The following is a long excerpt from an unpublished paper I wrote in 2012-2013, mostly before I was enmeshed in rationality-community ideas. The paper was a response to David Chalmers’ “hard problem of consciousness,” described well in “Facing up to the Problem of Consciousness” and “Consciousness and its Place in Nature.”
Chalmers gives various arguments for thinking that phenomenal consciousness isn’t reducible to merely physical facts. A complete reductive explanation must make it logically impossible for the reduced entity to differ in any way unless the thing you’re reducing it to also differs in some way. Chalmers argues that reductions of consciousness to physical facts can never be complete in this sense, because there is some aspect of consciousness that could in principle vary without varying any physical fact. This aspect is the first-person, subjective, phenomenal character of consciousness; what it actually feels like “from the inside” to instantiate conscious states.
I accept Chalmers’ arguments, for reasons I detail in an earlier section of the paper but won’t go into here. Rather, I agree with him that phenomenal reductionism is probably false; but Chalmers’ own position, which I call phenomenal fundamentalism (the idea that there are irreducible phenomenal states), also commits us to absurdities.
Eliezer Yudkowsky’s “Zombies! Zombies?” does a good job of articulating the core problem with non-interactionist fundamentalism, though I didn’t really understand this argument’s force at the time. Sean Carroll’s “Telekinesis and Quantum Field Theory” dispenses with interactionist fundamentalism.
By process of elimination, I conclude that phenomenal anti-realism, or eliminativism, is probably true: phenomenal consciousness is neither reducible nor irreducible (in our universe), because it doesn’t exist.
This idea seems absurd, so I endorse it only grudgingly: it’s absurd, but less absurd than the two alternatives.
There are a number of obvious objections to the idea. While I think some of these objections are partly successful, I think on the whole they aren’t successful enough to make eliminativism a worse option than reductionism and fundamentalism. Here, I’ll try to systematically address a large number of possible objections. In the process I’ll hopefully clarify for some people what I mean by “eliminativism.”
Be warned that the following is not my standard fare. It’s very much written for an audience of professional analytic philosophers, and is pretty relentless about pursuing fine distinctions and subtle counter-arguments. I think this is warranted by the fact that eliminativism is such a strange view. Philosophers to date have reasonably complained that anti-fundamentalists like Dennett and Yudkowsky have been needlessly sloppy and imprecise. My view is that the arguments of anti-fundamentalists have exhibited less rigor than those of fundamentalists for contingent historical reasons, and this shouldn’t be taken to indicate that the underlying idea is fragile and liable to collapse under close scrutiny.
I have suggested that there’s a good initial case to be made that physicalism is true. I’ve also said, contra ‘introspective’ and ‘skeptical’ phenomenal anti-realism, that eliminativism isn’t obviously true. But I’ve argued as well that reductionism is untenable. Given the difficulties phenomenal states raise, we should take seriously the hypothesis that there are no such states, even though there initially seem to be. I call this position ‘theoretical’ phenomenal anti-realism.
I will defend my own version of anti-realism, which grants a large number of the fundamentalist’s intuitions—that our introspection seems to give phenomenal realism some immediate plausibility, that phenomenal properties are invertible, etc.—but denies that any properties that could satisfy all these intuitions are instantiated.
In addition to modeling its environment, the brain models its own internal states; but systematic error is possible in the latter case just as it is in the former. Hypothesizing that a persistent error in internal representation is responsible for intuitions like “color qualia outstrip their functional roles in behavior and cognition” helps explain why fundamentalism could seem so plausible in a purely physical world.
This treatment of qualia as intentional objects is similar in several respects to Richard Hall’s treatment of phenomenal properties as “dummy properties”:
[I]t (the experience) doesn’t instantiate the phenomenal properties but rather represents – representsint – the perceived object as instantiating them. That is, the phenomenal properties only occur in the experience as content, as properties the object of perception is represented as having, not as properties instantiated in the experience itself. (Hall 2007, 205)
Hall’s account differs from my own mainly in that he treats qualia as properties of objects in one’s environment, whereas I treat them as properties of one’s experiences, as is more common in hard-problematic debates. Hall’s main argument is in fact compatible with fundamentalism, since his primary concern is with whether environmental objects have irreducible qualia—an easy idea to reject (Hall 2007, 205-206). He refutes certain representationalist arguments against physicalism, but Chalmersians are free to grant that our environment is devoid of the (Edenic) properties Hall thinks we represent it as having, while insisting that our experiences too seem to bear such properties—not just as intentional content, but as occurrent constituents (Chalmers 2006/2010a).
If both our environments and our minds seem to violate physicalism, it is consistent to say that the latter experience is veridical while the former is not. Hall’s view may also lead to broader epistemic worries than mine, since it plagues our ordinary perceptions of the physical world, not just our representations of our own and others’ minds (cf. O12 below).
Interestingly, fundamentalists like Chalmers can largely accept my account, as a theory of how our zombie duplicates come to express belief in fundamentalism. But Chalmers will insist that I have failed to explain our own epistemic situation. I will seek to further unpack and defend my view in the course of responding to this objection and a variety of others, which I’ll number “O1” through “O20”.
O1: Seeming. You grant that prima facie we seem phenomenally conscious, while insisting that this is an illusion. But phenomenal properties’ being just is their seeming. If Mary seems to see a table, she may be mistaken; there is a metaphysical and inferential gap between her mental image of a table and the worldly referent. But if she seems to experience phenomenal redness, this semblance just is phenomenal redness. There is no gap between a mental appearing of phenomenal character and the character itself, because the appearances are the phenomenal character. For qualia, there is no appearance-reality gap. So your view, ‘We don’t undergo phenomenal states, although we seem to undergo them’, is internally inconsistent.
My response is that this objection borrows a lot of its plausibility from an ambiguity in terms like ‘seeming’ and ‘appearance.’ Consider, by analogy, an optical illusion such as Shepard’s Tables (Shepard 1990):
When first confronted with this illusion, a reasonable observer is likely to be disposed by the overall evidence to believe—with good reason—that the horizontal table is shorter than the vertical one on the page. The horizontal table thus seems or appears shorter. However, an observer might then learn all about the illusion and its mechanisms, enough to correct for it in her beliefs in many other circumstances. It no longer seems to such an observer, in the aforementioned sense, that the horizontal table is shorter—the available evidence disposes the observer to think that the tables are equally long. Yet there is a sense in which the horizontal table still seems or appears longer to this observer, provided her vision works like most people’s. The tables still look as though they have different lengths, even after one has adjusted one’s beliefs so as to no longer misinterpret the visual impression.
We might say that the former kind of seeming is doxastic, while the latter is perceptual. Thus it might make sense to suppose that an oar partly submerged in water continues to perceptually seem bent—our visual experience still represents it that way—even after it ceases to doxastically seem bent, i.e., after the balance of evidence ceases to indicate that it is in fact bent (cf. Augustine c. 417/2002, 190-191).
I am happy to grant that there is a coherent notion of seeming we might call phenomenal seeming—a perceptual or quasiperceptual appearing analogous in some ways to sensory or somatic appearing, but characteristic of phenomenal consciousness qua phenomenal. And I am happy to grant that there is no gap between phenomenal seeming and phenomenal being; it seems to be part of our concept of phenomenal states that undergoing such states has two inextricable aspects, epistemological and ontological. That is, phenomenal properties involve a primitive, pre-doxastic (perception-like) seeming as well as a being, and these two always co-occur. But, this being the case, my view is simply that phenomenal seeming, like phenomenal being, is an illusion. Nothing in reality actually phenomenally seems any way to any subject.
Hence it is unproblematic that such seeming would necessarily go hand-in-hand with the actual instantiation of the indicated phenomenal properties. Instead, the theoretical eliminativist will insist, at a minimum, that we sometimes epistemically seem to instantiate phenomenal properties, while denying that epistemic seeming of this sort entails phenomenal seeming.
I feel especially confident in denying that if we doxastically seem to instantiate some phenomenal property, we therefore instantiate it. In conversation, Kirk Ludwig noted a clear example of this: A subject is told that she’ll be touched from behind with a hot poker, then is touched instead with an ice cube (cf. Smithies 2012, 265). It can epistemically seem to such a person, misled by his expectations, as though she has a burning sensation, i.e., qualia characteristic of being burned. This is possible even if she’s really instantiating cold qualia and her perception straightforwardly (though not particularly distinctly) represents it as cold.
O2: (Quasi)perception. Epistemic seeming isn’t enough. Our prima facie evidence for our own phenomenal consciousness doesn’t stem from a description, as in the case of inferentially mistaking a cold sensation for a hot one; it stems from something perception-like, some sort of immediate acquaintance. Eliminativism can account for the subjects’ tendencies to form certain beliefs about their consciousness. But the perceptually or quasiperceptually apparent presence of phenomenal states is a further datum, a further explanandum, not exhausted in any attendant judgments, beliefs, or verbal reports (Chalmers 1996, 186-189). (We can imagine, for instance, neurosurgically altering a subject to consistently form false judgments about her experiences’ phenomenal character or perceptual content, without thereby altering her predoxastic
perceptions to correspond to these spontaneous judgments.) Since eliminativism ignores
this datum, it cannot provide an intellectually satisfying response to the hard problem.
I think this is a very good argument against some forms of phenomenal anti-realism. Dennett, for instance, at times endorses something like perceptual anti-realism, asserting that nothing is introspectively evident except belief-like states. Writes Dennett:
Since I hold that we have privileged access only to judgments, and since I cannot make sense of any claim to the effect that something to which I do not have privileged access is an element of my immediate conscious experience, I am left defending the view that such judgments exhaust our immediate consciousness, that our individual streams of consciousness consist of nothing but such propositional episodes, or better: that such streams of consciousness, composed exclusively of such propositional episodes, are the reality that inspires the variety of misdescriptions that pass for theories of consciousness, both homegrown and academic. […]
You may be wondering if you even have judgments. Typically these episodes are the momentary, wordless thinkings or convictions (sometimes misleadingly called conscious or episodic beliefs) that are often supposed to be the executive bridges leading to our public, worded introspective reports from our perusal or enjoyment of the phenomenological manifold our reports are about. My view, put bluntly, is that there is no phenomenological manifold in any such relation to our reports. There are the public reports we issue, and then there are episodes of our propositional awareness, our judgments, and then there is—so far as introspection is concerned—darkness. What lies beyond or on the interior of our judgments of the moment, what grounds or causes or controls them, is wholly a matter for science or speculation—in any event it is not a matter to which we have any privileged access at all. (Dennett 1979, 95)
The infallibilist line on qualia treats them as properties of one’s experience one cannot in principle misdiscover, and this is a mysterious doctrine (at least as mysterious as papal infallibility) unless we shift the emphasis a little and treat qualia as logical constructs out of subjects’ qualia-judgments: a subject’s experience has the quale F if and only if the subject judges his experience to have quale F. We can then treat such judgings as constitutive acts, in effect, bringing the qualia into existence by the same sort of license as novelists have to determine the hair color of their characters by fiat. We do not ask how Dostoevski knows that Raskolnikov’s hair is light brown. (Dennett 1988, 55)
You seem to think there’s a difference between thinking (judging, deciding, being of the heartfelt opinion that) something seems pink to you and something really seeming pink to you. But there is no difference.
There is no such phenomenon as really seeming – over and above the phenomenon of judging in one way or another that something is the case. (Dennett 1991, 364)
Although this view is not incoherent, I see no reason to endorse it. Phenomenologically, our inner lives seem to divide (though perhaps not cleanly or crisply) into propositionally organized states (beliefs, desires, concerns, suspicions…) and non-propositional ones (pains, pressures, emotions, sounds, smells…) (Chalmers 1996, 189-191). This gives us prima facie reason to accept such a division, just as prima facie we have introspective warrant for phenomenal realism — but without a hard problem to problematize the former division. Accordingly, I think that some of our experiences are not structured like beliefs in all respects, though they may share some important features in common with beliefs. (For instance, perceptual states may represent states of affairs, albeit more in the manner of paintings than of sentences.)
My form of anti-realism allows that our doxastic prima facie warrant for phenomenal realism may be grounded in a pre-doxastic, perception-like apparent phenomenality. Phenomenal anti-realism does not entail perceptual anti-realism. No hard problem is raised if a computer processes information sometimes in a linguiform fashion, sometimes in a pictiform (or tactiform, or…) fashion. The same holds for a brain. We both doxastically (prima facie) seem and (quasi)perceptually seem to bear phenomenal properties, though we do not phenomenally seem to bear such properties.
To defend the distinction between phenomenal and perceptual seeming, I can note that zombies, to the extent that they have beliefs or perceptions at all, intuitively seem as though they would be susceptible to both doxastic and perceptual misrepresentation (e.g., they can succumb to optical illusions), though ex hypothesi they cannot phenomenally represent or misrepresent anything.
This is of course only an analogy. Visual perception is significantly unlike our quasiperceptual access to our own mental states. However, once it is granted that both doxastic and perceptual error is possible in the absence of the phenomenal, I see no principled way to rule out the possibility that zombies are capable of quasiperceptual metarepresentations as well, including erroneous ones. And such metarepresentations, I believe, do hold some promise of explaining the apparent presence of phenomenal properties in our stream of consciousness, as opposed to ignoring or flatly denying this pre-doxastic semblance.
O3: Phenomenality as explanandum. The eliminativist is still simply denying the data, not taking it seriously and attempting to explain it. Granting that we sometimes doxastically and (quasi)perceptually seem to be phenomenally conscious is not enough, so long as ‘doxastic seeming’ and ‘(quasi)perceptual seeming’ are given so thin a reading that even a zombie could attain this cognitive and epistemic standing. The intuition that we are phenomenally conscious is the intuition that we differ from zombies, and the project of responding to the hard problem is precisely the project of explaining why we aren’t zombies. Any characterization of the data that remains neutral regarding whether we are the same as zombies has not even begun to address the task at hand, which is to explain phenomenal consciousness itself, not to explain the blind cognitive processes correlated with phenomenal consciousness (Chalmers 1996, 107, 110-114; Chalmers 2002/2010, 112-113).
Here I concede that the eliminativist denies some of (what seem to be) the data. That much is unavoidable. Eliminativism calls for a Copernican revolution in our view of ourselves. This requires us to question our past givens just as Copernicus made us question as obvious a datum as ‘the Earth remains still while the Sun moves across the sky’. If we replace my formulation of the hard problem with ‘the hard problem given phenomenal realism’, then it trivially follows that the anti-realist, as anti-realist, cannot respond to this problem. But as long as it is possible that we have misconstrued the data itself, we cannot rule out all hypotheses that imply such a misconstrual.
Theorists who are already fully convinced of their views have no reason to carry on a dialogue (cf. O7-8). It is only the more moderate realists and anti-realists, who are open to being convinced, to whom discussions like these are of interest. These discussants much allow, at least for argument’s sake, a theory-neutral interpretation of the data. Dismissing anti-realism because one’s first impulse is to treat phenomenal properties as givens amounts to a refusal to seriously engage with anti-realism, just as dismissing Berkeleian idealism because one initially counted mind-independent rocks and horses as undisputed data would amount to a refusal to seriously engage with idealism. In the case of phenomenal anti-realism, serious evaluation is warranted jointly by the arguments for physicalism and against reductionism.
O4: Non-inference. The phenomenal is known by some form of direct quasiperceptual acquaintance, not by inference (cf. Chalmers 2002/2010, 114). Hence this knowledge cannot be wrong.
This objection is refuted by non-inferential misrepresentations, e.g., optical illusions. To revive this objection, one must either deny that perceptions can misrepresent anything (O6), or deny that phenomenal knowledge-by-acquaintance is representational at all (O5).
O5: Non-representation. Our acquaintance with the phenomenal is not a perception-like representation, because it isn’t a representation. We are in some sort of metaphysically and epistemically intimate contact with the phenomenal, as opposed to being in contact with some sort of depiction of the phenomenal. Representational states can be illusory, but bare acquaintance cannot.
I think this is a reasonable initial objection. My first-person experience in general does initially seem to me to be simply and immediately ‘present’, as opposed to being a model, accurate or inaccurate, of lower-level patterns of neural activity. But I ultimately reject this idea on theoretical grounds. Our access-conscious states do seem to serve a self-modeling functional role, and treating the illusion of phenomenal consciousness as an artifact of this process provides a simple means for me to articulate and defend physicalism. If such a move is decisively off-limits, I gather that it must be for reasons involving O7.
O6: The possibility of misperception. Perhaps we have some perception-like representations of our own phenomenal properties. But there isn’t any such thing as ‘perceptual misrepresentation’. Only beliefs can misrepresent the world. For instance, there is no nonveridical element in our perception of Shepard’s Tables independent of our tendency to form false beliefs therefrom. (Perhaps we only think that there is some lingering misrepresentation because the false beliefs the tables inspire remain so vividly seductive even to those in the know.) Since our beliefs must then be the locus of any errors and illusions on our parts, O2 and O4 regain their full force as rebuttals of phenomenal anti-realism.
I disagree that non-doxastic states are always veridical (or simply non-representational). Such a view clashes with how we usually think about perceptual experience. Perhaps perceptual states are only veridical or nonveridical in virtue of the judgments they tend to produce or justify. For instance, perhaps the reason the pseudophenomenality-generating quasiperceptual self-monitoring mechanism I’ve posited counts as yielding misrepresentations is that it tends ordinarily to produce and/or ground relevantly false judgments. Or perhaps perception has its own, independent representational content (Pryor 2004, 355-357). I don’t have a settled view on which account of misperception is right, but as long as the distinction is granted at all, my view remains importantly distinct from the doxastic accounts vulnerable to O2.
O7: Self-evidence. Contra O4, the non-inferential character of our immediate phenomenal knowledge doesn’t establish its certainty. But, nevertheless, our immediate phenomenal knowledge is maximally self-verifying. It isn’t possible that we’ve misconstrued the data so radically that we’re deluded about phenomenal consciousness altogether, because the phenomenal character of experience is introspectively manifest, in the strong sense of being beyond any doubt. Our acquaintance with it may in some fashion depend on a perception-like representation, but there is simply no way for us to misperceive it. No theory, no matter how appealing, can be compelling enough to override the sheer obviousness of our own consciousness.
What does this perfect manifest accuracy consist in? If its presence means that I cannot be doxastically mistaken about my phenomenal states, then it is surely absent, as shown by the hot poker example. Moreover, some people accept phenomenal realism, others anti realism. That in itself shows that some people are deeply mistaken about their own phenomenology.
Perhaps the phenomenal inerrantist is appealing to a stronger version of O5. Or perhaps she means that my (quasi)perceptual representations of my phenomenal states are inerrant, even if none of my subsequent judgments about my phenomenology are inerrant. In O1-2, the realist’s point was that one cannot give up phenomenal being without giving up phenomenal seeming too. Here, the claim is much stronger: One simply cannot reasonably give up phenomenal seeming. The phenomenal’s manifest presence rationally obliges us to assent to realism. Of course, Chalmers’ paradox of phenomenal judgment shows that a zombie would reach effectively the same conclusion in the absence of the phenomenal. But that doesn’t change the fact that we just do have an immediate acquaintance with our phenomenal consciousness (Chalmers 1996, 172-189). The fallibility of judgment can do nothing to undermine that, provided we do not reduce phenomenal quasiperception itself to a kind of judgment.
Like most of the objections here, I think this one has intuitive force, even for physicalists. As such, I want to do as little violence to the underlying intuitions as possible. Just as I granted that we quasiperceptually seem to have phenomenal properties—and quasiperceptually seem to phenomenally seem to have them!—I will also grant that these properties quasiperceptually (and doxastically, prima facie) seem manifest, self-evident, pretheoretically obvious, infallibly self-revealing. But just as I denied that we really do phenomenally seem to have phenomenal properties, I also deny that the phenomenal really is manifest. I grant a watered-down semblance, and seek to explain it. But I do not grant as much as the realist wishes.
The independent plausibility of eliminativism (via physicalism and anti-reductionism) is not the only reason to doubt this sort of self-evidence, however. Inerrant self-knowledge is an extraordinarily strong claim, and does not seem to hold for any of our other perceptual or self-monitoring faculties. Humans in general have turned out to be much more error-prone than introspection and intuition originally indicated, in a wide variety of cases. For instance, people’s introspective judgments about the contents and dynamics of their memory and reasoning processes (Nisbett and Wilson 1977), as well as their perception and imagination (Schwitzgebel 2011; Dennett 1991, 344-362), are not consistently reliable. I remain skeptical that naturalistic fundamentalists can devise a plausible metaphysical mechanism for an epistemic status of this sort. Human consciousness is a product of many complex working parts, an evolved system; it would be astounding if such a system, even if accompanied by irreducible mental properties, could achieve a perfectly error-proof reckoning of itself, much less one where the error-proofness itself is introspectively manifest in an error-proof way.
Even if phenomenal properties are manifest in our experience, I doubt that the fact of quasiperceptual manifestness is itself quasiperceptually manifest. As I type this, I perhaps seem to be inerrantly acquainted with my own phenomenal states. But I don’t seem to inerrantly know by acquaintance that inerrancy itself; I don’t even notice this fact, if fact it is, until I have reflected about my experience and formed a preliminary theory of epistemic access and representation. So, on reflection, it is possible that the initial quasiperceptual indicators of my immediate acquaintance with the phenomenal are delusive, even if we grant that such a delusion initially doxastically or quasiperceptually seems impossible.
O8: Self-evidence as realist sine qua non. Inductive arguments for eliminativism are
question-begging. For O7’s doctrine of phenomenal self-evidence is indispensable to realism. If phenomenal properties occur at all, then they manifestly and with certainty occur. In doubting that phenomenal properties are perfectly self-evident, the eliminativism-entertainer will thereby have rejected the very existence of phenomenal properties. Inductive arguments will then be superfluous. All the work is done at the stage of deciding whether or not anti-realism is even an epistemically open possibility, one for which any subsequent evidence or argumentation could conceivably matter.
1. If someone is phenomenally conscious, then it is not epistemically possible (for that person) that he or she is not phenomenally conscious.
2. The eliminativist must assume, at the outset, that inductive arguments are relevant to the issue of phenomenal realism, i.e., that they can weaken one’s initial conviction in realism.
3. Therefore the eliminativist must assume, at the outset, that it is epistemically possible (for us) that we are not phenomenally conscious.
4. Therefore the eliminativist must assume, at the outset, that we are not phenomenally conscious.
If we have strong prior reason to accept realism, this then shows that anti-realism can’t be rationally justified. And even if we lack such reason, this argument shows that realist and antirealist are stuck in a stalemate. There is no way to non-question-beggingly argue for either view.
My primary response is that the truth of phenomenal realism does not entail the manifest epistemic impossibility of phenomenal anti-realism. Some phenomenal realists might claim inerrantism, but realism also admits of epistemically moderate views. Even Chalmers tends to speak of phenomenal anti-realism only as unmotivated and preposterous, rather than as outright impossible. Responding to type-A views such as eliminativism, he writes,
The obvious problem with type-A materialism is that it appears to deny the manifest. [… I]t seems to be a further truth that we are conscious, and this phenomenon seems to pose a further explanandum, one that raises the interesting problems of consciousness. To flatly deny the further truth, or to deny without argument that there is a hard problem of consciousness over and above the easy problems, would be to make a highly counterintuitive claim that begs the important questions. This is not to say that highly counterintuitive claims are always false, but they need to be supported by extremely strong arguments. So the crucial question is whether there are any compelling arguments for the claim that, on reflection, explaining the functions explains everything. (Chalmers 2002/2010, 112)
Chalmers does sometimes argue as though eliminativism is not an epistemically open possibility (e.g., in O10). Yet he seems to welcome new eliminativist arguments, and to entertain at least the abstract possibility that one could be successful!
I reject [eliminativism] as being in conflict with the manifest facts. Perhaps an extraordinary argument could establish that conscious experience does not exist, but I have never seen an argument that comes remotely close to making this case. (Chalmers 1996, 164)
If realism is supposed to be immune to any degree of reasonable doubt, then this openness is hard to understand. One explanation, suggested by Ludwig, is that while Chalmersian realists may suspect that they know realism with certainty, they don’t claim to know with certainty that they know realism with certainty. Some degree of confidence or warrant is lost in the shift to meta-knowledge. Alternatively, perhaps certitude is lost in the shift from immediate phenomenal knowledge-by-acquaintance to explicit phenomenal judgments.
I conclude that it is reasonable to insist on phenomenal fallibilism, at least for ordinary introspecters. Perhaps some wondrous philosophical or attentional feat could make us infallible introspecters; perhaps some discovery could truly, definitively, and non-question-beggingly rule out our zombiehood. But anyone willing to take anti-realism at all seriously should insist that this is not our present epistemic situation. This does not beg the question in anti-realism’s favor, because not all realisms appeal to phenomenal certitude, and even those that do may restrict this certitude to idealized introspecters, first-order knowledge, or some non-doxastic bare insight.
O9: Introspective evidence. The only way to seriously entertain eliminativism as a hypothesis is to deny that we are immediately acquainted with phenomenal properties, since granting that we are immediately acquainted with such properties would immediately rule eliminativism out. But if she denies this acquaintance, then the eliminativist will have already ruled out the possibility of recovering any evidence for phenomenal realism. As Chalmers notes, the only reason to think we aren’t zombies is that we do seem to be immediately acquainted with phenomenal properties (Chalmers 1996, 102). O8 insisted that doubting this self-evident acquaintance is ipso facto to rule out phenomenal realism. Here we instead make the weaker claim that excluding the phenomenal’s apparent self-evidence as evidence for phenomenal realism unfairly excludes any possible way, according to the realists themselves, of arriving at supporting evidence for phenomenal realism.
My response is that I accept that apparent manifestness is good evidence for phenomenal realism. I think things really do initially epistemically seem that way; and this is why I grant phenomenal realism the pre-theoretic advantage, even though I think that in the endgame, when we weigh the merits of the best realist and anti-realist theories, the anti-realist’s view will win out. Although my inductive arguments do require us to grant at the outset that it is an open epistemic possibility that we are zombies, they do not require us to completely exclude introspective appeals as relevant evidence. We can then weigh how compelling those appeals are against the virtues and shortcomings of our best physicalist theories.
O10: Zombic skepticism. We can’t rationally entertain radical error in our phenomenal
(quasi)perception as a skeptical hypothesis. Skeptical hypotheses by their very nature are about holding constant how things appear, while varying how they (otherwise) are. Thus we can entertain the hypothesis that we’re a brain in a vat because the existence of our experiences is logically compatible with two transexperiential causal stories: authentic sensation, and complex simulation. But this clearly does not hold for eliminativism. It is not consistent to hypothesize the world continuing to appear as it does, while we are not conscious of it.
I think that this argument can be reduced to O1-3. Skeptical hypotheses hold constant how things seem; but if they can legitimately rely on holding constant how things doxastically, perceptually, or quasiperceptually seem, without always assuming that there is a certain way things phenomenally seem for us to hold constant as well, then eliminativism will be a coherent hypothesis in the same sense as the others. (For instance, one could speak of a brain-in-a-vat scenario in which we hold constant the nonphenomenal component of our judgments and perceptual states, ignoring whether the phenomenal component is also preserved.)
If this is not viable, it will only be because of the strength of O7 and O8: the eliminativist hypothesis isn’t epistemically possible because the manifest experiential presence of the phenomenal is overwhelmingly, irresistibly obvious.
O11: All knowledge requires consciousness. Our access to any information about the world must be causally mediated by phenomenal consciousness. Alternatively: Our warrant for believing anything depends on phenomenal consciousness. The idea here is, perhaps, that the phenomenal dimension of mind is precisely what I have access to. I cannot know about my unconscious mind, except inasmuch as I can draw reasonable inferences about it via the conscious mind. The world beyond my experience is a mystery, except inasmuch as it in some fashion affects my experience. But it then follows that if we were completely unconscious, we would be completely oblivious of everything. Our unconscious mind might skillfully model and respond to its environment, but I would be totally in the dark regarding that fact—indeed, in the dark regarding everything.
I think this objection borrows a lot of its attraction from blurring the lines between phenomenal consciousness and access consciousness. A fundamentalist who makes use of Absent Qualia arguments must grant that the phenomenal is causally unnecessary for our cognitive functioning. Inasmuch as the conscious/unconscious divide is of great significance for global self-monitoring brain states that a zombie might possess, and not just for phenomenal consciousness, the eliminativist may appeal to it to justify the zombie’s beliefs. Since there will in general be a precise correspondence between the reliability of the zombie’s and angel’s cognitive faculties, and between the zombie’s and angel’s fulfillment of doxastic duties (though cf. O20), it will be very difficult to find points of asymmetry between the zombie and the angel that do not simply presuppose that knowledge is impossible without phenomenal consciousness.
(An “angel” is an otherworldly being that resemble us in all respects, except that it’s phenomenally conscious.)
If there are deeper reasons to think that zombies are epistemic voids, then I suspect that they stem from a general sense that a zombie double would lack my mind (O13 and O17), or a general difficulty imagining the (seemingly) phenomenal as a wellspring, not of epistemic insight, but of epistemic impoverishment or confusion (O18).
Smithies writes that perceptual experience plays a central role in epistemic justification
in virtue of its phenomenal character. It is because perceptual experience has the phenomenal character of confronting one with objects and properties in the world around me that it justifies beliefs about those objects and properties. (Smithies forthcoming, 8-9)
Smithies’ goal is to explicate our epistemic concepts in light of their relation to phenomenal realism, not to defend phenomenal realism by appealing to its ties to epistemology, so it may be a stretch to treat his arguments as arguments for realism. (Cf. Searle’s argument in O14.) But regardless, I do not find his arguments convincing.
Smithies suggests that access consciousness is not necessary for epistemic justification, because the contents of our perceptual fields can justify beliefs on our part even if we aren’t actively attending to them; and that it is not sufficient, because a case of “super-blindsight”, where one spontaneously formed detailed and accurate beliefs about one’s visual environment in a blind spot, would constitute access-consciousness but not a source of (immediate, non-inferential) justification (Smithies forthcoming, 13-14). But this only shows that a more fine-grained functional notion of epistemic access is needed here, one that not only demands that information be “poised to be used” in reasoning and deliberative action (Block 1995) but that privileges certain ways to be so poised.
In particular, we can privilege functionally normal (i.e., non-blindsighted) perceptually poised-for-use information. In this way, an epistemic asymmetry between perceptual and non-perceptual access to information can be functionally grounded, as can an epistemic continuum from information that’s both perceived and attended to, to information that’s perceived but not attended to, to information that isn’t perceived at all.
In general, it would be extremely surprising if any demonstration could be made that functional states are categorically non-justifying, because this would imply (quite
uncharitably!) that zombies lack any warrant for any of their beliefs.
O12: Most knowledge requires consciousness. Zombies (or at least zombie eliminativists) lack warrant for most of their beliefs, because if the phenomenal illusion is so all-pervading, then succumbing to (or, perhaps, discovering) it will dramatically undermine our justification for a variety of other beliefs. For instance, I believe that tomatoes refract light in the 620–740-nm range in large part because I believe that my mental images of tomatoes are phenomenally red. Thus anything that undermines the latter would partly undermine the former; and if a sufficiently large number of my phenomenal beliefs were undermined at once, it would have a cascading effect on my nonphenomenal beliefs. Since anti-realism is being justified largely on inductive, scientific grounds, this threatens to make anti-realism self-undermining.
In response, I deny that phenomenal properties are particularly important to our everyday beliefs, particularly about the physical world. Although it is true that my beliefs regarding the objective color of tomatoes depend in part on my beliefs regarding the subjective color of my visual experiences of tomatoes, the relevant aspects of this subjective experience are its functional relationships to other experiences and to the world, not its phenomenal character.
What matters most for our overall world-view is not that tomatoes produce this experience—as I attempt to ostend a red quale—but that they produce the same experience as that produced by stop signs, the opposite of that produced by a clear afternoon sky on an HSV color wheel, etc. This can be shown by considering someone born a color invert. Intuitively, the invert would have perfectly ordinary, veridical beliefs about tomatoes, erring only in her beliefs about others’ phenomenal states. If Inverted Qualia doesn’t endanger our general beliefs about the external world, about our functional mental states, etc., then I see no reason why Absent Qualia should vitiate the zombie’s nonphenomenal beliefs.
O13: Mind requires consciousness. The ordinary assertions of eliminativists, like ‘I believe in eliminativism’ or ‘I’ve seen that article’, depend for their truth on the existence of phenomenal states like belief and vision. Since eliminativists deny phenomenal properties, their routine assertions—perhaps even their assertions of eliminativism itself—commit them to a performative contradiction (cf. Wallace 2007, 14-15).
Whether states like belief seem on introspection to have a phenomenal character remains actively debated. Whether they have an essentially phenomenal character is even more questionable. However, even in cases like ‘sight’, the phenomenal anti-realist can simply stipulate that she has in mind a functional state in the vicinity of the relevant phenomenal one. Thus when I speak of perception or memory in ways that take these faculties for granted, I have in mind the functional states phenomenal realists and anti-realists both agree exist, semantically devoid of any essential appeal to the phenomenal. Whether as it happens such faculties metaphysically depend on phenomenal consciousness will then be left open to substantive investigation and debate.
Thus I establish the theory-neutrality of much of my mentalist talk by stipulation. I’m perfectly happy to grant that this practice may be idiosyncratic—that is, the ordinary English-language senses of these words may indispensably presuppose phenomenal consciousness. I adopt it anyway, for pragmatic reasons.
O14: Intentionality requires consciousness. Intentional states in general depend on phenomenal consciousness. Thus, in denying phenomenal consciousness, eliminativists are rationally committed to the meaninglessness of their own utterances.
There is no good reason to think that intentionality depends on phenomenal consciousness. Chalmers notes that Searle argues for the dependence of intentionality upon consciousness in The Rediscovery of the Mind (Chalmers 1996, 371). Searle argues as follows:
1. Intentionality occurs and is not reducible to the physical.
2. Intentionality is correlated with consciousness.
3. Consciousness occurs and is not reducible to the physical.
4. Therefore the simplest explanation of (1) is that intentionality’s physical irreducibility derives from that of consciousness.
5. Therefore intentionality depends on consciousness.
But there are good reasons to doubt these premises. On 1: It seems much easier to argue for the reducibility of intentionality than to make the corresponding case regarding phenomenal consciousness.
On 2: It’s true that we generally speak as though intentionality depended on
consciousness. But if we presuppose that intentionality is a nonfunctional, nonphysical
phenomenon, then we seem thereby to lose any empirical grounds for retaining this assumption. After all, how would the world look any different if rocks had intentional objects? Panintentionalism is easily motivated by premise 1, for the same reason panpsychism is easily motivated by phenomenal fundamentalism.
It’s true that if you already accept phenomenal fundamentalism, you thereby gain some inductive support for the fundamentality of other difficult-to-reduce phenomena, including intentionality. But the dependence of intentionality upon consciousness seems to derive its plausibility here from a mixture of background phenomenal fundamentalism and a Moorean rejection of views like panintentionalism. In contexts where we cannot presuppose phenomenal fundamentalism, we lose much of our reason to suspect that intentionality is likewise physically irreducible and ineliminable. In particular, nearly all of the arguments for the inexplicability of the intentional in physical or functional terms also suggest the inexplicability of the intentional in phenomenal terms.
O15: Charity. People simply can’t be guilty of such radical, pervasive error. A guiding principle of semantic theories is that they should make speakers’ statements come out mostly true. But if anti-realism is true, then speakers are pervasively wrong about their mental lives. Any semantic theory of what terms like ‘phenomenal’ and ‘experience’ mean that has such radical implications is more likely to be false than is the body of assertions in question. Inasmuch as we wish to reject perverse semantic theories, we must calculate tradeoffs between these theories’ simplicity, their predictive force, and how many statements of the language come out true. If we exclude the last criterion, then we are likely to end up with otherwise simple and explanatory-seeming theories that assign absurd meanings to ordinary-language terms.
In response, I note that there are many cases in which the best explanation for human behavior, including human language use, is that people are consistently wrong about a topic. Historically, most people thought the Earth flat and immobile, viewed macroscopic objects as mostly solid, and assumed an objective, reference-frame-independent temporal simultaneity. Reinterpreting those language users’ words for ‘Earth’ or ‘now’ or ‘solid’ to make their statements literally true is not merely misguided; it is entirely beside the point. The strict truth of statements is often irrelevant to most plausible hypotheses for why statements mean what they do. What matters is some relevant respect in which they are true. For instance, the mereological nihilist can explain why the word ‘table’ means what it does because atoms arranged tablewise share nearly every important feature in common with metaphysically real wholes. Likewise, as I noted in response to O12, which phenomenal properties we instantiate—indeed, to some extent that we instantiate them at all—is irrelevant to how and why people use their language, including even their mentalist language.
This is not to deny that phenomenal anti-realism has far-reaching implications for the truth of our statements and theories. It is only to suggest that strict ‘truth’ is too simple and one-dimensional a concept to capture an important constraint on semantic theories.
O16: Phenomenal concepts. You claim that we quasiperceptually meta-represent our first-order mental states, nonveridically presenting the first-order states as possessing phenomenal properties. Phenomenal property instances or states are intentional objects (like leprechauns or infinitesimals) that do not actually occur. But to represent such intentional objects will require grounding the apparently phenomenal in a physical substrate. Replacing actual phenomenality with mere depictions thereof accomplishes nothing, because these representations will either fail to explain why the phenomenal illusion seems exactly the way it does, or will be physically inexplicable themselves, raising a new explanatory gap. Ludwig frames the argument like so:
1. Your brand of anti-realism claims that we quasiperceptually represent phenomenal redness, and that no property satisfies this representation.
2. Representing phenomenal redness requires some phenomenal concept.
3. To possess a phenomenal concept, one must have actually instantiated the phenomenal property in question.
4. So your brand of anti-realism entails a contradiction.
An analogous argument can be given regarding our doxastic representations of phenomenal properties, one that generalizes to all eliminativists.
This is a powerful objection, because it appeals to the same intuition underlying Secret Qualia. If we cannot understand a bat’s phenomenology, then a fortiori zombies cannot either. If Mary lacks the phenomenal concept ΩQR, then a fortiori Mary’s zombie twin does too.
The physicalist who gives in to this argument might adopt a quietist or non-cognitivist approach, granting that she has suddenly grasped the meaninglessness of ‘phenomenal property’ (and therefore of ‘phenomenal anti-realism’) but biting the bullet in a Wittgensteinian fashion (Wittgenstein 1922/1999, 108). However, I think this drastic step can be avoided.
On my theory, Mary does not acquire new knowledge when she first observes red objects firsthand. She acquires a new concept—she can represent (quasi)perceptually, imaginatively, and verbally, in a way that she could not before—but this concept is empty.
My view can accommodate the notion that she seems to encounter new, irreducible, ineffable properties. Similarly, it can accommodate the notion that these properties, upon reflection, seem of the sort that a physical system could not represent them. Intuitively, redness is too specific for the purely physical and functional even to falsely portray. Reapplying the Inverted Qualia intuition, it seems as though our pseudoquale
could have (falsely) seemed to have a different phenomenal character, and the rest of my account remain just as adequate in every detail.
But, while I grant that this seems intuitively so, once again I must deny that this semblance is veridical. A basic feature of pseudophenomenality, as I understand it, is that the properties themselves seem to have a qualitative richness that outstrips any functional specification. ‘Functional specification’ not only includes human theories, but also patterns of neural firing. For the same reason that no possible description seems up to the task of fully expressing phenomenal redness, no possible computer seems up to the task of picking it out in a model or simulation, whether veridically or unveridically.
My anti-realism treats both intuitions as illusory. If we are in the grip of an illusion that that which is quasiperceptually represented is present and irreducible, then we are also in the grip of an illusion that that which is quasiperceptually representing is present and irreducible. I deny the representatum’s presence, whereas I deny the representational content’s reducibility.
It may seem that I should either eliminate both or eliminate neither. But eliminating the representational content would overcomplicate any representationalist response to the hard problem, and might even lead to a regress. The representational content itself does not seem present in our experience; rather, it is a theoretical posit of mine. Since it is a novel posit, and since it does not seem in its own right to have a (quasi)perceptually vivid ‘aura of irreducibility’, I feel freer to specify my concept of it in such a way that I ensure its physical reducibility.
The opposite is the case for the phenomenal. I take ‘phenomenal property’ to have an established meaning along the lines of ‘those properties whose instances I ostend in my field of awareness that are (or persistently appear to be) ineffable, intrinsic, self-revealing, qualitative, etc.’. Since I think that there are no such property instances, but I think that there’s some aspect of experience that naturally leads us to think that there are (and that we can ostend them), I take the ostensive acts to be successfully ‘directed toward’ such instances (or at the ‘spots’ where they would be). They are unsuccessful only inasmuch as the phenomenal property instances aren’t actually there.
I could have redefined ‘phenomenal’ to refer to some state that I do think occurs, but then I would have had no word for the kind of property philosophers like Nagel, Chalmers, Searle, Levine, and Jackson falsely think they instantiate. So too, I would have had no word for the illusory aspect of our second-order experience that so powerfully motivates the dualist’s intuitions.
The higher-order representational content itself, on the other hand, is at best known only indirectly, and inasmuch as it too seems on reflection to be irreducible, it is only indirectly (as a side-effect of the irreducibility of the phenomenal itself) and inferentially.
Suppose that the fundamentalist digs her heels in and insists that it is a conceptual truth,
for her concept of ‘phenomenal property’, that bearing the concept requires one to have
instantiated the property. My response would be that that’s not a concept anyone actually has, given that the property in question is irreducible and nonphysical. Insofar as I believe in physicalism, I must deny that we have such concepts. We can certainly talk about such concepts, but only in the abstract, since we cannot possess them.
These quasiphenomenal properties should be distinguished from the kinds of phenomenal properties we zombies seem quite able to discuss (though we do not, on my view, instantiate either). To help keep them straight, while I call the phenomenal properties I’ve been discussing so far Q-properties, I’ll call hypothetical phenomenal properties that are inconceivable to zombies X-properties.
Ex hypothesi, as far as my view is concerned, we have no concept of Xhood, though perhaps we have a concept of having a concept of Xhood. If even that is closed to us, then we might at least employ a Ramseyfied version of the property, speaking of properties that are like phenomenal properties but with the added constraint of being unintelligible to non-angels. The realist will then attempt to assert ‘∀Q∃X(Q=X)’, while I will insist that the attempt is a failure and will negate structurally similar Ramsey sentences.
At this point, having clarified to what extent such a dispute is even expressible from our respective perspectives, a stalemate will seem to have been reached, and the question of whether phenomenal properties are also X-properties will likely need to be bracketed so that other points for and against realism can be contested.
(Footnote: Previously, I noted that phenomenal seeming is the sort of epistemic status that one can only be in if one actually instantiates the phenomenal properties one thereby seems to have. By definition, denying phenomenal seeming requires one to deny the occurrence of phenomenal states, and vice versa; what we instead possess is a pseudophenomenal seeming, a quasiperceptual, introspective nonveridical awareness of having phenomenal properties—and a corresponding nonveridical awareness of phenomenally seeming to have them.
One interesting way to further cash out phenomenal seeming, and to further incorporate the fundamentalist’s intuitions into the theory, is to say that phenomenal seeming, if we instantiated it, would involve quasiperceptually seeming to instantiate X-properties. The pseudophenomenal seeming we actually undergo, on the other hand, involves quasiperceptually seeming to instantiate Q-properties. The pseudo-presence of Q-properties would then be built into the illusion of phenomenal being, while the pseudo-representedness of X-properties would be built into the illusion of phenomenal seeming. This isn’t necessary for distinguishing the two kinds of seeming, but it might be an added theoretical benefit of the distinction, if the X-state intuition seems sufficiently strong and basic.)
O17: First-person zombiehood. It is widely accepted that we can conceive of someone’s being a zombie, in the third person. But one cannot conceive of oneself’s being a zombie, in the first person. In O7 and O10, it was claimed that my zombiehood is not a live epistemic possibility. Here, it is instead claimed that my zombiehood is in some sense inconceivable. I cannot even entertain the idea. I can imagine someone who looks and acts like me being a zombie, but I cannot imagine actually being one, for the same reason I cannot imagine being a rock: by assumption, there is nothing it’s like to be a zombie.
Nagel’s ‘what it’s like’ locution is standardly taken to pick out phenomenal properties. But in that case the anti-realist will simply respond that there’s nothing it’s ‘like’ to be anything. We have no actual experience at all with ‘what it’s like’ even to be ourselves, in that sense.
Possibly this sounds excessively strange; possibly ‘what it’s like’ shouldn’t be treated as theory-laden or illusion-prone to the same extent as ‘phenomenal property’. Perhaps, for instance, we should reinterpret ‘what it’s like’ to pick out our core epistemic situation, or our perception-like states, or everything to which we have cognitive or epistemic access. In the latter senses (on my view) functional duplicates cannot vary in ‘what it’s like to be them’—whereas phenomenal what-it’s-like-ness can vary independently of function, but we have only an illusory acquaintance with what it’s phenomenally like to be anything. We should say that this is ‘what it’s like’ to be a zombie, to the extent that ‘what it’s like’ is not a fundamentally mistaken idea.
Once we extricate phenomenality from our more metaphysically unassuming self-concepts, we may find that we have much more immediately important subjective content and structure in common with zombies than we had initially thought. Even fundamentalists should expect most things about the subjectivity of some organism that are likely to be put into words to be functionally specifiable—the perspectival nature of experience, its privacy, its ownership, its perceived self/other divisions, its cognitive and epistemic immediacy, its internal structure, its nonphenomenal representational content, its transparency, its rich network of causal and functional-similarity relationships to other states, and many (if not all) of its behaviors concerning the phenomenal (Chalmers 1996, Tye 1995). For the same reason that phenomenality seemed so difficult to reduce and so irrelevant to our world’s causal order, phenomenal eliminativism can retain the rest of our folk-psychological notions essentially intact. This may help alleviate the sense of alienness we ordinarily attribute to the zombie’s existence.
O18: Veridical quasiperception. In O17, the concern was that I can’t imagine what it
would like to be a zombie. Here, the concern is that I can’t imagine what it would like to be a physical near-duplicate of myself that lacks any illusion of phenomenality. If the apparent phenomenality of experience is nonveridical, then, Ludwig argues, it should be possible to represent e.g. visual scenes veridically, i.e., nonphenomenally. It should be possible to strip away the delusive ‘paint’ and perceive things as they really are, clearly and correctly. But when we imagine methodically subtracting the phenomenal from experience, we don’t seem to be left at the end of this process with an accurate experience. Rather, we seem to be left with no experience at all.
It is not surprising that we can’t imagine what it would be like for our imagination itself to be free of a persistent illusion. The more interesting concern here seems to be with hashing out, at least in the abstract if not in fleshed-out imaginative exercises, how a being like us would differ if her phenomenology lacked any indicators of the phenomenal.
Presumably she would still have introspective access to her internal states, but if asked whether these states are so rich in subjectively evident content that they appear partly ineffable and irreducible, the veridical introspecter either wouldn’t understand the question, wouldn’t have any idea as to the answer, or would know that the answer is ‘No’.
(If the second possibility seems incredible, note that there are innumerable natural questions about our phenomenology that introspection alone has a difficult or impossible time answering (Schwitzgebel 2011). Our experience does not consist of cohesive fields of instantiated properties, all specified in richly interlocking detail and ripe for immediate introspective verification (Dennett 1991, 344-356; Noë 2002, 3-8). If our visual representations can underdetermine what is in our blind spot, then I see no reason why a being’s metarepresentations could not underdetermine whether its own mental life consists of functional-seeming states. Our actual experience, however, is not with such underdetermination—phenomenal properties readily reveal themselves upon reflection to be irreducible plena, and this appearance remains consistent across intuitions and thought experiments.)
One assumption that seems to be lurking in the back of this objection, and of a number of others, is that our experience or cognition is wholly phenomenal. But Chalmers’ project of differentiating functional psychological concepts from phenomenal ones, I think, largely dispenses with that mistake (Chalmers 1996, 11-12). The only times when it seems as though our psychology is reducible to our phenomenal state is when we’ve already assumed at the outset that phenomenality is a necessary criterion for our cognition, perception, etc.
The objector may also be assuming that since the phenomenal is dispensable to our mental life (because it isn’t there!), the pseudophenomenal illusion is equally dispensable to our mental life. But the illusion itself is a real brain process. It plays a major causal role at the very least in our philosophizing, and possibly in more common behavioral and perceptual events as well. We should not be surprised if its functioning turns out in practice to be inextricably linked to other neural processes, even if in principle we could construct a cognizer (probably with a substantially different high-level neural architecture) that lacked this illusion.
O19: Evolution. Succumbing to a pervasive quasiperceptual illusion seems maladaptive.
It is presumably computationally expensive to go to the trouble of simulating innumerable
properties that are simply not present; so illusionless subpopulations ought to have outcompeted the deluded long ago. How could so complex and pervasive an illusion ever evolve?
Coming from a fundamentalist, this objection is very vulnerable to a tu quoque response;
the problem of explaining how we could evolve to have phenomenal consciousness of the sort we seem to is, if anything, even more difficult. Still, this is an important concern, and I don’t have a satisfying answer yet. Of particular relevance is Chalmers’ discussion of why we should expect complex self-modeling reasoning computers to think of the variations between their perceptual states as “brute facts”.
When we ask how it knows that it sees the red bicycle, an efficiently designed system would say, “I just see it!” If such a system were reflective, it might start wondering about how it is that things look red, and about why it is that red just is a particular way, and blue another. (Chalmers 1996, 185)
This is compelling, but I do not think it tells the whole story. We could build a machine that matches the above description and that in effect has super-blindsight. Such a being might very well come to wonder about the nature and origin of its brute visual-environment intuitions, and might very well come to false conclusions as a result, but it wouldn’t do so in the manner that we do.
Chalmers specifies that the errors of the reflective machine he has in mind arise from its perceptual states, dismissing the super-blindseer’s cognition as “curiously indirect”. But we can also imagine a normally sighted phenomenally unconscious automaton that can’t help spontaneously forming judgments about the irreducibility of perception, but whose perception on immediate introspection does not warrant such judgments. (At most, it merely causes them.)
On my view, the key missing element in both of these cases is that the illusion of
phenomenality is perception-like. The ‘zombic hunch’ is not a hunch. It is not a free-floating intuitive judgment, not even an irresistibly compelling one. I seem rather to have been immediately acquainted with a phenomenal manifold from an early age, long before I started reflecting on or theorizing about my consciousness. I seem to have concrete epistemic and metaphysical access to my mental life, in a way analogous to—but stronger than—my epistemic and metaphysical access to physical objects in my perceptual environment. And this seeming itself is grounded (and perhaps consists) in my quasiperceptual modeling of my mental states as phenomenal states. It is unclear to me that such a sophisticated illusion falls out of Chalmers’ self-modeling automaton so naturally.
(The above is perhaps why fundamentalists find fundamentalism intellectually satisfying. Their hypothesis seems to amount to an explanation of their observations. In acquiescing to realism, it feels as though one is acknowledging a natural phenomenon of some sort; it does not feel as though one is merely giving in to a vague ‘hunch’.)
We do not know how computationally intensive or structurally complex the phenomenal illusion’s mechanism is. As far as we know, pseudophenomenality is more efficient than the alternative, if not from a top-down engineer’s perspective then at least from within the constraint of building a reasoner via incremental cumulative random mutations and selection.
Our quasiperceptual awareness of our own experiential states in many ways mirrors our perceptual awareness of our environment; possibly the former system is modeled after the latter, and the illusion of the phenomenal is a vestige of some mechanism that was adaptive in the case of perception but is not adaptive in its higher-order incarnation. I find it extremely likely that much or all of the phenomenal illusion stems from random genetic drift or is a spandrel. Still, some explanation is indeed owed for the origin of the raw material of which the illusion is a vestige or side-effect.
Adaptive explanations of all or part of the illusion’s underpinnings are difficult to make, because many of the properties we intuitively associate with phenomenality also have functional counterparts. For instance, our experience could have appeared vivid, manifest, virtual, and private without appearing so content-rich as to outstrip its functional roles.
Still, in the interest of stimulating further research, I’ll suggest a very tentative hypothesis: Marking our experience in introspection as having a manifestly functionless dimension encourages humans to overextend their theory of mind in a strictly nonveridical but evolutionarily advantageous way.
Evolutionary theorists have hypothesized that we possess a Hyperactive Agent Detection Device, driving us to err on the side of detecting predators and other agentive threats very readily (Barrett 2000). Perhaps our experiences’ apparent phenomenality helped further exaggerate this device’s effect, in a fashion that had further selective advantages. Perhaps, for instance, our ancestors had an easier time forming fruitful empirical hypotheses about apparent agents than about mindless phenomena. If protoscientific hypotheses like ‘find shelter when the sky spirits become angry’ lent themselves to our progenitors’ reasoning and motivational drives more readily than did their nonmentalist counterparts, then a mechanism for encouraging such hypotheses may have been adaptive. A readiness to anthropomorphize external phenomena may also have been advantageous for persuading or cooperating with other agents, e.g., in religious contexts. Hence — still indulging here in wild historical speculation — it might have been useful to suffer from the illusion that one’s mental states abound with fully autonomous property instantiations, if the result was that one’s theory of mind could more readily generalize to entities that are, structurally and behaviorally, thoroughly inhuman. The conceivability of zombies and inverts, the epistemic gap, etc. would then be side-effects of a more general perception-like mechanism imbuing our mental states with flexibility as to their nonmental correlates.
This is all assuming, perhaps implausibly, that a simple and intuitive functional explanation is likely to hold here. By directing attention to this particular problem, my hope is that readers may devise more sophisticated accounts that likewise avoid conflating the different explananda.
Non-interactionist fundamentalists can join me in theorizing about possible explanations along these lines, since they will agree that a completely functional explanation can be given of the appearance of phenomenality inasmuch as this differentially affects our cognition and behavior. The only disagreement will, once again, concern whether explaining this appearance explains everything.
O20: Ethics. A final objection is to the moral implications of my view. It is natural to suppose that organisms are moral patients in large part because they are conscious (Harris 2010, 32-33; cf. Joyce 2007/2009, §5). In evaluating whether it is wrong to torture a robot, what matters is not whether it superficially behaves as though it is in pain, but whether it actually experiences pain. By eliminating phenomenal consciousness, your view leads to the unacceptable consequence that we must also eliminate morality.
One might argue theoretically here, from the truth of moral realism to the truth of phenomenal realism. Or one may argue at the practical level, claiming that our moral conduct would be hindered by a widespread belief in phenomenal anti-realism. In the former case, I see no incoherence in a phenomenally anti-realist moral realism. I concede that the idea is counterintuitive, but this is a drop in the bucket relative to phenomenal anti-realism’s internal bizarrery! Insofar as we do have reason to believe in moral facts that depend on phenomenal facts, I think this reason largely derives from our prior reasons to believe that we are phenomenally conscious, and not the other way around.
I am more sympathetic to the practical concern, and I’d like to spend more time thinking about the social and psychological ramifications of eliminativism someday. It can be bracketed for the moment, as the potential harmfulness of eliminativism is independent of its truth, and this philosophy paper is probably dry and dense enough to bear little risk of Internet virality.
The importance we ascribe to the phenomenal in our own lives, and our difficulty in disentangling phenomenal and nonphenomenal mental concepts, explains why we appeal to the phenomenal-nonphenomenal distinction to provide a clear and intuitive line dividing the proper objects of our concern from Cartesian automata. But if we did not have a prior reason to associate ourselves—and as a consequence our prudential and moral concerns—with a phenomenal dimension, then I do not think that anything basic to morality itself would have forced this association upon us. The notions of fairness, obligation, and harm make no explicit appeal to phenomenal consciousness, and to the extent that the two are implicitly associated it is not obvious that they could not come to be extricated once phenomenal realism itself came to be doubted.
I take the task of extricating moral and prudential norms from the phenomenal aspect of affect and volition to parallel the task of extricating epistemic norms from the phenomenal aspect of belief and perception. In both cases I believe that the original concepts are either inclusive enough to encompass zombies, or malleable enough to allow us to unproblematically expand their scope if the anti-realist thesis becomes sufficiently credible. It is not as though phenomenal anti-realism, whether morally realist or not, has the consequence that one stops preferring some things over others. As long as we in fact have values and concerns and interests, we will continue to have norms and value systems in practice.
Some of the above objections to eliminativism, I believe, are reasonable enough to give fundamentalists grounds to be wary about abandoning their foundational assumptions. But I think that my responses, though cursory, show that phenomenal eliminativism deserves much more consideration than it has received to date—particularly by those who do not consider O7 (self-evidence) conclusive.
I suspect that a number of my responses will not feel very satisfying to those who found the objections initially plausible. In a number cases I do little more than say, ‘Your premise, if true, would refute my view; but I can consistently affirm that your premise only seems true.’ Such arguments by their very nature can’t get eliminativism off the hook. They neither dispense with the semblance nor explain it—not in robust cognitive detail, at any rate.
My goal in such cases is not to completely defeat every objection to eliminativism; it is only to show that they are not fatal to eliminativism, and that the sorts of claims a theoretical eliminativist must assent to are interesting and novel enough to warrant much more careful investigation in the literature.
I have at least given moderate fundamentalists reason to take phenomenal anti-realism seriously, and not to dismiss it as manifestly absurd. For my main audience, committed physicalists, combining this account with my earlier argument against phenomenal reductionism indicates that eliminativist views are at present their best hope for fending off fundamentalism.
Fully explicating a plausible anti-realist account of human cognition is a large project, and I do not think that the arguments for physicalism are so strong, Chalmersian epistemology so bizarre, or the internal problems for eliminativism so minor that it is manifestly unreasonable to reject eliminativism; but I do think that the balance of evidence currently favors it.
On my anti-realist account, the intuitions that make the hard problem so hard can be explained quite straightforwardly.
Zombies and inverts seem possible because they are possible—but this poses no problem for physicalism if the angelic properties that zombies lack and that inverts interchange are not instantiated in our own world (Beisecker 2010).
Bats’ minds seem ineffable to us because we mistakenly think that they, like we, are phenomenal angels; once this mistake is corrected in our own case, we lose any reason to impute unphysical properties to other minds.
Mary’s understanding of color seems incomplete to us for the same reason—either Mary is a zombie and we have misunderstood her epistemic situation for the same reason we misunderstand our own, or Mary is an angel and the failure of physicalism in her world does not endanger physicalism in our own.
In all of these cases, the intuitive force, epistemic basicness, and experiential pervasiveness of this anti-physicalist evidence is best explained by a representational defect in a perception-like self-modeling faculty.
Fundamentalists, having more or less succeeded in vanquishing reductionism, should move on to the more difficult and foundational task of accounting and arguing for the special epistemic status of the phenomenal, so as to defend realism itself. At the same time, physicalists who appreciate the severity of the hard problem should shift their focus from defending reductionism to constructing eliminative theories.
I have only begun to spell out one such theory above, but it’s already forced me to reconsider my position on a number of foundational disputes in a variety of philosophical fields. Until more of phenomenal anti-realism’s consequences for philosophy as a whole have been drawn out, the overall merits of the theory will be difficult to estimate. Still, it’s already easy to see that the most radical break with philosophical tradition that is called for by the anti-realist is a rejection of the phenomenological tradition, following Descartes, of taking assorted experiential appearances as self-evident and self-verifying. Although the anti-realist is free to admit phenomenology as a legitimate, and frequently reliable, source of evidence, it will be very difficult ever to recapture a gem of true certitude from experience, once a phenomenon as epistemically basic as phenomenality itself has relinquished this status.
I’d expect those who find this step an easy one to take to be those who already felt disposed to grant that the reach of moderate skepticism is, if not deep, at least exceptionally wide.
- Augustine of Hippo (c. 417/2002). On the Trinity. Cambridge University Press.
- Barrett, Justin (2000). “Exploring the Natural Foundations of Religion”. Trends on Cognitive Science 4:29-34.
- Beisecker, David (2010). “Zombies, Phenomenal Concepts, and the Paradox of Phenomenal Judgment”. Journal of Consciousness Studies 17(3-4):28-46.
- Block, Ned (1995). “On a confusion about the function of consciousness”. Behavioral and Brain Sciences 18:227-47.
- Chalmers, David J. (1996). The Conscious Mind: In Search of a Fundamental Theory. Oxford University Press.
- ————— (2002/2010). “Consciousness and its Place in Nature”. In D. Chalmers (ed.), Philosophy of Mind: Classical and Contemporary Readings (pp. ??-??). Oxford University Press. Reprinted in D. Chalmers, The Character of Consciousness (pp. 103- 39). Oxford University Press, Inc.
- ————— (2006/2010b). “Phenomenal Concepts and the Explanatory Gap”. In T. Alter & S. Walter (eds.), Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism (pp. 167-194). Oxford University Press. Reprinted in D. Chalmers, The Character of Consciousness (pp. 305-336). Oxford University Press.
- Dennett, Daniel C. (1979). “On the absence of phenomenology”. In D.F. Gustafson & B.L. Tapscott (eds.), Body, Mind, and Method. Kluwer.
- ————— (1988). “Quining Qualia”. In A.J. Marcel & E. Bislach (eds.), Consciousness in Contemporary Science (pp. 42-77). Oxford University Press.
- ————— (1991). Consciousness Explained. Boston: Little Brown.
- ————— (2001). “The zombic hunch: Extinction of an intuition?” In A. O’Hear (ed.), Philosophy at the New Millennium. Cambridge University Press.
- Hall, Richard J. (2007). “Phenomenal Properties as Dummy Properties”. Philosophical Studies 135:199-223.
- Harris, Sam (2010). The Moral Landscape: How Science Can Determine Human Values. Free Press.
- Jackson, Frank (1986). “What Mary Didn’t Know”. The Journal of Philosophy 83(5):291-295.
- Joyce, Richard (2007/2009). “Moral Anti-Realism”. In E.N. Zalda (ed.), The Stanford Encyclopedia of Philosophy (Summer 2009 Edition).
- Levine, Joseph (1983). “Materialism and Qualia: The Explanatory Gap”. Pacific Philosophical Quarterly 64:354-61.
- Nagel, Thomas (1974). “What Is It Like to Be a Bat?” The Philosophical Review 83(4):435-450.
- Nida-Rümelin, Martine (1996). “What Mary Couldn’t Know: Belief about Phenomenal States”. In T. Metzinger (ed.), Conscious Experience (pp. 219-241). Allen Press.
- ————— (2002/2010). “Qualia: The Knowledge Argument”. In E.N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Summer 2010 Edition).
- Nisbett, Richard E. & Wilson, Timothy DeCamp (1977). “Telling More Than We Can Know: Verbal Reports on Mental Processes”. Psychological Review 84(3):231-259.
- Noë, Alva (2002). “Is the Visual World a Grand Illusion?” Journal of Consciousness Studies 9(5-6):1-12.
- Pryor, James (2004). “What’s Wrong with Moore’s Argument?” Philosophical Issues 14(1):349- 378.
- Schwitzgebel, Eric (2011). Perplexities of Consciousness. MIT Press.
- Shepard, Roger N. (1990). Mind Sights: Original Visual Illusions, Ambiguities, and Other Anomalies. WH Freeman and Company.
- Smithies, Declan (2012). “A Simple Theory of Introspection”. In D. Smithies & D. Stoljar (eds.), Introspection and Consciousness (pp. 259-294). Oxford University Press.
- ————— (forthcoming). “The Phenomenal Basis of Epistemic Justification”.
- Tye, Michael (1996). Ten Problems of Consciousness: A Representational Theory of the Phenomenal Mind. MIT Press.
- Wallace, Meg (2007). “Mental Fictionalism”. Unpublished draft.
- Wittgenstein, Ludwig (1922/1999). Tractatus Logico-Philosophicus. Dover Publications, Inc.