Symposium on Boyd Millar’s “Naïve Realism and Illusion”

I’m pleased to introduce our second Ergo symposium, featuring Boyd Millar’s “Naïve Realism and Illusion” with commentaries by Craig French (Cambridge) and James Genone (Rutgers). Thanks to each of the participants for their excellent work and to John Schwenkler for helping me put the symposium together.

According to naïve realism, perceptual experience consists in a subject’s conscious acquaintance with ordinary objects and properties of the environment. According to its proponents, only naïve realism can accommodate introspective evidence about the phenomenal character of perception as well as the explanatory role experience plays in an adequate account of empirical knowledge.

According to Millar, though, we should reject naïve realism, because it cannot accommodate illusion. If you look at a circular object through a special shape-distorting lens, e.g., the object will look elliptical to you. This is an example of illusion: you see the object and some of its properties (e.g., colour, distance), but misperceive at least one of its properties (e.g., its shape). Millar proceeds by considering the different ways naïve realists have tried to explain illusions and showing why each way is unsustainable.

Millar begins by considering disjunctivism about illusion. Disjunctivism holds, roughly, that veridical and non-veridical experiences are not mental states of the same type. A naïve realist adopting disjunctivism about illusions proposes that, whereas veridical perception consists in acquaintance with how things are, illusions consist in something else altogether.

Millar’s argument against the disjunctivist proposal has two steps: first, he argues that naïve realists should allow that illusions at least partly consist in acquaintance with perceived objects and properties, since the same considerations that favour naïve realism for purely veridical perception also favour naïve realism for the veridical aspects of illusion; second, he argues that if illusion at least partly consists in acquaintance, then, given the interdependence of phenomenal features, it must wholly consist in acquaintance. Disjunctivism about illusion is therefore unavailable to naïve realists.

Next, Millar considers two non-disjunctivist naïve realist proposals about illusion. According to the first proposal, in an illusion you are prevented from perceiving properties you would normally perceive and judge that the object has some property which it lacks. Looking at the circular object through a shape-distorting lens, e.g., you are prevented from experiencing shape and mistakenly judge the object to be elliptical. Error enters here only with judgment, not with experience.

Millar objects to this proposal by again invoking phenomenal interdependence. If colour phenomenology depends on shape phenomenology (and vice versa), then you cannot experience an object’s colour while failing to have any experience of shape. Consequently, when you see the circular object through the shape-distorting lens, your experience must (contrary to the above proposal) possess some shape phenomenology.

The second non-disjunctivist proposal attempts to satisfy the above requirement by appealing to an object’s look, understood as a visible mind-independent property of the object. In illusion, the perceived object has a look which is typical of a kind to which the object doesn’t itself belong. Seen through a shape-distorting lens, e.g., circular things have a look characteristic of elliptical things.

Millar offers two reasons to reject this proposal. First, he suggests there is no plausible account of what looks are that would allow them to fulfill the various explanatory roles naïve realism demands of them. Second, even if such an account can be provided, it won’t be sufficient to explain illusions. When you look at a tilted penny, e.g., you experience a look typical of elliptical things. Because of perceptual constancy, though, you don’t misperceive the penny as being elliptical. So, non-paradigmatic looks cannot explain illusion.

You can find the target article, commentaries, and Millar’s response below.

Target article:

Boyd Millar “Naïve Realism and Illusion”


Craig French “The Problem of Illusion for Naive Realism”

James Genone – Comments on ‘Naïve Realism and Illusion’ by Boyd Millar

Reply to Commentaries:

Boyd Millar “Reply to French and Genone”



  1. Hi Boyd,

    Thanks for the comments. I am wondering about this claim you make:

    “If the experience’s colour phenomenology is constituted by representation rather than acquaintance, and the experience gets its shape phenomenology partly due to its colour phenomenology, then the experience gets its shape phenomenology partly due to representation rather than acquaintance.” (p. 2).

    Some queries:

    (1) What is the argument for this conditional claim? (It seems non-obvious to me and in need of an argument).

    (2) Why think that experience gets its shape phenomenology “partly due to” its colour phenomenology (and what does that mean)?

    (3) If thinking this through on behalf of a naive realist, why have it that the illusory colour phenomenology be constituted by representation? Could it not be that it is constituted merely by some negative epistemic property (as in Martin’s account of hallucination)? If so, then the suppose we follow what we have above through, we get: the experience gets its shape phenomenology partly due to [insert negative epistemic story]. Suppose we go that far. Will we be in a position to then say *rather than acquaintance*? Will whatever negative epistemic story we’ve told rule out acquaintance? If so, why?


    • Boyd Millar

      Hi, Craig. Thanks for the questions.

      Regarding (1): The general principle that the conditional is an instance of seemed sufficiently plausible to me that I was simply taking it for granted. If an argument is required, perhaps the best I could do would be to consider and rule out purported counterexamples. I’m not sure if there would be a more direct argument for the conditional.

      Regarding (2): What I have in mind here is that you’re only able to see an object’s shape because you can discriminate its colour from the colour(s) of the background, and the precise nature of the shape is determined in part by where the colour of that object ends and the colour(s) of the background begins.

      Regarding (3): I didn’t mean to suggest that representation was the only option; I thought it would be the best example for the purposes of illustration. I mention in a footnote (p. 613, n. 12) that A. D. Smith extends the argument to cover Martin’s variety of disjunctivism.

  2. James Genone

    Hi Boyd,

    A brief follow-up to one of your responses: if I am acquainted with an F in virtue of being acquainted with its look L, and I cannot tell look L from look L* that Gs instantiate in these conditions, and I’ve only ever been exposed to Gs and never Fs, is that not a (defeasible) reason to judge that the object is a G?

    One other note: on my view, a circular object cannot instantiate a look that is characteristically instantiated by elliptical objects (as you state on p. 7 of your responses). It *can* instantiate a look that a perceiver _fails to distinguish_ from the look of an elliptical object, but that is a different matter.


    • Boyd Millar

      Thanks for the question, James.

      Assuming that there is some account of what looks are such that you can be acquainted with L and not be able to distinguish it from L*, and assuming that Gs typically instantiate L*, and assuming that you’ve encountered many Gs but never any Fs, it seems plausible to me to suppose you’d have a defeasible reason to judge that the object you perceive is a G. However, in the scenario you describe, because you are acquainted with an F, that reason is defeated (so you shouldn’t judge that the object is a G).

      My apologies for the confusion at the end of p.7 of my reply. Your variety of the “looks approach” is quite different from some of the others that I discuss in the paper, and I need to be more careful not to conflate them.

  3. Louise Moody

    Hello Boyd,

    Here is, an admittedly counter-intuitive, suggestion:

    Suppose, first, that we allow pure visibilia into the class of mind-independent things with which we can be immediately acquainted; that is, objects of sight that cannot be perceived by any other sense-modality – things such as rainbows, holograms, and the sky.

    Now, the Naive Realist might say this: the object that is seen through a distorting lens (in this case, the elliptical-thing) is a worldly item that is instantiated by molecules of air that refract and bend the light ways that are travelling through the lens in such a way so as to produce the elliptical-thing that is seen (in something like the way the colours of the rainbow are instantiated by an aggregate of raindrops). So it is not that one veridically sees the colour and misperceives the shape; one just sees a ‘new’ mind-independent thing, as it were.

    This could work in other cases, such as shapes that look inverted when placed behind a glass (consider the ‘Reversing Arrow’ illusion). In this scenario, you don’t misleadingly see the original shape; but you do see a pure visibilium that is instantiated by a group of water drops.

    My general point is that talk of misperceiving something through lenses and other distorting mediums is mistaken since, at least if we accept pure visibilia into our ontology, the subject is directly acquainted with a type of worldly object that is made visually detectable in *that* particular set-up (e.g. the position of the object relative to the distorting lens).

    Granted, this might all sound very weird; but it seems to me that the Naive Realist has options (I’m trying to work the above suggestion up into a paper, but we’ll see …).

    • Boyd Millar

      Thanks, Louise; that’s really interesting.

      The first point I’d make is that it’s one thing to suggest that we can see things such as rainbows, the sky, and shadows, and another thing to suggest that we can see the light refracted by a lens.

      Since it’s the lens that refracts the light in the “distorting lens” example, I take it that the elliptical thing that one sees in this case would be the pattern of light in the lens (or something like that). And I take it your suggestion is that in this scenario, one sees this pattern of light and not the circular object on the far side of the lens. A couple of difficulties with this proposal occur to me: First, presumably you would misperceive the light-pattern to be located where the circular object is located in this scenario (so rather than providing an account of the illusion, the proposal would replace a shape illusion with a depth illusion); Second, many people wear glasses, and the cornea and lens in your eye also refract light, so if the refraction of light generates a light-pattern that blocks the perceiver’s visual access to ordinary objects, then the result would be that none of us are ever visually acquainted with ordinary objects (and besides being prima facie implausible, I would assume that such a proposal would undermine the motivation for naïve realism). But perhaps you have ways of responding to these concerns in your paper.

  4. Louise Moody

    Hello, and thanks for taking the time to reply,

    Just to follow up:

    “it’s one thing to suggest that we can see things such as rainbows, the sky, and shadows, and another thing to suggest that we can see the light refracted by a lens.”

    Perhaps I am misunderstanding, but surely, a rainbow just *is* visible light that is refracted by a lens (raindrops functioning as miniature prisms)? I’m not sure I understand the salient difference here.

    I agree that there might be a ‘location’ problem for pure visibilia. However, a first-pass response is this: the elliptical-thing that is seen is just where it appears to be. To me, that looks to be somewhere behind the distorting lens, and not a pattern of light *in* it.

    Your last objection seems to be the most serious. I’m going to need to think about this very carefully. An initial response is that it makes sense to introduce pure visibilia in the strong-distorting-lens case owing to the fact that the way the light is refracted visually blocks us from seeing the circular-shape, whereas spectacles have the effect of enabling us to perceptually ‘latch’ onto *more* properties. The medium through which things are seen is different in each case (I suppose the question whether it is a significant enough of a difference to save the Naive Realist).

    Certainly, I’m in agreement with you that the Naive Realist needs to treat such cases as veridical experiences, otherwise she is sunk (them being so common). And if you want a treatment that does not separate shape/colour phenomenology, then a pure visibilium seems a strong candidate. You’ve given me a fair bit to think about, thanks!

    Thanks for your comments – I’ll think about them some more, and try to give a more worked out response.

    • Boyd Millar

      Thanks again, Louise.

      The first point I was trying to make was that we have independent reasons for thinking that we can see rainbows but no such independent reasons for thinking we can see the light-pattern objects you have in mind. For instance, there is a word for rainbows in English and we talk about them on occasion. There is also introspective evidence: when I reflect on what my experience is like it seems to me that there are bands of colour in the sky. Conversely, there is no word for the light-pattern objects at issue and we never talk about them (with the exception of this very conversation). And there is no introspective evidence: when I have a visual experience caused by the circular object on the other side of a lens, it seems to me that I can see a solid, elliptical object occupying a specific location in front of me.

      Regarding the location issue, I might be misunderstanding you, but I don’t see how the elliptical light-pattern can be located on the far side of the lens, since there isn’t an elliptical pattern of light until the light reflected by the circular object gets to the lens.


Comments are closed.