Concepts of Consciousness, Kinds of Consciousness, Meanings of ‘Consciousness’

MICHAEL V. ANTONY

CONCEPTS OF CONSCIOUSNESS, KINDS OF CONSCIOUSNESS, MEANINGS OF ‘CONSCIOUSNESS’*

(appears in Philosophical Studies 109(1), 2002, 1-16)

1. INTRODUCTION

A frequent claim within the consciousness literature is that there are several concepts (notions, conceptions, etc.) of consciousness. It is also held that there are diverse kinds (types, sorts, forms, varieties) of consciousness. And the literature is rife with claims to the effect that there are numerous meanings (senses, uses) of ‘consciousness’. My topic is these three kinds of claims and their interrelations. I shall not explicitly address the parenthetically mentioned variants on concepts, kinds, or meanings, since the discussion will apply to them relatively straightforwardly.

Let us represent the three kinds of claims uniformly as follows:

(K) There are distinct kinds of consciousness, k₁, k₂,…, k_n.

(M) There are distinct meanings of ‘consciousness’, m₁, m₂,…, m_n.

C, K, and M are meant to be representative of an indefinitely wide range of related claims like:

(1) These concepts of consciousness are often confused with one another.

(2) Jones distinguishes three types of consciousness.

(3) ‘Consciousness’ is multiply ambiguous.

In what follows, I often speak of C, K, and M as if they are the only claims made in the literature regarding multiple concepts or kinds of consciousness, or meanings of ‘consciousness’. Of course, C, K, and M are merely instances of the three broad types of claims I have in mind regarding concepts, kinds, and meanings.

What quickly becomes apparent to anyone reflecting on the uses of C, K, and M in the literature is that, for the most part, they are all the same: C, K, and M are used interchangeably; any one is deemed true or assertable if and only if the others are. More precisely, they are used interchangeably provided that c₁, k₁, and m₁ correspond to each other in an appropriate manner, and similarly for c₂, k₂, and m₂,…, and c_n, k_n, and m_n. What “appropriate correspondence” amounts to is hard to say, but the rough idea can be expressed in semantic terms. To illustrate, consider Block’s well-known distinction between phenomenal consciousness and access consciousness.[1] If this distinction can be reflected in two concepts of consciousness, two kinds of consciousness, and two meanings of ‘consciousness’, there will be two concept-kind-meaning triples such that each member of one triple is related to the semantic content phenomenal consciousness, and each member of the other is related to the content access consciousness. I shall not try to elucidate “correspondence” any further, since the intuitive idea will suffice for our purposes.[2]

Within the consciousness literature, examples abound in which ‘concepts of consciousness’, ‘kinds of consciousness’ and ‘meanings of ‘consciousness’’ are used interchangeably, or at least indiscriminately. In a recent book, Carruthers (2000) begins a section called “Some distinctions: kinds of consciousness” with the sentence “There are a number of different notions of consciousness and/or a number of different kinds of use of the term ‘conscious’ which need to be distinguished carefully from one another” (p. 9, emphasis mine). Or consider again Block’s distinction between phenomenal consciousness and access consciousness. Although Block commonly presents it as a distinction among concepts of consciousness, he sometimes describes it as among types, forms, or kinds of consciousness, or as among senses or meanings of ‘consciousness’. Others characterize Block’s distinction in their own favored ways.[3] Or take Rosenthal’s distinctions among state consciousness, creature consciousness, and transitive consciousness. Rosenthal typically avoids addressing the semantics of ‘consciousness’, but he sometimes characterizes his distinctions as among uses of ‘consciousness’, and also kinds of consciousness. Many of his interpreters construe his distinctions as among senses of ‘consciousness’.[4] Other theorists offer distinctions of their own, and either alternate between treating them as among concepts, kinds, or meanings, or say nothing about the terminological choices they make.[5] In sum, that C, K, and M can in all or most contexts be substituted freely for one another—call this interchangeability—is widely held within the consciousness literature, at least tacitly.

Interchangeability is problematic, however. As I argue in the next section, it can be made sense of in only two ways. The first involves interpreting C and K metalinguistically, and the second rests on certain semantically deviant, though coherent, literal interpretations of C and K. The trouble is that many researchers appear to use C, K, and M interchangeably without satisfying either way of doing so coherently. This generates serious problems of interpretation: if we charitably try to avoid taking theorists to be contradicting themselves, we will often be able to do no better than guess as to what their communicative intentions are.

2. MAKING SENSE OF INTERCHANGEABILITY

2.1. Interchangeability and Metalinguistic Interpretation

C, K, and M can coherently be used interchangeably if C and K are interpreted metalinguistically to mean something like:

(C^*) There are distinct concepts expressed by ‘consciousness’, c₁, c₂,…, c_n.

(K^*) There are distinct kinds expressed by (referred to by, etc.) ‘consciousness’, k₁, k₂,…, k_n.

Researchers sometimes seem have such interpretations in mind. Where creature consciousness and state consciousness (or phenomenal consciousness and access consciousness, etc.) are described as ‘distinct concepts of consciousness’, the idea often is not that there is some single phenomenon—consciousness—and two concepts of it. Rather the concepts are construed as of different things, which is much like saying they are distinct concepts expressible by an ambiguous ‘consciousness’. Where creature consciousness and state consciousness are characterized as ‘kinds of consciousness’, the same often holds: consciousness is not viewed as a general phenomenon of which there are two or more varieties or species—like human consciousness and bat consciousness. Instead the point is that there are distinct kinds of phenomena we happen to call ‘consciousness’.

Where metalinguistic interpretations of C and K are in play, interchangeability between C, K, and M makes sense, since C^* and K^* can for all intents and purposes be treated as synonymous (or at least materially equivalent) with each other as well as with M. We thus have one reasonably clear way of understanding interchangeability.

2.2. Interchangeability and Literal Interpretation: C and M

The second way to make sense of interchangeability involves non-metalinguistic or literal interpretations of C and K. Since the following arguments are independent of anything specific to consciousness or ‘consciousness’, I present them in general terms. I discuss first C and M, then K and M, and finally C and K.

How might C and M be used interchangeably if C is interpreted literally? Or, more generally, how might a literal interpretation of ^┌distinct concepts of f^┐be used interchangeably with ^┌distinct meanings of ‘f’^┐?[6]

Consider two concepts of space. It matters little what they are, so let one be Newton’s according to which space is an independently existing “container” in which objects are located, and the second Leibniz’s according to which space is the collection of all spatial relations among entities. Call these concepts absolute space and relational space.[7] Now I doubt that the English word ‘space’ means either absolute space or relational space, but it might have; anything can mean just about anything given the right conventions. So imagine a possible situation in which ‘space’ has those meanings. There would then be two concepts of space, absolute space and relational space, and two corresponding meanings of ‘space’, absolute space and relational space. And it seems unobjectionable in that situation to alternate freely between ‘two concepts of space’ and ‘two meanings of ‘space’’. We thus have a case in which a literal interpretation of ‘distinct concepts of space’ is interchangeable with ‘distinct meanings of ‘space’’.

The foregoing example is an imaginary one because I know of no real cases in which a literal interpretation of ^┌distinct concepts of f^┐ is interchangeable with ^┌distinct meanings of ‘f’^┐. Consider this question: Above, where ‘two concepts of space’ is interchangeable with ‘two meanings of ‘space’’, what does ‘space’ mean in ‘two concepts of space’? We stipulated that ‘space’ is ambiguous between absolute space and relational space. Can ‘space’ have either of those meanings in ‘two concepts of space’? Not if there is to be interchangeability. If it did, ‘two concepts of space’ would mean either two concepts of absolute space or two concepts of relational space. But then in neither case could the two concepts correspond to the meanings absolute space and relational space, so interchangeability would be impossible. Now since ‘space’ cannot have both meanings in ‘two concepts of space’,[8] it must have some third meaning. On reflection, one sees that it is a more general meaning, a meaning that—to resort to a traditional metaphor—is “contained” in the meanings absolute space and relational space. It is of space, understood in some such more general sense, that absolute space and relational space are concepts. Notice that the point is entirely general: if a literal interpretation of ^┌distinct concepts of f^┐is interchangeable with ^┌distinct meanings of ‘f’^┐, then the meaning of f in ^┌distinct concepts of f^┐must be distinct from and more general than the meanings referred to by ^┌distinct meanings of ‘f’^┐.

But such a state of affairs is a semantic rarity at best. Terms almost never have both general and specific meanings. An exception is ‘dog’, which means both canine and male canine. In contrast, ‘apple’ does not mean both apple and Macintosh apple and ‘restaurant’ does not mean both restaurant and Indian restaurant. How much less likely, then, that a term will have a general meaning and two specific meanings, as in the imaginary ‘space’ example? It is as if ‘dog’ were to mean, in addition to canine and male canine, also female canine! But that is exactly what is required of f if a literal interpretation of ^┌distinct concepts of f^┐is to be interchangeable with ^┌distinct meanings of ‘f’^┐. Since there are few if any such terms in natural languages, no actual practice of interchangeability between ^┌distinct concepts of f^┐ and ^┌distinct meanings of ‘f’^┐is likely to be understandable in this second way.

2.3. Interchangeability and Literal Interpretation: K and M

A similar story applies to interchangeability between K and M where K is interpreted literally, or more generally between a literal interpretation of ^┌distinct kinds of f^┐ and ^┌distinct meanings of ‘f’^┐. Imagine that ‘dog’, in addition to meaning canine, also meant both hunting dog and show dog. There would then be two meanings of ‘dog’, hunting dog and show dog, and two corresponding kinds of dogs, hunting dogs and show dogs. The expressions ‘two kinds of dog’ and ‘two meanings of ‘dog’’ would thus be interchangeable (for those meanings and kinds). Or suppose ‘space’ meant absolute space and relational space, as above, but also that the universe contained two kinds of space, absolute space and relational space (in different “parts” of the universe). There would then be two meanings of ‘space’ and two corresponding kinds of space; and ‘meanings of ‘space’’ and ‘kinds of space’ would be interchangeable.

In both examples, the meaning of f in ^┌distinct kinds of f^┐must be different from and more general than the meanings referred to by ^┌distinct meanings of ‘f’^┐. The meaning of ‘dog’ in ‘two kinds of dog’ can be neither hunting dog nor show dog, for neither two kinds of hunting dog nor two kinds of show dog could correspond to the meanings hunting dog and show dog. The meaning of ‘dog’ in ‘two kinds of dog’ is in fact close to the general meaning canine; and it must be some such more general sense if the kinds and meanings are to correspond in a way consistent with interchangeability. Similar points apply to the ‘space’ example. But, again, language does not seem to work that way: terms rarely if ever have both general and multiple specific meanings. So accounting for actual cases of interchangeability between K and M in this way is unlikely to succeed.

2.4. Interchangeability and Literal Interpretation: C and K

Interchangeability will be possible between C and K, or between ^┌distinct concepts of f^┐and ^┌distinct kinds of f^┐, where one is interpreted metalinguistically and the other literally. That is because such cases reduce to those already discussed—assuming, that is, that metalinguistic interpretations of ^┌distinct concepts of f^┐and ^┌distinct kinds of f^┐ are treated as synonymous with ^┌distinct meanings of ‘f’^┐. Interchangeability in such cases is of course equally problematic.

Is interchangeability possible where both ^┌distinct concepts of f^┐and ^┌distinct kinds of f^┐ are interpreted literally? Not so far as I can tell. Suppose we try to construct a case; the examples already discussed might furnish us with what we need. In one example we spoke of two concepts of space, absolute space and relational space, and in another we imagined the universe containing two kinds of space, absolute space and relational space. Are those not corresponding concepts and kinds? While they correspond in some sense, the correspondence is insufficient to underwrite interchangeability.

To see why, imagine an individual, Jones, who believes the universe contains two kinds of space, absolute space and relational space. For Jones, ‘space’ would have a general meaning associated with the genus of which absolute space and relational space are species. Could Jones think of absolute space as a concept of space in that general sense? It is hard to see how, since Jones believes that absolute space is just one kind of space. Absolute space for Jones is no more a concept of space in general than omelette is as a concept of food in general. For parallel reasons Jones would deny that relational space is a concept of space. Imagine now Smith, for whom absolute space and relational space are two concepts of space in general (as they were for Newton and Leibniz). Smith could not also view absolute space and relational space as distinct kinds of space. For Smith takes absolute space and relational space to be concepts of the very same thing; but if absolute space and relational space were distinct kinds of space they would be different things. So neither Jones nor Smith could use ‘two kinds of space’ and ‘two concepts of space’ interchangeably. In general, literal interpretations of ^┌distinct concepts of f^┐and ^┌distinct kinds of f^┐ cannot coherently be used interchangeably, since the concepts will always be of the same thing, and the kinds will be distinct species of that thing.

3. INTERCHANGEABILITY IN THE CONSCIOUSNESS LITERATURE

We have seen that the practice of using C, K, and M interchangeably is widespread within the consciousness literature. Can that practice be made sense of in either of the above two ways, or must we conclude that researchers are often mistaken or even confused in their uses of C, K, and M? I believe there is considerable confusion and error. At the very least, due to interchangeability, it is often exceedingly difficult to interpret authors’ claims. I briefly illustrate that in the writings of three philosophers, chosen almost at random: Armstrong, Carruthers, and Church.

3.1. Armstrong

Armstrong (1981, 1999) distinguishes between minimal consciousness, perceptual consciousness, and introspective consciousness. Is this distinction among kinds, concepts, meanings? Armstrong writes: “The notion of consciousness is notoriously obscure….It is not even clear that the word ‘consciousness’ stands for just one sort of entity, quality, process, or whatever” (1981 p. 55). Given the second quoted sentence, and the fact that Armstrong expresses his distinction as among senses of ‘consciousness’ (pp. 58-60), one might expect occurrences of C or K in his writings to be metalinguistic. So where he describes introspective consciousness as a sort of consciousness (p. 60), for example, perhaps he means a sort of thing ‘consciousness’ stands for. But then how is his reference to the notion of consciousness to be understood? He cannot mean, metalinguistically, the unique notion expressed by ‘consciousness’, since he takes ‘consciousness’ to be ambiguous. Might he just mean a general concept expressed by ‘consciousness’? If so he could maintain that ‘consciousness’ has a general sense and three specific senses. And, then, in accordance with the second way of making sense of interchangeability, he could treat introspective consciousness as literally a sort of consciousness. This interpretation gains support from Armstrong’s theoretical account of perceptual consciousness and introspective consciousness: “Consciousness, then, both perceptual consciousness and introspective consciousness, is representation” (1999, p. 119). Here it looks like perceptual consciousness and introspective consciousness are conceived as two species of consciousness in some more general sense of the term. (But what about minimal consciousness?) But then Armstrong is committed to the sort of highly implausible semantic claims discussed in the previous section.

It is hard to know what Armstrong is proposing. Does he take consciousness to be a single phenomenon that comes in different forms or varieties? Or does ‘consciousness’ ambiguously refer to distinct, more or less unrelated phenomena that we tend to confuse with each other? Or (somehow) both? These are very different things. It should perhaps be noted that in Armstrong’s (1999) recent book he refrains from formulating his three-way distinction as among senses of ‘consciousness’. In any event, until we know which of the above alternatives, if any, Armstrong has in mind, a proper evaluation of his theory is impossible.[9]

3.2. Carruthers

We saw that Carruthers (2000, p. 9) uses C, K, and M interchangeably at the start of his section “Some distinctions: kinds of consciousness.” Throughout the section (e.g., pp. 12-13) he speaks primarily of notions of consciousness (C) and forms of consciousness (K). It was argued above that there are two ways of understanding interchangeability between C and K: interpret both metalinguistically, or interpret one metalinguistically and the other literally. Is either way applicable to Carruthers’s discussion?

As with others who use C, K, and M interchangeably, it is hard to know what exactly Carruthers’s distinctions are meant to distinguish. Metalinguistic interpretations of both C and K can probably be ruled out, since Carruthers often seems to use K literally. That is evident in his proposed taxonomy, which to all appearances classifies three kinds of creature consciousness and three (four?) kinds of state consciousness:

Creature-consciousness 1 – intransitive

Creature-consciousness 2 – transitive

Creature-consciousness 3 – self-consciousness

State-consciousness 1 – phenomenal

State-consciousness 2 – functional

State-consciousness 3 – standing versus occurrent

If Carruthers does not interpret both C and K metalinguistically, and if he uses K literally, then his talk of distinct notions of consciousness must be metalinguistic. In that case intransitive consciousness, transitive consciousness, and self-consciousness (etc.) are all meanings of ‘consciousness’. But if intransitive consciousness, transitive consciousness, and self-consciousness are kinds of consciousness, ‘consciousness’ must also have a more general sense (in this case creature consciousness, given his taxonomy). But now ‘consciousness’ has both specific meanings and a general meaning, and so Carruthers is committed to the same implausible semantic theses to which Armstrong may be committed.

If , contrary to what I have supposed, Carruthers treats K metalinguistically, then he either must interpret C literally, or both C and K metalinguistically. Literally interpreting C entails the same incredible semantics. If both C and K are construed metalinguistically, however, Carruthers needs to explain passages that prima facie involve literal interpretations of C or K. One example is his taxonomy which, as we said, strongly suggests a literal interpretation of K. Another is where he speaks of “notions of consciousness and/or…kinds of use of the term ‘conscious’” (p. 9). Recall that where C and K are interpreted metalinguistically, C, K, and M can for all intents and purposes be treated as synonymous. But where two expressions f and y are synonymous, it is hard to know what to make of ^┌f and/or y^┐. It begins to appear as though Carruthers’s text will contain some errors or confusions no matter how it is interpreted.

3.3. Church

The first two illustrations concerned interchangeability in the works of individual theorists. I turn now to problems that arise from researchers’ uses of C, K, and M interchangeably when interpreting others.

Notice first that all points about interchangeability within individuals’ writings apply across theorists as well. If Jones speaks literally of three kinds of consciousness, but does not believe that ‘consciousness’ has both general and specific senses, Smith errs in taking Jones to speak of three senses of ‘consciousness’. Other errors and confusions, however, are characteristic of interchangeability across theorists, specifically where theorists evaluate others’ claims. This is seen in a paper of Church’s (1998) in which she discusses Block’s distinction between phenomenal consciousness and access consciousness. Church lists various ways of attacking Block’s distinction, two of which are these:

First, one can seek to discredit one of these types of consciousness…by simply denying that one of the types is a type of consciousness…. [Alternatively], one may seek to show that phenomenal consciousness and access consciousness are one and the same by showing that, as a matter of fact, the two concepts pick out the same property, that they have the same extension. (pp. 58-59)

Whether such criticisms can succeed depends crucially on whether Block’s distinction has been correctly interpreted. Take the first way of attacking the distinction. Seeking to discredit access consciousness or phenomenal consciousness by denying that one is a type of consciousness is to the point if Block’s distinction is literally between types (kinds, sorts, etc.) of consciousness. However if it is between senses of ‘consciousness’, then if one sense is shown to be irrelevant to consciousness in some other sense of the term, nothing of interest follows. Consider now the second line of attack. Showing that phenomenal consciousness and access consciousness are the same property is damaging to Block if his distinction is indeed among types. But if it is among concepts of consciousness, the argument misses its mark since there can be distinct concepts of a single type or property (e.g., the concepts water and H₂O).

Block uses C, K, and M interchangeably, and Church assumes that Block’s distinction involves literal interpretations of K. If that assumption is false, both lines of attack fail. If, on the other hand, Block does interpret K literally, the two criticisms are to the point but Block’s own discussion is likely to contain errors or confusions somewhere, since he uses C, K, and M interchangeably. I shall not examine further which of Church or Block have erred; it is enough to see that at least one of them almost certainly has.

4. CONCLUDING REMARKS

There is a strong tendency to imagine—even after having come this far—that interchangeability is perfectly harmless. If theorists’ uses of C, K, and M interchangeably are strictly speaking mistaken or incoherent, then it must be that they are speaking loosely rather than strictly (like one who speaks of two things being identical). Such theorists, however, are not confused, they know what they mean, they know whether their interpretations of C or K are metalinguistic or literal, and so on.

I have tried to argue, however, that that is simply false, an illusion. Even loose talk, if rational, must admit of coherent interpretation, but where there is interchangeability frequently no such interpretation is available. Though it is sometimes difficult to identify errors or confusions, we can often be certain that they lie somewhere. That is so both within the writings of individual theorists, and across theorists’ interpretations of others’ work. At the very least, because it is often far from apparent what a given distinction is meant to mark, fully understanding a researchers’ theory is impossible. Much confusion, then, accompanies interchangeability.

One might respond by admitting that confusion sometimes exists, but it is of a limited sort of which researchers are aware. More specifically, where theorists have no firm conviction as to whether their distinctions are among concepts, kinds, or meanings, their uses of C, K, and M interchangeably simply reflect their desire to leave things open. Notice that this response requires that researchers’ interpretations of C and K be literal. If they were metalinguistic, C, K, and M could be treated as synonymous, but where f and y are synonymous it makes no sense to leave open which of f and y are true. Now I believe that researchers sometimes do wish to leave open which of C, K, and M are true, and so use them interchangeably.[10] However, that just makes matters worse for the defender of interchangeability. For researchers never say they are using literal interpretations of C, K, and M interchangeably to leave things open. So unless everyone always interprets C, K, and M literally to leave things open—which is incredible—we just have a further possibility to consider when attempting to interpret uses of C, K, and M interchangeably.

How, then, should consciousness researchers proceed? I offer three suggestions. First, researchers should remain aware of the potential for confusion and error due to interchangeability. Doing so could have the added benefit of leading to new insights about one’s own distinctions and those of others. Second, where there is risk of confusion, theorists should state whether their uses of C, K, or M are metalinguistic or literal, or how matters are to be left open. Third, metalinguistic interpretations of C and K might best be avoided where possible, since they generate confusion and are of little benefit. Since they are more or less equivalent to M, why not just use M?[11] Moreover, interpreted literally, ‘concepts of consciousness’, ‘kinds of consciousness’, and ‘meanings of ‘consciousness’’ have importantly distinct meanings which if kept separate could be put to more efficient theoretical use. The literal interpretations of ^┌concepts of f^┐, ^┌kinds of f^┐, and ^┌meanings of ‘f’^┐, after all, are troublesome enough without any additional complications introduced by interchangeability and the metalinguistic interpretation.[12]

NOTES