Is ‘Consciousness’ Ambiguous?
(appears in Journal of Consciousness Studies 8(2), 2001, 19-44)
Michael V. Antony
Department of Philosophy
University of Haifa
Haifa 31905, Israel
Abstract: It is widely assumed that ‘consciousness’ (and its cognates) is multiply ambiguous within the consciousness literature. Some alleged senses of the term are access consciousness, phenomenal consciousness, state consciousness, creature consciousness, introspective consciousness, self consciousness, to name a few. In the paper I argue for two points. First, there are few if any good reasons for thinking that such alleged senses are genuine: ‘consciousness’ is best viewed as univocal within the literature. The second point is that researchers would do best to avoid the semantics of ‘consciousness’, since resorting to “semantic ascent” typically serves no clear purpose in the case of consciousness, and confuses matters more than anything else.
…if we always insisted on precise definitions we all would be speechless almost all the time. Definitions and precise theoretical constructs are the final product, not the starting point of inquiry.
—Lawrence Weiskrantz (1988, p. 183)
As anyone acquainted with the consciousness literature well knows, it is common for works on consciousness to contain—usually somewhere near the start—a discussion of the semantics of the word ‘consciousness’ (and its cognates: ‘conscious’, ‘consciously’, etc.). And in spite of there being little widespread agreement among researchers on most matters falling within the domain of consciousness studies, when it comes to ‘consciousness’ it is virtually universally agreed is that the term is multiply ambiguous. The idea, moreover, is not merely that lay people use the term with various meanings, or that they use it differently from how researchers do, or that the term has had different meanings throughout its history. It is that researchers themselves currently use the term with several distinct meanings or senses. Some examples of such alleged senses are these:
(1) Block’s (1991; 1993; 1994; 1995a,b) phenomenal consciousness, access consciousness, monitoring consciousness, reflective consciousness, self consciousness.
(2) Rosenthal’s (1986; 1993a,b; 1997; forthcoming) state consciousness, creature consciousness, transitive consciousness, introspective consciousness.
(3) Armstrong’s (1981) minimal consciousness, perceptual consciousness, introspective consciousness.
(4) Tye’s (1995) higher-order consciousness, discriminatory consciousness, responsive consciousness, phenomenal consciousness.
(5) Nelkin’s (1989; 1995; 1996) C1, C2, CS.
The expressions in (1) - (5), and many others beside them, are often taken to express what the word ‘consciousness’ alone expresses on various occasions of its use; that is, they are construed as synonymous with different senses of ‘consciousness’.
The practice of addressing the semantics of ‘consciousness’ in works on consciousness is in many ways odd. Lists of the term’s senses vary considerably from theorist to theorist (though there is also much overlap), and in many cases researchers offer their own idiosyncratic lists with little constraint and without defense or criticism of others’ proposals. However if semantics is to be taken seriously—as a branch of linguistics say—one cannot simply choose a term’s meaning(s) in a natural language. Semantics is not that easy! What should also give one pause is how poorly motivated such terminological discussions often are. Their relevance to the works that contain them is frequently unclear at best, and in most cases any important points can be made simply by employing the word ‘consciousness’ and making distinctions about consciousness when deemed necessary, but ignoring semantics entirely. What gets said about the term’s alleged senses, moreover, is often either highly controversial or unilluminating—all of which suggests that there is something rather amiss with the practice. This will become clearer as we proceed.
In what follows I argue for two main points. The first—which occupies us for most of the paper, and involves an excursion into linguistic matters—is that not only is it not obvious which are the multiple meanings of ‘consciousness’ within the literature, but it is also questionable whether the term as currently used by consciousness researchers is ambiguous at all. While it will turn out that the opposite thesis—that ‘consciousness’ is univocal, or has a single meaning, within the literature—is also open to doubt, I shall suggest that that is the rational default position to adopt. My answer to the question in the title—‘Is ‘consciousness’ ambiguous?’—therefore, will be a qualified ‘no’. I call the thesis that ‘consciousness’ is ambiguous within the current consciousness literature the ambiguity view, and the thesis that ‘consciousness’ is univocal within the literature the univocality view. Second, I shall argue that there are few if any good reasons for researchers to concern themselves with the semantics of ‘consciousness’, and a number of reasons not to, the primary one being that doing so generates considerable confusion. The philosopher W. V. Quine advocated resorting to “semantic ascent” (Quine 1960, §56): focusing on expressions used to refer to philosophically perplexing phenomena (like ‘truth’ and ‘knowledge’) instead of the phenomena themselves (like truth and knowledge). In this paper I urge semantic descent for ‘consciousness’; even semantic dissent.
Homonymy and Polysemy
In arguing for the second point—that researchers should avoid the semantics of ‘consciousness’ wherever possible—part of my strategy will be to show how complex many of the issues surrounding ambiguity are. To that end I introduce a distinction commonly made by linguists between two kinds of ambiguity: homonymy and polysemy. Though the distinction is an intuitive one that lay people make naturally (see Lehrer, 1974; and Panman, 1982), it is has proven remarkably difficult to ground theoretically. My main reason for discussing the distinction, however, is connected with the first point to be argued for, that ‘consciousness’ is univocal. In examining cases of polysemy one quickly discovers an ill-understood borderline region between polysemy and univocality, and hence also between ambiguity and univocality. There are many expressions—‘consciousness’ included, I shall argue—that have associated with them a kind of semantic variability that is explicable, prima facie at least, both in terms of polysemy and univocality. Seeing that sets the stage for an account of ‘consciousness’ as univocal.
Intuitively, homonymous words are those whose meanings are “unrelated” to each other, like the bark of a tree and a dog’s bark; and polysemous words are those whose meanings are more closely related, like the verb ‘open’ with its senses concerning unfolding, expanding, revealing, making openings in, and so on. A word can be both polysemous and homonymous: the adjective ‘right’ has several senses concerning correctness and righteousness, but also senses concerning the right-hand side (Hirst 1987, p. 5). Within lexical semantics, homonyms are often treated as distinct lexical entries (lexemes), or words, in the mental lexicon; and cases of polysemy are taken as distinct senses associated with a single lexical entry.
As was said, making sense theoretically of the homonymy-polysemy distinction has proven extraordinarily difficult; indeed, all attempts I have encountered fail. The difficulty becomes apparent when one reflects on what might be involved in clarifying the distinction between “related” and “unrelated” meanings. Even the two senses of ‘bark’, after all, are somewhat related—both concern living things—so the challenge of providing a principled account of which are the right relations for polysemy, and how many are sufficient, is bound to be formidable. The problems with the homonymy-polysemy distinction notwithstanding, if we assume that the distinction is real, and ask in which category the alleged senses of ‘consciousness’ belong, the answer seems clear: they are cases of polysemy if ambiguity at all. For the alleged senses are closely interrelated; at the very least they are all intimately related to the mind. They are much more like ‘open’ than ‘bark’. It is polysemy in particular, therefore, that will concern us in what follows.
Within the linguistics literature on ambiguity, not only is getting clear about the homonymy-polysemy distinction problematic, so is understanding the nature of the boundary between polysemy and univocality. Pustejovsky (1996, pp. 31-32) offers the following cases (among others) as examples of polysemy. (The words in square brackets help fix the intended senses of the italicized terms.)
(6) a. Mary broke the bottle. [container]
b. The baby finished the bottle. [contents]
(7) a. The window is rotting. [frame]
b. Mary crawled through the window. [opening]
(8) a. The newspaper fired its editor. [company, organization]
b. John spilled coffee on the newspaper. [product]
Some readers may doubt whether (6) – (8) involve real ambiguity, and I think there is room for doubt. But in any event Pustejovsky offers a further example the status of which he himself is uncertain:
(9) a. a good car
b. a good meal
c. a good knife
Here ‘good’ seems to point to different properties of the car, meal, and knife, and so in that sense may appear to involve distinct meanings. On the other hand, it might be thought that in each case ‘good’ simply expresses a positive evaluation of some sort, and so is univocal. If one takes that route, one ignores prima facie subtle shifts in meaning. If such differences are acknowledged, one risks saddling ‘good’ with an open-ended number of senses, since the cases can be multiplied indefinitely: a good conversation, a good breeze, a good friend, etc. This is just one example, among many others, suggesting that the boundary between polysemy and univocality is also theoretically problematic.
I now approach the polysemy-univocality divide from the other side, as it were—from a phenomenon many theorists of language take to involve univocality, in spite of its superficial resemblance to polysemy. Following the lexical semanticist D. A. Cruse (1986), I call that phenomenon modulation; and in this section I explain what modulation is. That done, I argue in later sections that the intuitions about the semantic variability of ‘consciousness’ that many take to support the ambiguity view are equally well explained in terms of modulation—and hence in terms of univocality.
Cruse’s (1986) discussion of modulation is brief. Forgoing the homonymy-polysemy distinction in his theory (p. 80), Cruse contrasts modulation with ambiguity simpliciter. He writes:
a single sense can be modified in an unlimited number of ways by different contexts, each context emphasizing certain semantic traits, and obscuring or suppressing others….This effect of a context on…[a] lexical unit will be termed ‘modulation’… (p. 52)
He offers two examples:
First, some part of an object (or process, etc.) may be thrown into relief relative to other parts. For instance, ‘The car needs servicing’ and ‘The car needs washing’ highlight different parts of the car [e.g., the engine and exterior, respectively]….Second, it is commonly the case that what is highlighted…is an attribute, or range of attributes….For instance, ‘We can’t afford that car’ highlights the price of the car, ‘Our car couldn’t keep up with his’ highlights its performance… (p. 53)
Cruse’s first example can be supplemented with further cases:
(10) The car needs painting. [exterior—excluding chrome, glass, wheels, etc.]
(11) The car needs vacuuming. [floor, mats, seats, etc.]
(12) The car needs loading. [trunk, back]
(13) The car needs filling. [gas tank]
Note that according to Cruse it is a single sense that gets modulated in an unlimited number of ways across contexts, so no ambiguity is involved.
Other theorists have pointed to similar phenomena. They have been explored most extensively perhaps by the philosopher of language Charles Travis, according to whom
Given words may have any of many semantics, compatibly with what they mean. Words, in fact, vary their semantics from one speaking of them to another. In that case, their semantics on a given speaking cannot be fixed simply by what they mean. The circumstances of that speaking, the way it was done, must contribute substantially to that speaking. (Travis 1997, p. 100, my emphasis)
One of Travis’s examples involves the sentence
(14) The leaves are green.
and a story:
Pia’s Japanese maple is full of russet leaves. Believing that green is the colour of leaves, she paints them. Returning, she reports, “That’s better. The leaves are green now.” She speaks a truth. A botanist friend then phones, seeking green leaves for a study of green-leaf chemistry. “The leaves (on my tree) are green,” Pia says. “You can have those.” But now Pia speaks falsehood. (p. 89)
According to Travis, the story illustrates two ways in which it can be said of leaves that they are green. However he argues that ‘are green’ is not ambiguous between one sense that is true of leaves painted green, and another sense that is false of leaves merely painted green. For there is no reason to suppose there is any limit to the ways in which (14) can be understood. Imagine a fluorescent green mould growing on the leaves. Does that correspond to a third sense of (14)? Presumably not. In any event, even if (14) does involve ambiguity, that cannot help, since the phenomenon infects language to its core:
If ‘green’ has, say, thirteen senses, there are, for each of them, various possible (and invokable) understandings of what it would be for leaves to be green in that sense. If so, then ambiguity is not a way of avoiding the present conclusion. (p. 90)
To illustrate, consider the suggested “paint counts” sense of ‘are green’, and
…suppose [the leaves]…are painted, but in pointillist style: from a decent distance they look green, but up close they look mottled. Is that a way of painting leaves green? It might sometimes, but only sometimes, so count. (p. 90)
So ambiguity, it seems, cannot account for the kind of semantic variability under discussion. Nor, Travis argues, can proposals that appeal to ellipsis, conversational implicature, functions from parameters involved in speaking words to what is said on such speakings, and so on.
In the rest of this paper I use Cruse’s term ‘modulation’ to refer to the general phenomenon that both Cruse and Travis discuss: a kind of open-ended semantic variability relative to a fixed meaning. It is important to note, however, that the phenomenon is broader than what is suggested by Cruse’s ‘car’ example. In that example different parts or attributes of the same object with its fixed properties are highlighted across contexts. In Travis’s examples, in contrast, the leaves’ properties vary across contexts (painted with green paint in one context, painted with non-green paint in pointillist style in another, covered with green mould in another). But that difference hardly seems crucial. One could imagine some of those properties—say, being covered with green paint, being covered with green mould, and containing chlorophyll—as co-instantiated in the same leaves. Various of those properties could then be highlighted across speakings. Now once one has gone that far, modulation can also be recognized where properties and objects vary across speakings. For example,
(15) That is green
said of (a) a normal leaf containing chlorophyll, (b) a car painted green, and (c) a rock covered with green mould, involve the very same modulations of ‘is green’ as do Travis’s examples. Moreover, this seems to be precisely what is happening in Pustejovsky’s ‘good’ example:
(9) a. a good car
b. a good meal
c. a good knife
In one case the tastiness of the meal is highlighted, in another the sharpness and durability of the knife, and so on. As I use the term, then, modulation can occur either where (a) both object and properties remain fixed across contexts, (b) only properties vary, or (c) both object and properties vary.
Below I argue that the purportedly distinct senses of ‘consciousness’ within the consciousness literature are interpretable as modulations of a single sense of ‘consciousness’, thus making the univocality view a serious contender alongside the ambiguity view. For modulation to explain the relevant semantic phenomena, however, it must be antecedently plausible that there is some single meaning of ‘consciousness’ that gets modulated across contexts. It seems, however, that there is a general sense of the term that is widely used among researchers—a sense, incidentally, that is acknowledged by many adherents of the ambiguity view. It is that sense, or something close to it, that I believe plausibly gets modulated across contexts.
To see that there is a general sense of ‘consciousness’ that enjoys widespread use, consider titles containing the word ‘consciousness’ that apply to a broad range of theoretical and methodological approaches—like ‘Journal of Consciousness Studies’ or ‘Toward a Science of Consciousness’. Both titles apply to topics from diverse fields: cognitive science, neuroscience, analytic philosophy, physics, phenomenology, transpersonal psychology, and so on. What is the meaning of ‘consciousness’ in those titles? A natural thing to say is that it is a general sense of the term that applies to all or most of the topics covered by the titles. Two further options, it might be thought, are these: First, the word does not apply to many of those topics. But that is absurd, surely; or at least it is false. The second is that the term is multiply ambiguous and simultaneously expresses its various senses—as might the word ‘bank’ in the title of an offbeat article about financial institutions and rivers’ edges. However, such phenomena are typically assumed by linguists not to occur. As Cruse (1986, p. 62) puts it, an ambiguous expression “resists as it were, the simultaneous activation of more than one of its senses.” All of this should leave us feeling fairly confident that there is a general sense of ‘consciousness’ within the literature. Notice furthermore that the meaning of ‘consciousness’ in the above titles seems identical to its meaning in titles like Baars’s (1988) ‘A Cognitive Theory of Consciousness’, Crick and Koch’s (1990) ‘Toward a Neurobiological Theory of Consciousness’, Chalmers’s (1995) ‘Facing Up to the Problem of Consciousness’, and so on for numerous others. And occurrences of ‘consciousness’ within such works typically have the same meaning as in the works’ titles. The general sense of ‘consciousness’ thus begins to appear pervasive indeed.
To show that modulation can explain the semantic phenomena others take to support the ambiguity view, it must be shown for each alleged sense of ‘consciousness’ that modulation indeed accounts for the relevant phenomena. Due to the numerous proposed senses of ‘consciousness’, however, treating each individually is impractical. I shall thus focus on the well-known distinctions of Ned Block and David Rosenthal, and then deal with the remaining purported senses en masse.
In an important series of papers, Ned Block distinguishes two concepts of consciousness: phenomenal consciousness (P) and access consciousness (A). And those he distinguishes from monitoring consciousness, reflective consciousness, and self consciousness, each of which I shall touch on later. On phenomenal consciousness Block (1994) writes:
There is no non-circular definition to be offered; the best that can be done is the offering of synonyms, examples and one or another type of pointing to the phenomenon….I used as synonyms ‘subjective experience’ and ‘what it is like to be us’. In explaining phenomenal consciousness, one can also appeal to conscious properties or qualities, e.g. the way things seem to us or immediate phenomenological qualities. Or one can appeal to examples: the way things look or sound, the way pain feels and more generally the experiential properties of sensations, feelings and perceptual experiences. I would also add that thoughts, wants and emotions often have characteristic conscious aspects… (p. 211).
In contrast, access consciousness is a ‘cognitive or intentional or functional’ concept of consciousness (p. 213). A state is access conscious if
in virtue of having the state, a representation of its content is (a) inferentially promiscuous, i.e. freely available as a premise in reasoning, and (b) poised for rational control of action and (c) poised for rational control of speech. (p. 214)
What is Block’s view of the semantics of ‘consciousness’? As noted earlier, he takes there to be a widely used general or undifferentiated sense of the term. However he also takes P and A to correspond to more specific, distinct senses of ‘consciousness’. While it is unusual for a term to have both general and specific senses, such cases do exist. The word ‘dog’ has a general sense canine and a more specific sense male canine, so something similar might be true of ‘consciousness’. On whether P and A are distinct senses of ‘consciousness’, Block’s opinion has shifted. Early on (1991, p. 670) he held that ‘consciousness’ is ambiguous between P and A. Later (1994, pp. 216-17) he maintained that it is not ambiguous but suggested that it should be. At that point he took the general, undifferentiated concept to be a cluster concept. Finally (1995a, p. 236) he returned to the position that ‘consciousness’ is ambiguous, and took the concept to be a “mongrel” concept. Block explained:
To call consciousness a mongrel concept is to declare allegiance to the ambiguity theory, not the cluster theory. An ambiguous word often corresponds to an ambiguous mental representation, one that functions in thought as a unitary entity and thereby misleads. These are mongrels. (1995a, p. 236)
Block’s primary reason for adopting the ambiguity view in the end, it appears, was that he thought it would best serve his aim of exposing certain fallacies. In such fallacious reasoning a function is ascribed to consciousness on the basis of cases in which both the function and consciousness are absent. (See §6 of Block 1995a.) Block maintains that in such instances both P and A are typically missing, but researchers, conflating P and A, jump to conclusions about the function of P—which is a fallacy. Block settled on the ambiguity view with respect to P and A because, as he says, ambiguous words can mislead. Equivocation among senses of ‘consciousness’ thus presents itself as a natural explanation for the conflation underlying the fallacious reasoning.
My main concern here is with whether P and A correspond to distinct senses of ‘consciousness’ and not with the fallacies Block discusses. However, given that his primary reason for treating ‘consciousness’ as ambiguous is to account for the conflation underlying the fallacies, it is worth inquiring whether the conflation can be explained by other means. It seems it can. All that is needed are distinct mental representations, or distinct “elements” of complex representations, such that it becomes possible to unknowingly slide from one representation (or representational element) to another. While ambiguity suffices for that, it is unnecessary, since mental representations or concepts can be cut more finely than word meanings. Consider common sense and scientific concepts of water. Since they are different concepts, conflation is in principle possible, in spite of the fact that ‘water’ is univocal. One might imagine, for example, a careless chemist who conflates the common sense and scientific concepts of water, and fallaciously ascribes to a lay person a technical belief about water that the person lacks.
Conflation is also possible where there is modulation. Consider how the semantics of ‘full’ subtly shifts across contexts:
(16) a. a full bookshelf [no room across the shelf]
b. a full auditorium [all seats occupied]
c. a full balloon [stretched to its limit by a gas or liquid inside it]
d. a full swimming pool [all its volume occupied by water]
e. a full swimming pool [contains a maximum number of people, as determined by safety regulations, comfort, etc.]
d. and e. furnish us with a way of showing how modulation can result in fallacious reasoning based on conflation. Imagine a swimming pool that was scheduled to open on June 1, but by that date contained no water (and hence no people). And suppose an employee at the pool overheard the manager complaining on June 1 about the pool not being full. Finally, imagine the employee telling a friend how the manager was upset because the pool was not filled with swimmers on opening day, whereas all the manager cared about is that it be filled with water. Here both “kinds of fullness” are missing from the pool, and conflating the two the employee fallaciously infers that the manager was upset because the pool was not filled with people. This closely parallels the fallacious reasoning about consciousness that concerns Block. However the example shows that such reasoning, as well as the conflation on which it is based, can occur in the absence of ambiguity, since ‘full’ is univocal.
As a rule, then, ambiguity is unnecessary for explaining fallacies based on conflation. Block’s reason for treating ‘consciousness’ as ambiguous is thus undermined. But is ‘consciousness’ ambiguous between P and A? After all, even if Block’s argument for ambiguity is wanting, his conclusion may be true; and certainly others have followed Block in treating P and A as distinct meanings of ‘consciousness’. It is far from clear, however, that P and A are distinct senses of ‘consciousness’, since a plausible modulation-based account is available. The story would go something like this. Corresponding to the word ‘consciousness’ we have a rich conception of a complex and multi-faceted mental phenomenon. This phenomenon appears to comprise (something like) states and events with phenomenal or qualitative aspects that are temporally located, that enter into part-whole relations, that are related to intentional features of the mind, that enter into causal relations with one another and with unconscious mental states and events, and so on. This complex phenomenon seems intimately linked to wakefulness, attentiveness, and capacities for “inner reflection”—though perhaps none of those are necessary. Something like that, and much more, goes to make up our complex conception or picture of consciousness. If that is right, then it makes sense that given our varied theoretical and folk psychological interests, different features of the phenomenon on different occasions will be thrown into relief or emphasized—sometimes phenomenal features, sometimes functional or cognitive ones, etc.—while others are suppressed, just as with Cruse’s ‘car’ example. But just as ‘car’ is univocal, so is ‘consciousness’ on this story.
There is another reason for doubting that P and A are senses of ‘consciousness’. With genuine cases of ambiguity there is typically considerable agreement that ambiguity is indeed involved. However such agreement is sorely lacking with P and A. There is much controversy in the literature over whether A and P are distinct senses of ‘consciousness’, whether A is a kind of consciousness at all, whether P just is A, whether there can be A without P or P without A, etc. All of that makes it very far from obvious, and indeed highly theoretical, whether two meanings of the English word ‘consciousness’ are involved. All told, therefore, the modulation story with respect to P and A seems an attractive one. At least it has as much going for it as the ambiguity view, and perhaps a fair bit more.
Another philosopher whose work has generated much discussion about different senses of ‘consciousness’ is David Rosenthal (see, e.g., Rosenthal, 1986; 1993a,b; 1997; forthcoming). Rosenthal distinguishes state consciousness, creature consciousness, transitive consciousness, and introspective consciousness. He speaks only rarely of senses or meanings of ‘consciousness’, although he does write in one place that the “term ‘conscious’…conceals a certain ambiguity” (1993b, p. 197). Usually, however, he is careful to say only that we refer or apply the term to different phenomena or properties in our talk about consciousness. In spite of Rosenthal’s reluctance to enter into the semantics of ‘consciousness’, other writers frequently take him to have distinguished different senses of the term (for example, Dretske, 1993, p. 269; Block, 1995a, p. 232; Lycan, 1996, pp. 2-3; Güzeldere, 1997, p. 9). For that reason it is worth arguing that his distinctions (and certain variations thereof) need not be seen as corresponding to distinct senses of ‘consciousness’.
I begin with Rosenthal’s distinction between state consciousness and creature consciousness:
In one use [of the term ‘conscious’], we speak of mental states as being conscious or not conscious. Mental states…are conscious if we are aware of them in some intuitively immediate way. But we also apply the term ‘conscious’ to the creatures that are in those mental states. A creature’s being conscious consists, roughly, of its being awake and being mentally responsive. Plainly, this property of being conscious is distinct from the property a mental state may have. (1993b, p. 197)
Consider, first, creature consciousness. If we take seriously Rosenthal’s insistence that wakefulness (and mental responsiveness) is necessary and sufficient for creature consciousness, it is doubtful that any such sense of ‘consciousness’ is used within the literature. Most researchers, after all, would treat creatures that are vividly dreaming as having conscious states, and hence as being conscious; and it is hard to imagine many consciousness researchers counting creatures that have a sleep-wake cycle but lack the capacity ever to be in conscious states as conscious creatures, as does Rosenthal. So Rosenthal’s creature consciousness appears not to correspond to a meaning of ‘consciousness’ within the literature. Among lay people’s uses of ‘consciousness’, on the other hand, there are uses that are closely tied to wakefulness—for example, when one who has just awoken is said to be not yet quite conscious. But it would take much work to show that modulation cannot account for such uses: given that wakefulness frequently accompanies consciousness, it may simply be that wakefulness has been highlighted in such cases. In any event, such uses of ‘conscious’ are largely irrelevant to consciousness research (except perhaps in anesthesiology).
Other theorists, however, have interpreted Rosenthal’s notion of creature consciousness differently. Güzeldere (1997, p. 9), for example, says that the “creature sense of consciousness denotes an overall state one is in…[as opposed to] states…of one type or another.” Perhaps that might be thought a more plausible candidate for a sense of ‘consciousness’ within the literature. But consider the word ‘corrupt’ in the expressions ‘corrupt government’ and ‘corrupt politician’. ‘Corrupt’ refers to distinct properties when applied to organizations and individuals no less than does ‘consciousness’ when applied to creatures and states. But it surely does not follow that ‘corrupt’ is ambiguous between “individual corruption” and “organizational corruption.” A more plausible view is that ‘corrupt’ is univocal, and modulation is taking place, with different properties being highlighted across contexts. And so similarly with ‘consciousness’ vis-à-vis states and creatures. Furthermore, will there also be state and creature happiness, fear, ecstasy, worry, etc.—since these (or minor variations of them) can be predicated both of states and of creatures? Are all such psychological terms to be deemed ambiguous in a parallel fashion? Modulation seems a much more economical and hence plausible view.
I turn now to transitive consciousness. Rosenthal (1993a) writes:
When a creature senses something or thinks about some object, we say that the creature is conscious of that thing. A full description of a creature’s being conscious of something always involves mentioning the thing the creature is conscious of. So it is natural to call this property transitive consciousness. (p. 355)
If we stick closely to what Rosenthal says about transitive consciousness, then, as with creature consciousness, it seems he is not characterizing a sense of ‘consciousness’ within the literature. Instead it seems close to a non-technical, everyday use that means something like knowledge (“I’m conscious of the fact that the road ahead is long”), where such knowledge states need not be experiential at all. In fact, according to Rosenthal, something even weaker is required for transitive consciousness—mere mental representation: “We are transitively conscious of something by virtue of being either in an intentional or a sensory state whose content is directed upon that thing” (forthcoming, p. 5). And such intentional or sensory states can be unconscious: “transitive consciousness can occur without intransitive state consciousness” (1997, p. 737). Indeed, as Rosenthal points out (1997, p. 737), it is only because transitive consciousness need not involve conscious states that he can attempt to construct a non-circular explanation of state consciousness in terms of it. But clearly transitive consciousness so understood is not something researchers mean by ‘consciousness’. Any property that can be instantiated by creatures incapable of being in conscious states is not a property commonly expressed by ‘consciousness’ within the literature.
As with creature consciousness, one might depart from the details of Rosenthal’s discussion of transitive consciousness. Specifically, one might reject Rosenthal’s claim that transitive consciousness is consistent with the absence of state consciousness, and just focus on the intentional aspect—the fact that we sometimes speak of being conscious of things. Does that suffice for a distinct sense of ‘consciousness’? Notice first that we attribute transitive consciousness to creatures, not states. (John is conscious of Mary, but none of John’s mental states are conscious of Mary.) Next, put aside the reading of creature consciousness that is consistent with the absence of state consciousness, and focus instead on Güzeldere’s notion that applies to any creature in an overall experiential state. In that case there are not two phenomena—creature consciousness and transitive consciousness—just one; and the difference between the two uses of ‘conscious’ is that the transitive use refers to a particular aspect of the phenomenon: its representational content. In a similar vein, “John is worried” and “John is worried about Mary” refer to only one kind of worry, and involve only one sense of ‘worry’. It is just that the second use of the term is followed by an explicit reference to an aspect of the phenomenon (its content) that the first use ignores. Now if transitive consciousness so understood does not differ from creature consciousness (as Güzeldere construes the latter), then it is far from obvious that transitive consciousness and state consciousness correspond to distinct senses of ‘consciousness’, since, as was argued above, state consciousness and creature consciousness plausibly do not.
Consider finally Rosenthal’s introspective consciousness. Introspective and non-introspective consciousness, Rosenthal writes, are “two distinct forms” of state consciousness. A state is introspectively conscious when the “state is conscious and the way we are conscious of it is attentive and deliberate.” A conscious state is not introspectively conscious “when we are conscious of it in the casual, unattended way characteristic of ordinary conscious states”(1993a, p. 356). Rosenthal’s notion of introspective consciousness can be dealt with quickly, for in this case we are dealing with a kind or species of consciousness. As I have argued elsewhere (Antony 2000), such genus-species differences are typically not reflected in distinct senses of the general term. (The ‘dog’ example discussed earlier is an exception that proves the rule.) Just as Granny Smith apples and Macintosh apples do not introduce new senses of ‘apple’, there is no reason to think that introspective consciousness (visual consciousness, etc.) introduces a new sense of ‘consciousness’.
Other Purported Senses
There are several other purported senses of ‘consciousness’ within the literature aside from Block’s and Rosenthal’s, too many to be addressed individually. Accordingly, they must be treated collectively, and superficially. I shall take most of them as instances of a broad class of proposals involving a “two-tiered” structure of entities (typically but not always, mental entities) with an intentional relation between them. For each alleged sense, I shall argue that either it is not a plausible candidate for a meaning of ‘consciousness’, or it can be explained in terms of modulation.
First, however, I discuss a way in which some might be misled into erroneously thinking that ‘consciousness’ is ambiguous. The literature on consciousness contains many theories: theories of consciousness, theories of kinds or aspects of consciousness, theories of how the concept consciousness is to be analyzed, or the word ‘consciousness’ defined, and so on. One must guard against inferring too hastily that such theories introduce new senses of ‘consciousness’. For any such theory must presuppose a meaning of the term. After all, the theories are theories of consciousness (or of the concept consciousness or the meaning of ‘consciousness’), in some pre-existing sense of that term. I mention this because there are many two-tiered proposals of this sort. One is Rosenthal’s higher-order thought (HOT) theory, which is a theory of what (state) consciousness is, and so presupposes, rather than introduces, a meaning of ‘consciousness’. Similarly for “inner perception” theories like Armstrong’s (1981) that are intended as theories of consciousness. These and several other accounts in the literature can be put aside as irrelevant to the semantics of ‘consciousness’ for this reason.
On a related point, notice that so-called “stipulative definitions” do not in themselves issue in new meanings for natural language expressions. If a theorist proclaims, “In this paper I shall mean by ‘consciousness’…,” filling in the details of some idiosyncratic use, the English word ‘consciousness’ does not thereby acquire a new sense, not even a new “technical” sense—just as it would not merely because one had stipulated that ‘consciousness’ means milkshake. A stipulative definition is little more than a useful shorthand that is temporarily employed in a restricted linguistic setting. For a term to acquire a new meaning within a natural language, in contrast, at a minimum, a sustained (sub-)community-wide practice or use is required.
Let us turn now to some more plausible candidate senses of ‘consciousness’ within the literature that involve two-tiered structures—to proposals about phenomena that researchers purportedly use ‘consciousness’ to pick out. There are many such proposals. Tye (1995), for example, speaks of higher-order consciousness, or H-consciousness, as one of the concepts of consciousness we employ. He also calls this ‘introspective consciousness’. A state is H-conscious if there is a higher-order thought that one is in that state. Block’s (1993) reflective consciousness and Nelkin’s (1995; 1996) apperceptive consciousness (or C2) amount to much the same thing. Related two-tiered accounts are those in which inner perception is substituted for higher-order thought. While it is often difficult to know whether an account is intended as a theoretical proposal about consciousness in some pre-existing sense of the term, or as a proposal about a meaning of ‘consciousness’, Lycan’s (1997) account, as well as Locke’s (1690) famous statement, “Consciousness is the perception of what passes in a man’s own mind,” might be taken as involving claims about the meaning of ‘consciousness’.
Two-tiered accounts differ in other ways as well. Sometimes they move in the direction of greater generality, as with Block’s (1994; 1995a) monitoring consciousness, which covers both HOT and inner-perception theories. Or they tend toward greater specificity: Armstrong’s (1981) introspective consciousness (unlike Tye’s) not only requires higher-order thought, but also “careful attentive scrutiny of our own state of mind” (p. 63). Indeed, for Armstrong, the self enters into the introspected contents: “Introspective consciousness is consciousness of self.” (p. 65). For other theorists, like Block (1994; 1995a), the self makes its appearance only with self consciousness. Lycan (1996, p. 4), also discussing self consciousness, distinguishes two types, one of which is “having a sense of oneself as an individual separate from other individuals and the rest of one’s surroundings. In a stronger sense, it is consciousness of one’s self.”
There is a use of ‘consciousness’ from the beginning of the twentieth century that can also be viewed as involving a two-tiered picture of sorts, just that on this use consciousness is identified with the upper-level tier. Any contents of consciousness—objects, qualia, sense data, etc.—are not parts or features of consciousness, but simply what consciousness is conscious of. It is this use that enabled William James (1996) to deny the existence of consciousness while continuing to posit experience. (There is nothing that has the contents; there is just experience itself.) And it is because G. E. Moore (1903) took consciousness to be distinct from its contents that he thought he could refute idealism. A somewhat related view comes out of Eastern spiritual traditions such as Buddhism and Yoga, and involves a phenomenon called ‘pure consciousness’ (see Forman, 1990). The idea is that under certain special circumstances—for example, as a result of practicing meditation—there can be contentless conscious experiences, or pure consciousness. In such cases the claim is, as it were, that only one of the two tiers remains.
What is to be made of these various two-tiered proposals? Do any involve distinct senses of ‘consciousness’? In the history of the term, there may have been two-tiered senses: James and Moore may have been using ‘consciousness’ with a different meaning in the early 1900s. However, that sense (if it is one) is not currently in use. There also may be two-tiered senses in current, everyday use: some lay people may think ‘consciousness’ means awareness of self, or introspection, etc. But there are no clear cases in which researchers do (as opposed to believing that consciousness is to be explained in terms of self-awareness, introspection, etc.). Of course, researchers often do employ ‘consciousness’ when speaking of states that involve self-awareness or introspection. But such cases accord just as easily with modulation as with ambiguity. Regarding pure consciousness, finally, this involves a theoretical claim about a special type or form of consciousness, and so presupposes a meaning of ‘consciousness’. I am aware of no instances in which the word ‘consciousness’ on its own (as opposed to in combination with ‘pure’) is used to mean pure consciousness. So it appears that none of the two-tiered proposals provide any special support for the ambiguity view.
Having covered most of the candidate senses of ‘consciousness’ within the literature, I end this section by addressing what might be viewed as a further source of meanings of ‘consciousness’. I have in mind discussions of theories and conceptions from other cultures or traditions in which foreign terms are translated as ‘consciousness’. Since the mind or self is often conceived very differently in foreign cultures than in the West, it might be thought that ‘consciousness’ must have a different meaning in such cases. But that cannot be so. Consider any word (or words) from a foreign tradition that has been translated as ‘consciousness’. Since the translation is into English, ‘consciousness’ must bring with it its English meaning(s)—which is why, after all, ‘consciousness’ was chosen rather than ‘mouse’ or ‘comet’. So a pre-existing meaning of ‘consciousness’ must be presupposed in any such translation. Further, once a translation is adopted, any differences among relevant beliefs of the foreign culture and our own must be viewed as differences in beliefs about consciousness in our sense of the term. If our beliefs and theirs differ so greatly that when speaking their native language they could not be referring to what we do through our uses of ‘consciousness’, then all that follows is that none of their terms are translatable as ‘consciousness’ after all. Such cases, therefore, introduce no new meanings of ‘consciousness’.
In Defense of Univocality
Although not all alleged senses of ‘consciousness’ have been discussed, I suspect that any remaining proposals either
(a) are identical or similar to those already considered,
(b) apply only outside the current literature: in the term’s history or in everyday non-theoretical usage, or
(c) are theoretical proposals that presuppose a pre-existing meaning of ‘consciousness’.
For each candidate that seemed as though it might correspond to a distinct sense of ‘consciousness’ within the literature, I attempted to show that modulation could explain the relevant semantic data. I have of course not provided conclusive arguments for modulation; that has not be my aim. My purpose has just been to show that a modulation-based account of ‘consciousness’ is prima facie plausible, and that correlatively the truth of the ambiguity view is far from obvious. As was noted at the outset, proponents of the ambiguity view typically simply assume that ‘consciousness’ is ambiguous, and offer nothing by way of defense of that assumption. And the few brief arguments that do exist (for example, in Block’s, Rosenthal’s, and Nelkin’s writings) were found wanting. I think it can be safely concluded that the univocality view is no less plausible than the ambiguity view.
That being so, which story about ‘consciousness’ ought to be embraced, ambiguity or univocality? I believe that considerations of simplicity, combined with the fact that there is nothing even approaching widespread agreement on which are the distinct senses of ‘consciousness’, decide the matter in favor of the univocality view. For the univocality view is clearly the simpler of the two: any account that ascribes one meaning to a word is simpler than any that ascribes many meanings to it (ceteris paribus, of course). So the univocality view should be adopted as the default position, at least provisionally.
In response, one might offer the following two-pronged objection: First, modulation also involves semantic variation, albeit of a kind that is consistent with an expression’s meaning remaining fixed. Why should that form of semantic variation be thought any simpler than the differences in meaning posited by the ambiguity view? Second, granted, there is little agreement among ambiguity theorists on which are the distinct senses of ‘consciousness’. However there is bound to be as much controversy with modulation, for example, regarding questions like: What is the single sense that purportedly gets modulated? Which modulations exactly are possible? What does it really mean for a fixed meaning to be “modulated”? I take the two parts of the objection in turn.
As Travis convincingly argues, modulation is ubiquitous. There is nothing exceptional about the words ‘car’, ‘green’, ‘leaves’, ‘full’, or ‘good’; examples could have been constructed from indefinitely many others. Furthermore, as was noted when discussing the suggested “paint counts” sense of ‘are green’, modulation occurs even with ambiguity: each sense is subject to modulation. It follows that even if ‘consciousness’ is ambiguous, modulation is likely to occur with each of its senses. On the univocality view, in contrast, only a single sense gets modulated. Since any proposal about one modulated sense is simpler than any about many modulated senses (again ceteris paribus), the univocality view remains the simpler of the two positions.
Turning to the second part of the objection, it is undoubtedly true that there is no significant theoretical understanding of modulation—of the principles that determine how and when and in what sense meanings are modulated. It is also true that I have said little about the general sense of ‘consciousness’ that is modulated. However, there is a marked asymmetry between the ambiguity and univocality views on this matter, one that strongly favors the univocality view: it is possible on the univocality view, but not on the ambiguity view, to ignore semantics entirely. As a result, our ignorance of the semantic details is a problem for the ambiguity view alone.
Here is why. Suppose ‘consciousness’ is ambiguous within the literature. Then anyone employing the term must specify which sense is in play; otherwise one’s audience might assume a different sense is intended. Specifying one sense of ‘consciousness’ among many, however, requires saying something about it: characterizing it to some degree, distinguishing it from the other senses, and so on. But given the vast range of opinions about which are the distinct meanings of ‘consciousness’, it is hard to see how that could be done responsibly without defending one’s characterizations and distinctions. So semantics is unavoidable. If the univocality view is assumed, on the other hand, semantics is unnecessary. Since ex hypothesi everyone means the same thing by ‘consciousness’, one can simply use the term, relying on the fact that one is a competent speaker of English. As with other scientific and everyday terms that are univocal, there is no need to provide an analysis or definition as a prerequisite for using the term. Nor need we be concerned with details about modulation. Due to the “open-ended” nature of modulation, cataloguing each of the ways in which a sense might be modulated may well be impossible. At any rate, there is no need to do so. If one has in mind some highlighted part or property, and suspects one’s audience may not know what it is, one can simply say what it is, adding detail as necessary. (“The car needs vacuuming; and it’s the back seat I mean.”) Listing other possible ways in which the sense might have been modulated, as if to prepare one’s audience for any eventuality, will typically not be to the point. (“Along with what I in fact meant by ‘The car needs vacuuming’, one also might have meant: the floor mats, the crumbs under the passenger seat, the floor in the back, the trunk, the fur on the dashboard…”) The upshot is that our semantic ignorance presents a problem for the ambiguity view, but not for the univocality view.
The conclusion thus stands: the univocality view should be adopted as the default position. Accordingly, the burden of proof must lie with anyone wishing to defend the ambiguity view. However to do that—it should now be clear—one must enter the tangled maze of complex issues surrounding the homonymy-polysemy distinction, the nature of modulation, and so on. Clearly, no serious claims about the ambiguity of ‘consciousness’ can responsibly be made without addressing such semantic issues. But is that really what consciousness researchers want to, or ought to, be doing?
Resisting Semantic Ascent
I believe that consciousness researchers would do best not to delve into the semantics of ‘consciousness’ in the hope of deciding between ambiguity and univocality. Though no one to my knowledge is actually doing that, one might be tempted to if one were convinced by the above arguments. But there are two reasons for resisting the temptation. First, decisions about word meanings—for example whether a word is univocal or ambiguous—are almost never adopted throughout language communities (including limited academic communities) as a result of semantic argument. Terms widely agreed to be ambiguous are typically obvious cases, and are usually homonyms. True, decisions about senses of polysemous words are sometimes adopted in light of lexicographical considerations, but semantically such arguments are often highly suspect (see, e.g., Weinreich, 1964). In any event, researchers who insist that ‘consciousness’ is ambiguous is ambiguous almost never bring lexicographical considerations to bear. None of this, of course, is surprising since semantics is extraordinarily difficult; and it would be astonishing if someone untrained in linguistics were temporarily to enter the field and clear everything up on ambiguity, modulation, and so forth. In sum, given the current state of semantics, there is simply no reason to think that semantic argument could convincingly decide between the ambiguity and univocality views.
The second reason researchers would do best not to focus on the semantics of ‘consciousness’ is that even if a decision about univocality versus ambiguity could be reached in that way, it is doubtful that any benefit for consciousness research would result. For regardless of which view is right, any theoretically useful distinctions could be made independently of how the semantics turns out. The distinction between phenomenal and access consciousness, for example, can be construed either as among highlighted aspects of a single meaning of ‘consciousness’ if modulation is right, or across distinct meanings of the term if the ambiguity view is correct. And similarly for any of the other distinctions discussed above. So for the purposes of theorizing about consciousness, if a distinction is a useful one, it can be made without addressing the vexed question of how it might map onto word-meanings. The semantic issues, as we have seen, are highly complex and controversial, and must by their very nature add further confusion to discussions of consciousness—and with no apparent benefit. The semantics of ‘consciousness’, I submit, is thus best avoided wherever possible.
At this point one might try a different tack in defending semantic ascent. One might agree that, strictly speaking, the univocality view should be adopted as the default position; and also that consciousness researchers should leave it to linguists and other theorists of language to sort out the semantic details. Still, one might object, the practice of speaking of distinct meanings ‘consciousness’ is no cause for alarm. For all researchers really intend by so doing is to bring out distinctions considered important for theorizing about consciousness. That such distinctions get expressed in semantic terms is unimportant, since researchers’ real interest lies not with semantics but with consciousness itself. They might just as easily have spoken of aspects of consciousness that are highlighted through modulation (or alternatively, of various kinds of consciousness, or concepts of consciousness). But why bother, given that modulation is no less puzzling than ambiguity? The upshot is that while talk of distinct senses of ‘consciousness’, strictly speaking, may be false or unjustified, it is entirely harmless if properly understood.
It is true that researchers often just wish to bring out distinctions they believe need making. Indeed, that they offer their own idiosyncratic lists of meanings with little defense, or criticism of others’ lists, supports the current proposal that their main concern is with the distinctions themselves and not the semantics. On the other hand, it is unlikely that many regard semantic ascent merely as a “convenient manner of speaking” that is strictly speaking false or unjustified; most believe that ‘consciousness’ really is ambiguous. Nor are researchers always guided by their theoretical purposes in distinguishing among senses of ‘consciousness’—though they undoubtedly are sometimes. In short, there appears to be much confusion among researchers about why exactly they engage in semantics. That notwithstanding, it must be admitted that often no harm is done in particular cases. However, that is not to say that the practice of resorting to semantic ascent is unproblematic, as I shall now try to show.
One way of showing that would be to offer examples in which researchers arrive at mistaken beliefs about consciousness as a result of semantic ascent. Though I believe such mistakes are made, demonstrating that is difficult, and for two reasons. First, it requires that mistakes be identified. But given the broad disagreement on most matters regarding consciousness, doing so non-tendentiously is unfeasible: it demands an independent defense of some theory of consciousness relative to which mistakes can be categorized—something I cannot undertake here. Second, even if mistakes about consciousness could be identified non-tendentiously, it would still remain to be shown that the mistakes resulted from semantic ascent, and would otherwise not have occurred. But as was noted above, any distinctions among the alleged meanings of ‘consciousness’ could be drawn by a univocality theorist among highlighted aspects of consciousness. As a result, the etiology of any mistake will often be explicable in terms of the content of the distinction itself, with no appeal to the fact that it was expressed in semantic terms. Suppose now that one could point to some mistakes that were uncontroversially due to semantic ascent. It would still be open for one to respond: “So what? So a few people erred. That’s no reason to reject the entire practice.” It is thus unclear that pointing to mistakes would be of much help anyway. All of this places me in a somewhat uncomfortable position dialectically. Still, I shall press on.
That ‘consciousness’ is ambiguous, it has been argued, is at best unjustified and at worst false. That in itself, one might have thought, is sufficient reason to recommend that such statements not be made. For even if one is uncertain about which errors, if any, would result from their making, it would seem to be a well-grounded methodological principle, if anything is, that unwarranted and false statements are best avoided. Another such principle aims at efficiency in one’s conceptual apparatus: ceteris paribus, one’s stock of theoretical concepts should be kept to a minimum, since inefficiency and redundancy breed error. But semantic ascent is inefficient in just this sense: it employs an additional set of conceptual tools—namely, semantic ones—to express what can already be expressed more clearly and accurately non-semantically. In my view, these points on their own are reason enough to shun the semantic idiom in consciousness studies.
But we can go further. For not only is it inefficient to mirror aspects of consciousness in terms of senses of ‘consciousness’, doing so leads to a profusion of distinctions and discussions within the literature that serve no clear purpose. This is more than inefficiency; it is conceptual clutter. Semantic ascent leads to this because, as was said, many researchers do not view the semantic talk as just a manner of speaking; they treat it literally and seriously. Such individuals, as a result, often think that they too must say something about the range of meanings of ‘consciousness’, and specify which meanings they have in mind. So they attempt to characterize (define, analyze, etc.), if only partially, those meanings, distinguish them from other possible meanings, and so forth. All of that might be quite detached from any theoretical purposes they have. Indeed, it often seems that researchers have no clear reasons for entering into semantics aside from the fact that others do, and perhaps some prior belief that one must “define one’s terms” at the start of an inquiry. The result is several distinct, incomplete, and poorly motivated terminological discussions about ‘consciousness’. And that contributes significantly to the overall level of confusion in the literature. Semantic ascent is thus not harmless.
Attempting to list and characterize senses of ‘consciousness’ can lead to difficulties of a more serious kind. Once one sets out to distinguish among a term’s (alleged) meanings—especially where the issues are controversial—one will often try to characterize particular meanings in order that distinctions among meanings might stand out more sharply. To the extent that one’s motivation for distinguishing among the alleged meanings is unclear to begin with, so too will be one’s reasons for characterizing the meanings as one does. What can result are premature and “forced” (partial) definitions or analyses that hamper research by too rigidly constraining what counts as relevant to one’s subject matter. Now an inquiry of course needs some form and limits; but for that we ought not depend on artificial constraints provided by inadequate analyses of the semantics of ‘consciousness’. Better simply to decide on how we want our studies to be constrained and then design them accordingly. Nor need we worry that without an initial characterization of the meaning of ‘consciousness’ we might drift aimlessly and find ourselves studying, say, sailboats or the stock market. Any definition or analysis, after all, would be guided by our (fallible) pretheoretical intuitions about the meaning of ‘consciousness’, so why not just let those intuitions guide the investigation of consciousness itself, and inform us of what is relevant and interesting and what it not? We should thus follow those who insist that definitions and the like must come, if at all, only at later stages of inquiry—like Weiskrantz (1988), as expressed in the epigraph to this essay, and Block (1995b, p. 277):
…the demand for…definitions is misplaced. Especially at the beginning of a scientific enterprise there is no alternative to going by the seat of one’s pants.
Because of the difficulty of demonstrating a tight connection between semantic ascent and errors about consciousness, my response to the objection of this section has been rather abstract and hypothetical. I thus close with what might be a real error about consciousness due to semantic ascent. I have in mind familiar arguments that begin with the premise that ‘consciousness’ is ambiguous and end by concluding that there is no single thing consciousness; or even that consciousness does not exist. Though the conclusion frequently is conflated with the truism that there is no single problem of consciousness, my interest lies with the metaphysical claim. Kathleen Wilkes (1984), to take one example, upon surveying four purportedly distinct uses of ‘conscious’, declares:
‘consciousness’ and ‘conscious’ are terms…which…should not figure in the conceptual apparatus of psychology or the neurosciences’….[C]onsciousness is not the sort of thing that has a ‘nature’ appropriate for scientific study. (pp. 236-237)
And because ‘consciousness’ in ordinary language “covers a heterogeneous range of phenomena,” she infers that “no adequate truth-conditions can be imposed upon ascriptions of consciousness” (p. 239). She concludes with the suggestion that
just as science can dispense with the concept of consciousness and loose thereby none of its comprehensiveness and explanatory power, so too could ordinary language….The suggestion is…[that] the proposition that consciousness is an important or real phenomenon should be dropped. (p. 241, my emphasis)
Wilkes appears to have reached quite radical metaphysical conclusions about consciousness partly on the basis of her belief that ‘consciousness’ is ambiguous. But the conclusion is unwarranted. For if something like a univocality-cum-modulation view is right, according to which different aspects of a complex phenomenon are highlighted across contexts, consciousness might be a “single thing” after all (though of course not a simple thing). I leave it to the reader to judge the extent to which Wilkes’s conclusions, and those of others who argue similarly, result from semantic ascent.
Summary and Conclusion
Much of the paper
was directed against the widely held view that ‘consciousness’ (and its
cognates) is multiply ambiguous within the current consciousness
literature. Appealing to
modulation, it was argued that intuitions that support the ambiguity view can
be explained on the assumption that ‘consciousness’ is univocal; and also that
the univocality view should be adopted as the default position, with the burden
of proof falling on the ambiguity theorist. It was also argued that researchers should not take on that
burden, since there would be little hope of settling the linguistic issues,
and, anyway, settling them would not help. Instead, distinctions about consciousness should be made
when one’s theoretical purposes warrant doing so. Sticking to consciousness itself, and relying on one’s
intuitions about the usefulness of any proposed distinction, one can manage
perfectly well in the absence of definitions and semantic analyses. Here, keeping modulation in mind can
help: due to the “open-ended” nature of word-meanings, attempting to list all
possible ways in which a sense might be modulated is useless. Now if one believes that one must
resort to semantic ascent—and for all I have said there may sometimes be good
reasons for doing so—one should be very clear, both to oneself and to one’s
audience, about exactly why one is doing so and what benefits one hopes to
derive. Finally, we considered an
objection that the semantic idiom is no more than a harmless manner of
speaking. It was argued that
although often no harm is done, the overall practice of resorting to semantic
ascent generates considerable confusion, and can lead to errors and misguided
directions of research. The upshot
is that semantics is best avoided in consciousness studies. When it comes to consciousness, one
should strive to say what one means without speaking about meanings.
Allport, A. (1988). "What Concept of Consciousness?" In A. J. Marcel & E. Bisiach (Eds.), Consciousness in Contemporary Science. Oxford: Oxford University Press.
Antony, M. V. (1999). "Outline of a General Methodology for Consciousness Research." Anthropology and Philosophy 3(2), 43-56. <http://research.haifa.ac.il/~antony/papers/Methodology.htm>
Antony, M. V. (2000). "Concepts of Consciousness, Kinds of Consciousness, Meanings of 'Consciousness'," unpublished manuscript.
Armstrong, D. (1981). "What Is Consciousness?", in his The Nature of Mind. Ithaca: Cornell University Press.
Baars, B., J. (1988). A Cognitive Theory of Consciousness. New York: Cambridge University Press.
Block, N. (1991). "Evidence Against Epiphenomenalism." Behavioral and Brain Sciences 14, 670-2.
Block, N. (1993). "Review of Dennett: Consciousness Explained." Journal of Philosophy 4, 181-93.
Block, N. (1994). "Consciousness." In S. Guttenplan (Ed.), A Companion to Philosophy of Mind . Oxford: Blackwell.
Block, N. (1995a). "On a Confusion About a Function of Consciousness." Behavioral and Brain Sciences 18, 227-47.
Block, N. (1995b). "How Many Concepts of Consciousness?" Behavioral and Brain Sciences 18, 272-84.
Brentano, F. (1995). Descriptive Psychology (Benito Muller, Trans.). London: Routledge. From lectures given by Brentano in 1890-91.
Chalmers, D. (1995). "Facing Up to the Problem of Consciousness." Journal of Consciousness Studies 2, 200-19.
Churchland, P. S. (1983). "Consciousness: The Transmutation of a Concept." Pacific Philosophical Quarterly 64, 80-95.
Crick, F., & Koch, C. (1990). "Towards a Neurobiological Theory of Consciousness." Seminars in the Neurosciences 2, 263-75.
Cruse, D. A. (1986). Lexical Semantics. Cambridge: Cambridge University Press.
Davies, M., & Humphreys, G. (1993). "Introduction." In M. Davies & G. Humphreys (Eds.), Consciousness . Oxford: Blackwell.
Dretske, F. (1993). "Conscious Experience." Mind 102, 263-83.
Fauconnier, G. (1997). Mappings in Thought and Language. Cambridge: Cambridge University Press.
Forman, R. (Ed.). (1990). The Problem of Pure Consciousness . New York: Oxford University Press.
Geach, P. (1956). "Good and Evil." Analysis 17, 33-42.
Goguen, J. (1999). "An Introduction to Algebraic Semiotics, with Application to User Interface Design." In C. Nehaniv (Ed.), Computation for Metaphors, Analogy and Agents . Berlin: Springer.
Goldman, A. (2000). "Can Science Know When You're Conscious? Epistemological Foundations for Consciousness Research." Journal of Consciousness Studies 7(5), 3-22.
Güzeldere, G. (1997). "The Many Faces of Consciousness: A Field Guide." In N. Block, O. Flanagan, & G. Güzeldere (Eds.), The Nature of Consciousness. Cambridge, MA: MIT Press.
Hirst, G. (1987). Semantic Interpretation and the Resolution of Ambiguity. Cambridge: Cambridge University Press.
James, W. (1996). "Does 'Consciousness' Exist?" in his Essays in Radical Empiricism. Lincoln, Nebraska: University of Nebraska Press. Originally published in 1904.
Lehrer, A. (1974). "Homonymy and Polysemy: Measuring Similarity of Meaning." Language Sciences 32, 33-9.
Locke, J. (1690). An Essay Concerning Human Understanding. (2 ed.). Oxford: Oxford University Press.
Lycan, W. (1996). Consciousness and Experience. Cambridge, Mass.: MIT Press.
Lycan, W. (1997). "Consciousness as Internal Monitoring." In N. Block, O. Flanagan, & G. Güzeldere (Eds.), The Nature of Consciousness. Cambridge, MA: MIT Press.
Lycan, W. (forthcoming). "The Plurality of Consciousness." In J. M. Larrazabal & L. A. Perez Miranda (Eds.), Proceedings of the Sixth International Colloquium on Cognitive Science : Kluwer Academic Publishing.
Lyons, J. (1977). Semantics (Vol. 2). Cambridge: Cambridge University Press.
Moore, G. E. (1903). "The Refutation of Idealism." Mind 12, 433-53.
Nelkin, N. (1989). "Unconscious Sensations." Philosophical Psychology 2, 129-41.
Nelkin, N. (1995). "The Dissociation of Phenomenal States From Apperception." In T. Metzinger (Ed.), Conscious Experience. Paderborn: Schoningh.
Nelkin, N. (1996). Consciousness and the Origins of Thought. Cambridge: Cambridge University Press.
Panman, O. (1982). "Homonymy and Polysemy." Lingua 58, 105-36.
Pustejovsky, J. (1996). The Generative Lexicon. Cambridge, MA: MIT Press.
Putnam, H. (1999). The Threefold Cord. New York: Columbia University Press.
Quine, W. V. (1960). Word and Object. Cambridge, MA: MIT Press.
Rosenthal, D. (1986). "Two Concepts of Consciousness." Philosophical Studies 49, 329-59.
Rosenthal, D. (1993a). "State Consciousness and Transitive Consciousness." Consciousness and Cognition 2, 355-63.
Rosenthal, D. (1993b). "Thinking that One Thinks." In M. Davies & G. W. Humphreys (Eds.), Consciousness. Oxford: Blackwell.
Rosenthal, D. (1997). "A Theory of Consciousness." In N. Block, O. Flanagan, & G. Güzeldere (Eds.), The Nature of Consciousness. Cambridge, MA: MIT Press.
Rosenthal, D. (forthcoming). "State Consciousness and What It's Like," in his Consciousness and Mind. Oxford: Oxford University Press.
Thomson, J. J. (1994). "Goodness and Utilitarianism." Proceedings and Addresses of The American Philosophical Association 67(4), 7-21.
Travis, C. (1981). The True and the False. Amsterdam: John Benjamins.
Travis, C. (1989). The Uses of Sense. Oxford: Oxford University Press.
Travis, C. (1991). "Annals of Analysis." Mind 100, 237-264.
Travis, C. (1997). "Pragmatics." In B. Hale & C. Wright (Eds.), A Companion to the Philosophy of Language. Oxford: Blackwell.
Tulving, E. (1993). "Varieties of Consciousness and Levels of Awareness in Memory." In A. Baddeley & L. Weiskrantz (Eds.), Attention: Selection, Awareness, and Control . New York: Oxford University Press.
Turner, M., & Fauconnier, G. (1995). "Conceptual Integration and Formal Expression." Journal of Metaphor and Symbolic Activity 10(3), 183-204.
Tye, M. (1995). "The Burning House." In T. Metzinger (Ed.), Conscious Experience (pp. 81-90). Paderborn: Schoningh.
Weinreich, U. (1964). "Webster's Third: A Critique of its Semantics." International Journal of American Linguistics 30, 405-9.
Weiskrantz, L. (1988). "Some Contributions of Neuropsychology of Vision and Memory to the Problem of Consciousness." In A. Marcel & E. Bisiach (Eds.), Consciousness in Contemporary Science . New York: Oxford University Press.
Wilkes, K. V. (1984). "Is Consciousness Important?" British Journal for the Philosophy of Science 35, 223-43.
Ziff, P. (1960). Semantic Analysis. Ithaca: Cornell University Press.
 I shall thus be unconcerned in this paper with the meaning(s) of ‘consciousness’ in the term’s history and in laymen’s everyday usage. Note also that I do not assume a sharp line between the consciousness literature and other literatures, or between consciousness researchers and laymen. Just as proponents of the view that ‘consciousness’ is ambiguous within the literature do not focus primarily on exceptional uses of ‘consciousness’, or on individuals on the fringes of the field, my claims about the univocality of ‘consciousness’ should be taken as applying to the overall character of the field, and not as excluding the possibility of unusual or borderline cases.
 Witness also the protracted twists and turns that Lyons (1977) goes through in attempting to theoretically ground the distinction in terms of the “co-ordination test” (also called the ‘reduction test’ or ‘identity test’). See also Panman (1982) and Cruse (1986).
 Though several others have discussed ‘good’, e.g., Geach (1956), Ziff (1960), and Thomson (1994), there is still no agreement on how it is to be treated.
 A further indication of the unclear boundary between polysemy and univocality is that lexicographers’ decisions about how to individuate word-senses often seem arbitrary. Weinreich (1964, p. 405), e.g., describes how the senses of the verb ‘turn’ jumped with little justification from a reasonable number in Webster’s 2nd Edition to an unruly 115 in the 3rd.
 See, e.g., Travis (1981; 1989; 1991; 1997). Others who discuss such phenomena (often citing Travis) are Putnam (1999, p. 87 ff., 124 ff.), and researchers in the field of cognitive semantics like Fauconnier and Turner, employing their notion of a blend (see, e.g., Turner and Fauconnier 1995; Fauconnier 1997). The concept of a blend has recently been formalized by Goguen (1999).
 I cannot elaborate or defend these points here. See Travis (1997) for a concise presentation of the arguments, and Travis (1989; 1991) for further details.
 Block (1994, p. 217), for example, maintains that we have an “undifferentiated ‘conscious’ [that]…works well for most purposes,” and Lycan (1996, p. 2) asserts that “[b]oth psychologists and some philosophers still use the word [‘consciousness’] univocally.” Both Block (1995a,b) and Lycan, however, take ‘consciousness’ to be multiply ambiguous within the literature.
 The biennial conferences at The University of Arizona, Tucson.
 See Antony (2000) for a discussion of a related idea that I call ‘The Single Meaning Principle’: An ambiguous term contributes at most one of its meanings to any one meaning of any complex expression in which it is used.
 See Block (1991; 1993; 1994; 1995a,b). Block also speaks of kinds (1995b, p 274), forms (1994, p. 213), and types (1995a, 9. 235) of consciousness, as well as senses of ‘consciousness’ (1991, p. 670; 1995a, p. 241). He is not alone in doing so, however; the use of expressions like ‘concepts of consciousness’, ‘kinds of consciousness’, and ‘meanings of ‘consciousness’’ interchangeably is ubiquitous within the literature. I have argued elsewhere (Antony 2000), however, that the only way to make sense of that practice—which I believe also generates confusion in the literature—is to interpret ‘concepts of consciousness’ and ‘kinds of consciousness’ metalinguistically to mean concepts expressed by ‘consciousness’ and kinds expressed by ‘consciousness’, respectively—both of which, for all intents and purposes, can be taken as equivalent to meanings of ‘consciousness’. In what follows, accordingly, I treat researchers’ uses of ‘concepts of consciousness’ and ‘kinds of consciousness’ as synonymous with ‘meanings of ‘consciousness’’.
 See note 7.
 Block actually goes further than claiming that ambiguity explains the conflation; he characterizes ambiguity such that conflation entails it: “The distinction between the ambiguity and cluster concept hypotheses can be drawn in a number of equally legitimate ways; there is some indeterminacy in the distinction. I favor drawing the distinction in terms of conflation: if there can be a conflation, we have ambiguity….” (Block 1995a, p. 263. See also 1995b, p. 282.). However if one takes semantics seriously as a scientific enterprise, meaning presumably will have some nature, and one will not be free to characterize ambiguity however one chooses. And even if there is some indeterminacy, it is hard to avoid seeing Block’s construal of ambiguity in terms of conflation as ad hoc, given that it is conflation (of P and A) that he wishes to explain by appeal to ambiguity. In any event, I shall presently argue that Block is mistaken in thinking that conflation requires ambiguity.
 See, e.g., the Open Peer Commentary following Block (1995a), and Block’s responses (Block 1995b). And Block himself says, “I find that many critics wonder why I would count phenomenal consciousness as consciousness, whereas many others wonder why I would count access- or monitoring- or self-consciousness as consciousness” (1995a, p. 235).
 See, e.g., Rosenthal (1993a, p. 355; 1993b, p. 197; 1997, p. 729 ff.; forthcoming, p. 1). Note that, indexicals aside, reference or application of a term to distinct phenomena or properties across uses of the term is insufficient for ambiguity. Nor does the dissociability of such properties guarantee ambiguity, as Nelkin (1996, p. 147 ff.) maintains. For both scenarios are possible with modulation. As a single sense is modulated across contexts distinct phenomena or properties are highlighted. And dissociation is also possible, as the ‘full’ example (and Travis’s ‘leaves’ example, etc.) shows: a pool can be full of water but not people and full of people but not water, but ‘full’ is univocal nonetheless. So neither consideration support the ambiguity view. In response one might claim that different “highlightings” do not amount to differences of reference or application; but then one owes a non-question-begging account of the distinction.
 Rosenthal (forthcoming, p. 2, fn. 3) claims that such intuitions are indecisive: “Are very vivid dream states conscious states? Must we be conscious when we’re in them? Since it’s far from clear what to say about these matters, it may well be that conscious states can occur without the creature itself being conscious.” Rosenthal is of course right that such matters—indeed most matters regarding consciousness!—are unclear, and so things may be as he says. The important question for our purposes, however, is whether there is an established use of ‘consciousness’ among researchers along the lines he suggests, and it seems there is not.
 On creatures whose mental states are never conscious (Rosenthal suggests frogs and turtles as possibilities), Rosenthal (forthcoming, p. 3) says: “it’s plain that when none of a creature’s mental states is conscious, there is nothing puzzling about what it is for the creature to be conscious.”
 One might claim that there also is a use of ‘consciousness’ that means knowledge within the literature. For that claim to be at all plausible, however, any such use would have to refer to states that are experiential. But then modulation could explain such uses: that a given conscious state involves knowledge could sometimes be highlighted or thrown into relief.
 Dretske (1993, p. 270), in effect, takes this line in maintaining that transitive consciousness implies state consciousness.
 Indeed, the transitive use makes explicit reference to that content, so modulation is not here required.
 One might think that the mere syntactic difference suffices for distinct senses. Lexicographers, after all, distinguish senses of verbs on the basis of the transitive/intransitive distinction. However this syntactic criterion is problematic since it “leads to an uneconomical repetition of definitions” (Weinreich, 1964, p. 408). Moreover, with ‘consciousness’ (‘conscious’, etc.), we are of course not dealing with verbs.
 To the extent that one endorses meaning-holism, and believes that any differences in theories or beliefs entail differences in meaning, the question whether ‘consciousness’ is ambiguous or univocal of course loses interest. Moreover, the practice of listing meanings of the term in works on consciousness must be seen as thoroughly misguided.
 Brentano (1995) also appears to uses ‘consciousness’ in this way.
 See, e.g., Forman (1990), and several articles in Journal of Consciousness Studies.
 Or at least the variation seems semantic. I am not committed to its being so, however. What is important is just the existence of the phenomenon itself—even if it turns out to be pragmatic, or something else.
 Though Travis does not use Cruse’s term ‘modulation’.
 It is often suggested that there are several concepts of access, and even phenomenality; and that lends support to the suggestion in the text—that is, even if P and A are distinct meanings of ‘consciousness’, modulation may well occur with each of them. On different construals of access, see Davies and Humphreys (1993) and Goldman (2000), among others. On phenomenality, Block writes: “there are somewhat different notions of phenomenal consciousness that are legitimate for some purposes…” (1995b, p. 283).
 Note that I beg no questions here by assuming that modulation exists, since I do not assume that modulation explains the semantic phenomena others take to support the ambiguity view. For all I have said, the alleged distinct senses of ‘consciousness’ may be real. Modulation theorists after all do not deny that ambiguity exists. They would insist, however, that if ‘consciousness’ is ambiguous, each sense is likely to be subject to modulation. But that is not to assume the univocality view.
 A personal reason for preferring the univocality view is that it accords with a methodology for studying consciousness that I favor. According to that methodology, our concept (in the psychological sense) or “picture” of consciousness is to be studied in order to make progress on consciousness itself. Adopting the univocality view allows one largely to avoid, at the start of the inquiry, the question: “Which concept?” For an informal sketch and defense of the methodology see Antony (1999).
 Lycan (1996, forthcoming), e.g., distinguishes between introspection, subjectivity, qualia, etc. for the purpose of “dividing and conquering,” and begins by claiming that ‘consciousness’ is ambiguous among those senses. But why is the semantic claim necessary? Any dividing and conquering can proceed without it. (In personal communication Lycan informs me that he is not strongly committed to the semantic claim.) Rosenthal, for example, for the most part manages perfectly well without engaging in semantics. From this it should be inferred that nothing in this paper is meant as a denial (or defense) of the theoretical utility of any of the distinctions discussed.
 Note that in science generally the semantics of theoretical terms is rarely if ever investigated, and science seems no worse off for that. Even where there is conceptual work to be done, the issue is not what words mean but how best to conceive of relevant phenomena. Why should consciousness studies be different?
 But see note 10.
 To clarify: Suppose that in one’s theory one mistakenly ignores a phenomenon (say, access relations) because one deems it to be associated with an “unimportant” meaning of the word ‘consciousness’. It would be hard to prove that the ambiguity view per se led to the error. For even if one thought that ‘consciousness’ is univocal, one still might have ignored aspects of consciousness one took to be unimportant (like access relations). Of course, one might not have; but settling the matter—assuming that to be possible—would, at a minimum, require subjecting the individual to psychological testing.
 I hope it is now relatively clear why I have not attempted to semantically characterize the single sense of ‘consciousness’ I claim is employed in the literature. See also Antony (1999).
 For related claims and arguments, see Churchland (1983), Allport (1988), Tulving (1993), Lycan (1996), among others.
 While several of the main themes in this essay may have application outside of consciousness studies, e.g., in other areas of philosophy, I must leave consideration of that matter for another occasion.
 A shorter version of this paper was presented at Toward a Science of Consciousness, Tucson 2000. My thanks to audience members who participated in the question period, as well as to Jonathan Berg, Joseph Goguen, and two anonymous referees for helpful comments or discussion.