...the phrase ‘what it’s like’ is now worse than useless (W. G. Lycan)
Baby You don’t know What It’s like to be me (Britney Spears)
The terminology surrounding the dispute between higher-order and first-order theories of consciousness is piled so high that it sometimes obscures the view. When the debris is cleared away, there is a real prospect—to borrow a memorable phrase from Mark Johnston—that this is one of those genial areas of enquiry in which the main competing positions are each in their own way perfectly true.
This paper has the following three goals, in ascending order of ambition. First, to show that there is a genuine dispute in the vicinity. Second, to rebut arguments against first-order theories based on a distinction (to be explained shortly) between “experiences” and “conscious experiences”. Third, to show that higher-order theories are mistaken. Even if none of these goals is met, at least the paper can serve as a terminological guide for the perplexed.
The dispute (as understood here) concerns theories of phenomenal consciousness—section 1 is a quick refresher on that topic. Section 2 describes the competing positions, and the crucial distinction between experiences and conscious experiences. Section 3 recounts the crucial distinction as explained by, respectively, Carruthers, Lycan, and Rosenthal. These three higher-order theorists use the distinction in somewhat different ways to argue against first-order theories; section 4 examines these arguments. Section 5 presents a brief argument against higher-order theories.
Here is how Block introduces the notion of phenomenal consciousness:
P-consciousness [phenomenal consciousness] is experience. P-conscious properties are experiential properties. P-conscious states are experiential, that is, a state is P-conscious if it has experiential properties. The totality of the experiential properties of a state are “what it is like” to have it. Moving from synonyms to examples, we have P-conscious states when we see, hear, smell, taste and have pains. (Block 1995: 230)
As a simplification, this paper restricts attention to perceptual experiences (hereafter simply called ‘experiences’), which will be taken to be events. The issue of which states are phenomenally conscious will accordingly be sidestepped, as will the issue of whether other events—e.g. episodes of thinking—are phenomenally conscious.
For what follows, an example of a Blockean “experiential property” will be useful. Consider visual experiences (in good light) of cucumbers, limes, green peppers, grass, and so on. These experiences saliently resemble each other in respect of “what it’s like” to undergo them; they accordingly share an experiential property, call it ‘G’. So-called spectrum inversion is—at least—a case where experiential properties are systematically permuted. When an inverted subject looks at a tomato, his experience has G. Sometimes a property like G is called a ‘phenomenal character’, a ‘qualitative character’, or a ‘quale’. The last two expressions are used in this way in Levine 2001; in the additional terminology of that book, G is “greenishness”.
There is a pair of distinctions that mark important differences between accounts of phenomenal consciousness. The first distinction is between intentionalism (or representationalism), and phenomenism. According to intentionalism, phenomenal consciousness is entirely intentional or representational. Less imprecisely, and restricting attention to perceptual experiences, intentionalism implies that facts about the representational content of an experience (together with facts about the representational content of the subject’s other mental events or states) fix or determine the facts about its phenomenal character. In other words, intentionalism implies that phenomenal character supervenes on representational content. Phenomenism rejects this supervenience thesis. One classic argument against intentionalism is based on an inverted spectrum thought experiment which is claimed to be a case of same representational content, yet different phenomenal character (Shoemaker 1981, Block 1990).
The second distinction is our topic: first-order (FO) vs. higher-order (HO) theories of consciousness. According to FO theories, an event may be phenomenally conscious even though it is not represented by one of the subject’s mental states/events. In other words, higher-order representations are not necessary for phenomenal consciousness. In particular, we may take the first-order theorist to hold that a certain first-order condition—to be explained in a moment—is sufficient for having phenomenal character G. Take a phenomenally conscious perceptual experience e of a cucumber in daylight. If e is in fact represented by one of the subject’s mental states/events, then consider e*, an event as similar to e as possible, except that e* is not represented by one of the subject’s mental states/events. (Hence, if e is not in fact the target of a higher-order representation, then e*=e.) Repeat this process for other phenomenally conscious experiences of cucumbers, limes, green peppers,..., resulting in other (possible) experiences e**, e***,... Then the first-order theorist’s sufficient condition for G is this: being mentally exactly like e* (or like e**, or like e***,...).
According to HO theories, on the other hand, higher-order representations are necessary for phenomenal consciousness. On this view, e* is not phenomenally conscious (and hence does not have G). That is, the higher-order theorist claims that there can be events that are mentally exactly like phenomenally conscious experiences, except that they are not targeted by higher-order representations, and thus are not phenomenally conscious. (The higher-order theorist also gives a sufficient condition; this will be explained in a moment.)
We now face a terminological decision. Suppose—for the rest of this paragraph—that the higher-order theorist is correct, and that e* is not phenomenally conscious. e* is not e, but presumably it’s pretty much like it. Like e, e* represents the subject’s environment as containing an elongated green thing; like e, e* causes the subject to believe that an elongated green thing is before her, and so forth. e*’s similarity to e counts in favor of labeling it a perceptual experience. On the other hand, given that phenomenal consciousness is often equated with experience (cf. Block’s slogan “P-consciousness is experience”), calling e* ‘an experience’ has something to be said against it. Since one of our chief protagonists—Carruthers—comes down on the side of using ‘experience’ broadly, to include events like e*, it will be convenient to follow his usage here. (See Carruthers 2000: 13-14; 18-20.) So e and e* are both experiences; the former is conscious and the latter isn’t. Block’s slogan, recast in this terminology, is ‘P-consciousness is conscious experience’.
These two distinctions—intentionalism vs. phenomenism, FO theories vs. HO theories—are independent. FO theory may be combined with intentionalism or phenomenism; likewise for HO theory. For example, Block and Levine are FO phenomenists; Dretske (1995) and Tye (1995) are FO intentionalists; and Carruthers and Lycan are HO intentionalists—albeit with a qualification in Lycan’s case. Rosenthal’s position has some hidden complications. This will be explained more fully later (section 3.3); for now the following enigmatic remark must suffice: Rosenthal is an HO intentionalist, but if he were to be converted to FO theory, he would be a phenomenist.
To avoid getting bogged down with too many qualifications, it will be helpful to assume (in agreement with Carruthers, Lycan, and Rosenthal) that intentionalism is correct. Nothing of substance will turn on this assumption. Thus (in the usual terminology) the dispute concerns first-order representational (FOR) and higher-order representational (HOR) theories of phenomenal consciousness.
We may think of the FOR theorist as maintaining that the following condition is sufficient for phenomenal consciousness (in particular, sufficient for G): being exactly like e* (or e**, or e***,...) in all representational or intentional respects. This unwieldy formulation will abbreviated as follows: the FOR theorist thinks that being an experience of green is sufficient for G. The HOR theorist disagrees, holding that some higher-order representation is necessary.
What about the HOR theorist’s sufficient condition for G? This will depend on the type of HOR theory. HOR theories come in two main flavors, depending on whether the relevant higher-order representations are (a) thoughts or (b) perceptions (or, at any rate, importantly similar to perceptions and importantly unlike thoughts). Rosenthal holds the HOT (higher-order thought) version; Carruthers and Lycan hold the HOP (higher-order perception) version.
Take the HOT version first. Since the HOR theorist is an intentionalist, she holds that there is an intentional sufficient condition for having G. Since she is a HOT theorist, she holds that this intentional sufficient condition involves the presence of a thought about the first-order experience: in particular the higher-order thought whose content is that one is having an experience of green.
Now take the HOP version. The HOP theorist holds that the intentional sufficient condition for G involves the presence of a (quasi-) perceptual experience directed on the first-order experience: in particular the higher-order perceptual experience whose content is that one is having an experience of green.
So, we have three candidate sufficient conditions for G, as follows:
FOR: being an experience of green
HOT: being an experience of green, accompanied by the thought that one is having an experience of green.
HOP: being an experience of green, accompanied by an experience that one is having an experience of green.,
FOR theories and HOR theories have been explained here in their weakest versions. The FOR theorist gives a sufficient condition for having G, and accordingly denies that a higher-order representation is necessary. Against this, the HOR theorist claims that a higher-order representation is necessary, and that the FOR theorist’s sufficient condition must be strengthened by the presence of a certain higher-order representation. Neither theorist need offer necessary and sufficient conditions for phenomenal consciousness, or the having of particular phenomenal characters. Further, neither theorist need offer naturalistic, physicalistic, or reductive conditions; or conditions that explain phenomenal consciousness.
As it happens, FOR theorists like Dretske and Tye, and HOR theorists like Carruthers, Lycan and Rosenthal, are more ambitious. They offer necessary and sufficient conditions (or implicitly give recipes for specifying such conditions). Second, these conditions are broadly physicalistic (at least, suggestions are made for how the conditions as officially stated could be given a physicalist reduction). Third, and connectedly, these conditions are intended to give some sort of explanation of phenomenal consciousness.
According to the FOR theorist, necessarily any experience of green is (phenomenally) conscious; according to the HOR theorist, there could be experiences of green that are not conscious. Of course, the FOR theorist will add other sufficient conditions for phenomenal consciousness (in our abbreviated formulation: being an experience of blue, being an experience of triangularity, and so on); again, the HOR theorist will claim that there could be such experiences that are not conscious. It is important to note that the FOR theorist is not committed to denying a certain consequence of the HOR theorist’s view, namely that there could be non-conscious experiences. For example, the FOR theorist could grant that we frequently have perceptual experiences of the disposition of our limbs (proprioceptive experiences) that are not phenomenally conscious.
The distinction that figures prominently in certain arguments against FOR theories offered by Carruthers, Lycan, and Rosenthal is that between experiences and conscious experiences. They draw the distinction in somewhat different ways, and their arguments based on it are quite different—as will be explained later, Lycan’s “argument” is, by his own lights, just a relatively uncontroversial observation. To complicate the picture further, they each coin their own terminology: “worldly subjectivity” and “experiential subjectivity” (Carruthers); “lower-order” and “higher-order” readings of ‘what it’s like’ (Lycan); “thin phenomenality” and “thick phenomenality” (Rosenthal).
The first order of business is to show that these three distinctions are basically the same. Along the way, important differences between Carruthers, Lycan, and Rosenthal will emerge. The next item (section 4) is to examine why the distinction is supposed to be real, as opposed to merely notional, and the subsequent arguments against FOR theories.
Carruthers’ distinction between worldly subjectivity (what the world is like for the subject) and experiential subjectivity (what the subject’s experience is like for the subject) is one between properties:
...what the world (or state of the organism’s own body) is like for an organism...is a property of the world (or of a world-perceiver pair, perhaps).
...what the organism’s experience of the world (or of its own body) is like for the organism...is a property of the organism’s experience of the world (or of an experience-experience pair) (2000: 127-8)
Explaining this distinction in an earlier paper, Carruthers asks which of these two sorts of subjectivity “deserves the title ‘phenomenal consciousness’”, and replies that he is “happy whichever reply is given” (1998: 209). In his later book, the question gets a more opinionated answer: “The subjectivity of experience, surely” (2000: 129, n.7). Since ‘Phenomenal Consciousness’ is the title of Carruthers’ book, we may safely presume that Carruthers takes himself to be using the term in the same way as Block—its inventor. It is also clear that Carruthers takes the FOR theorists Dretske and Tye to be offering accounts of phenomenal consciousness, as Carruthers understands the term. (These points might seem too trivial to mention. But as will transpire shortly, not every HO theorist agrees with Carruthers on these exegetical issues.)
The distinction between worldly and experiential subjectivity may be further clarified as follows. Suppose an organism has color vision. Then what the world is like for the organism is, inter alia, colored. In other words, the organism’s perceptual experience represents the world as colored. So we may say that a worldly subjective property (relative to this organism) is simply a property represented by the organism’s perceptual experience: green, for example. This is a property of objects in the organism’s environment (at least if its color experiences are veridical). Experientially subjective properties, on the other hand, are just Blockean experiential properties, like our old friend G. (See Carruthers 2000: 13.) G is a property of experiences, and hence is not the property green—worldly subjective properties are plainly different from experientially subjective properties.
An experience of green that there is nothing it’s like for the subject to undergo is said by Carruthers to “possess worldly subjectivity but [to] lack experiential subjectivity” (2000: 147); an experience of green that there is something it’s like for the subject to undergo possesses both forms of subjectivity. So the distinction between experiences and conscious experiences (as explained in the previous section) is the same as the distinction between experiences that only have worldly subjectivity, and those that have both worldly and experiential subjectivity.
According to Lycan (this volume), a distinction that he has made in a number of places is the same as, or at any rate very similar to, Carruthers’ distinction. Lycan’s distinction, he explains, corresponds to an “ambiguity” in philosophers’ talk of “what it’s like” (1996: 77).
Lycan claims that on one (“lower-order”) reading of ‘what it’s like’, the phrase picks out what Lycan—following C. I. Lewis—calls a quale. A quale is an “introspectible monadic qualitative property of what seems to be a phenomenal individual, such as the color of what Russell called a visual sense datum” (1996: 69). Lycan holds a “representational theory of qualia”, which basically amounts to the view that qualia are properties that visual experiences represent objects as having, and which in some cases may not in fact be properties of such objects. One may seem to see a green thing, according to Lycan, even though there is nothing green to be seen—not even a sense datum. (We may take this view to be common ground among all our protagonists.) Notice that Lycan’s qualia are the same as the worldly subjective properties of the previous section.
According to Lycan, on the lower-order reading of ‘what it’s like’, there is something it’s like to experience green and, in general, something it’s like to undergo a perceptual experience (Lycan 1999a). So talk of what it’s like in Lycan’s lower-order sense is straightforwardly translatable by Carruthers’ terminology of ‘worldly subjectivity’.
On the other (“higher-order”) reading of ‘what it’s like’, Lycan claims that this expression picks out “a higher-order aspect of the quale, namely, what it’s like to experience that quale” (1999b: 128; see also Lycan this volume); so talk of what it’s like in Lycan’s higher-order sense is straightforwardly translatable by Carruthers’ terminology of ‘experiential subjectivity’. Lycan takes these higher-order aspects or properties to be properties of qualia (i.e. properties of colors, shapes, odors, etc.). The higher-order “what it’s like” properties are accordingly not the same as the experientially subjective properties of the previous section (which are properties of experiences), but there is a simple correspondence between them. Let g be the Lycanian higher-order “what it’s like” aspect of “green qualia”. Then experience e has G iff it is an experience of a quale that has g.
Since the expressions ‘qualitative character’, ‘phenomenal character’, and ‘phenomenal consciousness’ are often explained using ‘what it’s like’, it is no surprise that Lycan finds an ambiguity here too (Lycan 1999a). This allows us to avoid clumsy locutions such as ‘higher-order “what it’s like” properties’, and to rephrase Lycan’s distinction in Lycanian terminology thus: lower-order qualitative character vs. higher-order qualitative character.
Lycan himself prefers to use ‘what it’s like’ with its higher-order reading (this volume). He avoids ‘phenomenal consciousness’, but if forced would presumably have a similar preference. So in this respect we may take him to be in terminological agreement with Carruthers. However, unlike Carruthers, Lycan thinks Block’s use of ‘phenomenal consciousness’ picks out qualitative character of the lower-order kind (Lycan 1999a: n. 1). Further, Lycan takes FOR theorists like Dretske and Tye to be merely offering representationalist theories of lower-order qualitative character (1999a; this volume). That is, according to Lycan, Dretske and Tye are simply giving representationalist accounts of perceptual experience—for example, “perceiving a [green] object as such”, or (in the ‘qualia’ terminology), “registering” a “green quale” (Lycan 1996: 76).
Carruthers thinks that the dispute between Block, Dretske and Tye, on the one hand, and Carruthers, Lycan and Rosenthal, on the other, is genuine; Lycan, however, thinks it is largely terminological. According to Lycan, when Dretske, for example, inveighs against HOR theories, he is correctly claiming that higher-order representations are not needed in accounts of lower-order qualitative character. And when Carruthers touts the virtues of HOR theories, he is correctly claiming that such representations are needed in accounts of higher-order qualitative character. For Lycan, the philosophical action is to be found further down the road, in the dispute between HOP and HOT.
Rosenthal makes a distinction between two kinds of phenomenal consciousness, or, in Block’s (2001) currently preferred terminology, “phenomenality”:
One kind consists in the subjective occurrence of mental qualities, while the other kind consists just in the occurrence of qualitative character without there also being anything it’s like for one to have that qualitative character. Let’s call the first kind thick phenomenality and the second thin phenomenality. Thick phenomenality is just thin phenomenality together with there being something it’s like for one to have that thin phenomenality. (2002a: 657)
In order to explain what “the occurrence of qualitative character” is supposed to be, more details of Rosenthal’s view are needed. Rosenthal thinks that experiences have introspectible properties that he elsewhere calls sensory qualities, or mental qualities (Rosenthal 1999). One such mental quality is characteristic of perceptions as of green objects, and Rosenthal calls this quality mental green. An “occurrence of qualitative character” is an experience that has some sensory or mental quality.
Mental green is the same as the Lycanian property of being a registering of a green quale and, as noted, Lycan identifies this with the intentional property of being an experience of green. But Rosenthal would refuse to make this identification, because he thinks that mental green is a non-intentional property; in this, he is in agreement with Block and Levine. (‘Mental green’ is used here is a neutral way, so that it is an open question whether mental green is non-intentional.) If Rosenthal converted to FO theories, he would think that mental green, and not any intentional property, was sufficient for G. Section 2 claimed that if Rosenthal were an FO theorist, he would be a phenomenist, and this is the reason why.
Solely for convenience, let us assume that Rosenthal is wrong about mental green, in which case it may be identified with the property of being an experience of green. Hence, an experience of green is an example of the “occurrence of qualitative character”.
Since thin phenomenality is “the occurrence of qualitative character without there also being anything it is like for one to have that qualitative character”, it appears that thin phenomenality is the same as what section 2 called ‘non-conscious experience’. However, the last sentence of the quotation implies that thin phenomenality does not preclude there being something it’s like (see also the quotation below). It is clear that Rosenthal’s intent is best served by ignoring the stipulation that there is nothing it is like to undergo thin phenomenality. Thin phenomenality, then, will be taken simply to be “the occurrence of qualitative character”, in which case it may be identified with worldly subjectivity and lower-order qualitative character.
Rosenthal’s explanation of thick phenomenality is in essence the same as Lycan’s explanation of the higher-order reading of ‘what it’s like’; so thick phenomenality may be identified with higher-order qualitative character, and experiential subjectivity.
According to Rosenthal, “Block describes phenomenality in different ways that arguably pick out distinct types of mental occurrence” (655), these “distinct types” being thin and thick phenomenality. But Rosenthal thinks that, charitably interpreted, Block has thin phenomenality in mind:
If we bracket the issue about how to understand the admittedly vexed phrase ‘what it’s like’, Block’s view seems to be that phenomenality is simply thin phenomenality, and what I’m calling thick phenomenality is phenomenality plus reflexivity [i.e. plus an appropriate higher-order representation]...
Terminology aside, this fits neatly with my own view of things. (2002a: 657; see also Rosenthal 1997a)
Rosenthal, then, shows some sympathy with Lycan’s ecumenicism. Like Lycan, and unlike Carruthers, Rosenthal takes Block’s use of ‘phenomenality’ and ‘phenomenal consciousness’—at any rate when charitably interpreted—to refer to thin phenomenality (i.e. worldly subjectivity or lower-order qualitative character). Thus, according to Rosenthal, when Block claims that Rosenthal’s higher-order thought theory is inadequate as an account of “phenomenal consciousness” (i.e. thin phenomenality), Block is right: thin phenomenality does not require higher-order thoughts. But, Rosenthal protests, the higher-thought theory was never intended as an account of thin phenomenality. Rather, it is an account of thick phenomenality—the higher-order kind of qualitative character, or experiential subjectivity.
So far, we have seen three accounts of what is—at least after some shoehorning—the same distinction: worldly vs. experiential subjectivity, lower-order vs. higher-order qualitative character, and thin vs. thick phenomenality. This explosion of terminology will be all sound and fury if the distinction turns out to be merely notional; if, that is, there can’t be non-conscious experiences. Now the task is to examine why our three HOR theorists think that the distinction is real, and their related reasons for thinking that FOR theories are false. To anticipate: it will be argued that the reality of the distinction has not been demonstrated, and that the associated case against FOR theories fails.
Recall an earlier complication: the reality of the distinction between experiences and conscious experiences does not entail the falsity of FOR theories (at least as explained here). According to the FOR theorist, subtracting higher-order representations from a phenomenally conscious experience does not subtract phenomenal consciousness. That is consistent with the claim that there could be experiences that are not phenomenally conscious.
In Carruthers’ presentation, the argument for the reality of the distinction is only the first stage in a much longer argument against FOR theories and for HOR theories. In contrast, it is an immediate consequence of Rosenthal’s and Lycan’s explanations of why the distinction is real that FOR theories are false.
Given the Lycan/Rosenthal interpretation of Block (sections 3.2 and 3.3), ambiguity threatens to break out in this paper. Temporarily setting aside the question of what Block actually means, ‘phenomenal consciousness’ is used in this paper in Carruthers’ sense: phenomenal consciousness, experiential subjectivity, higher-order qualitative character, and thick phenomenality are all the same.
One place where Rosenthal argues at length for the reality of the distinction between thick and thin phenomenality is in a paper responding to Block. According to Block (2001), in cases of visual extinction (where the subject reports he cannot see the stimulus, yet by other measures perceives it), it may be that the subject is undergoing a phenomenally conscious experience of the stimulus. That is—as Block would be happy to rephrase it—it may be that there is something it’s like to perceive the stimulus. Rosenthal objects:
As Nagel stressed in the article that launched that phrase, what it’s like to have an experience is what it’s like for the individual that has the experience. When a person enjoys the taste of wine, thereby enjoying gustatory phenomenality, there is something it’s like for that person to experience the taste of the wine.
Not so in cases of visual extinction; there is nothing it’s like for an extinction subject to have a qualitative experience of the extinguished stimuli. That’s why seeing visual extinction as the having of phenomenality without knowing it does not fit comfortably with the explanation of phenomenality in terms of what it’s like to have an experience. (Rosenthal 2002a: 656)
This passage claims that an extinction subject is undergoing a visual experience that is not phenomenally conscious (because there is nothing it’s like for the subject to undergo the experience). If correct, this would immediately establish the reality of the distinction between experiences and conscious experiences. (As just noted, more work is required to establish that FO theories are false.) Since this is basically Carruthers’ style of argument for the reality of the distinction, discussion of this will be postponed to the section after next.
However, the second paragraph makes another point, one that is potentially more powerful. It suggests that if there is something it’s like for the subject to undergo an experience, the subject is aware of having the experience, a suggestion that is repeated in the surrounding paragraphs. (‘Knowing it’ in the last quoted sentence is evidently interchangeable with ‘being conscious of it’; see p. 658.) If that is correct, then FOR theories are false: being aware that one is having an experience is a necessary condition for the experience to be phenomenally conscious.
Rosenthal’s argument for this necessary condition turns on the observation that there is something it’s like to be a rock, or—his example—a table. (“What it’s like to be a table, for example, is roughly something’s having the characteristic features of tables” (656).) This “more general, nonmental” use of ‘what it’s like’ is distinguished from the “special use to describe subjectivity” by the addition of the phrase ‘for so-and-so’. There isn’t something it’s like for the table to be kicked in the leg; there is something it’s like for Mr. N.N. to be kicked in the leg. The crucial step is this:
...conscious[ness] of oneself...must in any case occur if there is something it’s like to have the experience. We’re not interested in there being something it’s like for somebody else to have the experience; there must be something it’s like for one to have it, oneself. Without specifying that, what it’s like would be on a par with what it’s like to be a table. (656, my italics)
This passage only purports to explain why awareness of oneself is a necessary condition for having a phenomenally conscious experience (as opposed to awareness that one is having the experience), so some additional steps are needed to show that FOR theories are false. But let us waive this point.
There is certainly a distinction between there being (a) something it’s like for Mr. N.N. to have an experience; (b) something it’s like for Ms. M.M. to have an experience; (c) something it’s like for a person (no particular person) to have an experience. Rosenthal is claiming that the correct account of these distinctions will imply that if there is something it’s like for so-and-so to have an experience then so-and-so is aware of himself (and/or his experience). However, there is also a distinction between there being (a) something it’s like to be Tim-the-table; (b) something it’s like to be Tom-the-table; (c) something it’s like to be a table. What’s more, Rosenthal evidently agrees: “there is something it’s like...even to be this very table” (656). Clearly the correct account of the latter distinctions will not imply that if there’s something it’s like to be Tim-the-table, then Tim is aware of itself. So why think that the correct account of the former distinctions will have a similar implication?
In fact, the quoted passage doesn’t really offer a reason: it seems to be simply repeating the claim that if there is something it’s like for so-and-so to have an experience then so-and-so is aware of himself. And this is far from obvious.
Consider a specific example:
(*) There is something it’s like for Mr. N.N. to see a cucumber.
(*) is equivalent to ‘For Mr. N.N. to see a cucumber is like something’ which in turn is equivalent to ‘For a cucumber to be seen by Mr. N.N. is like something’. This illustrates the fact that in (*) ‘for’ has no particular attachment to ‘Mr. N.N.’; it is instead the complementizer of the infinitival clause ‘Mr. N.N. to see a cucumber’. (Unlike, for example, ‘for’ in ‘The police are looking for Mr. N.N.’.) Hence there is no syntactic reason to think that (*) will have some exciting entailment solely about Mr. N.N.—say, that he is aware of himself. Perhaps more strongly, it is doubtful that ‘There is something it’s like for so-and-so to φ’ has some “special use to describe subjectivity” (dialects of analytic philosophy aside). ‘What was it like for the car to be driven in the desert?’ appears to be an intelligible and literal question, that one might ask a driver returning from the Paris-Dakar Rally.
In any case, even if we grant for the sake of the argument that if there is something it’s like for Mr. N.N. to see a cucumber, then Mr. N.N. is aware of something, it is unclear why this needs to be Mr. N.N. himself, or his experience. An analogy: there is something it’s like for Mr. N.N. to read Wittgenstein’s Poker. That might be true, even though Mr. N.N. is so captivated by the book that he is not aware of his reading; instead, he is aware of the dramatic encounter between Wittgenstein and Popper at the Moral Sciences Club, and other events the book describes. Similarly, there might be something it’s like for Mr. N.N. to see a cucumber, simply because Mr. N.N. is aware of the cucumber—not of his seeing of the cucumber.
In short: there seems little prospect of deriving the falsity of FOR theories (and the consequent reality of the distinction between experiences and conscious experiences), from the semantics of ‘what it’s like for so-and-so to φ’, considered as a phrase of ordinary English.
As section 3.2 explained, Lycan thinks that ‘what it’s like’, in its philosophical usage, is ambiguous, and consequently that the participants in the debate are mostly talking past each other. Carruthers, remember, thinks that Tye is offering a first-order account of phenomenal consciousness, and accordingly expends much energy on the attempt to demonstrate its falsity. Lycan thinks that Carruthers is wrong about the exegetical claim, and that there is no disagreement once the ambiguity in ‘what it’s like’ is pointed out:
Now, “what it’s like”: that phrase certainly is usually used in Carruthers’ rich way, but it has also been used to mean just a bare phenomenal quale, such as the phenomenal property of redness that Dretske, Tye and I explicate representationally and that figures in what Carruthers calls a percept. What Tye means, presumably, is that there are “phenomenally-conscious” states in the weak [i.e. lower-order] sense, states having qualitative character (so there is “something it is like” to be in them, in the weak [i.e. lower-order] sense of that expression), but of which the subject is unaware (and so there is nothing “it is like” for the subject to have them, in the strong [i.e. higher-order] sense of that expression). (Lycan 1999a)
Notice that the last sentence moves from ‘unaware of the experience’ to ‘nothing it is like [in one sense]’ as if the transition is entirely unproblematic. But the transition is not at all obvious.
Earlier, Lycan’s distinction between lower-order and higher-order qualitative character was identified with the distinction between experiences and conscious experiences—that is, with the distinction between experiences, and experiences that there is something it’s like for the subject to undergo. Now we may take Carruthers to hold that ‘something it’s like’, as it appears in the preceding sentence, is not ambiguous (2000: 13); likewise for Rosenthal. So both Carruthers and Rosenthal should be comfortable with the explanation of the distinction between experiences and conscious experiences in terms of ‘what it’s like’.
On the other hand, by Lycan’s lights, this explanation of the distinction is seriously ambiguous (as is much else in this paper). Resolving the ambiguity one way, the distinction is merely notional: in the lower-order sense, for any experience, there is something it’s like for the subject to undergo the experience. Resolving it another way, and assuming that there are experiences of which the subject is unaware, the reality of the distinction is shown by actual cases: in the higher-order sense, there is nothing it is like to undergo some experiences.
FOR theories (as explained in section 2) claim that there could be phenomenally conscious experiences (i.e. experiences with higher-order qualitative character) of which the subject is unaware. According to Lycan, from the fact that an experience has higher-order qualitative character it immediately follows that the subject is aware of the experience; hence, on Lycan’s view, FOR theories are trivially false. (This is of course why Lycan thinks that Tye and Dretske are not FOR theorists in the sense of this paper.)
Let us examine Lycan’s claim of ambiguity first. The quoted passage highlights a point noted in section 3.2, namely that Lycan thinks that ‘what it’s like’ in its lower-order sense picks out a quale—greenness or squareness, for example. Taking this at face value, Lycan is saying that when some philosophers (notably Tye and Dretske) ask ‘What is it like to see cucumbers?’, or ‘What are experiences of cucumbers like?’, they are using these expressions in such a way so that the correct answer is ‘Green’. Now—as Lycan would agree—there is clearly no natural way of understanding these questions on which this is the right answer. ‘Green’ is a right answer to the question ‘What is the cucumber like?’; extraordinary contexts aside, it is never a right answer to the question ‘What is seeing the cucumber like?’. But, further, ‘what it’s like’ is not taken by Tye or Dretske to be a technical expression, whose meaning is a matter of explicit or implicit stipulation. Unlike, say, ‘intentionality’, ‘what it’s like’ is typically used in the literature without any special explanation or definition. Therefore there is no reason to postulate an ambiguity.
And if there is no ambiguity, then Lycan’s position reduces to Rosenthal’s. Namely, that if there is something it’s like for the subject to undergo an experience (in the univocal sense of ‘something it’s like’), then the subject is aware of the experience. But the previous section found no persuasive argument for that conclusion.
Progress so far can be summarized in four points. First—from the previous section—there is no ambiguity of the relevant sort in ‘what it’s like’. Second—from section 4.1—there is no easy route from the semantics of ‘what it’s like for so-and-so to φ’ to the falsity of FO theory. The first two points support the third and fourth. A version of the FOR/HOR dispute was explained successfully in section 2, and has no obvious resolution. Lastly, because of the lack of ambiguity, and the fact that FO theories are live options, Block, Tye and Dretske are FO theorists, and hence are genuinely opposed to HOR theories.
Section 2 confidently proclaimed that Lycan and Rosenthal endorsed HOR theories; some qualifications and reservations are now needed. Because of Lycan’s view that he and his opponents are mostly separated by terminology, one might question whether Lycan is an HOR theorist of phenomenal consciousness, given that this position really is controversial. And in fact, Lycan’s usual way of explaining HOR theories is not remotely equivalent to the characterization given in section 2: “HOR theories are...theories of ‘conscious states’, in the sense of states whose subjects are aware of being in them...HOR theories...[suppose] that what makes a state a conscious state in this sense is that it itself is represented by another of the subject’s mental states” (2001a: 3; my italics). In Rosenthal’s case, although there is little doubt that he endorses an HOR theory of phenomenal consciousness, a significant amount of what he says under the rubric of ‘a theory of consciousness’ does not depend on this commitment.
Carruthers, on the other hand, can be labeled an HOR theorist with little reservation or qualification. He rests his case for non-conscious experiences on a series of examples, of which the following are representative:
Armstrong’s driver, his mind on other matters, “comes to” and can’t recall seeing anything on the road he has been driving on for the last half an hour. Yet the driver navigated the road safely, so he must have seen the traffic. (Armstrong 1981; Carruthers 2000: 148-9.)
Milner and Goodale’s patient D.F. (suffering from visual form agnosia) can’t recognize the orientation of a slot in a disk—whether the slot is vertical, horizontal, etc., but can use visual information to post a letter through it (Milner and Goodale 1995: 128-9; Carruthers 2001: 161). Similar dissociations (between the so-called “what” and “how” pathways) can be found in normal subjects. In the Titchener illusion (illustrated in Milner and Goodale: 168; Carruthers: 163), two equally sized discs look different in size; yet when normal subjects are asked to pick up the discs, their grip aperture is approximately the same for both. (See also Rosenthal’s discussion of visual extinction quoted in section 4.1 above.)
Carruthers takes many extra steps from the reality of the distinction between experiences and conscious experiences to the falsity of FOR theories, and many more from that to his own brand of the HOP theory. Here only the first step will be disputed: the conclusion that these examples show that the distinction is real.
Take Armstrong’s driver first. It is not implausible that the driver is undergoing experiences of the road. (Absent-minded academics spend most of their waking lives in such a condition.) Was there something it was like for him to see a green traffic light? Well, why not? After all, variants of the example are familiar, for example when some time after “coming to” one recalls that the light was green, remembers that the experience was like something, and in particular what it was like. If one remembers this, then there was something it was like for one to undergo the experience, and so the experience was phenomenally conscious. Suppose one is not able to remember what it was like: that does not show that the experience was not phenomenally conscious. Armstrong’s driver is not a convincing exhibit. (In fact, Carruthers himself places little weight on examples like A.)
What about cases involving a dissociation between the “two visual systems”? Here there seems little pressure to say that the subjects are having the relevant kinds of experiences in the first place, and a fortiori little pressure to admit that the subjects are undergoing non-conscious experiences. If D.F. can see (or at any rate perceive) the orientation of the slot, then she is aware of the orientation of the slot. In the Titchener illusion experiment, if the subject sees (or at any rate perceives) that the discs are the same size, she is aware that the discs are of the same size. Yet it does not seem very plausible that D.F. is aware of the orientation of the slot—she is unable to say what it is, or make plans that depend on this piece of information. Similarly with the normal subjects in the Titchener illusion experiment, and Rosenthal’s example of visual extinction.
The above points do not presuppose that the ability to report that p is a necessary condition of perceiving that p, merely that the inability to report is sometimes good evidence that the subject is not perceiving that p. And, of course, it should be admitted that the crucial information (that the slot is horizontal, that the discs are the same size) is playing an important role in the subjects’ cognitive economy, in particular in guiding the subjects’ behavior. However, the information that p may play such a role without the subject being aware that p. If there are such things as subjects or selves (i.e. persons, in the case of adult humans), then this can hardly be denied. Finally, the above points do not presuppose that the subject or self is a primitive irreducible entity. Perhaps some neo-Humean reductionist account of the self is correct: as Block puts it, we are “loose federations of centers of control and integration” (1997: 162). That is entirely consistent with the claim that (for example) D.F. does not perceive the orientation of the slot.
The topic of the self hardly deserves such cursory treatment, but will receive it due to space limitations. Still, we may conclude that—at least prima facie—Carruthers’ examples do not add up to a convincing case for the reality of the distinction between experiences and conscious experiences.
In fact, this conclusion leaves most of Carruthers’ case against broadly reductive FOR theories unaffected. In the chapter where the distinction is put to the most work, Carruthers is arguing against FOR accounts that may be approximately schematized as follows:
(†) Event e is phenomenally conscious iff e has (first-order) functional role R.
And, whether or not examples like those in B are cases of non-conscious experiences, one can easily see how they could pose problems for the theories that are instances of (†). Maybe one theory predicts (implausibly, or so we may grant) that the Titchener illusion subjects do have phenomenally conscious experiences of the discs being the same size—which of course they are unable to report. Maybe another only avoids this prediction by making an ad hoc and otherwise ill-motivated assumption. And so on.
Still, although Carruthers may have shown that particular reductive theories have serious problems, this is not to impugn FOR theories across the board.
The previous sections have not argued that HOR theories are mistaken; this final section gives a short argument for that conclusion.
What is sufficient for one to know what one’s experience is like, in the sense relevant to phenomenal consciousness? Suppose that one knows—via “introspection”, whatever that is exactly—that one’s present experience has property P, and one knows that certain past experiences had property P. Suppose that one knows that yet other past experiences had other properties, Q, R,... Further suppose that experiences possessing these properties thereby saliently resemble each other, and that one knows these facts about similarity. So, for example, one may know that one’s (present) experience is more similar (in this salient respect) to a Q-experience than to an R-experience, and so on. This appears to be sufficient for one to know what one’s (present) experience is like. If it is not sufficient, it is unclear what else needs adding. (Insisting that P, Q, R,... must be intrinsic, for example, would make it very hard for a typical intentionalist to maintain that one can know what one’s experience is like.) Assume, then, that it is sufficient. If all these facts about one’s experience obtain, let us say that one’s experience is in condition C. Knowing that one’s experience is in C, then, is sufficient for knowing what one’s experience is like. Hence, if one’s experience is in C, then one’s experience is like something (in the relevant sense), and therefore one’s experience is phenomenally conscious.
Now observe that such a condition may be constructed from purely first-order materials. The roles of P, Q, R,... can be played by the properties of being an experience of green, being an experience of turquoise, being an experience of blue, and so forth. Everyone can agree that one can know (at least sometimes) by introspection that one’s experience is an experience of green. Recall that we are assuming that Rosenthal’s “mental green”, an especially striking and “qualitative” property of experiences of limes, grass, etc., is identical to the property of being an experience of green (section 3.1 above). Under this assumption, at least some of the salient similarities and differences between experiences are due to what they represent about one’s environment: because of their content, experiences of green are saliently more similar to experiences of turquoise than to experiences of blue, and so on. If, on the other hand, Rosenthal is correct, and mental green is non-intentional, then the desired condition may be constructed with the non-intentional but first-order properties mental green, mental turquoise, and so forth, playing the roles of P, Q, R,...
Now distinguish knowing what one’s experience is like from one’s experience being like something. Setting misguided Cartesian doctrines about the self-intimation of phenomenology aside, one’s experience may be like something (in the sense relevant to phenomenal consciousness), even though one does not know that one’s experience is like something. (This is just an instance of the general principle that any contingent proposition p may be true and not known.)
Take condition C as constructed in the paragraph before last. From the point just made, we can conclude that one’s experience may be in C, even though one does not know that one’s experience is in C. Indeed, we can go further, and conclude that one’s experience can be in C even though it is not the target of any higher-order representation at all. Setting aside Rosenthal’s view for simplicity, this last step basically amounts to saying that an experience of green that is not the target of a higher-order representation is more similar in salient respects to experiences of turquoise than it is to experiences of blue, which HOR theories can hardly dispute. But it was argued that any experience in C is a phenomenally conscious experience. HOR theories deny that, and so they are in error.
This argument does not show that FOR theories, as explained in
section 2, are true. For all that has been said, maybe higher-order
representations do have something to contribute to phenomenal
consciousness. Perhaps knowing what one’s experience is
completely like (in the sense relevant to phenomenal
consciousness) requires knowing that one’s experience is in
some higher-order condition (cf. Carruthers 2000: 183-4). In
particular, perhaps knowledge that one’s experience has our
old friend G is not first-order knowledge. Further argument is
needed to foreclose this possibility. But if we take the
explanation of phenomenal consciousness in terms of ‘what
it’s like’ seriously, then higher-order representations
are not necessary for phenomenal consciousness.
[*] Many thanks to Peter Carruthers, Rocco Gennaro, Alec Marantz, Sarah McGrath, and an audience at ASU. I am especially indebted to Bill Lycan and David Rosenthal who kindly supplied extensive comments on earlier drafts.
 1996: 77.
 See Johnston 1992: 221.
 This broadening of the supervenience base is needed to accommodate higher-order theories.
 Lycan adds facts about“functional organization” (1996: 11) to the intentional supervenience base for phenomenal character. On a more careful statement of intentionalism than the one given here, this is an intentionalist view. See Byrne 2001: 204-6.
 Intentionalism or representationalism comes in a variety of inequivalent formulations. On some of these, as Rosenthal (forthcoming, section 4) points out, he is not an HO intentionalist.
 Why not necessary and sufficient? Here is one reason: some episodes of mental imagery have G (at least on a natural way of understanding the introduction of ‘G’), and these are not perceptual experiences. The complicated question of how the FOR theorist should state a necessary and sufficient condition for G need not be examined here.
 As does Gennaro (1996).
 Carruthers also holds a dispositional version of the HOT theory but, for reasons that need not concern us here, thinks this is equivalent to a HOP theory (see Carruthers 2000; this volume).
 Two qualifications concerning Rosenthal’s view should be noted. First, according to Rosenthal the higher-order thought specifies one’s experience in partly non-intentional respects, as having a property Rosenthal calls mental green (see section 3.3 below). Second, to avoid certain counterexamples, Rosenthal adds that the higher-order thought must “[rely] on neither inference nor observation” (Rosenthal 1997b: 738). For our purposes. these qualifications can be ignored.
 Carruthers (2000: 248) takes the experience with the lower-order content to be the same as the experience with the higher-order content.
 One might wonder what the HOR theorist should say if the higher-order representation misrepresents the lower-order one. See Byrne 1997: 119-24; Neander 1998; Lycan 1998; Levine 2001: 104-111; Rosenthal 2002a; Rosenthal forthcoming; Carruthers this volume; Gennaro this volume.
 A first-order theorist might say that such proprioceptive experiences are phenomenally non-conscious because they lack a certain special sort of content, or lack a certain functional role. Smith (2002: 165) claims that experiences of resistance to one’s bodily movements have a “unique non-sensory nature”, which is arguably equivalent to the claim that they need not be phenomenally conscious.
 Cf. the alternative use of ‘quale’ noted in section 1 (for yet another use, see Carruthers 2000: 15).
 More cautiously: a (perceptual) experience has mental green iff it is a registering of a green quale. It may be that, in Rosenthal’s intended usage, there are some events that have mental green but which aren’t perceptual experiences (and hence not “registerings”); in particular, some “sub-personal” events in the subject’s cognitive system. See the following note.
 In fact, Rosenthal seems to intend thin phenomenality to be somewhat broader than the category of worldly subjectivity/lower-order qualitative character, including occurrences of qualitative character at early stages of perceptual processing (see 2002a: 660). This is another complication that can be set aside.
 Nagel, incidentally, seems to take ‘what it’s like to be X’ to be equivalent to ‘what it’s like for X’ (1974: 436). This is quite doubtful. Note that ‘What is it like to be X?’ is complete in a way that ‘What is it like for X?’ is not. In order for the latter to express a question, some activity or state needs to be specified: ‘What it is like for X to eat olives/drink gin/be drunk?’ A more plausible equivalence is between ‘what it’s like to be X’ and ‘what it’s like for X to exist/to be’.
 One of Lycan’s reasons for thinking that there are experiences of which the subject is unaware is that we speak of “unfelt pains” (1996: 16). See also Rosenthal 1997b, 731-733; 2002b: 411-412.
 Dretske does say (in effect) that anyone who knows what green is knows what it’s like to see green (1995: 81-93), which might suggest a Lycanian “lower-order” interpretation. But Dretske’s point is that knowing what green is suffices (given a certain conceptual sophistication) to know that seeings of green represent green—that is what such experiences are like. See also Dretske 1999.
 It appears that, for Lycan, only the lower-order sense of ‘what it’s like’ is a special technical one; the higher-order sense is just the ordinary sense. In support of this interpretation, notice that Lycan himself uses ‘what it’s like’ in explaining the higher-order sense (see the fourth paragraph in section 3.2 above).
 See also Lycan this volume, and the following comment on Siewert 1998:
My theory of...consciousness is, in brief: (1) Phenomenal/qualitative character is a combination of one or more (Lewis-) qualia and some other components...(2)...(3)...(4)...If sensory experience (and Siewert’s Phi-consciousness) presuppose awareness, something further needs to be added.
(5) I add...awareness!... (Lycan 2001b: section 5)
 See, e.g., Rosenthal 1986, 1997b, 2002b. I should emphasize that this paper simply ignores the usual arguments that Rosenthal gives for his “theory of consciousness”, none of which turn on the phrase ‘what it’s like’.
 For an excellent discussion of this example, see Lycan and Ryder 2003. See also Rosenthal 2002b: 407.
 See Carruthers 2000: 168-79.
 It is assumed
here that to “know what one’s experience is like”
is to have propositional knowledge (see Lycan 1996: 92-4). For the
contrary view, see Lewis 1990.
The addition of ‘in the sense relevant to phenomenal consciousness’ is meant to accommodate the context-sensitivity of ‘know what one’s experience is like’, not to suggest any ambiguity. One may use a cerebroscope to know what one’s experience is like (in various neural respects)—that does not imply that one’s experience is phenomenally conscious.
Armstrong, D. M. (1981). What is consciousness? In D. Armstrong, The Nature of Mind and Other Essays (55-67). Ithaca, NY: Cornell University Press.
Block, N. (1990). Inverted earth. Philosophical Perspectives, 4, 53-80.
Block, N. (1995). On a confusion about a function of consciousness. Behavioral and Brain Sciences, 18, 227-247.
Block, N. (1997). Biology versus computation in the study of consciousness. Behavioral and Brain Sciences, 20, 159-166.
Block, N. (2001). Paradox and cross purposes in recent work on consciousness. Cognition, 79, 197-219.
Byrne, A. (1997). Some like it HOT: consciousness and higher-order thoughts. Philosophical Studies, 86, 103-129.
Byrne, A. (2001). Intentionalism defended. Philosophical Review, 110, 199-240.
Carruthers, P. (1998). Natural theories of consciousness. European Journal of Philosophy, 6, 203-222.
Carruthers, P. (2000). Phenomenal Consciousness. Cambridge: Cambridge University Press.
Carruthers, P. (this volume). HOP over FOR, HOT theory.
Dretske, F. (1995). Naturalizing the Mind. Cambridge, MA: MIT Press.
Dretske, F. (1999). The mind’s awareness of itself. Philosophical Studies, 95, 103-124.
Gennaro, R. (1996). Consciousness and Self-Consciousness. Philadelphia, PA: John Benjamins.
Gennaro, R. (this volume.) Higher-order thoughts, animal consciousness, and misrepresentation: a reply to Carruthers and Levine.
Johnston, M. (1992). How to speak of the colors. Philosophical Studies, 68, 221-63.
Levine, J. (2001). Purple Haze. Oxford: Oxford University Press.
Lewis, D. (1990.) What experience teaches. In W. Lycan (Ed.), Mind and Cognition (499-519). Oxford: Blackwell.
Lycan, W. G. (1996). Consciousness and Experience. Cambridge, MA: MIT Press.
Lycan, W. G. (1998). In defense of the representational theory of qualia (replies to Neander, Rey and Tye). Philosophical Perspectives, 12: 479-487.
Lycan, W. G. (1999a). A response to Carruthers’ ‘Natural Theories of Consciousness’. Psyche 5 <http://psyche.cs.monash.edu.au/v5/psyche-5-11-lycan.html>.
Lycan, W. G. (1999b). Dretske on the mind’s awareness of itself. Philosophical Studies, 95, 125-33.
Lycan, W. G. (2001a). A simple argument for a higher-order representation theory of consciousness. Analysis, 61, 3-4.
Lycan, W. G. (2001b). Have we neglected phenomenal consciousness? Psyche 7 <http://psyche.cs.monash.edu.au/v7/psyche-7-03-lycan.html>.
Lycan, W. G., & Z. Ryder (2003). The loneliness of the long-distance truck driver. Analysis, 63, 132-136.
Lycan, W. G. (this volume). The superiority of HOP to HOT.
Milner, A. D., & M. A. Goodale. (1995). The Visual Brain in Action. Oxford: Oxford University Press.
Nagel, T. (1974). What is it like to be a bat? Philosophical Review 83, 435-450.
Neander, K. (1998). The division of phenomenal labor: A problem for representational theories of consciousness. Philosophical Perspectives, 12, 411-434.
Rosenthal, D. (1986). Two concepts of consciousness. Philosophical Studies, 49, 329-359.
Rosenthal, D. (1997a). Phenomenal consciousness and what it’s like. Behavioral and Brain Sciences, 20, 156-157.
Rosenthal, D. (1997b). A theory of consciousness. In N. Block, O. Flanagan, & G. Güzeldere (Eds.), The Nature of Consciousness (729-53). Cambridge, MA: MIT Press.
Rosenthal, D. (1999). The colors and shapes of visual experiences. In D. Fisette (Ed.), Consciousness and Intentionality (137-69). Dordrecht: Kluwer.
Rosenthal, D. (2002a). How many kinds of consciousness? Consciousness and Cognition, 11, 653-665.
Rosenthal, D. (2002b). Explaining consciousness. In D. Chalmers (Ed.), Philosophy of Mind (406-421). Oxford: Oxford University Press.
Rosenthal, D. (forthcoming). Sensory qualities, consciousness, and perception. In D. Rosenthal, Consciousness and Mind. Oxford: Oxford University Press.
Shoemaker, S. (1981). The inverted spectrum. Journal of Philosophy, 74, 357-381.
Siewert, C. (1998). The Significance of Consciousness. Princeton, NJ: Princeton University Press.
Smith, A. D. (2002). The Problem of Perception. Cambridge, MA: Harvard University Press.
Tye, M. (1995). Ten Problems of Consciousness. Cambridge, MA: MIT Press.