«PANIC theory and the prospects for a representational theory of phenomenal consciousness URIAH KRIEGEL ABSTRACT Michael Tye has recently argued that ...»
PHILOSOPHICAL PSYCHOLOGY, VOL. 15, NO. 1, 2002
PANIC theory and the prospects for a
representational theory of phenomenal
Michael Tye has recently argued that the phenomenal character of conscious experiences
is “one and the same as” (1) Poised (2) Abstract (3) Non-conceptual (4) Intentional Content
(PANIC). Tye argues extensively that PANIC Theory accounts for differences in phenomenal
character in representational terms. But another task of a theory of phenomenal consciousness is to account for the difference between those mental states that have phenomenal character at all and those that do not. By going through each of the four quali ers of PANIC, we argue that PANIC Theory fails to account for this difference in genuinely representational terms. We suggest, furthermore, that the reasons it fails are likely to be endemic to all representational theories of phenomenal conscious- ness.
1. Introduction: PANIC theory The Representational Theory of Mind (RTM) holds that “all mental facts are representational facts” (Dretske, 1995, p. xiii). RTM is attractive to naturalists, since they believe that all representational facts are natural facts . If all mental facts are representational facts, and all representational facts are natural facts, then all mental facts are natural facts.
A powerful objection to RTM is that facts about phenomenal consciousness are not representational facts. What are these facts? They fall into two groups: (i) facts about a mental state x having a different phenomenal character from a mental state y, and (ii) facts about a mental state x having phenomenal character at all. In the rst group are facts about phenomenal differences among different conscious experi- ences. In the second group are facts about the difference between mental states with no phenomenal character—e.g. belief, desire, and the other propositional attitudes —and mental states that do have phenomenal character. An example of (i) is the fact that Smith’s auditory experience of a piano an hour ago was phenomenally different from her auditory experience of a bagpipe yesterday. An example of (ii) is that both her auditory experiences had a phenomenal character at all, as opposed to Uriah Kriegel, Department of Philosophy, Brown University, Box 1918, Providence, RI 02912, USA, email: email@example.com ISSN 0951-5089/print/ISSN 1465-394X/online/02/010055–10 Ó 2002 Taylor & Francis Ltd DOI: 10.1080/09515080120109414 56 U. KRIEGEL her belief that Jones is a good pianist, which does not have any phenomenal character.
Call facts of these two types phenomenal facts. Proponents of RTM have recently attempted to argue that phenomenal facts are representational facts after all. Perhaps the most comprehensive defense of such a representationalist theory of phenomenal consciousness has been developed by Michael Tye. In a pair of books , Tye develops and defends the PANIC Theory of phenomenal consciousness, according to
which phenomenal character is “one and the same as” a speci c kind of representational content. The relevant content has four distinguishing properties. It is:
(a) Poised (b) Abstract (c) Non-conceptual (d) Intentional We will see what these properties are as we go along. Tye calls Poised, Abstract, Non-conceptual, Intentional Content PANIC. The thesis of PANIC Theory is
PT: The phenomenal character of a mental state S is “one and the same as” S’s PANIC.
PT entails—and perhaps is de ned by the conjunction of—the following pair of
(1) Two mental states S1 and S2 have different phenomenal characters iff S1 and S2 have different PANICs; and, (2) S1 has a phenomenal character and S2 does not have a phenomenal character iff S1 has a PANIC and S2 does not have a PANIC.
(1) and (2), if true, would account for phenomenal facts of both groups mentioned above. Smith’s auditory experience of a piano and her auditory experience of a bagpipe have different phenomenal characters in virtue of the fact that they have different PANICs; and Smith’s auditory experiences have phenomenal character at all, whereas her belief that Jones is a good pianist does not, in virtue of the fact that her experiences have PANICs at all whereas her belief does not.
For Tye, the purpose of theses (1) and (2) is to substantiate the idea that all phenomenal facts are representational facts; that phenomenal consciousness is not a counter-example to RTM. (1) is supposed to accommodate phenomenal facts of group (i) and (2) phenomenal facts of group (ii). That is, they are supposed to
substantiate, respectively, the following theses:
(3) For any states S1 and S2, the fact (when it is a fact) that S1 and S2 have different phenomenal characters is a fact about the representational properties of S1 and S2; and, (4) For any states S1 and S2, the fact (when it is a fact) that S1 does and S2 does not have phenomenal character at all is a fact about the representational properties of S1 and S2.
PANIC THEORY(3) accounts in representational terms for similarities and differences of phenomenal character; (4) accounts in representational terms for the difference between those mental states that have phenomenal characters and those that do not. Together, (3) and (4) substantiate—and perhaps de ne—a representationalist theory of phenomenal consciousness. In this way, PT protects RTM against the objection from phenomenal consciousness.
I am going to argue that this way of defending RTM fails. In particular, PT fails to defend (4). I shall argue that, when we look closer at what properties (a)–(d) are, we nd that they are all inadequate for a defense of (4). That is, I shall argue that each of these properties fails to distinguish, in genuinely representational terms, mental states with phenomenal character from mental states without phenomenal character.
Before starting, it is important to get clear on what we mean here by “representational fact”. By “fact” I will mean the instantiation of a property by a particular object, event, or state. A representational fact is the instantiation of a representational property by such a particular. A representational property is a property the particular has in virtue of its representational content. The representational content of the particular is what that particular represents (not, it is important to note, what does the representing). So a representational fact is the instantiation by a particular object, event, or state of a property the particular has in virtue of what it represents. To say that all mental facts are representational facts is therefore to say that all instantiations of mental properties by states of a subject are instantiations of properties these states have in virtue of what they represent.
2. An a priori suspicion Representationalists typically focus on defending (3): they try to show that phenomenal differences among mental states can be captured in terms of differences in representational content. But (4) is just as crucial to the viability of a representationalist theory. Such a theory must show how the difference between mental states with and without phenomenal character is also a difference in representational content. That is, it must show how the difference between the whole class of phenomenal states and the whole class of non-phenomenal states can also be captured in representational terms . This paper will focus on this latter task of representational theories of consciousness.
The problem is particularly acute because there are a priori reasons to suspect that no representational theory could account for the difference between phenomenal and non-phenomenal states. A representational theory must claim that the difference between phenomenal and non-phenomenal states is a difference in what those states represent. Therefore, a representational theory would have to identify certain environmental features that all and only phenomenal states represent. But prima facie it seems that every environmental feature can be represented either consciously or non-consciously. To suppose otherwise is to af rm the existence of environmental features which only lend themselves to conscious representation. It is implausible that the world should happen to contain such features.
58 U. KRIEGEL Moreover, the representational theory entails that if these environmental features did not exist, we would not be phenomenally conscious—in the sense that we would never harbor phenomenal states—since we would be unable to represent these features. Thus a world without the right features in the environment would necessarily be a zombie world. That, again, is rather implausible.
Call this set of considerations the “a priori suspicion”. This is the suspicion that no representational theory could ever account for the difference between phenomenal states and non-phenomenal states, because that would involve positing environmental features whose representation is necessarily phenomenally conscious. In what follows, we will approach Tye’s PT with the a priori suspicion in mind, and try to see what strategies he deploys in attempt to overturn it.
3. Abstractness and intentionality The burden of the present section is that properties (b) and (d)—abstractness and intentionality—cannot account for the difference between phenomenal and nonphenomenal states. This is because the representational contents of non-phenomenal states, such as beliefs, are also abstract and intentional in Tye’s sense.
What is it for a content to be abstract? The requirement of abstractness is
introduced by Tye to account mainly for illusions and other cases of misrepresentation (Tye, 2000, p. 62):
[One] condition is that the relevant content be abstract, that is, that it be content into which no particular concrete objects or surfaces enter. This is required by the case of hallucinatory experiences … A content is abstract just in case only abstract entities can “enter” into it. I will understand “entering into a content C” to mean being a constituent of C. Here is a genuine attempt to deal with the a priori suspicion: there is a certain class of objects—abstract objects—which phenomenal experiences represent.
The obvious problem with abstractness is that it provides a poor candidate for distinguishing phenomenal from non-phenomenal states. The reason Tye ascribes abstractness to the content of phenomenal states is that such states can misrepresent. But beliefs can misrepresent as well: x can believe that p when in fact not-p.
If anything, it is the propositional content of beliefs that is paradigmatically abstract, since propositions are abstract entities if anything is. So the difference between states with phenomenal character and states without phenomenal character cannot be that the former have abstract content whereas the latter do not .
What about intentionality? Content is intentional, we are told, in the sense that it is intensional (Tye, 2000, pp. 54–55). When we speak of propositional content, content is intensional in that it does not sustain existential generalization and substitution salva veritate. But PANIC, being non-conceptual, is non-propositional content. So what is it for non-propositional content to be intensional? Tye’s idea is to de ne the intensionality of the content of phenomenal states by reference to analogous logical features: from the fact that x perceives a yellow patch it does not follow that there is a yellow patch which x perceives (analogously to failure of 59
PANIC THEORYexistential generalization); and from the fact that x perceives a yellow patch it does not follow that x perceives a patch to re ect a certain percentage of the color spectrum (analogously to failure of salva veritate substitution).
Whether or not this account of “experiential intensionality” can work, we encounter here the same problem we encountered with abstractness: the content of beliefs and the other propositional attitudes is intensional as well. So the difference between phenomenal states and the propositional attitudes cannot be that the content of the former is intensional whereas that of the latter is not.
4. Non-conceptuality The difference between phenomenal states and non-phenomenal states—in particular, belief and the other propositional attitudes—must therefore reside in the property of poise and/or the property of non-conceptuality. This is what Tye himself
tells us (Tye, 1995, p. 138):
PANIC Theory entails that no belief could have phenomenal character. A content is classi ed as phenomenal only if it is nonconceptual and poised.
Beliefs are not nonconceptual, and they are not appropriately poised.
If we grant that phenomenal states have non-conceptual content, then non-conceptuality would seem to make the perfect candidate for distinguishing them from the propositional attitudes, since the content of propositional attitudes is obviously conceptual.
According to Tye, the content of phenomenal experiences must be non-conceptual because a subject x can experience today a speci c shade of red—say, red17— even if x lacks the concept of red17 and could not even recognize a sample of red17 tomorrow . I do not wish to contest this consideration here . But the fact that x can experience red17 even though x does not possess the concept of red17 does not show that the content of phenomenal states must be required to be non-conceptual.
At most, it shows that the content of phenomenal states must be allowed to be non-conceptual. For the content of phenomenal states to be required to be nonconceptual, it would have to be the case that x could not experience red17 if x did possess the concept of red17. But this is obviously false. Thus, if by some miracle x came to possess the concept of red17, x would not thereby lose the ability to have phenomenal experiences of red17.