What Phenomenal Intuition Gets Right

Where Brown and Kammerer disagree

Richard Brown’s discussion with Kammerer on Consciousness Live! brings the disagreement into focus. Kammerer does not think that adopting illusionism makes the world go dark. His account preserves perceptual discrimination, access, action, and introspective representations, and explains why ordinary subjects judge that their lives contain phenomenal experiences. But these real processes, he argues, fail to satisfy the concepts under which we grasp them as phenomenal. Brown replies that errors in how we characterize our epistemic relationship to consciousness need not eliminate consciousness. What calls for explanation does not disappear when a theory relocates redness from objects into the mind, or gives a physical account of it. Their exchange is substantially about whether correcting a concept preserves its target. Discussion at 1:31:00; on redness at 2:11:50.

Brown’s superblindsight challenge makes the difference vivid. A creature might discriminate and use visual information without a visually present world, and Brown does not see how adding mistaken consciousness judgments would supply what is missing. Kammerer replies that ordinary cognition differs from superblindsight in many functional respects, including its production of the phenomenal illusion, while accepting that in the phenomenal sense neither has what the realist affirms. The question is not whether the mechanisms differ, but what their differences constitute. Discussion at 2:15:00.

There is a pressure on Brown’s side too. At one point he connects undergoing an experience directly with knowing what it is like, without requiring a further phenomenal concept. Kammerer challenges the apparent exemption of this knowledge from the fallibility of the rest of our cognitive apparatus. I want to separate two claims that can become entangled here: an experience, if it occurs, is an actual presence; a judgment that an experience occurs can be mistaken. The first is a constitutive commitment. It does not establish the second judgment’s infallibility. Discussion at 2:04:43.

Kammerer’s final response also matters. He presents illusionism as an inference to the best explanation, rather than making everything depend on deriving a formal contradiction from physical phenomenal consciousness. A realist does not defeat that argument merely by describing a coherent possibility. We need an account that explains both the actual organization and our ways of recognizing and misrecognizing it. Closing response at 2:19:36.

The occurrence is the vehicle

Our identity proposal begins with a distinction between a representational vehicle and its content. The vehicle is a concrete state or process doing the representing. In this framework, contents are abstracta: what is represented should not be confused with another concrete occurrence inside the system. An experience of red is the vehicle in its constitutive organization. It is not a further event supported by that vehicle, nor an abstract content promoted into concrete existence.

This accommodates unsuccessful representation. A red object need not exist for a dreamed red object to be experienced. The vehicle can be present, and can have representational character, while what it represents is absent or misdescribed. Phenomenal recognition could therefore get two things right even in hallucination: an occurrence exists, and it presents something as being some way. Neither truth requires the represented object to exist. In weaker failures, as in disorganized dreams or minimal states, the occurrence keeps its role as a vehicle in the representational economy even where no determinate content can be attributed to it without interpretive license. Presence and Actuality develops this case.

Our earlier appeal to acquaintance was intended to secure this actuality. If there is actual regulation of a particular vehicle, that vehicle exists. Merely representing something does not similarly guarantee its existence. But the existence argument is deliberately limited. It does not derive phenomenality from the bare fact of a relation. Thermostats, cooling systems, and blood supplies have actual relata too. The substantive proposal concerns regulation responsive to representational distinctions and their consistency within a minded system.

That regulation can maintain distinctions, preserve disagreement, assess reliability, and alter the persistence or coordination of the vehicles concerned. Consistency does not mean forcing everything into agreement. An organization that silences inconvenient input might become less epistemically adequate while becoming more internally uniform. Dreams and imagination also require distinguishing sources and modes of representation, rather than treating accuracy as correspondence between every active representation and the current external environment.

The scientific task is consequently rich. It could identify particular levels of sensory representation, temporal dependencies, and arrangements for binding and correction. An intermediate representational level could be constitutively important without being the sole level involved, much as in Jesse Prinz’s account of consciousness as attended intermediate-level representation.¹ Nothing about the initial existence argument decides these questions. It prevents an actual vehicle from disappearing into an abstraction; the account of regulation must explain what kind of occurrence it is.

This is also where our metaphysical ladder matters. Earlier essays distinguished organized information processing from representation within a mind, and graded mindedness by its own conditions.² Genuine cognition need not yet establish mentality: causally effective discrimination and representational processing need not amount to understanding or evaluating within a psychological life. Phenomenality and personhood are then proposed as distinct achievements requiring mindedness, without either being defined through the other. These are hypotheses about constitution, not names that manufacture the distinctions they describe.

Distinctness without a little inner picture

Return to the red circle and red square. Neither vehicle is literally circular or square: it is a content–vehicle confusion to infer the represented shape directly of the representing process. It would be another mistake to conclude that, because the vehicles do not instantiate the represented shapes, there is no genuine content-tracking difference between them. Any two vehicles differ somehow, as the words “circle” and “square” do; the difference that matters is structured by what is represented. Vehicles need not resemble their objects for the relations among the vehicles to mirror the relations among the objects. An experience of a circle can stand closer to an experience of an ellipse than to an experience of a square without anything in it being round. Roger Shepard and Susan Chipman called this second-order isomorphism.³

Our identity claim adds that the experiential difference is such a content-tracking difference, realized in the occurrences’ own concrete organization rather than merely in which stimuli happen to reach them. What must track the content is the causal structure of that organization: its operation should be explicable in terms of the represented geometry. Where its components sit in physical space is secondary. The components could be scattered without loss, provided their causal connections were preserved, although physical proximity is what ordinarily makes those connections cheap enough to build.⁴

Phenomenal intuition can register a difference which scientific inquiry subsequently explains. The intuitive discrimination need not already display the explanatory decomposition. Recognizing two occurrences as different is not the same achievement as identifying which processes constitute the difference.

I want to preserve an intrinsic aspect of this character, in a sense that needs care. The occurrence has a constitution of its own; it is not made into a different experience merely by receiving a different report or a different association elsewhere. But the relevant occurrence includes its constitutive relations, such as the regulation through which it is maintained, which causally affect the occurrence. Intrinsic here means fixed by the structure within the occurrence’s boundary, rather than by its further relations. It does not mean a categorical nature underlying all structure, of the kind Russellian monists place beneath physical dispositions; the metaphysics in the background is structuralist throughout. Nor does it mean detaching a small neural fragment from the dynamics sustaining it and demanding that it remain experiential. The boundary of the occurrence is part of the explanatory task.

The hypothesis explored in Can Two Minds Share One Experience makes this distinction consequential. One concrete experiential process might participate in two subjects’ cognitive economies while retaining its own qualitative character. The subjects could connect it with different memories and purposes. If those differences alter the process’s constitutive regulation, its character may change. If they do not, the existence of two surrounding perspectives need not multiply or recolor the shared occurrence. That possibility requires distinguishing constitutive dependencies from other cognitive connections.

This is stronger than saying that there are real vehicles behind a false appearance. I am proposing that their presence, representational character, and actual distinctness are among the things phenomenal intuition can get right. The realist commitment needs all of this positive content if it is to answer Kammerer’s concern that we are merely preserving an unspecified something and calling it experience.

Where strong illusionism would be true

A computational system can genuinely discriminate red and green without deriving its significance from an observer’s arbitrary interpretation. It can receive inputs produced by different wavelength ranges, preserve their differences, and use them to control subsequent activity. Its computation is a real pattern in the causal nexus. We should not call it merely pretend cognition because we doubt that it experiences anything.

But matching a cognitive function does not guarantee that the causal pattern our account identifies with experience obtains. An organization could reproduce the input–output profile and cognitive efficacy of ours, including the production of phenomenal judgments, while lacking the regulatory coupling we take to be constitutive. A recurrent process might be unfolded into a feedforward cascade that preserves its responses without preserving a persistent carrier for its corrections to act on; a component might replay a recorded input history instead of participating in the loop that would correct it. For such an organization, the illusionist’s account would be the correct one. It would deploy phenomenal concepts through a mechanism like the one Kammerer describes, and nothing would satisfy them.

The point is not about substrate. A simulation that reproduces the constitutive causal pattern, with its variables standing in the right counterfactual dependencies, realizes that pattern; calling it a simulation does not disqualify it. What cognitive equivalence leaves open is whether the pattern obtains at all and, where it does, its exact character. A symmetrical case shows the second gap. If a cognitively specified organization admitted an automorphism exchanging its red and green pathways, the cognitive description alone would supply corresponding positions connected to different stimuli, not a constitutive difference between the occurrences themselves. Whether actual color organization admits such a symmetry, and at what grain the constitutive causal pattern must be assessed, are questions for a forthcoming essay that takes up Eliezer Yudkowsky’s argument from implementation.

Here the contrast matters for a dialectical reason. Our realism does not say that every system deploying phenomenal concepts has what those concepts purport to track. It says that some organizations do, and that others could have the illusion without the presence. That difference in distribution is one of the things separating the position from a relabeled illusionism, as we will see.

Available through its character

We can give the positive proposal more shape through a computational analogy, without treating it as a literal computational specification. Experiential occurrences make themselves available to attention through their distinctness, in something like a content-addressable-memory arrangement. The organization can select a vehicle through what distinguishes it, rather than needing a separate symbolic address that says which internal item is intended.

The target is the concrete occurrence. Attention does not retrieve an abstract content and thereby turn it into an experience. Nor does it merely obtain a description which could be generated in the occurrence’s absence. In successful acquaintance, the distinctive activity participates in the processes that select, maintain, assess, and regulate that activity. Its character makes a difference to the manner of involvement.

This is a proposal about a constitutive organization, not a claim that every associative memory has experiences. A hardware content-addressable memory already performs selection by matching stored patterns. Calling that operation attention would not establish phenomenality. Our account requires identifying what mode of such availability occurs within a minded system and how it belongs to the ongoing regulation constituting the experiential vehicle.

Availability also differs from current focal attention. An occurrence might be maintained in the relevant organization without being selected for a deliberate judgment or verbal report. This lets us distinguish the conditions of experience from the conditions of reflecting on experience. It does not settle how extensive unattended experience is, or which maintenance processes suffice. Those are questions for the theory rather than consequences of the analogy.

This offers a naturalistic sense in which an occurrence can reveal itself. Its actual distinctive character enables a form of access to that very occurrence. Such access need not reveal every constituent or dependency, and it need not occur every time a phenomenal concept is deployed. It can nevertheless be access to something real and distinctive, rather than an empty demonstrative gesture toward whatever happens to cause a report.

What Kammerer explains

Kammerer’s Meta-Cartesian account connects phenomenal thought to our representation of evidence. His Cartesian Device characterizes certain states as primitive direct evidence. They are represented as self-presenting, self-revealing, and infallibly directly presented. Primitiveness is important here: the relevant nature is represented as simple and uncomposed, captured by the representation of it. Self-presentation can then be construed as revelation of that nature.⁵ Kammerer builds on Derk Pereboom’s primitivist version of the qualitative inaccuracy hypothesis, on which introspection represents a phenomenal property as having a nature entirely presented in our representation of it, and as metaphysically simple.⁶

He proposes an Evidence-by-Resemblance Mechanism to implement this architecture. Roughly, it redeploys a sensory concept such as RED in a metacognitive concept RED. Similarity between representations is used as a proxy for similarity between their referents. A state is treated as providing direct evidence for what it distinctively resembles. Applying this scheme to the state itself helps generate the apparent self-revelation and infallibility. The resulting metacognitive concepts represent phenomenal properties, rather than merely marking that a sensory process happened.⁷

This is useful to a realist. An account of how phenomenal thought acquires its apparent epistemic privileges need not be rejected because its author is an illusionist. Indeed, Kammerer explicitly distinguishes his architectural and implementation hypotheses from the additional claim that phenomenal concepts are unsatisfied. On their own, the first two remain neutral between realism and illusionism. His argument for elimination has to be supplied separately.⁸

Our account would not simply accept every detail of his proposed mechanism. In particular, actual regulation through distinctive vehicles might support successful recognition without requiring the resemblance relations his mechanism attributes. But his analysis helps formulate the question precisely. Which aspects of the apparent epistemic relation could be realized by an organized occurrence, and which arise from a mistaken characterization of it?

The project is to minimize that discrepancy. A concrete vehicle supplies actuality. Its constitutive distinctions supply character. Availability through those distinctions supplies a candidate basis for acquaintance and self-revelation. Regulation can supply epistemic value without providing an infallible grasp of a primitive nature. As this positive account becomes more specific, less of phenomenal thought has to be treated as fabrication. The remaining error might be an understandable cognitive mistake made in taking up a substantially veridical presentation.

The slip into simplicity

The central candidate mistake is this: because the constitutive complexity is not presented in recognizing an experiential character, the character is taken to have no constitutive complexity. The system can address the occurrence as one distinctive presence. It then treats that unity of access as evidence of simplicity in what is accessed.

There is no valid inference from the first to the second. A complexly constituted activity can be available as a unit. Its recognizability need not require recognition of each constituent. Nor must that activity conceal its real character merely because it does not reveal its decomposition. The character through which it is recognized may be the real character of the complex activity.

The form of this mistake is familiar. David Armstrong’s headless woman, shown against a black background with a black cloth over her head, looks headless: not seeing her head is taken for seeing that she has no head. Armstrong used the illusion in defense of materialism. Introspection does not present mental states as physical, and we slide from that to its presenting them as nonphysical.⁹ Kant diagnosed a structurally similar step in the Second Paralogism, where the formal unity of the “I think” is taken as evidence that the thinking subject is a simple substance.¹⁰ The present proposal finds the same pattern in our grasp of experiential character.

The slip is into simplicity, not into property talk. I no longer want to say that the mistake consists in sliding from an experience to a property. Ordinary property attribution can record a genuine difference in how occurrences are constituted. Nor is self-revelation itself necessarily the mistake. What I want to preserve is access through actual character, while denying that this entails a simple nature exhausted by that access. Kammerer’s richer package of self-revelation should not be assumed whenever we use the term in this more limited sense.

This separates the present proposal from the observation that a scientific description does not reproduce an experience. That observation is true but weak. Reading a description of viscosity does not make the page viscous either. The more specific hypothesis concerns an architecture that can recognize a concrete difference without representing the complexity constituting it, and can then misread this limited access as presentation of an uncomposed nature.

The error could arise before explicit philosophical theorizing. It might be automatic, reinforced by repeated successful discrimination, and deeply entrenched. Yet its being pre-theoretical does not establish that it is part of the content through which the occurrence is recognized. Whether primitive simplicity belongs to that presentation itself, or to a further uptake of it, decides a great deal. If it belongs to the presentation, the presentation is inaccurate at a central point. I think it belongs to the uptake. To argue for that, we need to look more closely at what simplicity would have to be.

What simplicity would have to be

Simplicity has a long philosophical career, and several notions travel under the word. Plato’s Phaedo argues that the soul is akin to what is incomposite, and therefore does not dissolve. Socrates’ “dream” in the Theaetetus describes primary elements that can only be named and perceived, never given an account; only their complexes can be known.¹¹ Locke’s simple ideas of color are uncompounded, contain “nothing but one uniform appearance,” and cannot be defined, only had.¹² Russell made acquaintance the relation through which we know simples. Sellars found the deepest difficulty for the scientific image in the ultimate homogeneity of a pink ice cube’s color.¹³

We can separate at least four senses. Something is mereologically simple if it has no parts. It is definitionally simple if it admits no account, only a name or acquaintance. It is homogeneous if it is uniform at every grain. And it is relatively simple if it has no parts under a given analysis. Wittgenstein, quoting the Theaetetus, argued that it makes no sense to ask absolutely whether something is simple: whether a chair, or the visual image of a tree, is composite depends on a way of dividing it.¹⁴ Pereboom’s notion of a primitive property combines two clauses: its nature is entirely presented in our representation of it, and that nature is metaphysically simple, not constituted by or analyzable into multiple properties.

A presentation can deliver relative simplicity. A red patch can be available without any presented parts in the articulation through which it is available. Absolute simplicity is a different kind of claim. To say that something has no parts under any analysis is to say that what is presented is all there is to it. No presentation can show that merely by lacking presented parts, any more than a dark stage can show that a woman has no head. It would need to carry a further clause, and that is all. That clause is the first half of Pereboom’s primitiveness: a nature entirely presented.

Nothing in perception seems to need such a clause. Discriminations, similarity orderings, recognitions, and the guidance of attention all proceed identically with or without it. The clause makes no difference to what the system can discriminate; it appears only in judgments about what an experience is, such as judgments of irreducibility. That suggests a criterion. Features of a presentation are those that make a difference to what the presentation enables the system to discriminate and do. On this criterion, distinctness, similarity, and much fine-grained character belong to the presentation. Exhaustiveness, and with it absolute simplicity, belongs to the uptake.

Even relative simplicity is not uniformly presented. Since Ewald Hering, orange has been described as looking reddish-yellow and purple as looking reddish-blue, while unique red looks like neither.¹⁵ Hume noticed the strain in his own theory: different simple ideas of color resemble one another to different degrees, blue resembling green more than scarlet, although their “perfect simplicity” seems to leave no respect in which they could resemble.¹⁶ Color appearance presents an ordering, and sometimes a composition, that a doctrine of simple ideas has to explain away.

Where simplicity is compellingly presented, it is presented as a feature of what is represented. Vision presents surface colors as simple, homogeneous qualities, what David Chalmers calls Edenic colors.¹⁷ Kammerer treats such color perception as a reliable misrepresentation of complex reflectance properties.¹⁸ The early modern relocation of sensory qualities then moved those simple qualities from surfaces into the mind.¹⁹ That move transfers a feature of represented content to the representing occurrence: the confusion with which this essay began. An experience of a circle need not be circular, and an experience of a homogeneous red surface need not be homogeneous or simple. On this diagnosis, Sellars’s grain problem depends on the transfer. Homogeneity belongs first to the represented ice cube; it needs a further argument to belong to the experience.

This also suggests one reason why the simplicity of experience is harder to set aside than the simplicity of colors. Kammerer notes that we readily accept that primitive colors do not exist in the world, while irreducibility intuitions about experience keep bothering us. On his account, this is because the Cartesian Device represents introspective evidence as trustworthy and essence-revealing.²⁰ The present account adds a further factor. The relocation needed a destination. Simple qualities displaced from surfaces could be relocated in experience; once they reside in experience, there is nowhere further inward to put them. Correcting the error then requires recognizing that the simplicity was a feature of content all along, rather than finding another bearer for it.

Kammerer has an objection ready for diagnoses of this family. He groups Armstrong’s diagnosis with views that explain mystery intuitions through introspective opacity. Such views, he argues, predict cognitively penetrable intuitions, since they derive the intuitions from inferences, whereas our intuitions about experience are not penetrable. He adds a counterexample: we do not know the causal history of our belief about our own name, yet that belief does not seem obvious in the way experience does.²¹ The present proposal does not locate the mistake in an inference. Uptake can be a default characterization supplied automatically, as impenetrable as a perceptual illusion, while still not belonging to the presentation it characterizes. And the counterexample lacks what the proposal requires. A belief about one’s name is not available through a distinctive character presented without parts. The slip needs an occurrence addressed as one distinctive presence, which is what availability through character supplies.

If this is right, the dispute with Kammerer narrows. In his own mechanism, primitiveness is supplied by the metacognitive concept, not by the sensory state to which it is applied. Both accounts therefore place the characterization as primitive in how the occurrence is taken up. The disagreement concerns whether that characterization is core to the concept of experience, so that rejecting it replaces the concept, or a revisable addition to a recognition that is otherwise largely correct.

Presence without infallible recognition

The sense in which experience cannot fail to be present needs similar care. On our identity proposal, an actual experiential occurrence cannot fail to exist or fail to be a representational vehicle, even where its content cannot be determinately reconstructed. A hallucinated object can be absent while the hallucination is an actual occurrence. This is the metaphysical security of presence.

It does not follow that an assertion of experience guarantees its target. A phenomenal concept can misfire. Deploying it need not bring the alleged experience into existence. The concept’s own deployment is a concrete cognitive process, but whether that process is experiential is another question. We must not manufacture an experience merely by observing that something real produced the claim.

The existence claim by itself is modest: any actual occurrence exists. The substantive part is the proposed identification of certain regulated representational vehicles as experiential presences, together with an account of how their actual distinctness enables recognition. This gives the recognition something specific to get right. It avoids treating bare actuality as a proof of phenomenality.

We can therefore distinguish an experience’s being present, its being available to attention, its being successfully recognized, and its being correctly described in every respect. These need not occur together. A system could misclassify a present occurrence, fail to attend to it, or claim one that is absent. None of these possibilities requires a genuine experiential presence to consist in a systematic illusion of presence.

This is where my position differs from defending consciousness through an unrestricted claim of certainty. I do not need every phenomenal judgment to be beyond correction. I need a positive account on which some recognitions genuinely concern an occurrence and correctly characterize aspects of it. That remains a substantive hypothesis, open to the sorts of evidence that could undermine its proposed mechanism.

Presence finer than recognition

Presence can outrun recognition in a less dramatic way as well. The exact character of an experience is finer-grained than anything the system’s cognition requires or retains. Two shades of a color may be indistinguishable when seen separately and distinct when they share an edge. A graded series makes the point without relying on edges: each shade can be indiscriminable from its neighbor while the ends of the series are clearly distinct. If looking the same were sameness of experiential character, that sameness would fail to be transitive; so some indiscriminable experiences differ in character.²² Diana Raffman has argued that our perceptual discrimination of shades far outstrips our capacity to identify or remember them, which explains why such differences elude memory and report.²³

The shade itself is represented by the sensory vehicle. What is not represented, in any form available to memory, comparison, or report, is the exact character of the experience or its difference from a shade no longer seen. That character is not known in the sense of being represented. It is present: undergone in the occurrence rather than claimed about it. Here the separation between presence and recognition does real work. Nothing is judged at this grain, so there is nothing at this grain to be infallible about.

The same consideration bears on substrate. If exact character is finer than cognitive function demands, a replacement of substrate that preserves cognitive function need not preserve exact character, although it plausibly preserves its type: an experience of red would remain an experience of red. Chalmers’s fading and dancing qualia arguments target this kind of possibility, but they rely on changes large enough that a rational subject should notice them.²⁴ Changes below the grain of cognitive function are exactly the ones a subject should be expected to miss.

This makes one traditional feature of phenomenality partly veridical. Experience is ineffable, not because its character is simple, but because it is finer than the concepts, memory, and report that would have to carry a description of it. The Theaetetus inference ran from having no account to being elementary. The fine grain of experience breaks that inference: what lacks an available account can still be complexly constituted. Ineffability of this kind may also be part of what makes each experience unrepeatable, and part of why it matters.

Why phenomenal character is not detachable

In Redness Is an Experience, particularly its postscript on the Brown–Coleman debate, I argued against treating phenomenal qualities as detachable from the experience of them. The opposing proposal distinguishes a quality from awareness of it: the very same redness could remain instantiated after awareness ceases. There is no immediate formal contradiction in that proposal. The argument against it rests on what constitutes the experience, independently of whether its character is simple or complex.

Our objection concerns constitution. If the regulatory intervention is part of what the experiential occurrence is, removing it changes the occurrence. Preserving a local activity pattern or the represented content does not preserve the full constitution while merely turning off access. What remains can still have causal powers, represent something, or contribute to nonphenomenal mental activity. Its continuity with the earlier process does not establish continuity of the phenomenal occurrence.

This must not be confused with saying that every lapse of focal attention extinguishes experience. The relevant withdrawal is the loss of constitutive regulatory involvement, not simply the absence of a report, a higher-order judgment, or deliberate inspection. Which interventions remove that involvement is precisely what the positive account must specify.

The fine grain of presence sharpens this. If exact character outruns access, why should it not also outrun regulation, and survive its removal? Because regulation is not access. Fine-grained character goes beyond what report, memory, and comparison can use; it does not go beyond the constitutive regulation in which the vehicle has that character. The question of detachment can therefore mark the boundary of the occurrence. What cannot be removed without changing the occurrence is constitutive of it; what can be removed, such as a report or an association, is not. Intrinsic character, in the sense defended earlier, is character fixed within that boundary.

A defender of detachable qualities could accept the causal changes and deny their constitutive significance. Brown’s challenge to acquaintance accounts also remains: what does the involvement contribute, rather than merely disclose? Our proposal answers that it sustains the particular form of availability and coordinated activity in which the vehicle has its experiential character. Whether that answer succeeds depends on an adequate theory of the organization. The argument cannot be completed just by insisting that the word experience implies someone undergoing it.

The two corrections to the earlier essay can thus coexist. We retain the claim that phenomenality is constituted in the undergoing, while abandoning a general objection to properties. And we replace a sweeping rejection of qualitative character with an attempt to explain why real, complex character can be mistaken for something simple.

The distinction illusionists can preserve too

An illusionist has resources to distinguish an actual quasi-phenomenal occurrence from a consciousness claim generated in some other way. Kammerer separates the distribution of quasi-phenomenality from the distribution of the phenomenal illusion. An animal might possess states of the relevant kind without representing them as phenomenal. Conversely, a sophisticated AI might have the requisite metacognitive illusion while lacking states sufficiently like those tracked in human phenomenal introspection.²⁵

This matters for agentic language models. Relaxing restrictions on consciousness claims could change their verbal behavior without establishing either the relevant introspective mechanism or the relevant underlying occurrences. Suppressing the claims would not establish absence either. These alternatives remain open even if the system’s cognition is genuine and even if it has some degree of mentality. A theory needs evidence about how the claims arise and what, if anything, they track.

We should therefore not claim a victory over illusionism merely because our phenomenal concepts can misfire. Nor does the disagreement consist simply in whether there is something real behind ordinary consciousness talk. Kammerer accepts real processes; Frankish explicitly distinguishes a positive account of quasi-phenomenal organization from an explanation of its introspective misrepresentation.²⁶

The stronger realist claim is that the relevant organization constitutes an experiential presence whose distinctive character can be correctly recognized. Kammerer can investigate that same organization while denying this identification. He also questions whether quasi-phenomenality marks one privileged kind: similarities in neural realization and similarities in functional organization can support different classifications. Our account must earn any stronger claim that its constitutive conditions identify a sufficiently determinate achievement.²⁷

The same discipline applies to ethics. Calling something phenomenal or quasi-phenomenal does not by itself explain why it matters. I take experiential organization and personhood to be distinct morally significant achievements, both requiring mindedness. Personhood does not wait for a phenomenal entrance qualification. Conversely, a being’s lack of personal self-government does not make its experiences insignificant. This plural structure survives the terminology dispute, while leaving the grounds and distribution of each kind of concern to be argued for.

Weak illusionism, verbal and substantive

Many readers will take the position defended here for what Keith Frankish calls weak illusionism: phenomenal consciousness exists, but introspection misrepresents some of its features. Frankish argues that this position is unstable. He defines a quasi-phenomenal property as a physical property that introspection typically misrepresents as phenomenal. A weak illusionist must then employ a concept of phenomenality stronger than quasi-phenomenality, on pain of saying that phenomenal properties are real only in the sense that psychokinesis would be real if people could produce the illusion of it. Yet the concept must be weak enough to yield to physical explanation. Frankish doubts that anything fits: “What phenomenal residue is left, once features such as privacy, intrinsicality, and ineffability have been stripped away?” He also argues that reductive identifications are covertly illusionist. They seem plausible because a system with the identified property would judge its states to be qualitative, not because those states would really have a qualitative dimension.²⁸

Kammerer sharpens the charge. A weak illusionist who denies all the features corresponding to the characterizations in phenomenal concepts is a verbal weak illusionist: what such a theorist calls phenomenal consciousness is exactly what the strong illusionist calls quasi-phenomenal consciousness.²⁹ Pereboom, whose primitivist hypothesis is closest to the proposal here, himself describes the difference between his view and strong illusionism as merely verbal.³⁰

Three features of the present account resist that classification. First, it does not strip away what Frankish expects. Measured against the features traditionally attributed to experience, it keeps most of them, reinterprets several, and rejects two:

• Presence is kept. An experiential occurrence is an actual presence, even when it is misdescribed or unrecognized.
• Intrinsic character is kept, in the structural sense: fixed within the occurrence’s constitutive boundary, not by reports or associations beyond it, and not by a categorical nature beneath structure.
• Ineffability is largely kept. Exact character is finer than concepts, memory, and report can carry.
• Direct apprehension is partly kept. An occurrence is available through its own distinctive character, though not infallibly.
• Privacy is mixed. Exact character cannot be communicated; but if one occurrence can participate in two subjects’ cognitive economies, it is not necessarily unshareable.
• Infallibility is rejected, as the account already declined it.
• Simplicity is rejected. The exhaustiveness on which absolute simplicity depends belongs to the uptake, not to the presentation.

Second, the account’s phenomenality and Frankish’s quasi-phenomenality are individuated differently, and they differ in extension. Quasi-phenomenal properties are identified by their role in triggering introspective representations. The constitutive pattern proposed here can be present without triggering any: fine-grained character makes no introspective difference at its own grain. And the triggering can occur without the pattern, in an organization that reproduces our cognitive function and our phenomenal judgments without the regulatory coupling. For that organization strong illusionism is true; for us, on this hypothesis, it is false. The same divergence answers the charge of covert illusionism. The identification is not made plausible by the judgments the property would produce, since it allows those judgments without the property.

Third, what remains is Kammerer’s own test: whether the rejected features belong to the core of phenomenal concepts.³¹ Infallibility is unlikely to be the decisive case, since realists widely reject it. Simplicity is. If primitiveness were part of how experience is presented, rejecting it might amount to replacing the concept. If it belongs to the uptake, a characterization that adds nothing to what the presentation enables us to discriminate, its rejection is closer to discovering that water is not an element: a revision of what we believed about the referent, informed by better access to it.

This does not settle the matter. Kammerer allows that the quasi-phenomenal kind might be fixed by similarities in realization rather than by the role of triggering introspection, and such a kind could coincide with ours. The disagreement would then turn on the status of primitiveness, which is why the argument about what simplicity would have to be carries so much of the weight.

What the realist still owes

The answer to Kammerer’s charge that we have changed the subject is therefore conditional. It must still be shown that phenomenal recognition gets right what we say it gets right, and that the mistake about simplicity can be corrected without losing its target. Presence, representational character, actual distinctness, fine-grained character, and availability through that distinctness are proposed as positive commitments. They are not meant as an empty label for whichever mechanism science eventually discovers.

There is also a psychological burden. Kammerer argues that familiar fallacy accounts fail to explain the persistent difficulty of positively understanding physical phenomenal consciousness. He does not merely demand that the intuition disappear after correction: his comparison with ordinary fallacies allows a misleading tendency to persist while the correct answer becomes clearly intelligible. Our account must therefore do more than remind the reader that cognitive errors can be stubborn. It must make a complex occurrence’s experiential recognizability intelligible and explain why the temptation toward simplicity remains.³² The absence of any further destination for relocated simplicity is part of an answer, not the whole of it.

The availability proposal is intended to help with that burden. If the same constitution supports real discrimination, attentional selection, and the limited exposure of its own complexity, we have a candidate common explanation of successful recognition and overreach. A useful inquiry would distinguish interventions that alter the occurrence’s constitutive character from interventions that alter only judgments about primitiveness or certainty. The theory would have to explain the resulting discriminations and reports without assuming that every report transparently measures experience.

The scientific burden is equally important. We must specify the organization well enough to distinguish experiential regulation from ordinary associative retrieval, genuine but nonmental cognition, and nonphenomenal mental activity. We must show why the constitutive differences we identify correspond to experiential differences. The contrast with organizations that match our cognitive function without the constitutive pattern motivates that task; it does not perform it. For color, this means that at the stage the account identifies as constitutive, the vehicles’ causal structure should realize the similarity geometry that experience presents. Early cone-opponent coding does not, while later cortical representations come closer; which of them belongs to the experiential occurrence is exactly what the theory must settle. If the proposed organization turns out to track only access, our identification would need revision.

The position is a physicalist realism about experience. It can fairly be called a fine-grained functionalism, provided that “function” means the actual causal organization at the relevant grain rather than the computed cognitive function. I avoid the label anyway, for reasons given in Phenomenal Consciousness as Mode of Being: it invites a slide back to the coarser reading. What the position opposes is the reduction of experience to the computed cognitive function. What it affirms is that experience is a mode of the organized activity realizing that function: a causal pattern that must actually obtain, whatever the substrate.

Whether it should also be called type B materialism depends on which experiential truths it leaves underivable a priori from a physical description. Some become derivable. If similarity and distinctness belong to the presentation and are realized as content-tracking structure, a complete description of the vehicles entails that an experience of red is more like an experience of orange than one of green. Others do not. Truths stated with recognitional concepts are available only to a system that has undergone the occurrence, which is what Frank Jackson’s Mary lacks, and exact character outruns even those concepts. An epistemic gap therefore remains, and the position is type B. But the gap is not the brute one Chalmers attributes to type B views. His objection assumes that phenomenal concepts reveal the full nature of their referents, leaving no difference between appearance and reality for an a posteriori identity to exploit.³³ If the clause and that is all belongs to the uptake, that difference exists. Phenomenal recognition fixes its referent through presented character, and the referent’s constitution is discovered, as water’s was. The hard problem is then reduced to an ordinary a posteriori identity and a gap in access that the theory itself predicts: structure cannot deliver like this, because like this is available only to a system that has undergone the occurrence.

Kammerer’s account is valuable because it lets us ask where the mistake lies with greater precision. We can investigate how a mind treats its own evidence as primitive without conceding that every aspect of experiential recognition fails. The realist project is to identify the concrete occurrence, explain how its character makes it available, and show how this success can encourage a mistaken estimate of its simplicity. If that project succeeds, we will have learned something about experience from illusionism without having learned that experience was never there.