No ghost, just a shell: A rejoinder to Keith Frankish

This essay is a rejoinder to Keith Frankish’s reply to my original criticism of illusionism, the notion that consciousness is an illusion.

An initial reflection

In criticizing illusionists such as Frankish, one is always faced with the dilemma of either writing with the general public in mind or the individual illusionist one is criticizing. The most effective line of reasoning is different in each case, for the public isn’t tied up in the conceptual and definitional knots illusionists create for themselves. Indeed, whereas the public—watching from a more objective, uncommitted vantage point—can easily grasp the blatant circularity and inconsistency of the illusionist argument, the illusionists themselves are too immersed in their own story to fathom any of it. Instead, one must first meet them where they are, otherwise they will choose to believe that their points are merely misunderstood by their critics.

Having always written with the general public in mind, it thus comes as no surprise to me that Frankish should feel certain that I do not grasp what he is saying. At no point does he seem to entertain the possibility that I actually understand perfectly well where he is coming from, why he thinks what he thinks, and yet still consider his story blatantly absurd.

As a matter of fact, in the early years of my career as a computer engineer, I wrestled intensely with the question of how to build computers that would consider themselves conscious even if not programmed to do so; that is, how to construct a machine that would not only perform calculations, but also spontaneously claim to experience these calculations, just as you and I experience the goings-on in our brain. This wasn’t armchair philosophizing for me, but a very concrete and practical question. And that’s precisely why I ended up wasting so much time on it: I never stopped to examine the implicit assumptions embedded in the very problem statement that motivated my search.

And so it was that, in the first years of the 21st century, Pentti Haikonen, a researcher at Nokia, came up with a computer architecture that would not only consider itself conscious, but—Haikonen thought—also in fact be conscious. Haikonen’s deeply insightful realization was two-fold: first, the original semantic anchoring of the input signals fed into the computer should be preserved—as opposed to being encoded into arbitrary binary symbols—if the computer is to consider itself conscious; second, feedback loops should be inserted in the architecture at strategic points, so as to allow the computer to introspect by re-representing its own computational activity.

Haikonen’s approach, which I recognized as brilliant and hold in very high esteem to this day, can be regarded as effectively elaborating—much more specifically and persuasively than Frankish himself—on Frankish’s claim that introspective (mis)representation is what leads to the belief that we are conscious. As such, and implausible as it may sound to him, I believe I actually understand why Frankish considers illusionism so compelling. I am very familiar with the thinking and motivations behind it, in a fairly high level of (engineering) detail.

Indeed, Haikonen creatively tackled many of the difficulties I had identified for building a machine that could spontaneously claim to be conscious. Alas, we would be unable to verify such claim, for the only way to know would be to be the machine. Yet, the claim alone would already be a remarkable engineering achievement, one I was very interested in contributing to.

When it comes to us, however, it’s not just a matter of making spontaneous claims: we actually know that we are conscious, for we are ourselves. In our case, therefore, we must address the hard problem of consciousness, which Haikonen’s architecture—despite his philosophically naive claims to the contrary—completely fails to do: instead of creating consciousness in the phenomenal sense, his approach merely presupposes it. And so does Frankish’s.

No amount of structure, complexity, feedback, recursion, re-representation, etc., can make a substrate presumed to be fundamentally unconscious produce experience, in the same way that no amount of added speakers can turn a stereo into a television, and that no amount of extra legs can make a centipede fly. Recursive re-representations can only complexify pre-existing experiential states, not create them from something fundamentally non-experiential. More specifically, what recursive re-representations can do is to make pre-existing phenomenality accessible to metacognitive introspection, not create it.

Conflating consciousness with metacognitive awareness

For Frankish, it is our ability to introspect by metacognitively re-representing our neural processes that characterizes what we call consciousness. He writes:

It is a mark of conscious experience that we are, or can easily become, aware of having it. We can direct our attention inward (‘introspect’) and think about the experiences we are having.

It is this introspection that, according to Frankish, creates the illusion of qualitative experience: “Our introspective systems monitor these [neural] processes but misrepresent them as a simple quality,” he says. “The illusion concerns the nature of these processes—the belief that they are simple qualia.” Later he continues: “It is this emphasis on the effects of introspection that makes the notion of illusion so appropriate here.”

The problem is that Frankish conflates phenomenal consciousness—that is, raw experience, ‘what-it-is-likeness’—with meta-consciousness. As Jonathan Schooler explained in his 2002 paper, Re-representing consciousness: dissociations between experience and meta-consciousness,

Periodically attention is directed towards explicitly assessing the contents of experience. The resulting meta-consciousness involves an explicit re-representation of consciousness in which one interprets, describes, or otherwise characterizes the state of one’s mind.

But phenomenal consciousness does not require meta-consciousness: if an experience falls outside the field of our attention, we have the experience without being aware that we are having it. For instance, we regularly experience our breathing without metacognitive representation. Moreover, as discussed by Jennifer Windt and Thomas Metzinger in their 2007 paper, The philosophy of dreaming and self-consciousness: What happens to the experiential subject during the dream state, dreams largely lack introspective re-representation, despite their undeniably experiential nature. Even the emerging ‘no-report paradigm’ in neuroscience rests on the understanding that experience can—and frequently does—occur without explicit introspective awareness, such as in the cases of blindsight that Frankish likes to cite.

In conflating consciousness with meta-consciousness, Frankish is failing to heed a key conceptual distinction already discussed by philosopher Ned Block in his important 1995 paper, On a confusion about a function of consciousness: whereas introspection requires metacognitive access, it doesn’t need to be phenomenal. Phenomenal consciousness, in turn, doesn’t need to be introspectively accessible in order to exist. These are two different things.

Therefore, Frankish’s appeal to introspective (mis)representation to explain experience away is based—as I originally claimed—on conceptual confusion: if experience isn’t there to begin with, we have no reason whatsoever to believe that introspective re-representations would be, or even seem to be, experiential either. Instead, everything would happen ‘in the dark,’ without the light of awareness.

Eating the cake and having it too

For Frankish’s position to have any relevance in helping tackle or circumvent the hard problem of consciousness, what he must deny is phenomenality, felt experience, qualia, ‘what-it-is-likeness.’ Anything else, despite potentially having some other philosophical application, would be irrelevant as far as the hard problem is concerned.

Unsurprisingly, thus, Frankish often emphasizes that what he denies is precisely phenomenality, qualia, experience. For instance, already in the subtitle of a recent essay, he wrote that “Phenomenal consciousness is a fiction written by our brains” (emphasis added). This doesn’t seem to leave much room for ambiguity, as philosophers use the qualifier ‘phenomenal’ precisely to specify, unambiguously, that what is meant by the word ‘consciousness’ is qualities, felt experiences, ‘what-it-is-likeness.’

But to deny the qualities of experience is to deny, for instance, that we feel pain; is to say that our agonizing screams under torture—as well as the accompanying physiological processes—are merely functional, useful for getting help; but that, from the inside, none of the dreadful qualities we associate with pain are actually felt. If Frankish denies the qualities of experience, presumably he wouldn’t mind undergoing torture, which I very much doubt to be the case.

And so, in his reply to me, Frankish already starts out by acknowledging that it is “utterly ridiculous” to claim that “people are not conscious, don’t have experiences.” He goes on to say that “illusionists don’t deny that we are conscious,” that there is a “sense in which we undoubtedly are conscious,” that “our lives are filled with conscious experiences.” Really? What is it, then, that illusionists do deny?

“What illusionists reject is a certain conception of what consciousness is,” Frankish claims. But then again, if illusionism is to have any relevance as far as the hard problem is concerned, the “conception of consciousness” that must be denied is precisely that entailing pain and emotion, felt experiences, which Frankish has just acknowledged to exist! Any other conception of consciousness—such as Block’s ‘access consciousness’ or Schooler’s ‘meta-consciousness’—is irrelevant for the hard problem: it still leaves us with having to explain how raw experience, whether metacognitively represented or not, arises from an allegedly non-experiential substrate.

Frankish thus faces an impossible dilemma, which he can only tackle by systematically contradicting himself. He deserves our sympathy, for the job of manufacturing even a smidgen of plausibility for what is the most incongruous maneuver conceivable to the human mind—that of denying itself—is not exactly easy. On the one hand, he must acknowledge that “illusionists don’t deny that we are conscious,” otherwise they would just be crazy. On the other hand, he also has to claim that

Illusionists reject the qualitative conception of consciousness. They hold that qualia, and the private show they constitute, are illusory; they seem to exist but don’t really. This is the core claim.

How are we to square this circle? Frankish seems to be making a distinction—which he conspicuously doesn’t elaborate upon—between experience or phenomenality on the one hand, and felt qualities on the other. He acknowledges the former while, bizarrely, denying the latter. Yet, to reject the “qualitative conception of consciousness” is to deny experience, phenomenality, phenomenal consciousness itself; after all, the latter is defined as entailing the felt qualities of experience.

Is Frankish playing some silly game of words? In what sense is he acknowledging that we have pain and emotion if he is denying the qualities that pain and emotion are? Granted, pain and emotion are associated with certain functions and behaviors, but this has nothing to do with consciousness. By focusing on function and behavior to the exclusion of qualia, Frankish is merely ignoring the hard problem, closing his eyes to it, not tackling or circumventing it in any meaningful sense.

When he reassures us that “illusionists don’t deny that we are conscious” and thus aren’t outright crazy, Frankish is appealing to our intuitive understanding of conscious states as felt qualities. But then, having accomplished that, he immediately turns around and rejects the “qualitative conception of consciousness” so as to portray his approach as relevant to addressing the hard problem, instead of something utterly trivial. Which one is it? He can’t have it both ways. Either he is sane, or his work is relevant when it comes to the hard problem.

Explanation by redefinition of terms

Contrary to what Frankish suggests, the qualitative dimension of experience isn’t a merely conceptual reality, but a felt and immediate one. It is very important that we keep this in mind.

You see, there are many entities in science whose only accessible reality is conceptual: think of imaginary numbers in mathematics or quantum fields in physics, for instance. The world behaves as though these conceptual entities existed and, as such, it is very useful to imagine that they do. But we have no immediate, felt access to them; all we know about them is our conception of them.

Therefore, these conceptual entities are perfectly amenable to being redefined, if doing so helps to make sense of things. For instance, it has been useful to redefine gravity as a curvature of spacetime, instead of an invisible force acting between two bodies from a distance. We have no direct acquaintance either with the curvature of spacetime or the invisible force, so we might as well feel free to redefine gravity based on theoretical convenience.

However, an analogous rationale does not apply to phenomenal consciousness, for the qualities of experience aren’t merely conceptual; they are immediately felt. By rejecting “a certain conception of what consciousness is” illusionists aren’t making these felt qualities disappear; they are merely ignoring them, pretending that they don’t exist.

Indeed, whatever definition of consciousness we choose to use in our conceptual games, there remains this thing—this undeniable thing immediately accessible to us prior to all conceptual reasoning—that will continue to exist whatever we call it. The hard problem of consciousness is essentially about this thing, not the word ‘consciousness.’ If you think the label ‘consciousness’ shouldn’t be applied to it, fine, I don’t care, call it something else; call it… well, the ‘thing.’ But the thing won’t cease to exist just because you renamed it. Even if we can’t appropriately define it in words, it won’t be affected; it will remain what it is and has always been. Terminology games don’t change reality, no matter how hard we wish they did.

Frankish, however, seems to think that he can make the felt qualities of experience—the thing—disappear simply by redefining terms. Consider the following passages from his reply:

“experiences are physical states of the brain”

“consciousness consists, not in awareness of private mental qualities, but in a certain relation to the public world”

“It is this global broadcasting and its effects that constitute consciousness”

“I am proposing that consciousness is this complex of informational and reactive processes”

Hey, I can play this game too. How about ‘consciousness is the involuntary microscopic twitching of my left big toe’? Or—just a little more seriously—‘consciousness is the collapse of the quantum wave function in the synaptic clefts of my prefrontal cortex’? In terms of explanatory power, are these statements really so different from “consciousness is a complex of informational and reactive processes”? Are they any less arbitrary? Do definitional statements have any explanatory power at all? Do they solve any problem?

The bottom-line is this: We know first-hand what consciousness is, regardless of how the word is defined. It doesn’t matter how often and how passionately Frankish repeats his statements of faith, it is this thing we know that matters; it won’t disappear because of semantic games. Substituting redefinitions of terms for actual argument just won’t do. Otherwise, I would have won the Fields Medal long ago by merely redefining yet-unsolved problems in such a way that the solution would be trivial.

Failure to grasp the criticism

Frankish’s central point is that our introspective re-representations of our own physical brain states are illusory in the sense that they don’t accurately portray said brain states. That’s why—in his view—we mistakenly think we have qualia, instead of just physical brain states: the latter are metacognitively misrepresented as seeming qualia, which is the illusion in question.

My original refutation of this argument was as simple as it was generic: if the misrepresentations seem qualitative, then the very seeming is already a quality, regardless of what the misrepresentations seem like. After all, an illusion is already a felt experience in and of itself, regardless of its lack of representational accuracy. The implication is that we do have qualia, not despite our re-representations being inaccurate in the way Frankish claims them to be, but precisely because of it.

Yet, Frankish failed to understand this simple point. He misconstrues and misportrays it as something unnecessarily more restrictive, which can be seen in the following passage (if you find it too difficult to follow his reasoning, no worries, I summarize it in simpler words below):

Could we seem to have qualia without really having them? Kastrup thinks not. ‘Good Lord,’ he exclaims, ‘the ‘seeming’ is already an experience in and of itself.’ Does this simple point blow illusionism out of the water, as Kastrup supposes? There’s one way it might. Suppose that … seeming to perceive a thing involves being aware of the mental qualities one would have been aware of if one were really perceiving it. Then, by analogy, seeming to introspect a mental quality would involve being aware of the mental quality one would have been aware of if one were really introspecting it. And that, presumably, is the very same mental quality. The illusion would involve a real instance of the thing that was supposed to be illusory! … The flaw in this objection is obvious: it assumes that experience involves awareness of mental qualities.

What Frankish is saying here is that my criticism holds if, but only if, our alleged misrepresentations of perceptual brain states correspond to “the mental qualities one would have been aware of if one were really [consciously] perceiving.” But such constraint is not at all necessary for my criticism to hold: whether the alleged misrepresentations match what would have been the actual qualia of perception or not is irrelevant, as long as the misrepresentations seem like something; anything; it doesn’t matter what. The seeming alone already entails felt qualities—whatever they may be—and, therefore, felt qualities must exist.

Even if what is misrepresented are physical brain states—as opposed to experiential qualities—the corresponding seeming, in and of itself, is already an experience entailing its own (illusory) felt qualities. Therefore, contrary to what Frankish claims, the only assumption my criticism makes is that there is seeming, which is precisely what illusionism requires (otherwise one cannot speak of illusions to begin with).

Final reflections

I earn my living doing corporate strategy in the high-tech industry, perhaps the most rewarding but also most unforgiving environment for analytic thinking. In that world—my world—even subtle and understandable failures of reasoning are very quickly—and often disproportionately—punished, either by management or by the market. Reality can always be counted upon to settle all questions in a rather brutal but objective manner; something I have grown to appreciate over the years, for it forces me to be constantly critical of my own narratives.

It strikes me, though, that in philosophy one seems to be able to get away with incoherent thinking indefinitely. If one cannot clearly and substantively argue for one’s own position, verbose misdirection, ambiguity and handwaving often sustain just enough doubt about whether one is actually wrong. And so, nonsense can survive ad infinitum. There is always the lingering doubt that, hidden behind impenetrably obscure language constructs and indecipherable conceptual acrobatics, there might just be some deep, non-obvious philosophical insight. Yet, often there is none; often things are precisely what they seem to be: very confused and self-contradictory thinking. Perhaps this is the reason why philosophy doesn’t seem to make progress.

Although I have made a deliberate effort in this essay to patiently meet the illusionists where they are—so as to do something different than just repeating my original criticisms, most of which Frankish didn’t even address, such as my claim that he falls for the fallacy of infinite regress—it remains my position that illusionism is the most ludicrous and self-defeating view conceivable. Nothing in the history of human thought is, or can be, more preposterous than it. That some otherwise intelligent people espouse it is, in my view, merely a psychological artifact—a desperate attempt to salvage an untenable metaphysics many have associated their very identity with—not the outcome of clear, rational thought.

My willingness to engage in an extensive, detailed and protracted debate about illusionism should, therefore, not be construed as a sign of respect for it; I have precisely none. In fact, I find it embarrassing to be in a position of having to argue against it. My doing so reflects merely a begrudging acknowledgment that our philosophy is in a lamentable state and that, if anything is to be done about it, it must be done from within our present circumstances.

In a less confused world, illusionism wouldn’t even be a joke. Perhaps we will get there one day. In the meantime, however, brutally honest, even scathing public criticism may be the only system of checks-and-balances available to preserve the sanity of philosophy. The present essay has been written in this ultimately well-meaning spirit.