Are Feels Real? Reflections on Frankish's Illusionism

Keith Frankish is perhaps the best known proponent of illusionism as a theory of consciousness, which he describes in two 2022 papers published in Human Affairs. Here I take up the question of whether on illusionism we can still legitimately talk of how experiences feel, and suggest the anwswer is no. Should illusionism carry the day as the settled account of consciousness, then talk of experiential feels, as in how pain feels, is in error. But Frankish himself declares feels to be real, so something has to give. 

Frankish's comments on this essay, and my reply, are here


It’s hard to deny that everyday conscious experiences, such as having a cup of tea, feel a certain way. The colors, shapes, tastes, sounds, and other sensations that constitute the experience of brewing and tasting tea all have a particular felt character, and we often talk of what they’re like in terms of various sensory qualities, e.g., sweet, bitter, hot. But illusionists such as Keith Frankish deny there are any such qualities (qualia in philosophical parlance), while at the same time agreeing we have conscious experiences. One wonders, therefore, how they would describe the tea experience, composed as it is of distinct elements varying within a smoothly evolving conscious gestalt.

To no surprise, they use the same vocabulary as the rest of us, talking about colors, tastes, textures, and other sensations; after all, there’s no alternative when it comes to saying what it’s like. And I surmise that Frankish’s experience is just as replete with what in my experience I’d call qualities – the various sensory feels of tasting tea – even though he insists there are no such things: there is no phenomenal aspect to experience. This denial and its logical sequelae, as he describes it in two recent articles from Human Affairs I quote below, put him in strange territory when it comes to consciousness.[1] To wit: not only is there is nothing qualitative about experience, there’s nothing essentially private about it either. As a purely physical-functional and non-qualitative phenomenon, it’s potentially all out there in public as an observable process. My pain, therefore, exists not just for me – a categorically subjective phenomenon – but is completely objective, just like a chair or table.[2]

Why qualia won’t go away

Illusionism, as explained by Frankish in his two papers (see note 1) and by other illusionists such as Dan Dennett, is an essentially negative thesis about consciousness, namely that the subjective qualities (qualia) that purportedly constitute phenomenal consciousness don’t exist; the project of explaining consciousness becomes explaining the illusion that they do. For illusionists, qualia are still the primary feature of conscious experience around which the explanatory project revolves, even though the ultimate conclusion is that they’re illusory. As Frankish also puts it, there’s nothing intrinsic to consciousness that can’t be accounted for by physical mechanisms (No trick, 324). The basic idea is that a higher order representational capacity (what he calls introspective consciousness) misrepresents first-order perceptual representational states (what he calls perceptual consciousness) as having qualities. All that really exists are the various neural goings-on that constitute representations, reactions, dispositions, reports, etc., all of which are in principle observable, physical, and thus objective happenings.[3] But note that this alleged misrepresentation produces the essential characteristics of consciousness that motivated the mind-body (“hard”) problem in the first place: its apparent subjectivity (privacy) and qualitativeness. Absent these, we wouldn’t be discussing consciousness as a phenomenon in need of explanation. Consequently, the exposition of illusionism mainly and necessarily adverts to subjective qualities, the original and perhaps primary conception of an experiential feel in consciousness studies. The majority of theorists, even physicalists (as is Frankish), are realists about qualities; but he is not:

We certainly have conscious experiences — episodes of attentively seeing, hearing, tasting, feeling pain or pleasure, and so on. Our lives are full of them. These episodes make a powerful impact on us, psychologically and behaviourally, and we find it natural to talk about what they are like, how they feel. Illusionists deny none of this. What they deny is a certain theory of what conscious experiences are — the theory that says they involve direct awareness of private mental qualities. (Interview, 302, underlining added)

Eliminating qualities

The question I want to raise is whether this very natural talk of how experiences – visual, olfactory, tactile, auditory – feel is allowable under illusionism. It seems (to me at any rate) that talk of experiential feels is necessarily to invoke the idea of qualities, however one might account for them, e.g., as mental or physical. To say that pain feels a certain way, but has no qualitative character, seems to me a contradiction. If qualities don’t exist in experience, there’s no such thing as the qualitative character of pain or any other sort of sensory experience, so talk about how things feel is logically barred under illusionism unless there’s another way to cash out the concept of a feel. It’s difficult to come up with a non-qualitative, non-subjective substitute that still holds feels to be real, although as we’ll see Frankish attempts just that. But first, here’s the basic negative thesis of illusionism again:

We can reject [phenomenal] realism and hold that the qualities do not really exist, either as properties of objects or as properties of our experiences…Illusionists argue that we shouldn’t try to explain mental qualities, since they do not exist; rather we should concentrate on explaining why we have the impression that they exist. (Interview, 301, original emphasis)

Clearly, physical objects like tables and chairs are not possessed of mind-dependent mental qualities as properties since they don’t possess minds. Yet as Frankish acknowledges, they are, at least on a first pretheoretical pass, presented to us in experience in terms of qualities:

We all start off as children with the naive view that qualities are properties of things in the world — that the blue quality belongs to the sky, not to our minds or brains. It is only when we realize the problems with that naive view that we adopt the theory that qualities are mental. We notice that we can experience qualities in the absence of their usual causes (as when we dream) and conclude that the qualities are really in our minds. (Interview, 302)

Many dispute the naïve realist view that the sky possesses the mind-independent, objective property of being qualitatively blue, or apples the properties of being red or sweet: in general, science doesn’t find sensory qualities – so-called “secondary qualities” – in the world (Galileo’s insight). So as Frankish says, we (less naively) move them into the mind as mind-dependent features of experiences, not the world. But on illusionism there is no such property to be moved in the first place, since as he puts it “…qualities do not really exist, either as properties of objects or as properties of our experiences” (Interview, 301). [4] Are we then to renounce talk of qualities altogether? If so, then illusionists are tasked with replacing qualitative talk about objects (the apple is red) and experiences (the apple looks red to me) with the non-qualitative, purely physical terms still available to them, e.g., electro-chemical signals, patterns of psychological reactions, behavioral dispositions, etc. Clearly this is not in the cards.[5] In which case, continuing (as we must) to talk about qualitative properties, whether of objects or experiences, amounts to a performative self-contradiction for illusionists. Unless, that is, they always qualify such talk as referring to illusions.

Metaphysical agnosticism about phenomenal consciousness

Frankish avoids this tiresome prospect by proposing a non-illusory but non-qualitative substitute for qualities, at least the mental variety. This permits talking of what experiences are like:

The key point is that the monitoring process [of introspective consciousness] doesn’t provide a detailed picture of the reactions it tracks. Rather, it creates a simplified, schematic model — a sketch of the overall reactive pattern… We simply have a sense that each experience has an indefinable but distinctive character. This is what we’re expressing when we talk about what our experiences are like. It is something that’s happening in us (to that extent the phenomenal realists are right), but it is not a private mental quality. Rather, it is a complex pattern of objective reactions which we mistake for a simple subjective quality. (Interview, 304, emphasis added)

Here Frankish acknowledges that experiences have particular characters, something we report when saying what it’s like to have them, for instance that the sky really looks blue in our experience: there’s nothing illusory about this appearance. But these indefinable but distinct characters (IDCs) are what non-illusionists like me would say just are basic sensory qualities such as sweetness, red, or pain: they are distinctive but not further definable in terms of more basic characteristics; they are thus simple and non-decomposable, the hallmark of qualitativeness. Why then does Frankish say we’re mistaken in calling such experiential characters qualitative? I’d suggest it’s because he supposes to do so necessarily commits us to the idea of a specifically non-physical substance or property, a commitment he thinks bars the scientific explanation of consciousness. But talk of qualities does not commit us to any particular metaphysics; rather, it’s the nature of the qualitative – of the indefinable but distinct character of sensory experiences, to use Frankish’s language – that’s the big open question. Having such a character just is what we mean by qualitative prior to any inquiry about what constitutes that character. It is not built into the concept of phenomenal consciousness to suppose, as Frankish does, that “…we tend to think that [experiences] have a private subjective aspect that can’t be explained in third-person scientific terms.” (Interview, 301, emphasis added). It is not obviously the case that “qualia realism excludes consciousness from scientific investigation” (No trick, 326). Were he not an illusionist, Frankish could simply take phenomenal qualities (qualia) as the explanatory target while offering his physicalist hypothesis about what they might be: higher order representations of lower level patterns of objective reactions (more or less Hakwan Lau’s approach in his recent book). This hypothesis might turn out to be correct.

But again, illusionism is committed to denying the existence of qualities, even though it’s hard not to talk about them. The following passage illustrates the near ineliminable role of quality talk in describing an intuitive, and then a more reflective, take on nature of conscious experience:

…although the nature of these qualities reflects our reactions, they seem, pretheoretically, to belong to the objects we are perceiving. The information supplied by introspection is bound up with that supplied by our senses, so that we perceive things as having a certain impact on us, as infused with potent qualities. This explains why qualities seem to have a dual nature — in objects yet dependent on us. And it explains why it is a mistake to ask where the qualities really are. They are not really anywhere. Thanks to the combination of perceptual and introspective processes in our brains, we experience objects as qualitatively charged. But the qualities don’t correspond to real, intrinsic features either of objects or of us. They are potent illusions, an imperative commentary, generated by our brains to express the significance things have for us. (Interview, 304, original emphasis)

I think it’s correct to say that qualities aren’t really anywhere, but we don’t have to be phenomenal eliminativists in order to make this claim. Adam Pautz (Brown University) and I, both qualia realists of a representational bent, have independently come to this conclusion.[6] We can accommodate and perhaps explain the fact that qualities seem, as Frankish puts it, “intensely real” (Interview, 303) without having to locate them in space or deny their reality. If we take them to be representational contents, and possibly distinct from their representational vehicles (my conjecture, not Pautz’s) then their non-locatability makes sense since as a general rule we don’t find the terms of representation (e.g., concepts, propositions, numbers, or qualia) in the world as represented in those terms.[7] We’re not forced to think of qualia as partaking of any sort of mental substance (an “intangible mental essence” as Frankish puts it), so there’s no substance dualism necessarily implied by this content hypothesis. Rather, the idea (at least as I’ve suggested, see note 7) is that consciousness as phenomenal content constitutes a representational reality for the system and only the system, one that won’t be detected in the world the system evolved to represent (a represented reality), a world which includes its very own body. This is a naturalistic, non-spooky conjecture about the nature of experience that takes seriously the role of representation in cognition as a route to possibly explaining, not eliminating, the prima facie privacy and qualitativeness of consciousness. Whether or not this naturalistic brand of representationalism, should it pan out, can be accommodated under physicalism is, I think, an open question since naturalism is not necessarily equivalent to physicalism; I thus remain metaphysically agnostic about qualia. In any case, to bring consciousness within the scope of scientific investigation does not, as Frankish seems to think, require us to become illusionists about phenomenality.[8]

Can we objectify feels?

The content hypothesis concerning sensory qualities might be a way to explain, not eliminate, their prima facie subjectivity and privacy.[9] Illusionism, oppositely, in principle puts each person’s specific conscious experience out there in public space since there are no private mental qualities or anything else categorically private about consciousness, only physical goings-on. The physical events that on illusionism are identical to my experience of pain are in principle available to inspection by anyone in a position to see them (MRI scans, guided tours through the Leibnizian mill of the brain). The obvious objection here is that inspecting the physical goings-on that on illusionism constitute someone’s experience isn’t to feel what it’s like to have it. And the feel, after all, is what originally picks out a conscious sensory experience as such. So there still seems to be something unsharable, and thus categorially private, about my pain. As a feel, it only exists for me, not anyone else.

As noted above, Frankish allows that on illusionism we can still talk about what sensory experiences feel like; they have indefinable but distinctive characters (IDCs) which we then label with terms like red, sweet, pain, etc [update: Frankish actually denies experiences have characters, we only have the sense they do, see his comments on this essay]. But on illusionism, these characters, constituted by “complex patterns of objective reactions” are in principle visible: they are particular neurally instantiated processes which, given enough time, we might eventually pin down and observe in complete detail. As a result, there will be nothing undefined about these reactions, they’ll be very specific neural architectures and processes; nor, as neural events, will they have a distinctive character. Visually (more generally: observationally), ensembles of neural events are pretty much all of a piece, even though they vary in organization and function. So, in observing these events we’re not going to see anything that resembles an IDC. All told, even substituting IDCs for qualia, it seems a category mistake to think we can observe the feels of experience as any sort of public object. In which case, if we regard feels as real, they end up as essentially subjective: existing only for the system doing the experiencing, not for an outside observer. It is only from the inside – of being the system in question – that neural processes could have an IDC.

Consequently, it seems to me the only way to make experience public and objective, with no private subjective residue, is to deny that there are any feels involved; this means that talk of what experience feels like doesn’t actually refer to anything. And perhaps that’s the logical end point for illusionism: not only are there are no private mental qualities, there are no IDCs cooked up by introspective consciousness, only propensities to talk about such things. Feels aren’t real, but fictions that, at most, label the physical-functional states subserving certain cognitive capacities. This seems a bit radical, but it’s in the illusionist spirit of denying what seems dead obvious. We just have to accept that talk about what experience feels like, natural as it seems, is fundamentally in error. But this is not Frankish’s view:

I do not think that people are unfeeling machines. (Though I think that suitably complex machines could feel.) (“No trick,” 322)

So feels are real under his illusionism. But they’re objective, not subjective; that is, they are in principle public objects available to inspection. As argued above, this seems to me problematic. Neural processes as public objects are not possessed of IDCs – Frankish’s objective candidate for feels. If IDCs are real, they’re categorically subjective, not objective. So the way I see it, Frankish has to choose between realism about feels and his claim that experience is objective, that there’s nothing categorically private about it. Illusionism might win the day as a theory of consciousness – time, if we don’t run out of it, will tell. But if illusionism wins, I think we end up as fictionalists, not realists, about experiential feels.

October, 2022


[1] The articles are Illusionism and its place in contemporary philosophy of mind, an interview with Katarina Sklutova, and Illusionism is no trick, a reply to Paul Stenner, both of which appear in Human Affairs, 32, 321–327, 2022. In the text I’ll refer to the first as “Interview,” the second as “No trick.”

[2] Frankish says that experiences are subjective in a certain sense, but not private: “…conscious experience is constituted by a complex of activated sensitivities and reactions. These sensitivities and reactions are subjective — they are shaped by the subject’s biology, psychology, and personal history — but they are not essentially private” (No trick, 324). But given that experiences are not, at least on the face of it, observables, subjectivity as I think of it includes privacy: your experiences only exist for you, they are not publicly accessible. Experience as subjective in this sense is thus the rough opposite of what’s objective. The question, awaiting a settled theory of consciousness, is of course whether experiences actually are categorically private and non-objective, or whether consciousness can be fully objectified under illusionism or other brands of physicalism. I doubt it can, see Locating consciousness: why experience can’t be objectified, 2019 Journal of Consciousness Studies.

[3] Frankish calls these first order states “experiences,” so on illusionism the higher order representations (introspective consciousness) misrepresent experiences as having qualities (see Interview, 304). But of course, it’s the illusion of there being qualities that centrally picks out experiences on illusionism, so it seems the higher order representations are needed to have them.

[4] Frankish says “Consciousness involves reacting in a certain way to features of the world or one’s own body — the features we call colours, smells, sounds, pains, and so on. These features are physical ones (surface properties, airborne compounds, pressure waves, tissue damage, and so on), though this is not obvious to us” (No trick, 324). So, he identifies what we (naively, he would say) take to be sensory properties of experiences (putative “secondary qualities”) with objective physical properties (features). Experiences, as reactions to these features, themselves have no qualitative properties. So there are no qualities anywhere on illusionism. Rather, experiences are a “direct personal relation” to these features, a relation that results in the illusion of there being qualities.

[5] Patricia and Paul Churchland once recommended such a replacement, which for some reason has not caught on. See here for an amusing vignette of what it might be like.

[6] In Representationalism about consciousness, forthcoming in Uriah Kriegel (ed.), Oxford Handbook of the Philosophy of Consciousness. Oxford University Press, Pautz says “In sum, sensible properties (‘qualia’) cannot be plausibly located in the external world or in mindbrain. The best view may be that they are not located anywhere, even if they appear to be located in regions and surfaces in space.” In Locating consciousness (2019, Journal of Consciousness Studies), I say “There is a real, locatable apple with objectively specifiable features (mass, chemical composition, etc.), but it appears to us in conscious experience in terms of (real) phenomenology that isn’t itself anywhere.” 

[7] See section 5 of Locating consciousness: why experience can’t be objectified, Journal of Consciousness Studies, V 26, No.11-12, 2019.

[8] Frankish: “I have tried to set the record straight, showing that illusionism is a modest, ontologically deflationary view, which brings consciousness within the scope of scientific investigation” (No trick, 326). But of course there are several other approaches aimed at explaining consciousness within the scope of science that take phenomenal consciousness to be at least provisionally real, e.g., “higher-order theories, global workspace theories, re-entry and predictive processing theories and integrated information theory” (quoted from the abstract of Anil Seth and Tim Bayne’s Theories of consciousness, Nature Reviews Neuroscience, 2022).

[9] I’m not sure this proposal will pan out, but the work of Pautz and other representationalists who are also realists about phenomenal consciousness will help make that determination as they pin down the notion of representational content.