3.1 The Bent Stick: The Argument Stated

Picture a straight stick resting halfway in a glass of water. At the waterline it looks bent. You know the stick has not changed shape, but knowing that does not straighten the look. Now ask the awkward question: what are you directly aware of? Try answering before giving the bend a place to live.

The argument from illusion offers a compelling answer. Since the physical stick remains straight, something else must actually bear the apparent bend. That must introduces the extra premise. Philosophers later gave it a useful name, the Phenomenal Principle: whenever something appears to have a sensible quality, awareness must include something that really has that quality. Grant the principle and the first inner item has already entered the room. Call it a sense-datum, an appearance, an inner image. It bears the bend the stick lacks, and awareness lands on it rather than on the physical stick.¹

That gives the argument its first joint. The second spreads the result from illusion to perception as a whole. Grant, for the moment, that the bent-looking case involves an inner item. A veridical perception of a straight stick and an illusory perception of a bent one can shade into each other continuously: lift the stick slowly from the water and watch the bend ease out. No point in that progression announces a switch in the kind of thing you confront. Ayer and Price infer a common kind across the series.² If awareness lands on an inner item at the distorted end, consistency seems to place an inner item at the veridical end too.

Keep the two joints apart.

1. The Phenomenal Principle moves from the stick looks bent to something before awareness actually is bent.
2. The continuous-series argument then moves from an inner item appears in illusion to the same kind of item appears in every perception.

A critic can reject either move. Deny the first and no inner object enters. Grant it but resist the second, and the veil theorist can ask where, along the smooth slope, awareness switches from inner to outer. That question gains force only after the first joint admits an inner object.

If both joints hold, the conclusion lands hard. In every perception, veridical or otherwise, the immediate object of awareness lies within. The world stands at one remove, behind the representation, on the far side of the glass.

It also carries a modern cargo. The conclusion gives human perceivers and text-only language models one problem in common: neither would reach a scene directly. The comparison stops there. Human perception comes embodied, causally governed by its surroundings, and shaped by a long history of successful action; a text-only model has no comparable perceptual loop. Still, if every mind meets only an inner display, one contrast Part Four needs has already disappeared. More than a stick bends on the outcome.

---

3.2 The Argument from Hallucination

A sharper version of the same argument presses from the limit case, where the world drops out entirely.

Actual hallucinations show how much a mind can present when no matching object occupies the scene. People with Charles Bonnet syndrome may see faces, animals, or patterned figures with nothing corresponding before them. A bereaved person may briefly see a late spouse in the kitchen doorway. These cases deserve care: people with Charles Bonnet syndrome often retain insight into the source of what they see, and bereavement experiences vary widely in force and interpretation. Clinical examples establish that vivid objectless presentation occurs. They do not establish the philosopher’s stronger premise that hallucination can match perception perfectly from the subject’s point of view.³

The argument therefore asks us to consider a possible hallucination specified to meet that stronger condition: same felt character, same apparent scene, no external object, and no clue available from within the episode that anything has gone wrong. The subject does not feel himself generating an inner image. He feels himself seeing.

Two claims now do the work.

1. If a hallucination and a genuine perception can remain wholly indiscriminable from within, they must share the same fundamental mental kind. Call this the common-factor premise.
2. Whenever awareness presents an object and its qualities, something must exist that bears those qualities and serves as awareness’s immediate object. Call this the phenomenal-object premise.

Since no external object exists in hallucination, the bearer must lie within; the common factor then carries that inner object into genuine perception.

Indiscriminability alone produces no sense-datum. An intentionalist can grant perception and hallucination a shared way of presenting the world without turning the presentation into an object one sees. Only the phenomenal-object premise brings an inner item into the common factor.⁴

Hallucination remains the sharper challenge. Illusion leaves a real stick available for misrepresentation; hallucination removes it. Anyone who rejects the inner item must explain detailed presentation without quietly reinstalling the scene inside the head.

Now watch the two pressures compound. Trace the causal story: the world reaches consciousness only after a long relay, crowded with processing, as the previous chapter described. Examine the phenomenology: the same apparent scene could, the argument claims, occur without any scene answering to it. The first pressure concerns causal mediation; the second concerns what experience can present. Neither by itself proves that awareness terminates within. Together they make the wall look solid. Science presses from one side, phenomenology from the other.

---

3.3 The Two Escape Routes: Austin and Disjunctivism

Two lines of resistance expose the disputed moves without yet settling the case.

J. L. Austin thought the whole argument ran on a grammatical sleight of hand.⁵ He had spent the war in military intelligence and came back to Oxford with a plain habit: read the sentence people actually say. The sleight lives in a quiet trade: the stick looks bent, a remark on how perception presents its object, becomes I am aware of something bent, an existential claim about an inner item. Only the trade conjures that item into being. The conceptual order runs against it too: something must count as bent before we can say that it merely looks bent. Appearance-talk borrows its sense from object-talk and cannot supply the inner stuff from which the world gets rebuilt.

Disjunctivists press a different seam.⁶ The argument from hallucination assumes that states indiscriminable from within must share a fundamental mental kind. Deny that, and the inference never starts: genuine perception may involve a relation in which the world itself figures, while hallucination belongs to another kind entirely. A convincing forgery can defeat inspection without becoming the original. Likewise, your inability to tell two experiences apart reports a limit in you; it does not settle what, beyond that limit, the experiences amount to.

The hardest pressure on this route comes from a causally matching hallucination. Suppose an unusual intervention produces the same proximate neural conditions as a genuine perception, but without the perceived object. If a positive account of those neural conditions explains the hallucination’s apparent scene and its effects, why does the same account not also explain the matching perception, leaving the world-involving relation with no work to do? At Berkeley, M. G. F. Martin built his most demanding defense of disjunctivism around this pressure. He refuses that positive common characterization: the matching hallucination gets described through its indiscriminability from perception, not through a shared representational nature. The cost is explanatory, not contradictory; the move from matching causes to an identical total mental state still needs an argument, and Martin refuses it.⁷

Austin and the disjunctivists therefore leave different inheritances. Austin identifies the illicit promotion of a way an object looks into a second object of awareness. The Seeing and the Seen completes his diagnosis by separating the vehicle of presentation from its object. Disjunctivism preserves another result the book will keep: indiscriminability never settles fundamental kind. The book declines the further claim that perception and hallucination share no positive mental character, because representational content can explain what they share without turning content into an inner object.

An answer must explain objectless presentation, preserve the world’s role in genuine perception, and find no private scene along the way.

---

3.4 What the Argument Would Cost Us

The arguments have not forced a veil upon us. Consider, nevertheless, what adopting one would cost. Those costs need to be weighed against the explanation it offers.

Borrow a small case from epistemology. A man hallucinates a clock on the wall of his kitchen. The wall carries no clock; the space stands empty. But the hallucination is vivid and orderly, and at the moment he looks, the hallucinated clock reads 3:42. As it happens, the actual time is 3:42. He forms the belief that it is 3:42. The belief is true. He feels every bit as certain as he would with a real clock in front of him. Does he know what time it is?

No.⁸ The time played no part in producing his true belief. Let the time advance, stop every working clock in the house, or carry him across a time zone; the hallucinated dial follows its own schedule. From within the episode, he cannot distinguish this state from one produced by a clock and functioning eyes. Outside that perspective, the difference looks decisive. One belief answers to the time. The other merely agrees with it.

The veil theorist owes you no apology for that kind of luck. Her inner display tracks the world along lawful pathways: move the tomato and the display changes with it. Reliable indirect perception would therefore differ from the hallucinated clock in exactly the respect the example exposes. The analogy isolates causal disconnection; it cannot show that a lawfully driven state lacks perceptual contact. That leaves the real dispute: do we see the world by means of the state, or first see an inner bearer that depicts it?

One pressure concerns what perception consists in. If awareness terminates at an inner display, no improvement in fidelity removes the intermediary. Perfect correlation makes the display reliable; it does not remove it. A direct realist can therefore distinguish causal control from perceptual contact. The pressure disappears if content names the manner in which the tomato itself appears rather than a second object in its place. That possibility belongs to the next chapters.

The veil theorist replies that the world appears as content: an inner display of the tomato hardly leaves the tomato absent. The dispute turns on whether the world figures as the content of something seen or as what is seen. The Seeing and the Seen separates the vehicle of presentation from its object; The Identity Claim asks whether content needs any inner bearer with tomato-like qualities. Until then, the constitutive cost remains a charge to be tried, not a sentence already passed.

The second pressure concerns knowledge of the coupling. Every theory uses perception to investigate perception, so ordinary circularity proves nothing. Trouble begins only if the veil confines direct evidence to private inner items and requires an inference from them to the world. Retinal maps, psychophysics, and the stick pulled dripping from the water then arrive as further items inside the divide. Descartes secured the bridge through a divine guarantee. An epistemic externalist can refuse the demand: reliable coupling may yield knowledge even when the subject cannot prove it from within. The pressure therefore falls on the stricter veil account that treats private inner items as the whole evidential base.

Neither pressure refutes indirect realism. The veil theorist must explain why an inner object does not terminate awareness short of the world and how veiled evidence establishes a bridge beyond it. The direct realist must explain how mediated, fallible perception reaches the world without magic or infallibility. Nobody leaves the audit early.

Whatever perception turns out to involve, ordinary seeing presents itself as contact with the world, not acquaintance with a private stand-in. That suspicion, felt before anyone can prove it, supplies the thread the next chapters follow.

---

Chapter Summary

The claim. The argument from illusion needs two joints: the Phenomenal Principle places an inner bearer behind the bent appearance, and the continuous-series argument spreads that bearer across perception. Hallucination adds the possibility of an objectless but indiscriminable scene and a common-factor premise, but it reaches an inner object only by retaining the phenomenal-object premise.

Where the argument stands. Illusion and hallucination do not force an inner object: both arguments need a further premise about what presentation requires. The clock case isolates causal disconnection, not a defect shared by all indirect perception. Lawful coupling leaves the separate question of what awareness reaches.

What’s left open. Austin’s grammatical diagnosis awaits the vehicle/object distinction that completes it. The book keeps the disjunctivist insight that indiscriminability does not settle kind, while leaving its no-common-character metaphysics behind.

The hand-off. We need not accept the inner object, but we still owe an account of sensory presence without it. The next chapter begins with what experience presents when we attend.

---

Notes

1. Howard Robinson later labels the first inferential joint the Phenomenal Principle in Perception (London: Routledge, 1994), esp. 32: if something appears to a subject to have a sensible quality, then something of which the subject is aware must actually have that quality. The principle converts a manner of appearance into a further bearer of the apparent quality. J. L. Austin’s criticism targets that conversion directly in Sense and Sensibilia, reconstructed from manuscript notes by G. J. Warnock (Oxford: Oxford University Press, 1962), esp. secs. III and VII. ↩
2. The ancient pedigree concerns perceptual variation, not yet the modern inference to sense-data: Sextus Empiricus uses the oar that appears broken in water and straight outside it in Outlines of Pyrrhonism I.118–119. The epigraph appears in A. J. Ayer, The Foundations of Empirical Knowledge (London: Macmillan, 1940), 6. Ayer and H. H. Price, Perception (London: Methuen, 1932), give classic modern statements of the spreading argument. Austin reconstructs their continuous-series progression at Sense and Sensibilia, 60–62. That progression supplies the second joint; it has force only after the first joint has placed an inner item in the illusory case. ↩
3. The philosophical argument requires a perfect or subjectively indiscriminable hallucination, not merely a vivid clinical episode. Typical Charles Bonnet syndrome combines complex visual hallucinations with preserved cognition and, often, insight; see Dominic H. ffytche, “Visual Hallucinations and the Charles Bonnet Syndrome,” Current Psychiatry Reports 7, no. 3 (2005): 168–179. Sensory and quasi-sensory experiences of deceased people also occur commonly in bereavement and vary in modality and interpretation; see Evelyn Elsaesser, Chris A. Roe, Callum E. Cooper, and David Lorimer, “The Phenomenology and Impact of Hallucinations Concerning the Deceased,” BJPsych Open 7, no. 5 (2021): e148, https://doi.org/10.1192/bjo.2021.960. These examples establish objectless presentation but not exact phenomenological matching. ↩
4. Modern philosophical formulations of the stronger hallucination argument appear in Frank Jackson, Perception: A Representative Theory (Cambridge: Cambridge University Press, 1977), and Robinson, Perception (1994). Intentionalism can grant common representational structure while denying that the structure becomes an inner object of awareness; the later chapters on transparency and the identity claim develop that response. ↩
5. J. L. Austin, Sense and Sensibilia (1962), esp. secs. III and VII. Austin’s central charge is not merely that the argument selects unrepresentative examples (though he presses that too — perception consists overwhelmingly of cases where things are as they appear), but that it depends on what he treats as a grammatical confusion: the move from “x looks F” to “there is something that is F of which the subject is aware.” His priority claim is conceptual and linguistic rather than a claim about child development. For an extension of Austin’s approach into epistemology, see John McDowell, “Criteria, Defeasibility, and Knowledge,” Proceedings of the British Academy 68 (1982): 455–479. The decisive form of the diagnosis — recasting the confusion as a conflation of the vehicle of a representation with its object — is reserved for the later chapter that names that distinction. ↩
6. Disjunctivism denies that veridical perception and subjectively indiscriminable hallucination share a common fundamental mental kind. J. M. Hinton introduced the disjunctive conception in “Visual Experiences,” Mind 76 (1967), and Experiences (Oxford: Clarendon Press, 1973); Paul Snowdon, “Perception, Vision and Causation,” Proceedings of the Aristotelian Society 81 (1980–81): 175–192, John McDowell, Mind and World (Cambridge, MA: Harvard University Press, 1994), and M. G. F. Martin developed importantly different versions. For the broader debate, see Alex Byrne and Heather Logue, eds., Disjunctivism: Contemporary Readings (Cambridge, MA: MIT Press, 2009). ↩
7. Causally matching hallucinations supply the sharp pressure case. Martin does not simply deny a universal same-cause/same-effect law. He argues that if a matching hallucination receives a positive representational characterization sufficient to explain its salient features and effects, the same explanation threatens to screen off the naïve realist’s world-involving account of perception. His response characterizes the hallucination negatively, through its indiscriminability from perception; see “The Limits of Self-Awareness,” Philosophical Studies 120, nos. 1–3 (2004): 37–89, esp. 60–63. The book keeps the disjunctivist’s epistemic modesty but declines its denial of positive common representational character. ↩
8. The hallucinated clock adapts Bertrand Russell’s stopped-clock example in Human Knowledge: Its Scope and Limits (London: George Allen & Unwin, 1948), 170–171, a predecessor of the justified-true-belief counterexamples Edmund Gettier sharpened in “Is Justified True Belief Knowledge?”, Analysis 23 (1963): 121–123. Because the hallucinated belief remains causally uncoupled from the time, it fails influential safety and sensitivity conditions; see Ernest Sosa, A Virtue Epistemology (Oxford: Clarendon Press, 2007), Duncan Pritchard, Epistemic Luck (Oxford: Oxford University Press, 2005), and Robert Nozick, Philosophical Explanations (Cambridge, MA: Harvard University Press, 1981). Those diagnoses do not transfer automatically to the indirect realist’s perceiver, whose inner states covary lawfully with the scene. The main text therefore uses the case only to isolate causal disconnection and then asks what, beyond lawful control, perceptual contact requires. It does not treat perception as a species of knowledge or convict indirect realism of Gettier luck. ↩