98

Indeed, Rosenthal thinks that we have evidence of such a determination of what it is like by a HOT independently from any first order state, in a ‘change blindness’ experiment by James Grimes (1996). As Rosenthal (2009a: 162) puts it: ‘In one case of change blindness, a large parrot switches back and forth between being red and green… even when the parrot’s color changes, there is often no change in what it’s like for subjects; Grimes’s subjects often continue seeming to see red when the parrot is green.’ No doubt someone else has endorsed such a view of ‘change blindness’, but I haven’t seen it. And for good reason: in standard ‘change blindness’ experiments, items can change colour 50 times before the subject notices a change, and it would be a mystery why the subject would settle on one interpretation rather than another to maintain throughout those colour changes. Downloaded from analysis.oxfordjournals.org at Harvard University on July 22, 2011higher order approach to consciousness | 425 thought. In the example, there is only one higher order thought at t, and we can assume it is not self-referential. So there is no conscious episode at t after all. Thus, the sufficient condition and the necessary condition are incompatible in a situation in which there is only one non-self-referential higher order representation. The reason I keep repeating the ‘at t’ is to emphasize that because there is an appropriate occurrent higher order thought at t, there must be an occurrence of consciousness at t. An occurrence of consciousness must be the object of a simultaneous occurrent thought, but there is only one thought. One could suppose that it is its own object, but that would make the theory a version of the same-order theory and would not be a defence of the higher order theory. That is the advertised incoherence.5 Weisberg (2010) says what is conscious is always an intentional object; however sometimes the intentional object exists and sometimes not. (If any non-philosophers have gotten this far, the intentional object of a mental state is what it is ‘about’. Ponce de Leon searched for the Fountain of Youth and so the (non-existent) Fountain of Youth is said to be an intentional object.) One could put Weisberg’s point by saying that there is no such thing as a HOT with no object because there is always an intentional object that is a conscious state, though sometimes a non-existent one! Rosenthal has taken this line though somewhat less explicitly. For example (2002: 415), ‘Conscious states are states we are conscious of ourselves as being in, whether or not we are actually in them.’ And (2005a: 209): ‘…HOTs determine what it’s like for one to be in various conscious qualitative states. So erroneous HOTs will in this case result in there being something it’s like for one to be in a state that one is not actually in.’ However– and this is where ambition comes into play– this line of thought does not avoid the argument I gave, because it ignores the fact that according to the ambitious form of the view, if a subject has a higher order thought of the right sort at t, then there is something it is like for the subject at t, an episode or occurrence of what-it-is-like-ness at t. No doubt some defenders of the HOT theory will say that in the ‘empty’ cases, what-it-islike-ness and consciousness are both instantiated only in intentionally inexistent states. If they say this, they reveal that the what-it-is-like-ness they invoke is fake. To see this, consider the triplets again, but this time let the HOT be the thought that one is in intense, agonizing pain. As before, each of the triplets has a different first-order ‘sensory’ state: one triplet has agonizing pain, another has the sensation of tasting chocolate and the third has no relevant first-order state at all. What it is like for all the triplets is exactly the same 5 I say ‘incoherent’ rather than ‘contradictory’ because the theory does not literally contain or entail sentences of the form p and of the form p, but rather gives conflicting answers about one kind of case. Downloaded from analysis.oxfordjournals.org at Harvard University on July 22, 2011426 | ned block according to Rosenthal and Weisberg: (‘A higher-order awareness of a P state without any P state would be subjectively the same…’ (Rosenthal 2004: 132).) But how can it be that that the triplets all have states that matter in exactly the same way– in this case three states that are bad in themselves (or good in themselves) in the way that real agonizing pain is bad (or that intense pleasure is good)?6 If what-it-is-like-ness is supposed to matter in the same way whether it exists or not, that just shows that ‘what it is like’ is being used in a misleading way. Of course, the higher order theorist could avoid phony ambition by moving to the modest view that the higher order description of the three triplets as having the same higher order consciousness is compatible with very different kinds of what-it-is-like-ness in the three cases. Talk of intentional objects and intentional inexistents can serve some legitimate purposes, but such talk can be misleading. This is a case where it misleads. It will be useful to distinguish between the aboriginal higher order thought view and the new version, produced in response to the empty HOT problem. The aboriginal higher order thought view explained each conscious happening in terms of a pair of actual occurrences, one said to ‘accompany’ or to be ‘simultaneous’ with the other. For example, Rosenthal (1997: 741) encapsulates the theory as follows: ‘The core of the theory, then, is that a mental state is a conscious state when, and only when, it is accompanied by a suitable HOT.’ (My computer counted 15 uses of variants of the term ‘accompanied’ in this article.) Accompaniment is a straightforward and non-mysterious relation between distinct happenings. Rosenthal (2005b: 179) continues to introduce the theory in terms of the attractive idea of an extrinsic relation, as in: ‘The idea that a mental state’s being conscious is an extrinsic, or relational, property of that state fits strikingly well with the way we actually talk about consciousness.’ But he adds a footnote: ‘There will be reason to retreat on the claim that a state’s being conscious is strictly speaking relational…’ And it is the version to which he has retreated, the non-extrinsic relational view that is the source of the trouble. The aboriginal (extrinsic relational) higher order thought view says (for example) an actual pain is conscious if and only if it is the object of an actual appropriate higher order thought. Someone can hold this view without holding that an empty higher order thought all by itself makes anything conscious. Certainly, the modest higher order theorist can say ‘The way we talk about consciousness dictates an answer to some questions about which states are conscious, but the presence or absence of consciousness in a case of a higher order thought with no object simply is not decided by the way we talk.’ However, this is not a satisfactory response for a higher order 6 Those who take pain asymbolia (Grahek 2007) as showing that what it is like to have a pain is not bad in itself can substitute the experience of hurting or awfulness in the argument. Downloaded from analysis.oxfordjournals.org at Harvard University on July 22, 2011higher order approach to consciousness | 427 theorist who has metaphysical ambitions, especially of the persuasion that I have termed ‘ambitious’, since it exposes the fact that the modest theorist is not giving a metaphysical account of the nature of consciousness or what-it-is-like-ness. The aboriginal higher order thought view does not have the form of a claim about what consciousness (or what-it-is-like-ness) is, but rather about what the conditions are in which an actual state is a conscious state, leaving it open what to say about the ‘empty’ case. So there is intellectual pressure for the theorist with both kinds of ambition to move from the aboriginal theory to the new one by claiming that a higher order thought with no object can all by itself make for consciousness and what-it-is-like-ness. I raised the issue earlier of how it can be that an unconscious pain combined with an unconscious thought about it can yield consciousness. The new ambitious view has an answer to this, albeit an implausible one, in terms of some kind of sufficiency: an appropriate higher order thought is somehow sufficient for consciousness (and what-it-is-like-ness) to occur. However, this ambition is self-destructive, since the new version of the higher order view has to say, as I just noted, that what-it-is-like-ness that is non-existent can matter just as much as what-it-is-like-ness that exists, and that amounts to abusing the notion of what-it-is-like-ness. There is, however, a way out: to say that what-it-is-like-ness is a property of the HOT itself. That is, the HOT in a case of an empty HOT is the very state that is conscious and has what-it-is-like-ness. Neander’s triplets would now make sense: each is having the same HOT and so no wonder what-it-is-like for them is the same. However, it is not clear how this view can preserve the insight that a conscious state is a state one is conscious of. Perhaps the HOT is always self-reflexive– a state of consciousness of itself. But, as noted earlier, this would be a version of the rival ‘same-order’ theory. It may be said that the higher order approach can do without what-it-is-like-ness altogether. But the point I have been making applies to consciousness too. If a theory of consciousness is worth anything, it must be about consciousness in a sense that matters to us in the way that conscious agony or ecstasy matters. If a state of being conscious of agony is supposed to matter equally whether it exists or not, the supposed theory of consciousness is worthless. Thus, the higher order thought perspective faces a number of alternatives of which these are the most salient: (1) Adopt a modest view that prescinds from real metaphysical questions about the nature of consciousness. My objection does not apply to this option– but see the paragraph below for an alternative. (2) Swallow the consequence that non-existent conscious agony can be just as bad as existing conscious agony, in which case the view is about consciousness in a merely technical sense of the term. Downloaded from analysis.oxfordjournals.org at Harvard University on July 22, 2011428 | ned block (3) Move to a version of the same-order view. At the outset I mentioned that the higher order approach assumes that the phenomenon that higher order theories of consciousness are trying to explain is that a conscious state is one that the subject is conscious of. I mentioned an alternative point of view that emphasizes the phenomenal aspect of experience, ‘what it is like’ to have a conscious experience. The two points of view can be usefully combined if we hold that a reflexive conscious state is one that is phenomenally presented in a higher order representation of it (Block 2005a). Since phenomenal presentation of a state requires that it exist, there is no ‘emptiness’ problem. Since this account uses a phenomenal state in a representation of it, there is no issue of mistaken representation. Further, the higher order theory (in its ambitious form) is ad hoc (Block 2009; Byrne 1997; Van Gulick 2004). If consciousness is a matter of a relation between a higher order representation and a first-order representation, why should it matter what the causal origin of that higher order representation is? So why is it OK to stipulate that it cannot have arisen through observation (e.g. if I observe my own angry behaviour and so come to think about my anger) or inference? If the higher order view in question is a modest theory based in how we use the word ‘conscious’, it would make sense to read conditions off usage, but if the theory is meant to– ambitiously– go deeper into metaphysical reality, the theory should be one for which no ad hoc conditions are required. And this test is passed by the theory that says that a reflexive conscious state is one that is phenomenally presented in a higher order representation of it. The origin of the phenomenal presentation does not matter. Asimilar point applies to the fact that in the higher order account it has to be stipulated that the higher order representation occurs at the same time (or at least overlaps in time) and in the same mind as the first-order representation it represents (Block 2005a, 2009; Van Gulick 2004). No such stipulations are required in the theory that a reflexive conscious state is one that is phenomenally presented in a higher order representation of it, since a state cannot be presented in a representation of it without a kind of temporal overlap, and the presentation relation requires the same mind. The advocate of naturalization and demystification will baulk at the use of an unreduced notion of phenomenal, but the naturalistic credentials of the ambitious higher order perspective are themselves suspect given the upshot that conscious agony and ecstasy matter equally whether they exist or not.7