Contribution
It gives the position a name and a problem. Before it, denials of phenomenal consciousness travelled as “eliminativism”, “irrealism” or “qualia quining”. Frankish replaces these with “illusionism”, sets the “illusion problem” against the hard problem, and fixes the line between the strong view and the weak view many physicalists already hold. The article does not supply an illusionist theory. It argues that the programme is worth pursuing and maps the choices a theory would face.
Key insight
The dividing question is not whether introspection misrepresents experience. Most physicalists agree it does. The question is whether anything “feely” is left once the misrepresented features (ineffability, intrinsicness, privacy, infallible access) are removed. If what remains is only a physical property that causes representations of phenomenality, that is quasi-phenomenality, and calling it phenomenal is like calling stage conjuring psychokinesis. So conservative realism must either name a residue or become illusionism (§§1.2, 2.2).
Assumptions
- Physical is used broadly: properties identical with or realised in microphysical ones (p. 2).
- “Experience” is used functionally, for the direct outputs of sensory systems, so it is not definitional that experiences are phenomenally conscious (p. 2).
- The causal closure of the physical is assumed in the argument against radical realism (“as ours appears to be”, p. 11).
- A representational theory of mind: properties must be represented to have cognitive, affective or motivational significance (p. 14). The abductive argument and the reply to acquaintance theory both depend on it.
- A presumption of conservatism in science, stronger when the pressure for innovation comes from introspection (p. 11). Asserted as a methodological principle.
- Introspective representations are not themselves phenomenally conscious (p. 3), on pain of regress; footnote 13 records Pereboom’s alternative.
Key results
The article is philosophical argument; its “results” are distinctions and arguments.
- Three approaches (§1.1): radical realism, conservative realism, illusionism. The first is radical, the other two conservative.
- Strong vs weak illusionism (§1.2) and quasi-phenomenal properties (p. 4, “zero qualia” in Frankish 2012). Weak illusionism is “diet qualia”; classic qualia are “qualia max” (n. 4).
- Analogies (§1.3): Dennett’s user illusion, Rey’s projection of stable objects, Humphrey’s “ipsundrum” after the Gregundrum, Pereboom’s qualitative inaccuracy hypothesis, Rey’s child in the cinema. Introspection bundles complex physical detail into a simple feel. The illusion is cognitively impenetrable, as perceptual illusions are.
- Varieties (§§1.3–1.5): theories differ on which sensory states ground the illusion, on which of their properties are quasi-phenomenal, and on the form of introspective access. An “outward-looking” illusionism puts the illusory properties on distal objects (Hall 2007). Illusionism is distinct from the grand-illusion thesis about visual richness, though change blindness supports it.
- What it keeps (§§1.6–1.7): consciousness in an inclusive sense; a functional sense of “what it is like”; a fictionalist reading of everyday phenomenal talk (the Lear analogy). In the functional sense “there is something it is like to be a zombie” (p. 10).
- Against radical realism (§2.1): conservatism, and epiphenomenalism under causal closure. Russellian panpsychism avoids it only by “huge profligacy” (p. 11).
- Against conservative realism (§2.2): the phenomenal concept strategy still needs concepts of feels, so the pressure remains. Reductive theories that identify phenomenality with a functional property explain at most why subjects would judge their experience qualitative: “a conflation of phenomenality with the representation of phenomenality” (p. 13).
- For illusionism (§2.3): (i) the debunking argument, after Chalmers 1996 pp. 186–7; (ii) anomalousness is evidence of illusion, and since introspection cannot be checked from within, the explanatory gap and zombie conceivability support illusionism as much as radical realism. Further advantages: it lets phenomenal properties be both “magical” and potent, as intentional objects; and it opens an adaptive function, Humphrey’s “magic show”.
- Replies (§3): denying the data (§3.1, acquaintance is psychologically inert, needs a non-psychological subject, and posits “intrinsic subjectivity”); no appearance–reality gap (§3.2, true only weakly); who is the audience (§3.3, no inner observer, “multiple drafts of sensation”); representing the non-existent (§3.4, functional-role semantics or hybrid concepts combining a theoretical concept of the phenomenal with recognitional ones).
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Weak illusionism collapses into strong illusionism unless it names a phenomenal residue beyond quasi-phenomenality | moderate | §1.2, §2.2; argued by challenge, with the residue’s impossibility “doubted”, not shown (p. 4, citing Frankish 2012) |
| C2 | Radical realism faces epiphenomenalism | moderate | §2.1; conditional on causal closure, which is assumed |
| C3 | Reductive theories of phenomenality are covertly illusionist | moderate | §2.2; one worked case (higher-order perception), generalised by assertion |
| C4 | Our phenomenal claims give no evidence for phenomenal realism, given an anti-physicalist premise | moderate | §2.3; the debunking argument, from Chalmers 1996, conditional on anti-physicalism |
| C5 | Apparent anomalousness plus uncheckable introspection supports an abductive inference that phenomenality is illusory | weak–moderate | §2.3; the abduction’s strength depends on the prior weight given to illusion, not stated |
| C6 | Acquaintance makes phenomenal consciousness a psychologically inert datum | moderate | §3.1; follows from the representational-theory assumption |
| C7 | The no-gap objection adds nothing against illusionism | moderate | §3.2; turns on reading “seeming” as introspective representation |
| C8 | Illusionism needs no Cartesian theatre: appreciation of the illusion can be distributed over unintelligent subsystems | assertion | §3.3; Dennett’s argument for perception is extended to introspection by analogy |
| C9 | Representing non-existent phenomenal properties is no harder for illusionists than for realists | weak | §3.4; options sketched, none developed; the author calls these “preliminary remarks” |
Concepts
- Illusionism — the view that experiences do not have phenomenal properties and that the task is to explain why they seem to. Always the strong form in this article (p. 4).
- Illusion problem — explaining why experiences seem to have phenomenal properties, and why the illusion is so powerful (pp. 1, 21).
- Phenomenal properties — introspectable qualitative properties (“feels”) that determine what it is like to have an experience (p. 2).
- Quasi-phenomenal property — a non-phenomenal, physical property, perhaps gerrymandered, that introspection typically misrepresents as phenomenal (p. 4).
- Weak illusionism — real phenomenal properties, illusorily represented as ineffable, intrinsic, private and infallibly known; “it might equally be called weak realism” (p. 4).
- Introspective vs intrinsic subjectivity — awareness of experience produced by representational mechanisms, versus a subjective dimension that arises “simply from our being the things we are” (p. 17). The first is granted, the second suspected to be a fiction of the first.
- Representational zombie — a creature lacking introspective representations of its phenomenal properties, and so without cognitive access to them (p. 14).
Connections
- Dennett. Called the pre-eminent defender (p. 1). The user illusion (CE pp. 216–20, 309–14), heterophenomenology (n. 8), multiple drafts extended to “multiple drafts of sensation” (p. 20), the storyteller’s authority (p. 18), and change blindness as evidence (p. 8) are all his. “Quining Qualia” (LIT-433, NOTE-357) supplies the list of marks weak illusionists deflate. Dennett’s reply in the same issue (LIT-434) is filed unread.
- Frankish 2025 (LIT-103, NOTE-143). That commentary assumes this article: its “post-Cartesian paradigm” is illusionism, and its advice to drop “qualia” talk is this article’s §1.6 hedged the other way. Here Frankish allows that phenomenal terms “might” be reconceptualised as quasi-phenomenal ones but says that “could invite confusion”, and in scientific work it “might be wiser” to drop them (p. 9). The 2025 commentary makes the second option firm.
- Chalmers. The Conscious Mind is cited for the debunking argument, the “datum” objection and acquaintance; the 2003 phenomenal belief paper for acquaintance and quotational concepts. The meta-problem paper (LIT-425, NOTE-354) is the later reply to the programme this article launched. NOTE-354 records that Chalmers suggested the name and that §6 agrees that dissolving illusionists must be strong illusionists. LIT-425’s Moorean step is the “denying the data” objection of §3.1.
- Schwitzgebel. The Weirdness of the World (LIT-216, NOTE-188) cites Frankish 2016 when sharpening the eliminativist exit from the United States argument.
- Dennett 1980 (LIT-456, NOTE-371). Dennett’s reply to Stich holds that “there are no such things as pains, although of course people do feel pain”. That is §1.6’s distinction between phenomenal properties and the experiences that are said to have them, made 36 years earlier for pain, without the word “illusion”.
- Taylor 2024 (LIT-461, NOTE-379) reads Dennett as a real-pattern realist for whom qualia eliminativism is peripheral. This article reads him as the leading strong illusionist. Both can be true only if “consciousness is a real pattern” is read as Frankish’s inclusive sense of consciousness (§1.6), with phenomenality excluded.
Bearing on the record
- THEORY-043 (society of mind and group consciousness). No direct bearing: the article never discusses groups or composition. Its §3.3 claim that the illusion’s audience is distributed over “numerous unintelligent subsystems” (p. 19) is Dennett’s discharge of homunculi (LIT-393, unread, known through Stich) applied to introspection. If an illusionist reading of THEORY-043 is wanted, the question becomes whether a group has introspective representations of its own states that represent them as phenomenal. That is the reader’s framing.
- THEORY-023 (AI consciousness). No direct bearing; the article does not discuss machines. §1.7’s functional sense of “like something” makes the question for a machine one about introspective mechanisms.
- LIT-103’s standing. Confirmed: NOTE-143 reads “quasi-phenomenal properties” as Frankish’s (2016) neutral term, which is right (p. 4).
- No ML instruction. Nothing belongs in the anthology.
Limitations
- A programme, not a theory. The article says it “will not propose a detailed illusionist theory” (p. 1). The illusion problem is posed, not solved.
- The key challenge is a challenge. C1 depends on no one being able to articulate a stable weak-illusionist residue. Frankish “doubts” this is possible and cites his 2012 paper; the article does not show it.
- The abduction is underspecified. “If there is even a remote possibility that we are mistaken … there is a strong abductive inference” (p. 14) does not say how anomalousness is to be weighed against the cost of error about something introspection presents as certain.
- The representation of phenomenality is left open (§3.4), and it is the core of the illusion problem by the article’s own account.
- No empirical programme. §4 names the method (hypotheses about mechanisms across the cognitive sciences) but does not propose one.
Open questions
- Can any illusionist theory explain the content of the introspective representations: why they represent “feels”, richly and determinately, rather than mere introspectable somethings?
- Is “multiple drafts of sensation” (p. 20) compatible with Humphrey’s single evolved “magic show”? Frankish flags that Humphrey might not agree.
- Does the fictionalist reading of everyday phenomenal talk (§1.6) survive the fact that most speakers do not take themselves to be speaking within a fiction? Frankish compares them to “naïve theatregoers” (p. 9); Schwitzgebel’s charity objection to Dennett, reported by Taylor (NOTE-379), presses the same point.
Corrections
- No dossier existed, so there is no seeded skim to correct.
- Citation. The brief’s citation (JCS 23(11–12):11–39, 2016) agrees with the eprint’s header. The eprint is a later revision (“Version2, 6/4/26”), and “may differ in minor ways from the print version”.
- One reference slip in the paper: Randi (1997) is given as pp. “170–108” (reference list, p. 23); the range is impossible as printed.