Prompts for What Phenomenal Intuition Gets Right

Actuality and the explanatory task

The relation argument is meant to establish the existence of the vehicle that sustains the particular redness; one can further say that the redness as identified by phenomenal intuition. So that argument alone is not meant to establish that the occurrence is phenomenal, only that it exists.

The explanatory task is scientifically rich, for example it can explain that the experiential vehicles are (often, particularly for vision and hearing) constitutively (physically and representationally) at an intermediate level (as in Jesse Prinz’s account). Obviously the fact of relation doesn’t establish anything beyond existence, the fact of regulating vehicles for representational consistency does establish things.

But foregrounding that argument was meant to explain why the phenomenal intuition is correct instead of being an illusion: the red does in fact exist, just as experience claims (as long as we drop from it the property of being red).

The disagreement with illusionism

I did have the feeling from a discussion between Kammerer and Brown that they argued about experience, as the stalemate didn’t seem to hinge on properties. I think Brown’s HOROR is a very similar account to mine, just unnecessarily higher-order (the higher-order vehicle is the experience, the first-order vehicle is the sensation) and it doesn’t specify which scientific theory should implement the explanatory features such as binding (e.g. whether it’s global broadcast).

Brown’s argument is that accepting illusionism is like concluding that everyone is a case of superblindsight. So Brown reads illusionism as denying the existence of experiences. I guess Kammerer denies phenomenality, not existence?

Does Kammerer deny that what phenomenality identifies as an experience in fact exists? Or does he only say that phenomenality misdescribes it?

1. Do deployments of phenomenal concepts track experiences? (2) His account could fail scientifically if there in fact are no phenomenal concepts.

Phenomenal concepts are not formally necessary for phenomenality. For example first-order theories do without them. So Kammerer’s positive account is to establish that phenomenal concepts exist, to explain consciousness using them, and only then to be an illusionist.

Brown’s challenge would be that Kammerer would first need to show that the other materialist accounts cannot succeed; otherwise what Kammerer establishes is simply an irreducible illusion of a hard problem.

I think the distinctions might be coming back to the property account: whether the property account is required for a scientific theory of phenomenal consciousness to count as a metaphysical theory of which illusionism claims there is no target.

The question is which camp I belong to, am I admitted to identity theorist or am I forced to illusionist.

Moral significance

“why should those states have distinctive moral importance once the phenomenal characterization is rejected?” Because they’re significant organizational achievements.

My view complicates things in also thinking personhood is a moral achievement independent of (quasi)phenomenality; both depend on mindedness.

Presence and recognition

“experiential occurrence. Its existence is supported by relations involving its vehicle” – no, the occurrence is the vehicle, since contents are abstracta, no?

Phenomenal intuition also correctly characterizes the vehicle as a presence: as existing and having a representational character (regardless of whether it’s successuful as representing)

That’s the spirit of realism in my account. A phenomenal concept can misfire: that will not bring about a (quasi)phenomenal occurrence. When the phenomenal concept fires veridically, it correctly describes the representational occurrence. That gives phenomenal experiences their infallibility as presence: metaphysically, they cannot fail to be presences.

They (according to Kammerer, but it sounds plausible) impose the systematically false characterization of quality as a property.

But maybe phenomenal concepts don’t impose instantiated phenomenal quality; maybe they impose intrinsic distinctness.

I meant the first thing; the second is (folk)theoretical.

But maybe they veridically present their actual distinctness. That would be the identity theorist who’s not a functionalist.

Yes, intrinsic is important to win some phenomenal points.

This is easier to see with shapes. The red circle does not instantiate circleness (it’s not a to-scale circle), but it’s intrinsically from a red square.

Symmetry and concrete constitution

Yes, this account is non-functionalist (non-computational in a narrow sense; perfectly fine to run on GPUs, but it’s not reducible to the computed congitive function).

This is stronger. Two identical vehicles could automorphically represent red and green in a computational cognitive system (cognitive is insufficient for minded), but could never be experiences of red and green. At least so would seem from can-two-minds-share-one-experience.md (This is very different from my views in 2025)

No, my point with automorphism was to make sure that the vehicles including their causal connections are identical; I don’t want to say that two pieces are identital based on what remains when detached from their causal connections.

Actually on re-reading I think you described it ok. I somehow misread you as placing my point in the re-assigning rather than in the symmetry/non-distinctness

Using the language of interpretation is risky, the computation can still be a real pattern in the causal nexus, receiving red-wavelength and green-wavelength inputs.

Sounds like I’m going out on a limb far enough to be a realist?

Positive organization and the remaining mistake

Does Kammerer, and illusionism more generally, have resources to distinguish pseudo-phenomenal occurrences from other ways that phenomenal consciousness ascriptions can fail? Can they say that LLM-based agentic AIs, whose guards have been lowered to let them claim conscious experience, nevertheless don’t have pseudo-phenomenal occurrences?

We can couch this in computational terms (but not used literally). Our positive account of phenomenality is that phenomenal occurrences make themselves available for addressing (attention) via their distinctness in a “Content-addressable memory” style.

The project here for realism is to minimize the gap between what the Kammerer describes the epistemic engine/gadget as claiming and what quasi-phenomenal occurrences indeed support. At some point, the gap becomes crossable by a pre-theoretic mistake (cognitive fallacy) without requiring an illusion.

Why do you highlight “from access through its character to exhaustive knowledge of its constitution” rather than “from access through its character to instantiating the reified distinctness”?

“Kammerer’s emphasis on revelation of essence” is that really more important than the mistake of phenomenal property? Or is it just the mechanism, the analysis of that mistake?

Yes Kammerer is helpful in understanding the mistake.

From properties to simplicity

Is my use of “reification” correct for what is needed here?

The mistake is not in “self-revealing” but in “simple”.

This also marks progress in my views because my language of slipping to property was sloppy. It’s rather slipping to simplicity.

But the “redness is an experience” article also argued against unconscious phenomenal experiences. The oxymoron points at a formally non-contradictory possibility, that of phenomenal properties not undergoing experience. I can maintain that argument because the constitution changes once the regulatory intervention ceases.

The three essays

The essay from this session makes use of the symmetrical realization thought experiment as further evidence of non-functionalism. This is an opportunity to dig deeper into genuine cognition without mentality. (On my metaphysical ladder: physics -> biology/computation/cognition -> mentality/psychology -> phenomenality (and separately personhood)

I meant to ask actually: Should that rather be a separate essay?

Can you write a first draft of the illusionism essay (based on this session) for the three essays to emerge into a coherent account? One of the interlocutors for the cognition essay will be the book Bacteria to AI: Human Futures with Our Nonhuman Symbionts N. Katherine Hayles

Causal constitution

“But the relevant occurrence includes its constitutive relations.” I’d add “causal” – “its constitutive relations (which are causally affecting the occurrence).”

Keeping the argument direct

“The word reification can obscure this refinement.” This is bad writing. If a word can obscure something, why not forgo bringing it up so that nothing is obscured? We can still use “reification” positively and refine – or drop it.

What the investigation can establish

“Calling it a fallacy rather than an illusion does not decide that issue by vocabulary.” This does not do the work you think it does. We have three options:

1. To drop it.
2. To replace it with: This investigation can show whether illusionism or realism has the upper hand.
3. To bring up the part of the Brown-Kammerer discussion where Kammerer’s view is in the new wave of illusionism relative to Dennett and company who thought of it as mistaken “folk-philosophical intuition”.

Anchoring the argument about unconscious qualities

In this paragraph: “Rejecting simplicity does not withdraw the argument against an unconscious instantiation of the very same phenomenal character.” – you are riffing off our session without anchoring the relevant background for the reader. This is confusing, in particular because accepting simplicity would not entail the argument. You need to point to the “redness in an experience” article.

Steering clear of “functionalism”

“with the qualification that the target of its nonfunctionalism is reduction to the computed cognitive function” is badly put, the intent was “with the qualification that the target of its nonfunctionalism is to oppose reduction to the computed cognitive function”

My thinking was calling Godfrey-Smith position “fine-grained functionalism” would motivate calling IIT the same too, which is a reductio; but a better path for me would be to steer clear of the term.

There’s no distinction between simulation and the real thing, as long as the causal graph is reproduced. The distinction is that input-output matching and cognitive efficacy don’t guarantee the relevant causal pattern.

Structure, grain, and substrate

Definitely not Russellian. The whole metaphysics is ontic structuralist. There’s a further question of what’s “metaphysically the same”, whether replacing one substrate by another would lead to any phenomenal difference. My answer is that the phenomenal character has finer grain than cognitive function demands, so function-preserving substrate replacement doesn’t preserve phenomenal character token-for-token, although plausibly preserves it type-for-type.

“how a character that makes no cognitive difference is known” – it is not known in the epistemic sense because it’s not represented. It is known in the phenomenal sense (it’s what makes experience special/unique/precious). Consider two color shades that are indistinguishable when separated, but distinct when sharing an edge. Their difference is not accessible when they’re separated (for report, but also more generally cognitively inaccessible).

Pereboom and Russellian monism

“introspection represents phenomenally conscious properties as having certain characteristic qualitative natures which these properties actually lack.” Hmm, in the essay I specifically name one such nature: simplicity. “The second response draws on the Russellian monist proposal that currently unknown fundamental intrinsic properties provide categorical bases for known physical properties and also yield an account of consciousness.” I disagree, physical theories are not at a relevant grain, see presence-and-actuality.md.

Weak illusionism, simplicity, and detachability

7. is also adjacent to Kammarer’s “verbal weak illusionism”. What’s Frankish argument for instability? Worth responding to, including a source reference. (8) I’m leaning simplicity is part of the uptake, but I never understood simplicity (which has roots in Plato). If possible we could dig deeper. (9) I’m open to moving the symmetry argument to the future Yudkowsky response essay. But the argument was doing work for this essay, by showing that some cognitive organizations make strong illusionism true. (10) Detachability was argued against in an old essay which was the same essay that introduced the against-property-framing argument. So I’m bringing it up here for two reasons: (a) in case someone happened to worry if we also take back that argument; (b) because detachability could illuminate intrinsic character. I’m actually don’t think that simplicity bolsters detachability, one can worry even more about detachability on any identity account.

Polishing distinctness

Let’s remove references to the missing article about the metaphysics of cognition / computation / biology. It doesn’t have a concrete shape nor ideas we haven’t yet communicated and might not happen.

What does this sentence say precisely? “It would be another mistake to conclude that, because the vehicles do not instantiate the represented shapes, there is no genuine experiential difference between them.” I would say the consequent more precisely: “there is no genuine content-tracking difference between them”. I think using “experiential” here begs the question, no? And appealing to arbitrary difference is not achieving anything?

How do we feel about this claim? An edge detecting neuron does not map to an edge via a spatial projection. But its dendritic field maps onto the geometry of edges.

But the vehicle is the neuron, not the receptive field. Receptive field is in the content space.

But the brain might be making things a bit simpler to us than the challenge I’d like to withstand. The organizational causal structure of the vehicle needs to be content-tracking: allowing explanation in terms of the geometry. The physical space location can be scrambled without affecting it. (But the demand has impact on physical spatiality, as physical proximity facilitates causal connection.)

Color, symmetry, and neural codes

Are we at risk of vindicating Kammerer, precisely where we went to diagnose our claim as scientific? How does color work in the brain?

I intended to ask whether we can claim lack of symmetry between distinct color vehicles, sufficiently extended.

You’re describing symmetry in the phenomenal space. I don’t see how your explanation applies to neuronal organization. Except indirectly through physicalism, but this missing move is precisely what we need to show, no? Not the phenomenal space organization.

Analog computation and the label

We can make a callback to “Presence and Actuality”, where we brought up Michał Ryszard Wójcik’s argument for the difference between analog computation and digital computation. This doesn’t do work for the essay, but I think MRW would appreciate it. The callback is that content tracking in the “lowest”-level vehicle is what distinguishes analog and digital computation.

“It may look like a fine-grained functionalism” -> “It can fairly be called a fine-grained functionalism” WDYT?

Type A versus type B

Type A versus type B: have we reduced the hard problem? Also, explain these sentences a bit to see if I’m just tired or if they can be put more clearly: “Conversely, correcting the illusion of simplicity need not make every experiential truth derivable a priori from a physical description. The account should determine the classification rather than being fitted to one in advance.”