In Subjective Consciousness (2009) Uriah Kriegel presents an ambitious, original and closely argued theory of consciousness. Indeed he claims to give a strictly sufficient reductive analysis of what it is to have a phenomenal experience and to provide a full account of what he takes to be the two main elements of phenomenal character: its qualitative aspect and its subjective aspect. Both aspects are explicated by appeal to a naturalistic notion of representation, and each in its way involves a form of self-representation. The theory is meant to be reductive in that none of the features specified in his definition are themselves phenomenal. I am not convinced that Kriegel has succeeded—indeed as I will explain below I believe he has not—but that is hardly surprising given the ambitious nature of his goal.
Following Joseph Levine (2001) Kriegel treats phenomenal experiences as having both a qualitative aspect and subjective or “for-me-ish” aspect. My conscious experience of the sky on a sunny day has both a particular qualitative “blue-ish-ness” (a specific hue of blue) of which I am aware and also a phenomenally manifest subjective aspect that involves its being intimately present to me as such, its “for-me-ness”. Kriegel provides an account of each aspect, and then integrates them to produce his overall theory of phenomenal character. His primary focus is on the later subjective aspect, and he expresses some residual reservations about his account of the former qualitative aspect. Thus, I will say just a bit about his account of qualia and focus mostly on his account of subjectivity.
With respect to qualia, Kriegel endorses a specific form of representationalism, the general view that qualitative character is to be explained solely in terms of representational content. On Kriegel’s account, the relevant content involves response-dependent properties essentially constituted by the fact they produce a particular sort of response in a given range of perceivers. The basic idea goes back at least as far as John Locke (1690) who thought of secondary properties, such as the colors of objects, as powers to produce particular sensible ideas in the minds of those who viewed them. Given his reductive aim Kriegel, unlike Locke, excludes any reference to phenomenal properties from his account, and defines the relevant responses as purely neural in nature. He also adds a meta-representational element. An experience’s being qualitatively blue is not a matter of the content it has but of the content it is represented as having. An experience M’s being qualitatively blue is thus not constituted by M’s representing some object as having the relevant response-dependent property, but rather by M itself being represented as having that content.
Kriegel explicates subjectivity or “for-me-ishness” in terms of self-representation of a specific sort. In his succinct formulation: “(SR) Necessarily, for any mental state M, M has subjective character iff M is non-derivatively, specifically and essentially self-representing” (p. 164).
Applying the combined account, which he calls the “self-representational theory of phenomenal consciousness”, to the blue sky example, he writes, “The experience has its bluishness (qualitative character) in virtue of non-derivatively, specifically, and essentially representing itself to represent the specific response-dependent property it does represent itself to represent; and it has its “for-me-ness” (subjective character) in virtue of non-derivatively, specifically, and essentially representing itself at all.” (p. 165).
Concerns about what Kriegel calls the “Via Positiva” or positive epistemic argument for the self-representational view.
The issue of “intimacy” and whether the representational approach can account for the apparently tight link between our conscious states and our awareness of them.
The problem of how to individuate states relative to Kriegel’s model.
The question of phenomenal adequacy and whether or not the proposed conditions in fact provide a reductive and strictly sufficient condition for phenomenal consciousness.
1 The via positiva
We have evidence that all conscious states are represented;
If we have evidence that all conscious states are represented it must be direct phenomenological evidence;
We have direct phenomenological evidence that all conscious states are represented only if all conscious states are consciously represented; therefore,
All conscious states are consciously represented;
If all conscious states are consciously represented, then they are all self-represented; therefore,
All conscious states are self-represented (p. 129).
Kreigel offers collateral arguments to support the main argument’s premises, especially steps two and five. The argument that he gives for five seems sound, but his argument for two is problematic.
In support of five, Kriegel argues that the alternative HOT theory would lead to a vicious infinite regress, but the self-representational theory does not. If every conscious state M1 was itself represented by a conscious state M2 distinct from M1, it would follow on the HOT theory that there would need to be another conscious state M3 which represents M2, as well as a conscious state M4 that represents M3, and so on ad infinitum (unless the series were a loop which there is no plausible reason to believe). However, the self-representational theory has no similar problem. Since the conscious state M represents itself, it follows that M is consciously represented, i.e., represented by a conscious state. No further conscious meta-state is needed, and no regress is generated. Thus, if the main argument were sound up to step four, it would be difficult to resist premise five or the final conclusion six.
However, the HOT theorist could well challenge premise two, by denying that direct phenomenological evidence is the only way to support the belief that all conscious states are represented. There seem at least two alternative lines of support for that claim.
First, one might make a general appeal to contemporary empirical theories of cognition that take representation of x to be the essential basis for any awareness of x (Hill 2009). It would follow automatically that any mental state that was conscious in the sense of being a state of which one was aware, would be a state that was represented. A conscious desire, intention or perception would need to be represented in so far as it was a state of which we are aware. Kriegel might reply that his claim concerns states that are conscious in the phenomenal sense rather than in the sense of “state of which one is conscious”. However, the HOT theorist like Rosenthal might in turn deny that there is anything that is it is like to be in a mental state unless one is aware of it. If so, the general appeal to representation as a condition for awareness would suffice to show that phenomenal, i.e., “what is it like” states, must be represented.
A second line of support would involve an inductive argument from the fact that whenever we introspect our conscious states we find a higher-order representation of them to the conclusion that such representations are also present when we do not introspect our conscious states. Indeed HOT theorists like Rosenthal claim that in the normal everyday case in which we are merely having conscious experiences but not attending to our mental states per se, the conscious-making HOTs are themselves not conscious. However, whenever I direct my attention inward, I find a HOT of the relevant sort. Thus, one might argue that such HOTs are always there whether I am introspecting them or not. Indeed that might be the best explanation of why I find them whenever I look for them.
Kriegel considers and rejects this second line of support (p. 119). However, his reasons for doing so are not persuasive. He argues that any such induction would be invalid because it extrapolates from a biased sample. In particular he denies that it would be legitimate to inductively generalize from an evidence base consisting solely of introspected states to a conclusion about conscious states that are not introspected. He treats the inference as a case of enumerative induction and argues that such inferences are valid only when the base cases and the target cases are relevantly similar which he says they are not in the proposed argument.
Similarity is no doubt relevant, but the HOT theorist might reply that his own inference compares favorably with undeniably valid cases of induction. One’s base cases always consist exclusively of instances that have been observed while one’s target cases do not. In the induction proposed by the HOT theorist, introspection plays the role of observation. Whenever, we “look” (i.e., introspect) we find a HOT accompanying our conscious states. Similarly whenever, the scientist takes a water sample along the course of the river, she finds pollution. The most plausible explanation in both cases would seem to be that the unsampled/unintrospected cases are the same as those we have checked. Why do we find a HOT or pollution in every sampled case? The best explanation would seem to be that it is always there whether we check or not. Thus, the inductive inference is not a simple matter of enumerative induction, but of inference to the best explanation (IBE) reasoning of the sort that Kriegel himself endorses in another context (pp. 123–124).
Kriegel would no doubt deny the analogy and argue that, unlike the water-sampling, introspection should be regarded as creating a HOT rather than revealing a pre-existing one. But I do not see that the dialectic favors such an interpretation. Thus, there would be good grounds for the HOT theorist to reject premise two of the main argument, and the Via Positiva should be regarded at best as inconclusive.
2 The objection from intimacy
In Chapter 3.2 Kreigel considers an objection to any representational theory of phenomenal self-awareness which he terms the “objection from intimacy” and attributes to Joseph Levine (2001) and Benj Hellie (2007). He describes it as motivated by “the strong sense that the relationship between inner awareness and that of which it makes one aware is more intimate than the relationship between a representation and that which it represents” (p. 108). He adds, “The idea of intimacy is sometimes cast in terms of the thought that there could be no gap between how a conscious experience is and how it appears to one to be—between the qualitative properties the experience actually has and the qualitative properties one is aware of it as having…”(p. 108). He believes it is these intuitions that lead some philosophers to talk about being directly “acquainted” with qualia. Representation may seem too cognitive and indirect to account for the sense of direct “mental contact” with qualia, which seems to allow for no gap between appearance and reality.
Kreigel’s response is to accept the basic intuition and try to accommodate it within his theory. His central move is to treat the self-representation of a state’s perceptual content as constitutive of its qualitative character. According to Kriegel a state’s having a given qualitative character, e.g., sky-blue, consists in its being represented as having the content of representing some object as having that color (i.e., as having the relevant response-dependent property). Because the state’s qualitative character is constituted by how itself is represented as being, there is thus no “room” for any gap between what qualia it has and how it is represented as being. Its having the relevant qualitative character just is a matter of how it is represented to be. Appearance in this case is reality.
Kriegel’s theory of qualia differs from more standard versions of representionalism in identifying qualitative character not with the externally directed content of the perceptual state but with the content it represents itself as having. He introduces the notions of “schmalitative” character to refer to the more typical externally directed content. A perceptual state is schmalitatively sky-blue if it represents some object (e.g., the sky) as having the relevant response-dependent property. However, the state’s being qualitatively blue is constituted by its representing itself as having that schmalitative property. Kriegel thus aims to collapse the gap between appearance and reality with regard to qualia. A state by definition has whatever qualitative character it represents itself as having.
As Kriegel acknowledges, (p. 111) some of his critics will find this maneuver inadequate as an account of the intuitive intimacy that motivates appeals to acquaintance. They want the awareness of qualia to be immune to error in a way that involves a genuine cognitive achievement rather than definitional avoidance. In response, Kriegel disputes that stronger intuition. His sympathy for the intuition of intimacy extends only as far that which his theory can accommodate, and he rejects the demand for any stronger form of immunity.
As with so many clashes of intuitions, the dispute in this case is difficult to adjudicate. My sympathies lie more with his critics, and his account of intimacy seems a bit like philosophical slight of hand. In some respects it echoes what used to be said about the common sense or ordinary language use of “seems”. When one is looking at an object, one can be wrong about what shape it has, but not about what shape it seems to have. Such immunity to error supposedly followed from the rules for the use of “seems” which simply expresses a subjective view about which there is no objective fact of the matter (see also Dennett’s (1991) denial of objective seeming).
In so far as one feels that there is phenomenal sense of “seems” that goes beyond the merely epistemic sense, one is likely to be less than satisfied with Kriegel’s account of intimacy. On his theory, a state’s qualitative character by definition is fully constituted by what schmalitative properties it represents itself as having. There is no room for its having any qualia other than those it represents itself as having. Again how it seems is how it is by definition. Though I cannot produce any clear refutation of that view, I confess it leaves me like his imagined critics less than satisfied. My own inclination would be to account for intimacy by building a tighter and inter-dependent link between the state’s other-directed content (its content about the sky) and its self-directed content (its content about itself) as I have tried to do elsewhere (Van Gulick 2004, 2006). Consideration of such options leads naturally to my third concern about how to individuate mental states.
3 Individuating mental states
One of Kriegel’s major aims is to distinguish his theory from standard higher-order theories of consciousness. How to individuate or count states is critical to the dispute. On the HOT theory, the existence of a conscious requires two distinct mental states: the higher-order thought (HOT) and the conscious state that is its target or intentional object. On Kreigel’s self-representational theory there is just one state with multiple contents and parts, with one part reflexively representing the whole of which it is a part. Thus, the criteria for individuating states are of central importance.
Some standard ways of individuating mental states would seem unfavorable to Kriegel’s view. For example, states are often individuated on the basis of their contents. The individuation of contents is itself a controversial matter—e.g., coarse grained versus fine grained. But however, contents are distinguished, it is common to assume that two mental states that differ in content are distinct states. Identity of content may not be sufficient for identity of mental states, but it is typically regarded as necessary.
Individuating states by content would seem to favor the HOT theory. For example in the case of a conscious perception, the HOT theorist attributes an externally directed first-order content to the perception and higher-order or meta-intentional content to the HOT. There are two differing contents and two mental states. On Kriegel’s view, there is just one state, which has multiple quite different contents—again one of which is first-order and externally directed and one of which is meta-intentional. According to Kriegel, these contents are associated with two parts of one state, but individuating states on the basis of content would seem to imply they are two mental states rather than just one.
Kreigel will likely reply that on his view a conscious state is a complex each of whose parts has a distinct content, and the content of the whole complex is the sum of the contents of its parts. Perhaps so, but one can then ask what reason is there for regarding the combination as a single complex state rather than a set of distinct connected states as on the HOT theory. One answer might the Via Positiva, but as argued above that argument is less than persuasive.
Other means of individuating mental states might be considered, but it is not clear that any will be of much help to Kriegel. One could perhaps regards mental attitudes with differing contents as a single state if they have a common realization base, i.e., if they are implemented by the same physical or neural structures (where “same” here means the very “same neural tokens”). Even if one could resolve the thorny issue of how to individuate token neural structures, the proposal is unlikely to be of much help to Kriegel. The neural substrates for conscious states are likely to be quite global and widely distributed across the brain. Indeed the cortical regions associated with first-order and meta-intentional content are likely to be distinct, and thus not to provide any independent bottom-up reason to regard the supposed complex as a single mental state.
A similar problem would recur if one individuated states on the basis on functional roles, since again the first-order perception and the meta-intentional awareness clearly differ in their respective roles, and would thus seem to be distinct states.
It seems that any solution favorable to Kreigel’s view will need to acknowledge that the parts of his proposed mental complexes differ in content, function and realization basis, but that they nonetheless constitute a single mental state because of some way in which they mentally cohere or in which they are mentally unified. I do not think Kriegel gives a satisfactory account of that sort in Subjective Consciousness.
Nonetheless, I think one may be possible. Indeed I have myself suggested elsewhere that the key might be in showing that the multiple contents associated with such a conscious complex are essentially interdependent rather than additive. The content of my conscious experience of the desk before me is not best thought of a sum such as “there is a desk before me’ plus “this state is an (or my) awareness of a desk”. Rather the content of the experience is given by something like “there is a desk present before me as part of my experienced world”. The later formulation builds in the subjective or self-directed content as part of the world-directed content about the desk. Whatever the merits of that specific proposal, it seems some further account of mental unity is needed to support Kriegel one-complex-state view against the HOT theorist’s two-state alternative.
4 Phenomenal adequacy
Kreigel’s explicit goal is to give a reductive and strictly sufficient condition for being phenomenally conscious. I believe he fails to do so in two respects. First his conditions may not be as reductive as he claims, in so far as his account of the qualitative aspect of the phenomenal may rely on an implicit appeal to phenomenal properties. Second his account appears vulnerable to counterexamples of a familiar sort that would seem to challenge its sufficiency.
The concerns about reduction relate to how the relevant response-dependent properties are specified or identified. On a non-reductive theory, like Locke’s classic account, an object’s being blue is a matter of its having the power to produce blue sensory ideas in relevant perceivers. The appeal to blue ideas renders the analysis non-reductive since it appeals to phenomenally blue ideas. As noted above, Kriegel aims to define the required response-dependent properties in terms of the neural responses they produce in relevant classes of human perceivers.
The problem arises in specifying those classes of perceivers. Kriegel acknowledges the possibility of inverted or shifted qualia according to which not all humans need produce the same neural response in relation to the relevant external property. Where I see blue you might see green. Thus, the relevant external properties that are schmalitatively represented in experience need to identified in terms of the responses they produce in the right group of perceivers, i.e., in groups that respond in the relevantly same way to that property. But what is the relevant sense of sameness and how is it to be delineated? Since there is likely some room for neural variation and multiple neural substrates in how a given color experience is realized, one could not require that all the relevant perceivers produce the exact same neural response down to its smallest physical details. Thus, on what basis is one to count the various neural responses as of the same type? Here is where the possibility of a hidden implicit appeal to the phenomenal enters in. I suspect the answer is that the relevant neural responses count as the same just if all the members of the class underlie the realization of a phenomenally blue experience. Kriegel clearly makes no such claim and he would surely deny it. But if so, I believe he needs to give some alternative account of how he identifies and individuates the relevant classes of neural responses without compromising his reductive aims.
As to sufficiency, it seems one can imagine beings or cognitive systems that satisfy all of Kriegel’s conditions but that nonetheless lack phenomenal consciousness. For example imagine a robot that has detectors that enable it track the exact causal profile associated with the relevant response-dependent property SB that constitutes being sky-blue relative to a specific group of humans who have blue color experiences. The robot itself need not have the relevant neural structures nor produce the relevant neural response to SB. It need only be able to reliably detect SB based on its causal profile. The firing of the robot’s detectors would generate an internal representation R which carries the content that the scanned object X has the property SB. Imagine further that R occurs as part of a unified complex representation R*, another part of which R† carries the information that the complex R* as a whole represents X as having SB, and that R† does so in way that is specific, non-derivative and essential. I see no reason why such conditions could not be satisfied in the robot case, nor any reason to believe that the naturalistic theory of representation that Kriegel applies to human cognitive states could not be applied in the same way to the robot’s states. Yet intuitively one feels little or no inclination to believe the robot would have any qualitatively blue experience in the phenomenal “what-it-is-like” sense. Kriegel’s view appears open to such standard counter examples, challenging its claim to provide a strictly sufficient condition.
In his final chapter, Kriegel tries to undermine the basic intuition on which such objections rely: that phenomenal consciousness is not fully functionalizable and includes some essential nonfunctional aspect. Kriegel argues the basis of that intuition is epistemic rather than metaphysical and derives from the way in which we know about our own phenomenal experience. In particular, he argues that our knowledge of our phenomenal experience is constitutive rather than causal. He claims that our knowledge of our own experience is not the result of any causal impact it has on us, but is a constitutive aspect of the experience itself, its subjective for-me-ness aspect. According to Kriegel, it is because he know our experience in a way that is independent of its causal profile that it seems to us to involve something other than its causal and functional nature. He argues that once we recognize the source of the intuition, it is rational to regard it as illusory and as providing no real basis for believing that phenomenal consciousness is not functionalizable.
I agree that we should be cautious about the relevant intuitions and about any metaphysical conclusions based on them (Van Gulick 2003), but I find Kriegel’s particular attack on them less than convincing. Even if self-representation is a constitutive component of the complex state posited by Kriegel, it seems that component needs to track the other aspects of the complex, such as it schmalitative content, in order to carry information about it. Thus, how could our knowledge of our own phenomenal experience not have a causal basis? Though such knowledge is not typically based on any inference from observations of causal effects, it still seems it must have a causal basis. If so Kriegel’s attempt to explain away the relevant intuitions fails, and the standard counter examples remain a challenge to the sufficiency of his condition.
- Dennett, D. C. (1991). Consciousness explained. Boston: Little Brown and Company.Google Scholar
- Hill, C. (2009). Consciousness. Cambridge: Cambridge University Press.Google Scholar
- Kriegel, U. (2009). Subjective consciousness: A self-representational theory. New York: Oxford University Press.Google Scholar
- Locke, J. (1690). An essay concerning human understanding. London: Thomas Ballet.Google Scholar
- Rosenthal, D. M. (1997). A theory of consciousness. In N. Block, O. Flanagan, & G. Guzeldere (Eds.), The nature of consciousness: Philosophical debates. Cambridge, MA: MIT Press.Google Scholar
- Van Gulick, R. (2003). Maps, gaps and traps. In A. Jokic & Q. Smith (Eds.), Consciousness: New philosophical perspectives (pp. 323–352). Oxford: Oxford University Press.Google Scholar
- Van Gulick, R. (2004). HOGS (higher-order global states): an alternative higher-order model of consciousness. In R. Gennaro (Ed.), Higher-order theories of consciousness (pp. 67–92). Amsterdam and Philadelphia: John Benjamins.Google Scholar
- Van Gulick, R. (2006). Mirror mirror, is that all? In U. Kriegel & K. Williford (Eds.), Self-representational approaches to consciousness (pp. 11–40). Cambridge, MA: MIT Press.Google Scholar