “indiscriminability property” of being not introspectively knowably not a veridical perception (Martin, 2004, 76–81).12 This has become known as the “epistemic” account of the nature of hallucinatory experiences.
That an experiential episode has the property of being indiscriminable from a veridical perception is, according to the naive realist, sufficient to account for the explanatory role of hallucinatory experiences.13 Someone hallucinating a piece of fruit, for example, might reach out to pick it up. Why did she reach out? Because she wanted a piece of fruit and was having an experience indiscriminable from a veridical perception of a piece of fruit. That her experience is indiscriminable from a veridical perception means that it seems to her that a piece of fruit is there, and therefore she reaches out in the same way she would were she veridically perceiving a piece of fruit.14 The same kind of explanation can be given of anything that we would normally appeal to a hallucinatory experience to explain.
One might object that an account of hallucinatory experiences needs to explain how they can purport to present objects without anything actually being presented—the account needs to say what is it about those experiences that explains the fact that it seems to the subject of the experience that something is there. Some writers doubt that any such explanation is available to the naive realist. A. D. Smith, for example, expresses such a doubt in the following passage:
When . . . we turn to hallucination . . . to say that the subject is not aware of anything is surely to under-describe this situation dramatically. Perhaps we can make sense of there being “mock thoughts,” but can there really be such a thing as mock sensory awareness? . . . We need to be able to account for the perceptual attention that may be present in hallucination. A hallucinating subject may, for example, be mentally focusing on one element in a hallucinated scene, and then another, describing in minute detail what he is aware of. In what sense is all this merely “mock”? . . . The sensory features of the situation need to be accounted for. How can this be done if such subjects are denied an object of awareness? (2002, 224–225)
One way to understand this complaint is as an expression of the sense-datum theorist’s claim that an experience cannot purport to present something as being some way unless there is actually something that way that is presented by the experience.15 But both naive realist and representational accounts of experience reject that idea. What, then, are the “sensory features of the situation” that an account of hallucinatory experience needs to explain?
If a sensory episode is an episode that modifies our consciousness in some way—an episode that contributes to our stream of consciousness—then the objection might be that the epistemic account of hallucinations cannot explain what makes a hallucinatory experience a sensory episode. It cannot explain why there is something that it is like to experience a hallucination. This objection is persuasive only on the assumption that the epistemic account of hallucinations is intended—or is required—to explain the fact that hallucinatory experiences have a sensory character, to explain why hallucinations are sensory rather than nonsensory episodes. If such an explanation is required, then the objection is right that the epistemic account fails to provide it. But why does the naive realist have to accept that assumption?
To see how it might be rejected, we need to distinguish the task of explaining how it is that a mental episode has a sensory character at all, from that of explaining the particular sensory character that it has. Our stream of consciousness is made up of many different kinds of sensory episodes—perceptual experiences, emotions, sensations and bodily experiences, episodes of imagination, sensory episodes associated with various kinds of thinking, and so on. Even if we take it as an unexplained fact that these various different conscious sensory episodes occur, we can still ask how the different kinds of episodes should be characterized and what it is that explains the distinctive sensory features that they have. A satisfactory account of a particular kind of sensory episode will typically tell us what features are characteristic of the kind of episode in question, and it will explain in virtue of what episodes have those features. In doing so, it will tell us what makes a sensory episode an episode of that kind.
An account of pain experiences, for example, will typically characterize the features distinctive of pains: their apparent mind dependence, how they differ from itches, that they are unpleasant, lead to certain kinds of behavior, and so on. It will then attempt to explain how it is that sensory episodes have these distinctive features; what, for example, explains the unpleasant character of pains, what explains their mind dependence? Answers to these questions can take different forms and may appeal to representational properties or intrinsic properties or naive realist properties of the episodes in question (some accounts, for instance, claim that the features of pains are a consequence of the distinctive objects of pain experiences, others that they are properties of the experiences themselves, and so on). Suppose that we give a satisfactory account of pain experiences in these terms. Such an account will tell us what pains are, but it is not obvious that we will thereby have given an explanation of that in virtue of which pain experiences are sensory episodes, rather than an account of that in virtue of which pain experiences are a particular kind of sensory episode.
If that is right, an account of hallucinatory experiences doesn’t need to explain what in general makes a hallucinatory experience sensory; the naive realist account of hallucinations can take their sensory character for granted. What it must explain is what makes hallucinatory experiences the kind of sensory episodes they are. It must explain the features that are distinctive or characteristic of the sensory episodes that are visual hallucinations.
It is worth noting that non-naive realist accounts of hallucinatory experiences—representational accounts and accounts that appeal to sense-data—provide no more explanation of what it is for a sensory episode in general to occur than the naive realist does. Sense-datum theorists typically appeal to the idea that we apprehend or are aware of sense-data without further explaining what makes the relation of apprehension or awareness sensory. Representationalists typically appeal to the idea that we stand in a sensory or perceptual relation to a representational content without explaining what makes standing in that relation to that content sensory.16 Anyone who finds the naive realist’s account of the sensory character of hallucinatory experiences unsatisfactory will therefore find the alternatives equally unsatisfactory. That means that the dispute between the naive realist and non-naive realist accounts of hallucination cannot, or at least should not, be about accounting for the sensory features of the situation if this is understood as requiring an explanation of sensory episodes in general.
What is distinctive of the sensory character of hallucinatory experiences is that they seem relational but aren’t. When I hallucinate a red cube, my experience seems to present me with a red cube even when there is no red cube to present. It is this seeming relational character that an account of hallucinatory experiences needs to explain. It needs to explain why sensory episodes that don’t intrinsically involve mind-independent objects nonetheless seem to involve mind-independent objects.
One way to explain this feature of hallucinatory experiences would be to appeal to something they have in common with perceptual states. That is, we explain how hallucinatory experiences seem to involve the world by appeal to the same properties to which we appeal to explain how veridical perceptual experiences seem to involve the world. We can view representational accounts of experience as attempting to explain the seeming relational feature of hallucinatory experiences in this way. The representationalists’ explanation is the same as the explanation they give of the seeming relational character of veridical perceptual experiences. In both cases, an experience seems to present an object in virtue of the fact that it has representational content that represents an object as being some way. Since, according to the representationalist, veridical experiences seem relational without actually being so, this explanation of the seeming relational character of hallucinatory experiences is bought at the cost—from the naive realist’s point of view—of an inadequate account of the relational character of veridical experience.
The naive realist rejects the idea that the same explanation can be given of the seeming relational character of veridical and hallucinatory experiences. In the case of veridical experiences, the naive realist explains the fact that experience seems relational by appealing to the fact that it actually is relational. The same explanation cannot be given in the case of hallucinatory experiences. So why do we characterize a nonrelational sensory episode as seeming to present a mind-independent object? The naive realist’s answer is that we characterize it that way because it introspectively seems the same way a genuinely relational state seems. For example, if a sensory episode is such as to be indiscriminable from an experience that actually presents a red cube, then it must also seem to present a red cube even if it doesn’t. Although the sensory episode introspectively seems the same as a veridical experience, it is not the same; introspection misleads us about the phenomenal character of hallucinatory experiences. Note that this explanation of the seeming relational character of hallucinations only succeeds because veridical experiences are genuinely relational states. So the explanation is derivative—it depends on the naive realist account of veridical perceptual experiences. If not for the fact that we can explain the seeming relational character of veridical experiences by appeal to the fact that they are actually relational, we would not be able to explain the seeming relationality of hallucinatory experiences in terms of their introspective similarity to veridical experiences. That veridical experiences are actually relational therefore explains both why veridical experiences seem relational (they actually are so) and why hallucinatory experiences seem relational (they are indiscriminable from episodes that actually are so).
The epistemic account explains why two sensory episodes that are fundamentally different in kind—one of which constitutively involves mind-independent objects and features, and one that doesn’t—should nonetheless be characterized in the same way. But it also explains why sensory episodes whose intrinsic nature does not involve objects nonetheless seem to be directed on objects. They do so because they are introspectively indiscriminable from sensory episodes that are constitutively directed on objects. We characterize episodes that are not intrinsically world involving in world-involving ways as a result of the limits of introspection. There is no more general or deeper explanation of why hallucinatory experiences seem to present objects than that. There is nothing substantive in common to all hallucinatory experiences (there couldn’t be without undermining the argument for naive realism). All that they have in common is that, for someone enjoying such a sensory episode, it isn’t possible for them to tell introspectively that they are not undergoing a different kind of sensory episode—a perceptual experience.
If what I have been arguing is correct, then appealing to their indiscriminability from veridical perceptions is sufficient to explain why hallucinatory experiences have the features they do, in particular their seeming relational character. The naive realist account of hallucinations therefore does give an account of these “sensory features of the situation.”
One might object that this still leaves something out. Some sensory episodes—hallucinatory experiences—are such that, for someone who undergoes them, there seem to be objects present when there are none. The naive realist gives no explanation of this fact, and it is this that objectors have in mind when they say that the epistemic account of hallucinations doesn’t account for the sensory features of the situation. According to the epistemic account of hallucination, having an experience that has the indiscriminability property is sufficient for you to be having an experience that purportedly presents objects and their features. But how does an experience’s having the indiscriminability property explain that? It might seem that the explanation must go in the other direction: a hallucinatory experience’s having the indiscriminability property is explained by the fact that the experience purports to present objects and their features—explained, that is, by the sensory features of the situation—and it is the fact that, in having a hallucination, it seems to the subject that some object is there that explains the consequences, for action and judgment, of a hallucinatory experience.
The representational account of experience, in contrast, can claim to offer an explanation of the fact that hallucinatory experiences purport to present objects; namely, that hallucinatory experiences have the same representational properties that explain why veridical experiences purport to present objects. This kind of explanation is not available to the naive realist: the relational properties of experience that explain why veridical experiences present objects are not instantiated in the hallucinatory case. So here we have a substantial difference between the naive realist and representational accounts of experience. Both accept that there occur sensory episodes—hallucinatory experiences—that seem to be relational when in fact they aren’t; and both offer explanations of why these sensory episodes introspectively seem that way; but in addition, the representationalist offers an explanation of why it seems to someone who undergoes a hallucinatory experience that objects are present when in fact there are none.
How significant is this difference? The naive realist account of experience doesn’t rule out the possibility of giving “local,” that is, less than fully general, explanations of why it seems to someone who undergoes a particular sensory episode that there are objects present when in fact there are none. Such “local” explanations might include psychological explanations that appeal to the processes involved in vision, for example, or to states of the brain that play a role in perception. What the naive realist denies, however, is that there can be any explanation common to all sensory episodes that are hallucinatory experiences of why they purportedly present objects, in particular no explanation that appeals to the mental properties of the sensory episodes in question. So the naive realist can account for the sensory features of any particular hallucinatory situation, but not of hallucinations in general.
The dispute between the representational and naive realist accounts of hallucinatory experiences therefore turns not on whether they can account for the sensory features of hallucinatory experiences but on the generality of the explanations provided. The representationalist can offer a general account of hallucinatory experiences, at the cost of giving up a relational account of veridical perception; the naive realist gives a relational account of veridical perception at the cost of a less than fully general account of hallucinatory experiences.17
In the previous section, I argued that the epistemic account of hallucinatory experiences is adequate to explain the features characteristic of those experiences—that appealing to indiscriminability is sufficient to explain the fact that hallucinatory experiences seem to present objects. Here I argue that the naive realist can appeal to the fact that hallucinatory experiences purport to present mind-independent objects to answer objections that question the idea that two experiences can be indiscriminable without having some properties in common in virtue of which they are indiscriminable. Remember that the conclusion of the modified causal argument requires the naive realist to deny that hallucinatory experiences have a common phenomenal character; all that they have in common is that they are indiscriminable from veridical perceptual experiences. Now consider the following objections.
Dogs lack the cognitive abilities necessary for having introspective knowledge of their experiences, so there will be no perceptual experience that a dog can know to be distinct from a veridical perception. That means that all the experiences of a dog will be indiscriminable from each other, and hence each experience the dog has is of the same kind as any other experience it has. Examples of this kind can be generalized to other cognitively unsophisticated subjects, and to subjects who are for some reason—excessive alcohol or drugs, say—incapable on an occasion of acquiring introspective knowledge. All the experiences of such subjects will have the indiscriminability property, so each of their experiences, according to the epistemic account, will be of the same kind as any other. Such a conclusion is absurd (see Siegel, 2004, 2008; Martin, 2006, 379–387). Surely the dog’s experiences are different in virtue of having phenomenal characters that differ, even if the dog—and any other cognitively unsophisticated subject—lacks the cognitive abilities to appreciate that fact.
Hallucinations differ from one another—a hallucination of a red cube is different from a hallucination of a green sphere. According to the epistemic account, these differences consist in facts about discriminability. The phenomenal character of a red cube hallucination consists in its being not knowably not a veridical perception of a red cube, whereas the phenomenal character of a green sphere hallucination consists in its being not knowably not a veridical perception of a green sphere. Suppose you hallucinate a green sphere. You can presumably know, on the basis of introspection, that you are not perceiving a red cube; that is, the following seems true: if you are having a green sphere experience, then you can know that you are not perceiving a red cube.18 The consequent of this conditional ascribes to you potential knowledge of a positive fact about a hallucinatory experience. According to the epistemic account, all there is to the phenomenal character of the hallucinatory experience is that it is not knowably not a veridical perception of a green sphere; that fact about what you can’t know doesn’t imply that you are not perceiving a red cube, and the epistemic account of hallucinatory experiences seems not to have any further resources to explain how you can have positive knowledge of your experience (Siegel, 2008, sec. 5). We can explain such positive knowledge of our experiences only in terms of our knowledge of the phenomenal character of hallucinatory experiences.
“How,” Scott Sturgeon asks, “could satisfying a purely epistemic condition be sufficient for a phenomenal experience?” (2008, 134). He notes that a zombie—a creature that is a physico-functional duplicate of you but lacks consciousness—could satisfy the purely epistemic condition of not being able to know introspectively that its experience was not a veridical perception of a bowl of fruit, but would not, ex hypothesi, be having an experience that seemed to present a bowl of fruit. So either the naive realist must deny the possibility of zombies, or the epistemic account of hallucinations is inadequate, and hallucinatory experiences must have a phenomenal character, after all.19
These objections all describe things that it seems can only be explained by appeal to the phenomenal character of hallucinatory experiences, things that it seems the epistemic account—with its claim that there is no more to the phenomenal character of hallucinatory experiences than the indiscriminability property—cannot explain. If the objections are right, then the naive realist is caught between, on the one hand, giving a characterization of the phenomenal character of hallucinatory experiences that screens off the naive realist phenomenal character of veridical experiences and, on the other, not giving a characterization that is sufficiently robust to explain the things that hallucinatory experiences explain.
The objections are plausible if we assume a conception of introspection according to which introspection provides us with an awareness of properties of experience akin to our perceptual awareness of properties of the objects of experience. In the perceptual case, if two objects look the same or look different, then we must be aware of some visible properties of the objects, in virtue of which they look the same or look different. If introspection provides us with an analogous awareness of properties of our experiences, then the fact that two experiences introspectively seem the same, or seem different, would suggest that we are similarly aware of properties of the experiences in virtue of which they do so. According to the objections, it is in virtue of sharing such properties that the experiences of a cognitively unsophisticated subject are indiscriminable, whether or not the subject can appreciate that fact; knowledge of such properties is required to ground our knowledge of positive facts about hallucinatory experiences; and such properties of experience are absent in the case of the zombie.
The naive realist rejects this conception of introspection in favor of an account from which it follows that an experience can seem to have a phenomenal character even if it lacks one. That is possible because the fact that an experience purports to present some mind-independent object or feature is sufficient for it to introspectively seem to have a phenomenal character, and (as I argued in the previous section) we can explain how an experience purports to present an object or feature without appealing to a phenomenal character (Martin, 1998, 2006, 383–399).
When a subject has a perceptual experience of, say, a red cube, she has an experience in virtue of which a mind-independent object—the cube—seems a certain way to her. Saying how an object seems to the subject in having an experience is saying something about the subject’s experience: it says something about the way the subject is modified in having that experience. How do we characterize the phenomenal character—what it’s like for the subject—of such an experience? We can characterize what it’s like for a subject of a perceptual experience by saying how it seems to the subject to have the experience. In seeing a red cube, for example, we can say that it seems to the subject that she is aware of a red cube. In characterizing what a perceptual experience is like—in saying that it seems to the subject that she is aware of a red cube—we are characterizing the mental state that the subject undergoes in having that experience.20 So there is a connection between how objects seem to the subject and the subject’s mental states, in particular the states that are relevant to characterizing how things seem to the subject. It’s a connection between the way experiences seem to the subject of, or what it’s like to have, a perceptual experience and the way the objects of that experience seem. Given this connection, how can a subject come to know introspectively about the character of her own experiences? And how, in general, can we acquire knowledge of what our experiences are like?
When we introspect our perceptual experiences, we direct our attention to the objects and features the experience presents. In introspecting a visual experience of a red cube, for example, we attend to the cube and its redness. Since our experience is not itself red or cubic, what we attend to isn’t a property of the experience: it’s an object or feature that is independent of experience. Given the connection between the way objects seem and the way experiences seem, coming to know how an object presented by an experience seems is a way of coming to know about what that experience is like. In attending to a red cube, I can come to know about the cube (that it seems red), and I can thereby come to know that my experience seems to be a certain way (that it seems to be the experience of a red cube). That’s a way of coming to know what my experience is like. So just as our perceptual knowledge of objects is grounded in our awareness of how the objects or features presented by that experience seem, so our introspective knowledge of what experience is like is grounded in reflection on, and awareness of, how those objects and features seem; it is not grounded in awareness of properties of the experience.
In the case of veridical experiences, “attending to what one’s experience is like cannot be separated from exploring and attending to the features of the world as perceived” (Martin, 1998, 170). When we hallucinate, our experience doesn’t actually present any objects or features, so there is nothing to which we can attend. Nonetheless, in being indiscriminable from veridical experiences, hallucinatory experiences are sensory episodes that purport to present objects and features. For the subject of such an experience, it is as if an object is presented and as if there is something there to attend to. That fact is sufficient to explain why the subject judges—if she does—that there is a red cube there, and it is therefore also sufficient to explain her introspective judgment that her experience is such that it seems to her that there is a red cube there. I see no reason why such an introspective judgment can’t amount to knowledge.21 In the case of nonveridical experiences, then, attending to what one’s experience is like cannot be separated from exploiting and attending to the features of the world as purportedly presented by the experience.
If how experience seems is grounded in how objects (or purported objects) seem, then the two necessarily go together; it couldn’t seem introspectively to someone as if things seem a certain way to them unless his or her experience presents, or purports to present, things as being that way. A necessary condition of its introspectively seeming to you that you are experiencing a red cube is that you have an experience that purportedly presents a red cube, therefore its introspectively seeming to you that you have an experience of a red cube is sufficient for you to be having an experience that presents or purports to present a red cube. It therefore follows from this model of introspection that there is a constitutive connection between experience and introspective awareness: “If it is true of someone that it seems to them as if things seem a certain way, as if they are having a certain sense experience, then they are thereby having that experience” (Martin, 2006, 397). This is an epistemic connection that depends on the metaphysical connection between the way objects seem to a subject and how the subject’s experience seems.
It follows from this account of introspection that two experiences can introspectively seem the same without their having a shared phenomenal character. For them to have a shared phenomenal character would be for there to be a common explanation of the fact that they purport to present objects and their features. But according to the naive realist, there is no explanation common to both veridical and hallucinatory experiences, nor is there anything common to all hallucinatory experiences other than that they are indiscriminable from the veridical ones.22 Therefore, if the account of introspection sketched here is correct, the fact that two experiences are introspectively indiscriminable gives us no reason to think that there is a common explanation of the fact that they are.
In the light of this, consider again the three objections.
In the case of the cognitively unsophisticated, the subject fails to acquire knowledge through a failure to attend, or to make judgments on the basis of attending, to the objects purportedly presented by the experience. That means we can idealize away the subject’s failure and say that two experiences are indistinguishable if, were the unsophisticated subject to have the cognitive capacities necessary to attend to the purported objects of the experience and judge accordingly, he or she would not be able to discriminate them (Sturgeon, 2008, 127–129).
According to the epistemic account of hallucinations, there is no more to the phenomenal character of an experience than the facts about discriminability. That makes puzzling how we could come to have positive knowledge of hallucinatory experiences; but it is only puzzling if we suppose that knowledge of hallucinatory experiences must be grounded in awareness of the phenomenal character of the experience. If, as the naive realist claims, knowledge of an experience is grounded in awareness of the purported objects of the experience, then we can explain how a subject can have positive knowledge of an experience that lacks a phenomenal character.
The zombie problem questioned how satisfying an epistemic condition (something a zombie could do) could be sufficient for having sensory experience. According to the naive realist’s preferred account of introspection, having a certain kind of introspective knowledge is sufficient for having sensory experience; since the relevant introspective knowledge is knowledge that is acquired in a certain way, the naive realist is not committed to denying the possibility of zombies but committed only to denying that zombies can acquire knowledge in that way. Since the relevant way appeals to judgments grounded in attention to the objects of conscious awareness, such a denial is plausible.
Any account of perceptual experience must give an adequate account of the nature of hallucinatory experiences; much recent criticism of naive realism has focused on what it says about the nature of hallucinations. According to the critics, naive realism either gives no account at all of such experiences or gives an account that is plainly explanatorily inadequate. In this chapter, I have explained why the naive realist is constrained or limited in the positive characterization that can be given of hallucinatory experiences, and I have argued that the resulting “epistemic” account is, pace the critics, explanatorily adequate. This argument has two parts. First, I suggested that an account of hallucinatory experience is not required to explain the sensory features of experiences in general; and second, I argued that the right account of introspection shows how experiences can be introspectively indiscriminable without having a shared phenomenal character. If my argument is right, then the naive realist account of hallucinatory experiences is no less satisfactory than the accounts offered by other, non-naive realist, accounts of experience.
References
Byrne, A., & Logue, H. (2008). Either/or. In A. Haddock & F. Macpherson (Eds.), Disjunctivism: Perception, action, knowledge. Oxford: Oxford University Press.
Crane, T. (2006). Is there a perceptual relation? In T. S. Gendler & J. Hawthorne (Eds.), Perceptual experience. Oxford: Clarendon Press.
Farkas, K. (2006). Indiscriminability and the sameness of appearance. Proceedings of the Aristotelian Society, 106, 205–225.
Haddock, A., & Macpherson, F. (2008). Introduction. In A. Haddock & F. Macpherson (Eds.), Disjunctivism: Perception, action, knowledge. Oxford: Oxford University Press.
Harman, G. (1990). The intrinsic quality of experience. Philosophical Perspectives, 4, 31–52.
Hawthorne, J., & Kovakovich, K. (2006). Disjunctivism. Proceedings of the Aristotelian Society, 80 (Suppl. Vol.), 145–183.
Johnston, M. (2004). The obscure object of hallucination. Philosophical Studies, 103, 113–183.
Kind, A. (2003). What’s so transparent about transparency. Philosophical Studies, 115, 225–244.
Martin, M. G. F. (1998). Setting things before the mind. In A. O’Hear (Ed.), Current issues in philosophy of mind. Cambridge: Cambridge University Press.
Martin, M. G. F. (2002). The transparency of experience. Mind and Language, 17(4), 376–425.
Martin, M. G. F. (2004a). The limits of self-awareness. Philosophical Studies, 120(1–3), 37–89.
Martin, M. G. F. (2004b). Uncovering appearances. Available at http://www.homepages.ucl.ac.uk/~uctymfm/chapters.htm.
Martin, M. G. F. (2006). On being alienated. In T. S. Gendler & J. Hawthorne (Eds.), Perceptual experience. Oxford: Clarendon Press.
McDowell, J. (1998). Having the world in view: Sellars, Kant, and intentionality. Journal of Philosophy, 95, 431–491.
Pautz, Adam. (2007). Intentionalism and perceptual presence. Philosophical Perspectives, 21, 495–541.
Peacocke, C. 1993. Externalist explanation. Proceedings of the Aristotelian Society, 93, 203–230.
Robinson, H. (1994). Perception. London: Routledge.
Siegel, S. (2004). Indiscriminability and the phenomenal. Philosophical Studies, 120(1–3), 91–112.
Siegel, S. (2006). Direct realism and perceptual consciousness. Philosophy and Phenomenological Research, 73(2), 378–410.
Siegel, S. (2008). The epistemic conception of hallucination. In A. Haddock & F. Macpherson (Eds.), Disjunctivism: Perception, action, knowledge. Oxford: Oxford University Press.
Soteriou, M. (2007). Content and the stream of consciousness. Philosophical Perspectives, 21, 543–568.
Stoljar, D. (2008). The argument from diaphanousness. In M. Ezcurdia, R. Stainton, & C. Viger (Eds.), New essays in the philosophy of language and mind. Canadian Journal of Philosophy Supplementary Volume.
Sturgeon, S. (2006). Disjunctivism. Proceedings of the Aristotelian Society, 80 (Suppl. Vol.), 185–216.
Sturgeon, S. (2008). Disjunctivism about visual experience. In A. Haddock & F. Macpherson (Eds.), Disjunctivism: Perception, action, knowledge. Oxford: Oxford University Press.
Travis, C. (2004). The silence of the senses. Mind, 113(449), 57–94.
Tye, M. (1992). Visual qualia and visual content. In T. Crane (Ed.), The contents of experience. Cambridge: Cambridge University Press.
Yablo, S. (1992). Mental causation. Philosophical Review, 101, 245–280.
1. In this chapter, whenever I talk about perceptual experiences, it is usually visual experiences that I have in mind. Although the arguments may generalize to the experiences associated with the other senses, it needs further argument to show that they do.
2. Since purporting to present something doesn’t imply failing to present it, in what follows I usually shorten “present or purport to present” to simply “purport to present.”
3. If experience entails the existence of its objects, then it is relational; if it seems to entail the existence of its objects, then it merely seems to be relational (see Crane, 2006).
4. I think, for example, it would be accepted by anyone who accepts the claim that experiences are transparent. That claim is defended by, among others, Harman (1990) and Tye (1992); for a critical discussion, see Martin (2002), Kind (2003), and Stoljar (2008).
5. Naive realism is the view elaborated and defended in a series of papers by M. G. F. Martin (1998, 2002, 2004, 2004b, 2006), and the argument I present is my interpretation of the argument developed at much greater length and sophistication in these papers. Other views may be labeled “naive realism,” but I will not discuss them.
6. See Martin (2002, 389).
7. Note that the form of naive realism that I am discussing is not a view about the kind of representational content experiences have, and in particular it is not the view that perceptual experiences have externally individuated or object-dependent content (although naive realists need not deny that experiences have content in some sense, they do deny that having content plays any fundamental explanatory role; for an argument against the idea that perceptual experiences have representational content, see Travis, 2004).
8. Robinson (1994, 151) develops a version of this argument; see also Martin (2004, sec. 4) and Johnston (2004, 115–116).
9. This is the kind of asymmetric relation that is typical of the determination relation, but the relation in this case doesn’t seem to be one of determination: veridically perceiving is not a way of hallucinating in the way that being scarlet is a way of being red.
10. For a discussion and defense of this kind of principle, see Yablo (1992).
11. The condition “by introspection alone” is intended to rule out situations in which the subject knows, by virtue of having some collateral information, that her experience is not veridical.
12. For further discussion of this use of indiscriminability, see Farkas (2006), Siegel (2004, 2008), Byrne and Logue (2008), Sturgeon (2008), and Hawthorne and Kovakovich (2006).
13. Given this, it can be argued that the representationalist’s putative phenomenal character is nonexplanatory unless the representationalist can show that there couldn’t occur an indiscriminable experience that lacked that phenomenal character. For the argument, see Martin (2004, sec. 7); for a critical discussion, see Siegel (2004), Byrne and Logue (2008), and Hawthorne and Kovakovich (2006).
14. See Martin (2004, 68).
15. For Smith’s own account, see Smith (2002, chap. 9); for a critical discussion, see Siegel (2006). For a survey and criticism of some recent attempts to address Smith’s complaint within a broadly representationalist framework, see Pautz (2007).
16. Of course, representationalists often claim that there is something distinctive about the relation or the content that explains the sensory character of the experience; but simply asserting that the relation or content is of a distinctively sensory kind doesn’t provide any real explanation.
17. See Martin (2004, 50), for an argument that the commitment to a general explanation of hallucinatory experiences requires an implausible epistemology of introspection.
18. This seems plausible, but it is something that the naive realist is committed to accepting, since it is the contraposition of the following claim: “Not knowably not a veridical perception of a red cube → not having a green sphere experience,” which the naive realist would accept (see Siegel 2008, 218–219).
19. Sturgeon’s own suggestion (2006, 140) is that the disjunctivist should embrace the radical view that hallucinatory experience “brings no phenomenology with it at all.”
20. For an insightful discussion of the relation between such states and experiences, see Soteriou (2007, esp. 555).
21. More needs to be said here than I have space to say about why this is a way of acquiring knowledge of experiences in general and hallucinations in particular.
22. Suppose that we can provide some explanation, in terms of properties of the experiences, of what it is for a hallucinatory experience to purportedly present something. Given the account of introspection, the fact that an experiential episode has those properties would be sufficient to explain the way that an experience introspectively seems. If there are properties of hallucinatory experiences that are sufficient to explain how those experiences introspectively seem, then we can use the screening-off argument to undermine the argument for naive realism.