Perhaps, but then by "perceptual belief" I mean "a belief that the world is as it appears". — Michael

Thanks for making the definition explicit — that helps a lot.

I don’t accept that a perceptual belief should be defined as “the belief that the world is as it appears.” That definition already builds in the very thesis at issue in the direct/indirect realism debate, namely that phenomenal appearance is the standard against which perceptual belief is assessed.

By perceptual belief I mean something more ordinary and less theory-laden: they are beliefs about objects and states-of-affairs that are formed in ordinary perceptual contexts (e.g. “there is a ship”, “the screen is emitting orange light”, “the umbrella is wet”). Whether the world is “as it appears” is a further philosophical question about how such beliefs relate to phenomenal character, not what makes them perceptual beliefs in the first place.

Once that distinction is in view, my position is straightforward: perceptual beliefs can be justified even if phenomenal character does not mirror the qualitative character of the world. Justification does not turn on the world being “as it appears,” but on norm-governed judgment answerable to how things actually are, with experience constraining inquiry rather than supplying a standard of adequacy.

If one defines perceptual belief by stipulation as belief that the world is as it appears, then of course my view won’t count as addressing that debate. But that would mean the disagreement is about how to frame the problem, not about whether perceptual knowledge is possible.

Thanks, this helps clarify where the disconnect is.

I think the issue is that your formulation of (1) already presupposes a particular conception of justification — namely, that perceptual beliefs are justified if and only if the world is “as it appears”, where phenomenal character is taken to mirror the phenomenal character of the world (as in naïve colour primitivism).

That is precisely the assumption I’m rejecting. I don’t think perceptual justification turns on whether phenomenal character is the phenomenal character of the world, either successfully (direct realism) or unsuccessfully (indirect realism).

So I don’t agree with the indirect realist that perceptual beliefs are unjustified, nor do I agree with the direct realist about why they are justified. On my view, perceptual beliefs are justified by norm-governed judgment answerable to how things are, with experience constraining inquiry but not serving as the justificatory ground.

In that sense, I’m not denying perceptual justification; I’m rejecting phenomenal appearance as the criterion of it. That’s why my position doesn’t line up cleanly with either side of the traditional direct/indirect realism divide as you’ve framed it.

Thanks — this is very helpful, because it makes the structure of your view explicit.

I think the disagreement now turns on a single point. You want stage-two contents (“I am seeing orange”) to be conceptual but not truth-apt, and yet to function in inference to stage-three judgments (“the screen is orange”). But inference requires propositional, truth-apt premises.

That leaves you with a dilemma:

• If “I am seeing orange” is truth-apt, then it is already a judgment and your staged model collapses.
• If it is not truth-apt, then it cannot function as a premise, and the claim that stage-three judgments are inferred from it does not follow.

This is why I’ve insisted that perceptual judgments are not inferred from sensory contents. Sensory experience constrains inquiry causally and motivationally, but it does not supply inferential premises.

When you say that “seeing orange represents an orange screen,” you are reintroducing representation at the sensory level — precisely the move I’m resisting. On my view, representation, truth, and epistemic authority belong at the level of judgment, not sensation.

So the issue isn’t whether the senses mediate our contact with the world — I agree they do — but whether that mediation is inferential and representational, or whether judgment is norm-governed and answerable to how things are without being derived from inner items. That is the point at which we diverge.

I think the recurring confusion here comes from a difference in what we take the core epistemological problem to be.

As you frame it, the dispute between direct and indirect realism concerns whether phenomenal character presents mind-independent properties, and whether skepticism follows if it does not. Within that framework, I agree that rejecting phenomenal presentation pushes one toward indirect realism.

My claim is that both traditional direct and indirect realism share a deeper assumption: that phenomenal character is epistemically primary, and that justification for beliefs about the world must flow from experience outward (whether successfully or unsuccessfully).

I reject that assumption. On my view, phenomenal character neither succeeds nor fails at justifying knowledge of the world; it is not the kind of thing that plays that role at all. Epistemic justification belongs to judgment governed by norms of relevance, sufficiency, and answerability to how things are, not to phenomenal character.

That’s why my position doesn’t fit cleanly into the traditional direct/indirect realism framework. I’m not trying to resolve that dispute on its own terms; I’m questioning whether it’s framed at the right epistemological level to begin with. So the divergence is genuine, but it's aimed at a shared underlying assumption rather than at one side or the other within the traditional framing.

I agree that propositions like “I am experiencing such-and-such phenomenal character” are truth-apt and can function as premises in reflective reasoning. My claim was never that such inferences are impossible.

What I deny is that ordinary perceptual judgments are epistemically justified by inference from such introspective premises. In normal perception, John does not first judge “I am experiencing orange” and then infer “the screen is orange”; he simply judges that the screen is orange. Introspective propositions typically arise later, and for special purposes (disagreement, error-checking, theory-building).

Your reconstruction helps explain how John and Jane might reason about their perceptual differences, but it does not show that perceptual judgment itself is grounded in inference from introspective awareness. Those introspective premises are reflective and ad hoc with respect to perception, not epistemically basic.

So while P2 can play an explanatory role in some contexts, it does not follow that perceptual knowledge of the world is generally inferred from it.

I agree that background knowledge plays an essential role, and I also agree that differences in phenomenal experience help explain why different hypotheses are entertained or dismissed. Where I still disagree is in treating phenomenal character itself as an inferential input.

Even given background knowledge, phenomenal character is not truth-apt and cannot function as a premise. It does not count for or against a hypothesis in the way evidence does. Rather, it causally constrains inquiry by making certain hypotheses intelligible or salient and others not.

The inferential work is done entirely at the level of judgment, under norms of relevance and sufficiency, drawing on background knowledge and further perceptual judgments. Phenomenal character helps explain why those judgments arise, but it does not justify them or serve as a premise from which they are inferred.

So I don’t deny that phenomenal experience matters in inquiry; I deny that it plays the epistemic role you’re assigning to it.

I think part of what’s driving the disagreement here is that two different notions of “same size as” are in play, and they come apart precisely in the infinite case.

In everyday contexts, “same size” usually means something like this: if you subtract one collection from another and anything is left over, then they are not the same size. That notion is closely tied to finite counting, monotonicity, and the idea that proper subsets must be smaller than wholes. By that standard, it’s perfectly reasonable to say (for instance) that the natural numbers and the integers are not the same size.

What @Banno is appealing to, though, is a different notion that mathematicians use when working with infinite sets: sameness of size defined in terms of one-to-one correspondence. On that definition, “same size” no longer tracks what’s left over after subtraction, but whether elements can be paired without remainder. This isn’t meant to preserve ordinary quantitative intuitions; it’s meant to give a notion of comparability that still works once subtraction and counting break down.

So I don’t think the disagreement here has to be read as one side being confused or irrational. It looks more like a clash between two legitimate concepts that happen to share the same words. The intuitive notion works well for finite collections but doesn’t generalize cleanly; the mathematical notion is explicitly engineered to handle infinite cases, even at the cost of violating everyday expectations.

Once that distinction is on the table, the question isn’t really “who is right,” but what we want the concept of “same size” to do in this context. Mathematics answers that one way; ordinary language answers it another.

Thanks for laying this out so carefully — this helps clarify exactly where we disagree.

I want to focus on the role you assign to your stage two, since that’s where the inferential claim is doing all of its work.

You characterize stage two (“I am seeing orange”) as conceptual but not truth-apt, and therefore not a judgment. I’m happy to grant that description for the sake of argument. But then I don’t see how stage two can function as the basis of an inference to stage three. Inference requires premises that are truth-apt — something that can be correct or incorrect. If stage two is not truth-apt, it cannot play that role.

So either stage two is truth-apt, in which case it already is a judgment and your staged model collapses, or it is not truth-apt, in which case the claim that stage-three judgments are inferred from it does not follow.

More generally, I don’t deny that all inquiry is mediated by the senses. What I deny is that mediation entails inferential grounding. Sensory experience supplies data that constrains inquiry, but it does not supply premises from which judgments about the world are inferred. The epistemic work is done at the level of judgment itself, not by moving outward from inner representations.

For that reason, rejecting inferential mediation does not amount to Idealism. It does not deny a mind-external world, nor does it deny sensory mediation. It denies only the empiricist assumption that knowledge of the world must be constructed by inference from sensory contents.

So the disagreement isn’t about whether the “bridge of the senses” must be crossed — it’s about what crossing that bridge amounts to: inferential reconstruction from inner items, or norm-governed judgment constrained by experience but not inferentially derived from it.

Between the mind and any external world are the five senses. The mind only knows what passes through these five senses. Therefore, for the Indirect Realist, anything we think we know about any external world comes indirectly from “inference to the best explanation”. However, for the Direct Realist, we are able to transcend these five senses and directly know about any external world.

One question for believers in SDR is how they explain their judgements are able to transcend their phenomenal experiences — RussellA

I can only speak for myself on this, but I do not reject the idea that knowledge is mediated by the senses. What I reject is the idea that sensory content forms an epistemic base from which the rest of our knowledge about the world is inferred.

The key issue here is that sensation is not a normative act. This means it is not conceptual and is not truth-apt – it is simply not the kind of thing from which the rest of our knowledge could be inferred.

By contrast, judgment is conceptual and truth-apt. The act of judgment is part of the norm-governed process of inquiry. So, while judgments are constrained by sensory content, they are not inferred from sensory content. As we argued above, this would be impossible.

When we make perceptual judgments we are not making judgments about sensory content. We are making judgments about things in the world (“there is a ship”). That’s not to say that we can’t make judgments about sensory content – we can (“I am seeing red”) – but this is not what we ordinarily mean by the word “perception”. Instead, this is a reflexive, second-order kind of judgment more commonly referred to as “introspection”.

I would argue that perceptual judgments are neither inferred from nor justified by introspective judgments. If someone questions one of our perceptual judgments, we don’t try to justify it by appealing to introspective judgments. Instead, we appeal to background knowledge and other perceptual judgments.

Consider the example of John and Jane that provided. Jane makes a perceptual judgment (“the screen is orange”) and infers that the wavelength of the light is between 590nm and 620nm. Appealing to an introspective judgment (“I am seeing orange”) in order to justify her perceptual judgment simply won’t convince anyone, including herself. If she really wants to justify her judgment that the screen is orange, she’ll need to appeal to her background knowledge (optics, screens, color-blindness, etc.) and further perceptual judgments about her environment (current lighting, viewing angle, screen filters, etc.).

So when you ask how judgments are able to “transcend” phenomenal experience, my answer is that they can’t – at least not in the sense of bypassing or overriding the senses, nor by inference from inner representations. Rather, judgments were never justified by phenomenal experience to begin with. Sensory experience constrsins inquiry by supplying data, but epistemic authority belongs to judgment, which is governed by norms of sufficiency, relevance, and answerability to how things actually are. Once that distinction is in view, the need to “bridge” phenomenal experience via IBE largely dissolves.

I'm not so sure about this. While God is not seen as an efficient cause of entities, it is seen as their final cause, IIRC. Given this, I'm not sure how you can safely say that their intelligibility isn't rooted in the Unmoved Mover according to him. — boundless

That’s a fair point — you’re right that, for Aristotle, the Unmoved Mover functions as a final cause of cosmic motion, even though it is not an efficient cause. I don’t mean to deny that.

My claim is narrower: even granting divine final causality, Aristotle does not treat the Unmoved Mover as the source of the intelligible form of natural substances. Final causality explains why motion is directed toward an end; form explains what a thing is, and it is form that grounds intelligibility. On Aristotle’s own terms, those forms are intrinsic to substances rather than conferred by divine cognition.

So while Aristotle certainly affirms a divine intellect, the intelligibility of nature, as he understands it, does not consist in being thought by God, but in being formally structured in its own right. And on that much, I think we agree that intelligibility alone doesn’t prove the existence of God.

Note that Aristotle himself, however, endorsed the idea that a Divine Mind exists. I know that one can make an Aristotelian model without reference to such an Intellect, but it nevertheless is interesting that apparently Aristotle himself thought that the two ideas are connected. — boundless

That’s true, although it’s worth noting that Aristotle’s unmoved mover does not function as a source of the world’s intelligibility. As νοήσεως νόησις, it thinks only itself, and does not impose form or order on the cosmos. For Aristotle, the intelligibility of nature is intrinsic to substances themselves rather than conferred by a divine intellect contemplating or structuring the world.

Excellent job on the OP, as usual.

I think your critique of the “pre-history” objection is largely successful. In particular, I agree that appeals to cosmology often assume, without argument, that temporal succession is simply given as a fully determinate framework, independently of the conditions under which “before” and “after” have any sense. Your insistence that physics presupposes, rather than explains, temporal passage seems exactly right.

That said, I wonder whether the antinomy you describe really forces us to treat temporal succession as dependent on an actual standpoint or observer. There may be a middle position here, one that avoids both brute temporal realism and observer-dependence.

In a broadly Aristotelian tradition, the world is understood to be intrinsically intelligible. That is, it need not be thought of as intelligible because it is taken up by a mind; rather, minds are possible because the world is already ordered and determinate. On that view, structure and sequence are not imposed by understanding, but are what make understanding possible in the first place. This does not reduce order to mere physics, but neither does it make order depend on experience.

If something like this is right, then it seems important to distinguish physical change, lived temporality, and temporal order as such. Your argument shows convincingly that lived duration - in the Bergsonian sense - cannot be reduced to physical change, and that clocks and equations do not by themselves yield passage or continuity. But I would argue that it does not follow that temporal order itself requires an experienced point of view in order to be real.

One might say instead that the world prior to observers was not timeless, but unexperienced. The sequence of events was ordered and determinate, even though that order was not taken up or reflected upon by any subject. What emerges with consciousness is not temporal order itself, but the explicit presence of that order as order.

Framed this way, the tension you identify remains genuine, but it may not mark a final antinomy. Scientific accounts of a long pre-history and phenomenological accounts of temporality would then be addressing different aspects of the same reality: one describing ordered succession, the other describing how that succession comes to be experienced as passage.

But direct (naive) and indirect realism, as traditionally understood, are concerned with what sorts of things are phenomenally present to the mind (and the epistemological implications). — Michael

Yes, and I’ve acknowledged that. I’ve also acknowledged that my own view does not count as traditional naïve realism. My point is that it does not count as traditional indirect realism either.

…newer brands of “direct realism” … have fabricated a dispute with indirect realists that isn’t really there… — Michael

I agree that these other views don’t adhere to the traditional rendering, but that doesn’t mean their disputes with indirect realism are fabricated. Rather, they reject a set of core assumptions that have often driven both direct and indirect realism as traditionally understood, such as:

(1) Sensory content is the direct object of perception
(2) Sensory content misrepresents the world
(3) Knowledge of the world is inferred from sensory content
(4) (Therefore) our knowledge of the world is deeply uncertain

These are all claims that have come up in this discussion, and my responses have simply been directed at those assumptions as they’ve arisen.

I see what you mean, but I’d suggest that the position you describe differs from indirect realism as it was classically articulated by Locke, Hume, and later sense-datum theorists. Traditionally, sense contents were taken to be mental items that represented the world in some way, and knowledge of the external world was understood to be inferred from them in an epistemically basic way.

Your view seems to reject the representational aspect while still treating experience as epistemically primary, whereas I would want to reject both.

Best response I have read yet. — unimportant

That's interesting. I was thinking along opposite lines. While I agree that it gets many things right, I think it has some substantial problems. Here are some issues I see with 's post:

(1) Reduction is not explanation
To say “survival instinct explains why we go on living” is not the same as saying “survival instinct exhausts the reasons for living”. This collapses reasons into causes. While it may be that I am causally disposed to avoid death, it doesn't follow that my normative reasons for living are illusions. Calling reasons "cope" merely assumes, rather than argues, that reflective endorsement adds nothing over brute causality.

(2) Conscious choice cuts both ways
Niki wants consciousness to do two incompatible jobs. On the one hand, consciousness is dismissed as a post-hoc rationalizer ("cope"). On the other hand, consciousness is elevated as the faculty that allows us to override biology and choose death. You can't have it both ways. If reflective agency is real enough to negate survival instinct, then it is also real enough to generate reasons, commitments, and meanings that are not reducible to instinct.

(3) "Nothing matters cosmically" is a category mistake
This sounds deep, but it trades on equivocation. The absence of cosmic meaning does not entail arbitrariness unless one assumes - without argument - that value must be cosmic to be real. Sure, nothing matters to galaxies or black holes, but meaning was never supposed to matter at that scale in the first place. Meaning is agent-relative, but not arbitrary, and the demand for cosmic endorsement is a pseudo-standard. Meaning lives at the level of agents, practices and commitments - precisely where humans actually exist.

(4) The "ants" analogy fails
The analogy erases the very feature that is doing the argumentative work: self-interpretation. Ants do not ask whether their lives are worth living or frame their suffering as tragic or unjust. The moment suicide becomes a question, the ant analogy collapses. As Camus saw clearly, the absurd does not eliminate meaning, it forces the question of meaning into explicit consciousness.

(5) Perfomative nihilism is self-undermining
Niki's post is itself not neutral. It frames all positive valuation as illusion, but exempts its own evaluative stance from that diagnosis. Calling everything "cope" functions no less as a coping strategy, one that protects the speaker from vulnerability, disappointment, attachment and loss. This is not a moral criticism but a philosophical one. The stance tries to cut a "view-from-nowhere" that human agents cannot actually inhabit.

Yes, that’s broadly how I see it. Phenomenal experience is particular and non-conceptual, and for that reason it isn’t the kind of thing that can represent the world accurately or inaccurately. Representation, justification, and inference all belong at the conceptual level. Experience can condition and constrain judgment, but it doesn’t itself do representational or justificatory work.

Thanks, that’s helpful. I think where we still differ is that the argument you quote builds in a phenomenological notion of “direct presence” from the outset. On the view I’m defending, epistemic directness is not a matter of what is phenomenally present to the mind at all.

I don’t accept (1), but not because I think mind-external objects are phenomenally present. Rather, I reject the assumption that perceptual justification must be grounded in phenomenology in the first place. Directness, on my view, concerns what our judgments are about, not what appears in experience.

So I’m not accepting the Indirectness Principle; I’m rejecting the framing in which it is formulated. That’s why I don’t see my position as either naïve realism or indirect realism as characterized here.

Thanks for the clarification. I should just note that my earlier point wasn’t meant as a defense of indirect realism of any sort, nor as an argument that we never perceive external objects. The claim was narrower: given your own explanation of indistinguishability, an indirect realist can accept everything you say about mental imagery and hallucination while declining to posit direct perception in the good case on grounds of parsimony.

In other words, given your explanation of indistinguishability, mental imagery is sufficient to explain the phenomenal character in the hallucination case, and the presence of an external object makes no difference to that phenomenal character in the veridical case. That is what gives the indirect realist a foothold: they can accept everything you say while treating the external object as explanatorily superfluous with respect to phenomenal experience.

Yes, I’m familiar with Davidsonian triangulation, and I agree it’s crucial for explaining how content and interpretation get off the ground at all, so I wouldn't wish to deny its power or relevance. My worry isn’t that triangulation leaves the world out of the picture, but that it treats world-involvement as exhausted by the interpretive nexus itself. Triangulation explains how beliefs acquire content and stability, but it doesn’t fully explain how a belief that is perfectly triangulated, charitable, and norm-governed can nevertheless misrepresent how things are. When a paradigm collapses, is the old paradigm abandoned simply because we decided to chose new norms, or because reality itself made the old norms untenable? Triangulation explains error and correction in terms of re-stabilization of practice, but (in my opinion) fails to capture the full import of what it means to say "we were wrong about this". That’s the asymmetry I’m trying to preserve. In that sense, I see triangulation as presupposing answerability rather than replacing it — which is why I think it doesn’t yet settle the issue between us.

Then perhaps I haven't explained myself clearly, because indirect realism is the position that because perception of the world is not direct (i.e. its features do not manifest in phenomenal experience) the phenomenal character of experience doesn't justify our knowledge of the world, hence there being an epistemological problem of perception. — Michael

My understanding is that, traditionally, indirect realism has held that phenomenal experience (1) does not justify our knowledge because (2) it functions as an inaccurate representation of the world and (3) the rest of our knowledge is inferred from it. From this, (4) the problem of skepticism arises.

I accept (1) but reject (2), (3) and (4), so I wouldn't classify my view as indirect realism.

By contrast, my view is that phenomenal experience does not justify our knowledge because it does not function as a representation of the world at all. As a result, our knowledge is not inferred from it, and the skeptical problem you've described does not arise.

That's a nice reference. Perhaps we could say the debate here turns on how one interprets the role of the "favour of Nature" within regard to knowing.

That said, in light of some recent exchanges elsewhere, I suspect that @Banno and I may be closer in substance than it initially appeared, even if our vocabularies and points of emphasis differ. Terminology is always a challenge, and I’ve spent most of the last few years working in a more Continental register, which no doubt shows. I’m trying to correct for that in recent posts.

:up:

I’m glad to see that we agree on something. It makes sense to me that you would also reject any appeal to the phenomenal as epistemically foundational given how you've argued in other threads.

I also take your point about “judgment” lining up closely with what you call intent. In both cases, what matters is that we’re talking about norm-governed, world-directed acts rather than inner episodes. The Anselmian point about direction of fit seems especially apt here: judgment isn’t a matter of mirroring appearances but of committing oneself to how things are.

As you might have guessed, while I agree that our classificatory practices (“this counts as a ship”) are indispensable for discourse I’d also want to say that such judgments are not merely stipulative. But on the main point — that skepticism doesn’t arise from rejecting phenomenal foundations — it sounds like we’re very much on the same page.

Your proposal is something like, reality → judgment. I'm not proposing judgment→ reality, so much as judgment ↔︎ reality. This mutual dependence does not collapse misrepresentation into misuse — Banno

I agree that our practices are world-embedded and constrained by reality. My worry isn’t about whether reality plays a role, but about how. On your picture, reality constrains judgment only through the evolution of norms internal to practice. On mine, judgment is essentially answerable to how things are in a way that allows us to say that a practice-embedded, norm-governed belief nevertheless misrepresented reality. That’s the asymmetry I’m trying to preserve. If one accepts mutual dependence, that asymmetry disappears, and with it the distinction between being wrong about the world and merely revising our norms. At that point, the disagreement really is about whether truth outruns even world-embedded practice.

I think we can proceed by looking at the best theories we have of truth. And that's Tarski. We know that the semantic theory of truth is coherent. It also holds for any of the other more substantive theories. '"p" is true IFF p' is pretty much undeniable without a loss of coherence. — Banno

I don’t disagree about Tarski or Davidson. The semantic theory of truth is indispensable and coherent, but it answers a different question than the one I’m raising. Tarski tells us how “true” functions in a language, but he does not explain what it is for a judgment to misrepresent the world rather than merely fall out of favor within a practice. If I recall correctly, even Davidson explicitly acknowledged this gap: semantic theories of truth explain meaning, not epistemic success.

Rejecting a heavyweight correspondence theory doesn’t eliminate answerability; it only eliminates a particular metaphysical picture of it. My concern isn’t with “matching” reality as an external comparison, but with preserving the asymmetry that makes shared error intelligible. That judgments answer to how things are in a way that practices themselves do not exhaust. If that asymmetry is rejected, then error collapses into norm change. If it’s preserved, then truth outruns practice. That’s the divide I’ve been pointing to.

It is both the case that (a) the phenomenal character of experience is not truth-apt and the case that (b) we use the phenomenal character of experience to make inferences about the environment. (b) is exactly what John and Jane do in the example I gave; their assertions about the wavelength of light emitted by the screen are not made apropos of nothing — they derive their conclusion from the phenomenal character of their experience (coupled with their knowledge of the wavelengths of light that are usually responsible for such an experience). — Michael

Let me push this a little further. I would argue that not only is phenomenal experience not truth-apt, it is not even conceptually articulated. Raw phenomenal character — redness as-seen, loudness as-heard, sourness as-tasted — is not the kind of thing that can directly participate in inferential relations, whether to ground them or otherwise. Inference requires intelligible, conceptual content. Phenomenal qualities can only figure in inference once they have been conceptualized through an act of reflexive understanding, but the resulting conceptualization is not identical with the phenomenal qualities themselves.

Once phenomenal experience has been conceptualized, we can make judgments about it (e.g. “there is a red patch in my visual field right now”). These judgments can participate in inferential relations, but at no point do phenomenal qualities themselves participate in inference. So you are right that John’s and Jane’s assertions are not made apropos of nothing, but it is not correct to say that they derive their conclusions directly from the phenomenal character of their experience. Phenomenal experience can condition and constrain the formation of judgments in a causal and heuristic sense, and judgments about phenomenal experience can certainly play a justificatory or explanatory role within reasoning — but that is not the same thing as phenomenal experience itself functioning inferentially.

But at least with respect to colour (and other secondary qualities, à la Locke), the world just isn't this way. — Michael

I agree that color does not exist “out there” in the way the naïve realist insists, and in that sense I do not count myself among their number. I do consider myself a kind of direct realist, but only in the broader sense I’ve described in previous replies.

Any inference about the mind-independent nature of the world from these secondary qualities is open to scepticism. That's really all there is to indirect realism. — Michael

I don’t agree that this is all there is to indirect realism. Aside from the two required characteristics I pointed out in a previous post, I would also say your framing above seems to assume that our knowledge of the world is inferred from phenomenal character, as though phenomenal experience — or judgments about it — provides a justificatory foundation for all other knowledge. That is one of the assumptions I’m pushing back on. Phenomenal experience is not the kind of thing that can play a justificatory role, and even judgments about phenomenal experience are not epistemic bedrock.

Indirect realism, as you are presenting it, seems to depend on the idea that knowledge of the world is justified by first securing knowledge of phenomenal character and then inferring outward. Once that picture is abandoned — once experience is seen as conditioning inquiry rather than grounding justification — the skeptical pressure you associate with secondary qualities never arises.

I think your last reply makes the disagreement as clear as it’s going to get. On your view, certain judgments are constitutive of what is the case, norms and reality are mutually dependent, and truth is exhausted by what survives within practice. On that picture, error is ultimately a matter of deviance from shared criteria, not failure to answer to how things are independently of those criteria.

My resistance isn’t to circularity, constitutive cases, or practice-based explanation — I accept all of those. It’s to the consequence that misrepresentation reduces to misuse and that inquiry no longer answers to anything beyond its own norms. That’s a coherent position perhaps, but it’s one that reshapes the notions of truth and error in a way I ultimately can’t accept.

At that point, the disagreement isn’t about regress, limit cases, or examples like “EQV is reading.” It’s about whether truth is exhausted by practice or essentially involves answerability to how things are. I don’t really see a neutral way to adjudicate that any further.

I think your reply helps make the divergence between us very clear. You’re treating phenomenal character as something like an epistemic instrument - a reading from which we infer how the world is, much like a thermometer reading. To be fair, this is how many direct realists treat it as well. My point is that the thermometer analogy already builds in the representational role that I’m denying. On the view I’m defending, phenomenal character is not a “reading” at all. It is not truth-apt, not accurate or inaccurate, and not something whose reliability is assessed independently of judgment.

That’s why I don’t think the epistemic question is “can we trust that the world is as it appears?” Appearances don’t make claims, so they aren’t candidates for trust or distrust. Judgments make claims. Error and skepticism arise at the level of judgment, not at the level of experience as such.

For that reason, I don’t accept the biconditional tying direct realism to P2. My view doesn’t require that phenomenal character be explained by an object’s qualitative property manifesting itself in experience. What matters for my approach to realism is that the intelligible structures grasped in understanding and affirmed in judgment are the very structures instantiated in the world. Once that is in place, rejecting P2 doesn’t entail indirect realism - it entails the rejection of a particular kind of direct realism that is based on (what I consider to be) a faulty account of how experience secures knowledge.

In other words, I'm rejecting one of the key assumptions that the traditional dilemma is based on.

I agree with you that (1), (2), and (3) can all be true, and I also agree that in that context “red” and “orange” refer to phenomenal character rather than wavelength. Where I part company is with the claim that phenomenal character thereby functions as an epistemic intermediary.

On the view I’m defending, "phenomenal character" is not what John or Jane are making inferences about. Phenomenal character does not assert anything about wavelengths, nor does it justify any belief. What does the epistemic work is their background understanding of light, screens, and illumination conditions, together with a judgment about what is the case. Jane’s mistake is not located in her sensory experience - it is located in a false judgment about wavelength.

If phenomenal character were itself an epistemic intermediary, then error would have to be traced back to it as being inaccurate or misleading. But in your own example, nothing is wrong with the experience as such; what is wrong is the judgment made on its basis. That’s exactly why I resist treating phenomenal character as representational in the epistemic sense. It conditions inquiry, but it is not what our judgments are about.

The point I am making is that even if the environment has properties that resemble the properties that manifest in sensory experience (as naive colour primitivists would claim), and even if English grammar describes the interaction between the body and the environment as "seeing the environment", if there is such a thing as sensory content distinct from the environment then it's still indirect realism. — Michael

I see what you are saying, but I would argue that indirect realism has traditionally claimed something a bit narrower than that. I don't think indirect realism follows from the mere fact that "something" mediates the connection between mind and world. It seems to also require that this "something" has the following characteristics:

(1) It represents some aspect of the world
(2) It is itself the direct object of perception

In other words, this "something" needs to act as an epistemic intermediary rather than a merely causal intermediary, irrespective of how that epistemic role is theoretically cashed-out (e.g. representation, resemblance or something else).

For my part, I would deny both (1) and (2). Inherent to my denial of (1) is the denial that sensory qualities as-such ("redness", "sweetness", "loudness") represent or resemble features of the world. I don't think they need to. Instead, I would say that sensory qualities simply need to provide enough data for the intellect to grasp the structures, patterns, unities and dependencies that exist in the world. These are relational rather than qualitative, and the point is that the very same relations grasped by the intellect are instantiated in the world itself. That is my understanding of what it means for the mind to make direct contact with reality.

To push this a little farther, we could argue that this is what makes science possible. It enables us to accept that sensory qualities are not "out-there" in any naive sense while still maintaining that science has some theoretical purchase on the world.

You say: "What is judged to be the case and the presupposition are the very same."
I say: "The judgment and the fact that satisfies it are still distinguishable in kind."

You say: "Now how can one be in error about your reading this, here, now? There seems to be no such possibility."
I say: "Error is only impossible in this case because answerability is immediately fulfilled, not because it has disappeared."

You say: "Calling it a limit case is special pleading."
I say: "Calling it a limit case is correctly identifying it for what it is."

As explained in my previous reply, it is a limit case precisely because the answerability relation is immediately fulfilled - not eliminated. I explained how the representational gap widens again as you move away from the limit case. This is precisely how we would expect a limit case to function.

You say: "If you were right, truth would be normatively grounded in reality, not practice."
I say: "Norms of truth are constituted in practice as norms of answerability to reality."

Norms are made by practice, but are about getting things right - which is why practice can fail.

Practice and reality are not competitors here. They play different roles.

Thank you for the clarifications with regard to the existence of mental imagery. I think this will help us hone in on the core issue.

My view is that in the hallucination case I am perceiving mental imagery, whereas in the good case I am perceiving a mind-external object. I take it that our minds can copy good-case perceptual experiences and store these copies (and we call upon these copies in memory and imagination). And as these are copies, they - these mental images - can create in us an experience indistinguishable from perceiving the object they are depicting. — Clarendon

Thanks for confirming your view on these matters. You say that in cases of hallucination we perceive a mental image; in cases of veridical perception it is the external object itself that we perceive. So the difference between hallucination and veridical perception is the object of perception, not the perceptual relation itself. This makes sense.

Where I think you may still have an issue is in your treatment of indistinguishability. Your account requires that an external object and a mental image can lead to the same phenomenal experience. So what we have is:

Step 1: Hallucination case
Object of perception = mental image
Phenomenal experience = X

Step 2: Good case
Object of perception = mind-external object
Phenomenal experience = X

So on your own account, the same phenomenal experience (X) can be generated by both a mental image and an external object. The indirect realist will ask: "if hallucination achieves X via a mental image alone and veridical perception adds an external object without altering X in any way, then mental images are sufficient to explain X in all cases. If mental images are sufficient to explain X, what explanatory work is the external object doing in producing X?".

In other words, your account of direct realism does not rule out indirect realism or the "improper" forms of direct realism you were concerned to distinguish your view from, since the these others can leverage your own explanation of indistinguishability in support of their model.

Now, this doesn't show that your account is incoherent, nor does it show that you are forced to accept indirect realism. It only shows that the indirect realist can happily accept your account and, if they wish, eliminate direct perception in the veridical case on grounds of parsimony.

You might not want to describe the latter as "seeing a mental representation" but it would still be the case that sensory content is a mental representation, and I would say that that's all it takes for indirect realism to be true. — Michael

I don’t think it’s correct to say that sensory content is a mental representation. Representations, in the epistemic sense relevant here, have a normative valence - they can be accurate or inaccurate, correct or incorrect, better or worse. Sensory content does not function this way within cognition. It does not assert, refer, or purport to get things right. Representation, in that sense, is the job of judgment.

We then have an epistemological problem to address. If sensory content is a mental representation then can we trust that it is accurate, in the sense that the sensory content resembles the distal object. — Michael

You’re right that if sensory content were a mental representation, then we’d face the epistemological problem you describe. But if sensory content is not a representation in that sense, then it’s not the kind of thing that can be inaccurate or misleading to begin with. It is only it judgments that can be accurate or inaccurate. And while those judgments are conditioned and motivated by sensory contents, they are not about sensory contents, but about how the world is.

Then isn't a veridical experience the experience of imagery plus a true judgement? I believe Clarendon is just saying that the imagery (mental phenomena) that occurs when we hallucinate is indistinguishable from the imagery that occurs when we have veridical experiences. — Michael

Yes, and I agree. Where I would push back is on the idea that imagery itself is the object of judgment.

Take the example of hallucinating a ship. We have (at least) two acts on the part of the subject: sensation and judgment. While the judgment is dependent on the sensory content, it is not about the sensory content.

Furthermore, sensation is not truth-apt: it does not refer, assert, or commit. It does not make a claim about whether anything does or does not exist. That is what judgment does. Insofar as perception makes such a claim, I would say it includes a judgment of existence, and in that respect is distinct from mere sensation.

So it may be that the sensory content is the same in both veridical and non-veridical cases. What distinguishes them is the correctness of the judgment, which is determined by the way the world is, not by the sensory content. In the hallucinatory case, the judgment fails not because it is about a non-existent object, but because nothing in the world satisfies it.

I think it is a problem for @Clarendon specifically because the way that he leverages them in his model undermines his commitment to direct realism.

More generally, though, I agree with you - I don't think that we can sensibly reject the existence of mental images. But I do think we need to be careful about the epistemological and ontological roles we assign to them.

I can't claim to have this all sorted out, but I am wary of reifying mental images into objects of perception rather than treating them as features of experience that condition our judgments. Once images are treated as perceptual objects, they begin to play exactly the mediating role that indirect realists have historically relied on, which is a move that I am resistant to.

For that reason, I would tend to say that a hallucination is not the perception of an image, but the experience of imagery plus a false judgment.

I'll take a stab at this, since I think that @Richard B's critique is on the right track.

You've introduced "mental images" into your model in order to explain hallucination. This introduces an instability within your position that indirect realists have been capitalizing on for centuries in order to show that direct realism is untenable.

The problem is that you appear to be explaining indistinguishability in terms of identity within phenomenal experience (I.e. identical “appearing object”). This is ambiguous. If by "appearing object" you just mean an object within phenomenal experience - i.e. an object directly present to consciousness - then you've already collapsed into indirect realism since now the direct object of perception in both veridical and non-veridical experience is a phenomenal object.

If, on the other hand, the “appearing object” is not what is directly present to consciousness (i.e. objects within phenomenal experience), then they cannot secure identity within experience at all, and so cannot explain why hallucination and perception are indistinguishable as experiences. In that case the appeal to mental images does no explanatory work.

So, either you must give up on explaining indistinguishability in terms of the identity of "appearing objects" within experience, or you must give up on direct realism.

If we say "Rational inquiry", I think I can agree. — Philosophim

I think that’s a helpful way of putting it. If we restrict the discussion to rational inquiry - inquiry aimed at truth rather than persuasion, expression, or validation or anything else - then I agree that we’re talking about something governed by norms of rational correctness. That’s the sense of inquiry I’ve been trying to isolate.

I think your reply helps move the discussion forward, so thank you for spelling it out.

I don't think the “reading” case provides a counter-example to my claim that judgment presupposes answerability to how things are. On the contrary, even here the judgment is true because things are a certain way, and would be false if they were not. What the example shows is that there are cases where answerability is immediate - leaving no room for representational error - not that answerability is absent altogether. I take these as a limit-case, not a counter-example.

As soon as we move beyond such limit cases to claims about past events, theoretical entities, or explanations of why things are as they are, the distinction between misusing words and misrepresenting reality explicitly reasserts itself. In those cases, practice can fix criteria for correct application, but it does not itself make the judgment correct.

You could argue that all discourse can be inferentially grounded in limit-case claims, but since inference preserves only entailment rather than fulfillment of conditions, this wouldn't address worries about the regress of conditions.

My point is that inferential articulation does not exhaust the normativity of judgment. Inferential articulation explains how judgments are connected; it does not explain what it is for an inferentially licensed judgment to be wrong about the world.

That is why I resist the claim that practice can exhaust the notion of truth. The “reading” example merely shows what judgment looks like when answerability is transparently fulfilled, not that answerability can be replaced by practice. Inquiry in general presupposes that there is a way things are that judgments answer to or fail to answer to, and it is such failure that we mean by "error".

I agree with most of what you said here, especially at the level of psychology and development. People clearly differ in temperament, in how much they question their judgments, and in how they experience the relation between themselves and the world. And it’s also true that much of our engagement with the world is unreflective and action-oriented rather than constantly critical.

The distinction I am trying to draw is a little different. It’s not about the different ways that people can come to recognize the independence of reality, or the temperamental and development differences that lead them to engage with reality in different ways. It’s about what commitments are implicitly presupposed in the act of inquiry itself.

Consider the act of asking a question. It might seem at first that there’s not much to such an act, but I would argue that there is a lot that is implicit within it. For example, asking a question presupposes that there is something to ask about. It presupposes that we already know something about it, but also that we don’t yet know everything about it—otherwise there would be no point in asking.

In other words, there is a logic and a set of commitments that are implicitly presupposed in the act of asking a question. To say that these things are “presupposed” is to say that the act of asking a question would be incoherent without them; they are constitutive of what it means to ask a question. To say that they are “implicit” is to acknowledge the fact they generally remain out of conscious awareness while performing the act. It is only through philosophical reflection upon what it is we are doing when we ask a question that these presuppositions are made explicit.

So what I am arguing is that robust notions of truth, error and reality are implicitly presupposed within inquiry as norms governing correctness, and that these are not reducible to weaker notions such as endorsement, misuse or coherence without loss. When we engage in inquiry we are intrinsically oriented toward a reality that is determinate independently of our beliefs. If we weren’t, notions like truth, error and reality would lose their meaning and inquiry would become unrecognizable in comparison to what we actually do and say in practice.

If this is still unclear, no worries. I have really enjoyed our conversation. It has given me plenty to think about, and I hope it has for you as well.

I think you’re right about contradiction doing real work here, and I agree that it shows our judgments are not sovereign over reality. Where I’d want to press a little further is on what makes contradiction count as error in the first place.

My point isn’t just that we can be contradicted, but that when contradiction occurs, we take it to show that our judgment was wrong about how things are, not merely overridden by a new experience. That normative force doesn’t come from the contradiction itself, but from the fact that judgment already aims at a determinate way things are.

Put differently: contradiction doesn’t create objectivity; it reveals a failure relative to an objectivity that judgment already presupposes. Even in cases where no contradiction ever shows up, we still take our judgments to be answerable to how things really are, not merely to what has survived so far.

So I think we’re actually very close on this issue. I’d just want to say that the possibility of contradiction has its significance only because judgment is already oriented toward a reality that is determinate independently of our beliefs, not merely because we sometimes get corrected by experience.

So what I'm looking for in your response is "what judgment itself presupposes" so that a judgement can be true or false. — Banno

I’ve already addressed this point several times. Briefly: judgment presupposes answerability to how things are and the intelligibility of error. I don’t think restating this again in detail will move the discussion forward.

Is this concerning the branch of Agrippa’s trilemma that results in an infinite regress? — Banno

No. I’m not concerned with an infinite regress of reasons, but with an infinite regress of conditions—conditions of being right rather than reasons we happen to give.

So now you have two notions of truth... — Banno

I’m not introducing two notions of truth or explanation. What I’m saying is that if we deny that truth involves answerability to how things really are, or that explanation aims at getting things right about how things really are, then what remains is no longer recognizable as truth or explanation. Instead we are left with weaker surrogates (endorsement, acceptance, coherence) that cannot carry the same epistemological weight within inquiry.

Let's start with an example. I think that if I ask you if you are reading this post, here, and now, you would quite rightly judge that you are....So here, the condition and the judgement are the very same — Banno

I don’t think that follows. The condition is a state of affairs; the judgment is an act of affirming that state of affairs. They may coincide extensionally in this case, but they are not identical in kind.

I'm not suggesting that your judgement is based on some observation of yourself reading, but that what you are now doing counts as reading. — Banno

Practice determines what would count as being right, but it does not (and cannot) itself make a judgment right.

So someone who denies that you are reading isn't mistaken as to the facts, but as to the words we use to set them out. — Banno

That may be right for this particular case, but it doesn’t generalize. In many cases of disagreement—scientific, historical, or ordinary—we treat people as mistaken about how things are, not merely about how words are used. Reducing error to misuse doesn’t capture that distinction.

And here we have avoided the picture of "conditions all the way down". Our justification is this is just what we do. — Banno

Appealing to “what we do” avoids conditions all the way down only by treating truth as exhausted by acceptability within practice. But acceptability cannot do the same epistemological work as truth.

Notice also that it's not some "fact of the matter" that settles the discussion. — Banno

If no fact of the matter ever settles anything, then the distinction between misrepresentation and mere misuse disappears. That distinction is doing real work in inquiry and cannot simply be set aside.

The pattern here should be familiar. There's the intuition that there must be something firm - absolute, necessary, unconditional - upon which we build whatever it is we are building. — Banno

This mischaracterizes my position. There’s an important difference between (1) trying to build inquiry on an absolute foundation and (2) reflectively identifying what inquiry itself presupposes in order to function as inquiry.

Perhaps this should not surprise us, since we know that at least for the case of a simple formal system that is capable of doing counting, it might be consistent but it can never be complete... — Banno

I agree that we shouldn’t expect completeness, and I haven’t suggested otherwise. But invoking Gödel here actually cuts against a practice-exhaustive conception of truth rather than supporting it. Gödel’s result shows that (1) truth outruns formal derivability, (2) consistency does not collapse into completeness, and (3) there are truths that hold even though they cannot be proven within the system. Notably, this presupposes a notion of truth that is not exhausted by system-relative coherence.

------------------

I think we have reached the point in the discussion where further clarification is unlikely to be productive. Thank you for the interesting discussion.