In the scenario you describe, I agree that the eyes count as part of direct perception. And more generally, I’m happy to grant this: any system—organic or artificial—can count as constitutive of direct perception if it fixes the space of perceptual normativity for the subject at the time. What matters is not what the system is, nor which came first, but whether it defines what perceptual correctness even amounts to for that subject.

So if a subject initially has only the visor, then perception is direct relative to the visor. If the subject later acquires eyes that bypass the visor, then the eyes now constitute the perceptual capacity instead. In neither case is the constitutive system assessable as misrepresenting, because there is no independent perceptual standard against which its outputs could be evaluated. Altering or removing it would undermine perception itself, not merely change its outputs.

The contrast with the original visor cases is that there the visor operated against an already-defined perceptual capacity. That is what made misrepresentation intelligible. When a system intervenes on a perceptual capacity whose identity conditions are already fixed, its outputs become assessable as correct or incorrect relative to that background. That’s the sense in which the visor was instrumental in those cases and constitutive in the one you’ve now described.

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

At this stage, I think it’s worth noting that once the distinction is understood in these role-relative, normative terms, it’s inevitable that there will be borderline and hybrid cases. I don’t see that as a defect in the view; it’s exactly what one should expect if perceptual normativity is not reducible to causal structure alone.

What the edge cases are really testing is not whether a particular gadget counts as “direct,” but whether there is any principled distinction between systems that establish perceptual standards and systems that merely operate within them. If one thinks that any lawful causal mapping to neural states fixes intentional content, then my distinction will inevitably look arbitrary. If one thinks that perceptual capacities fix standards of correctness that other systems can intervene on, then the distinction is principled even if it resists sharp boundaries.

So I’m happy to keep discussing cases, but I think we’re now very close to a foundational disagreement about normativity versus causal covariance—one that further edge cases will merely illustrate rather than resolve.

So if a subject initially has only the visor, then perception is direct relative to the visor. If the subject later acquires eyes that bypass the visor, then the eyes now constitute the perceptual capacity instead. — Esse Quam Videri

Then referring back to this post, do you now accept that (1) is direct perception if and only if (4) is direct perception? Because to rephrase the quote above: if a subject initially has only eyes, then perception is direct relative to the eyes, but if the subject later acquires a visor that bypass the eyes then the visor now constitutes the perceptual capacity instead.

I think the “if and only if” formulation still overgeneralizes, and the reason is that it abstracts away from the role a system is actually playing at a given time.

What I’ve been claiming is this: at any given time, whatever system is fixing the space of perceptual normativity for the subject constitutes direct perception relative to that system. That does not imply a biconditional across all possible rewiring histories.

So yes:
– If a subject initially has only eyes, perception is direct relative to the eyes.
– If a subject initially has only a visor, perception is direct relative to the visor.

But it does not follow that (2) is direct perception if and only if (4) is direct perception in the original cases you described. In (4) as originally formulated, the visor was introduced as an intervention on an already-functioning perceptual capacity, not as the system that fixed perceptual normativity for the subject. That is exactly what made it instrumental rather than constitutive in that case.

Your rephrasing changes the scenario in a substantive way. If the visor genuinely bypasses the eyes and wholly replaces them as the system that fixes perceptual correctness for the subject, then yes, in that revised scenario, perception would be direct relative to the visor. But that is no longer the original (4). It is a different case with a different role-assignment.

So the disagreement is not about whether eyes and visors can ever be on a par in principle—they can. It’s about whether, in a given setup, a system is functioning as the constitutive basis of perception or as an intervention on one. Once that distinction is kept fixed, the “if and only if” claim does not go through.

More generally, this is why I’ve been resisting the idea that directness can be decided by gadget-swapping alone. Directness is not a property of devices considered in isolation, but of the normative role they occupy in the subject’s perceptual economy at a time. Changing that role changes the verdict; holding it fixed does not.

If the visor genuinely bypasses the eyes and wholly replaces them as the system that fixes perceptual correctness for the subject, then yes, in that revised scenario, perception would be direct relative to the visor. But that is no longer the original (4). — Esse Quam Videri

It is the original (4)?

4. The strawberry reflects 700nm light into a visor and the visor bypasses John's eye to stimulate his B neuron, causing him to "see blue".

Compare with:

5. The strawberry reflects 700nm light into John's eye and his eye bypasses the visor to stimulate his A neuron, causing him to "see red".

(5) is direct perception if and only if (4) is direct perception, and above you argued that (5) is direct perception.

This still seems like special pleading. You're arguing... — Michael

I think there’s a misunderstanding here about what I’m actually committed to, and it’s generating the appearance of an asymmetry that I don’t accept.

I’m not committed to the pattern you attribute to me. On my view, the symmetry holds at the level of replacement. In particular:

(1) Starting with only a visor → direct perception relative to the visor.
(2) Starting with only eyes → direct perception relative to the eyes.
(3) Replacing a visor with eyes, where the eyes now fix perceptual normativity → direct perception relative to the eyes.
(4) Replacing eyes with a visor, where the visor now fixes perceptual normativity → direct perception relative to the visor.

So there’s no device-based asymmetry here at all. Eyes and visors are on a par in principle. I've not been denying this.

What I’ve been denying is a different claim, which your formulation runs together with (4):

(4*) Adding a visor that intervenes on an already-functioning eye-based perceptual capacity—without replacing it as the system that fixes perceptual normativity → indirect perception relative to the visor.

The crucial distinction is not eyes vs visor, but replacement vs intervention. A system counts as constitutive of direct perception only when it defines what counts as perceptual correctness for the subject at that time. A system that operates against the background of an already-defined perceptual capacity is instrumental, and its outputs are intelligibly assessable as succeeding or failing relative to that background.

This applies symmetrically. If someone initially perceived only via a visor, and eyes were later added in a way that merely intervened on that visor-based capacity, then perception would remain direct relative to the visor and indirect relative to the eyes. There’s no privileging of biology here.

What’s been doing the work in our disagreement is that your original case (4) was underspecified between replacement and intervention, and I think that you and I have been reading it differently. Once that ambiguity is resolved, the alleged inconsistency disappears. The view is role-relative, not device-relative, and it treats eyes and visors in exactly the same way.

What determines whether or not (4) counts as replacement or as intervention?

If it helps, as it's pertinent to real life, all scenarios are fixed at birth.

1. Only eyes with eyes into brain
2. Only visor with visor into brain
3. Both eyes and visor, with visor bypassing eyes into brain
4. Both eyes and visor, with eyes bypassing visor into brain
5. Both eyes and visor, with visor into eyes into brain
6. Both eyes and visor, with eyes into visor into brain

Which count as direct perception and which count as indirect perception?

The criterion is normative, not causal: a system is constitutive if it fixes what perceptual correctness means for the subject, instrumental if its outputs are assessable against an already-defined standard. Here “fixes” is not meant causally, but normatively: it determines what counts as seeing correctly rather than incorrectly for the subject. Wiring diagrams underdetermine this, which is why your cases (3)–(6) can't be resolved by causal structure alone—each could go either way depending on which system, if either, fixes the perceptual norm.

So 'fixed at birth' and 'bypasses' don't answer the question; what matters is whether there's an independent standard against which the system's outputs are intelligibly assessable. If you reject that distinction—if you think causal covariance exhausts perceptual normativity—then the question dissolves, but so does the notion that any system could misrepresent rather than merely malfunction.

I don't understand what you mean by saying that the standard is normative. If you just mean that my perception is direct if this is what I normally see in such a situation then the people born wearing visors with a screen on the inside have direct perception of the world beyond the visor, because that it is how they normally see the world. They ought no more assess the "veracity" of the visor's output than they ought if they didn't have eyes and the visors interfaced directly with their brain. This is where I think you're guilty of special pleading, or moving the goalpost. Either the former is direct perception or the latter isn't.

:meh: