97

The higher order approach to consciousness is defunct NED BLOCK The higher order approach to consciousness attempts to build a theory of consciousness from the insight that a conscious state is one that the subject is Analysis Vol 71 | Number 3 | July 2011 | pp. 419–431 doi:10.1093/analys/anr037 The Author 2011. Published by Oxford University Press on behalf of The Analysis Trust. All rights reserved. For Permissions, please email: journals.permissions@oup.com Downloaded from analysis.oxfordjournals.org at Harvard University on July 22, 2011420 | ned block conscious of. There is a well-known objection to the higher order approach, a version of which is fatal (Balog 2000b; Byrne 1997; Kriegel 2003; Levine 2001; Mandik 2009; Neander 1998; Rey 2000; Van Gulick 2000, 2004; Weisberg 2010). Proponents of the higher order approach have realized that the objection is significant. They have dealt with it via what David Rosenthal calls a ‘retreat’ (2005b: 179) but that retreat fails to solve the problem. The fatal objection is short and simple, but some ground-clearing is required in order to state it. There are two quite different approaches to the nature of consciousness. One line of thought (that I have favoured) emphasizes the notoriously elusive ‘what it is like’ (Nagel 1974) to have an experience (Block 1978; Chalmers 1996; Levine 1983; McGinn 1989). Higher order theorists often use the phrase ‘what it is like’, supposing, controversially, that their theories are able to account for what-it-is-like-ness, and I will contest that claim, taking advantage of that common terminology. The second approach is the one that emphasizes that a conscious state is a state one is conscious of. This perspective, which is the topic of this article, is supported by one strand of ordinary usage of the term ‘conscious’. As William Lycan (1996: 25) says ‘I cannot myself hear a natural sense of the phrase ‘‘conscious state’’ other than as meaning ‘‘state one is conscious of being in’’ ’. Perhaps these two approaches are best understood as concerned with two different mental kinds that correspond to two different senses of the term ‘conscious’ (Block 1995; Lycan 2004a, 2008). Alternatively, perhaps they are best thought of as conflicting approaches to the same thing (Weisberg 2010). A third and I believe better alternative is that they are non-conflicting approaches that should be woven together into a single account that emphasizes both perspectives. (See the last paragraph of this article for an attempt in this direction.) It is, however, a mistake to suppose that the approach that emphasizes what-it-is-like-ness insists on a characterization of the phenomenon of consciousness as non-relational, i.e. intrinsic in one sense of that term (Dennett 1988; Harman 1990; Weisberg 2010). That characterization derives almost entirely from opponents of the view and should be suspect on that ground alone. Many of the theorists who emphasize what-it-is-like-ness are also scientific realists, and all the major recent accounts of what consciousness is in the brain have been heavily relational (Block 2005b, 2007, 2009). Higher order approaches divide into those on which the higher order state is perception-like (Armstrong 1968, 1978/1997; Lycan 1987, 1996, 2004b) and those on which the higher order state is a thought (Rosenthal 1986, 1997, 2005b; Weisberg 2010). The point I will make applies to both, and perhaps also to higher order ‘dispositional’ theories of consciousness (Carruthers 2000), but except in this sentence, I will be talking entirely about the non-dispositional versions. I will mainly be concerned with the Downloaded from analysis.oxfordjournals.org at Harvard University on July 22, 2011higher order approach to consciousness | 421 higher order thought (HOT) version of the view, with occasional mentions of the higher order perception view. According to (non-dispositional) versions of the higher order approach, what it is for a mental item– a happening or state– to be conscious is for it to be the object of another mental item– that has a particular causal origin. This ‘particular causal origin’ is meant to exclude the HOT about your sensations that you infer from observing your own behaviour or from a brain scan. Sometimes the restriction on origin is spelled out as: not observational or inferential (Lycan forthcoming) or as not involving inference or observation that satisfies a certain access condition (Rosenthal 2005a), a fine point that I will ignore. Thus (collapsing observation into inference for simplicity) the two versions of the higher order approach agree on the following biconditional claim: The higher order theory: A mental state is conscious if and only if the state is the object of a certain kind of representation arrived at non-inferentially. The notion of ‘object of’ is sometimes cashed out in terms of mere accompaniment and sometimes in terms of representation. I shall tend towards the latter formulation. In the case of the HOT version, the representation is an assertoric1 thought. It should be noticed that when I have a conscious pain, the second order representation in virtue of which the pain is conscious (or perhaps the third-order representation in virtue of which the second-order representation is conscious) can be expected to be unconscious. Humans do not routinely achieve ever higher and higher orders of representation. We now come to the most significant preliminary point, one that has not figured in previous formulations of the fatal objection: a higher order theory of consciousness can be held as an immodest (or ambitious) theory that purports to capture what-it-is-like-ness, or alternatively as a modest theory of one kind of consciousness, or of consciousness in one sense of the term, higher order consciousness. The modest version of the higher order approach recognizes another kind of consciousness (or consciousness in another sense of the term), what-it-is-like-ness or phenomenal consciousness. The contrast between the modest and ambitious forms of the view can be illustrated with regard to the following question: why should it be that when 1 ‘Assertoric’ is introduced by Rosenthal in part to try to avoid refutation by the case of unconsciously feeling guilty about, e.g. unconsciously wanting to kill one’s father and marry one’s mother, an unconscious higher order state that is about another unconscious state (Rosenthal 2005b: 185). It also is an attempt to avoid a problem raised by Kati Balog (2000a). In one form, Balog’s point is that if the HOT makes for consciousness all by itself, one would not have to bother arranging pleasurable experiences or even imagining them since merely thinking one has them would be just as good. Assertoric thinking is supposedly not so ‘easy’ to do (Rosenthal 2000). Downloaded from analysis.oxfordjournals.org at Harvard University on July 22, 2011422 | ned block you put together an unconscious pain with an unconscious thought about it, you get a conscious pain? On the modest view, this question gets the not very substantive answer that that is what higher order consciousness amounts to, and this answer allows that even when it is not an object of any higher order state, an ‘unconscious’ pain might nonetheless be conscious in the what-it-is-like-ness sense of the term. The ambitious theorist, in contrast, needs a substantive response. Is the modest version of higher order theory purely terminological? Not quite, since even if one regards it as an account of consciousness in the higher order sense of the word ‘conscious’, there are a number of alternative accounts of that sense of term– for example, the HOT and higher order perception accounts. In a recent defence of the ambitious higher order perspective, Josh Weisberg (2010: 18) makes much of the claim that the folk ‘will not consider a state conscious if the subject is in no way aware of it’. But this observation can be explained by the modest thesis that the higher order sense of the term is easy to evoke in folk usage. I believe that it is an unnoticed oscillation between the relatively innocuous modest version and the controversial ambitious version that fuels the fire of the higher order approach. Rosenthal has been clear in being an ambitious theorist and in the days before Karen Neander’s (1998) version of the objection that I will be elaborating here, Lycan sounded rather like an ambitious theorist. I quoted Lycan (1996: 26) on the verbal aspect of higher order approaches, but the point he is making in that passage is really that the verbal point cuts no philosophical ice. He says (26) ‘Although I cannot myself hear a natural sense of the phrase ‘‘conscious state’’ other than as meaning ‘‘state one is conscious of being in’’, the philosophical use of ‘‘conscious’’ is by now well and truly up for grabs, and the best one can do is to be as clear as possible in one’s technical specification.’2 I believe that the main advantage of the higher order approach is a demystifying perspective on the change from a conscious experience of a pain to an unconscious experience of it (or the reverse change). I have the conscious pain, then I am distracted, and when the pain is not accompanied by any higher order state about it, it is unconscious. However, note that this account is as available to the modest version of the view as to the ambitious version. The main competitor to the higher order account in its ambitious form (and, arguably in its modest form) has been the ‘same-order’ account, according to which conscious experience is reflexive in that it consists in part in an awareness of itself (Brentano 1874/1924; Burge 1997, 2006; Byrne 2004; 2 This attitude can be contrasted with Lycan’s (2001: 3–4), in which he gives an argument for a modest form of the view based on ordinary usage. Lycan tells me he was never an ambitious theorist. Downloaded from analysis.oxfordjournals.org at Harvard University on July 22, 2011higher order approach to consciousness | 423 Caston 2002; Kriegel 2005; Kriegel and Williford 2006; Levine 2001, 2006; Metzinger 2003; Ross 1961; Smith 1986). For the same-order account to explain the transition between conscious and unconscious states, it would have to supply an account of the difference between a pain having and lacking the self-referential property and why this self-referential property comes and goes with consciousness. I believe that this issue is of a piece with the famous ‘explanatory gap’ (Levine 1983), and that, as with the gap, no one has a glimmer of a clue of an idea of how to think about it. (But I don’t take this as a disadvantage of the same-order theory– the higher order account makes consciousness out to be less puzzling than it really is.) The fatal problem stems from a radical misrepresentation case in which the item that the HOT represents, the ‘target’ in Rosenthal’s terminology, does not exist. Can this happen? There are a variety of neurological syndromes in which patients confabulate– what they do or experience is at variance with what they think they are doing or experiencing. A prominent example is anosognosia in which blind patients think they see or paralysed patients think they are moving the paralysed arm. In some cases, it appears that the patient is hallucinating, but in other cases the disorder genuinely seems to be a disorder of thought, a matter of ‘confabulation’ (Fotopoulou 2010). Karen Neander (1998) introduced an example of triplets, each of whom has an assertoric thought expressible as ‘I have a sensation of green’. One has a sensation of green; another has a sensation of red; and in the empty higher order thought case, the triplet has no relevant sensation at all. Ambitious higher order theorists such as Rosenthal say that what-it-is-like-ness for all three triplets is exactly the same. As he puts it: Suppose my higher-order awareness is of a state with property P, but the target isn’t P, but rather Q….A higher-order awareness of a P state without any P state would be subjectively the same whether or not a Q state occurs. The first-order state can contribute nothing to phenomenology apart from the way we’re conscious of it. (2004: 32) Or, as Rosenthal put it more recently: If one has a sensation of red and a distinct HOT that one has a sensation of green, the sensation of red may nonetheless be detectable by various priming3 effects. But what it will be like for one is that one has a 3 Unconscious exposure to a stimulus can facilitate later responses to the same or related stimuli. That is unconscious priming. Downloaded from analysis.oxfordjournals.org at Harvard University on July 22, 2011424 | ned block sensation of green. Similarly if one has that HOT with no relevant sensation at all. (2009b: 249) 4 What is ambitious about ambitious higher order theories of consciousness is that a state of consciousness is a state of what-it-is-like-ness. However, the ambitious higher order account is committed to the converse as well. We can understand this converse link by considering sensory qualities such as what distinguishes the experience of red from green or pain from thirst. According to some theorists who emphasize what-it-is-like-ness (me for example), these sensory qualities have a kind of consciousness in themselves (namely, what-it-is-like-ness– what I call ‘phenomenal consciousness’). In contrast, higher order theorists think that those sensory qualities can exist without any kind of consciousness. And to preserve the significance of the higher order theory, they say that those unconscious qualities have no what-it-is-like-ness either. I say ‘preserve the significance’ because conscious pain and pleasure matter to us in a way that depends on what-it-is-like-ness; if the putatively unconscious pains had what-it-is-like-ness, it would be a mystery how they could be unconscious. Lacking consciousness requires lacking what-it-is-like-ness and so a state of what-it-is-like-ness is a state of consciousness. As Rosenthal (1997: 411) says ‘…what it’s like for one to have a pain, in the relevant sense of that idiom, is simply what it’s like for one to be conscious of having that pain’. Here is the brief argument I have been building up to. I’ll put it in terms of the HOT theory, though I believe it applies equally to the higher order perception account. Suppose that at time t, I have an assertoric higher order thought to the effect that I am experiencing seeing something green, but in fact I am having no visual representation at t: the thought is ‘empty’. Let us suppose further that the higher order thought is not arrived at inferentially. Also, I have no other higher order thoughts at t. The theory supplies a necessary and sufficient condition for a conscious episode. An episode is conscious at t if and only if it is the object of an assertoric higher order thought at t, arrived at non-inferentially. The sufficient condition dictates that this thought at t is sufficient for a conscious episode at t. By the necessary condition, that conscious episode at t is the object of a simultaneous higher order