Tuesday, January 19, 2010

A Better Statement of the Puzzle for Reid on Color Perception

Here is a much shorter, clearer statement of the puzzle I see for Reid on color perception:

1) A perceiver observing a uniformly blue sphere would only perceive something uniformly blue if one sees the sphere as a 3-dimensional object.
2) An original perceiver observing a uniformly blue sphere would not see the sphere as a 3-dimensional object
3) So, an original perceiver observing a uniformly blue sphere would see something variably colored, rather than uniformly blue.
4) No external object being perceived is variably colored.
5) So, either a) the original perception is of the sphere, but is not correct., or b) the original perception is correct, but not a perception of something external.
6) Reid's direct realist account of perception requires that the original perceptions are correct, so, not (5a).
7) Reid's direct realist account of perception requires that the original perceptions are of external objects, so, not (5b).

Reid is committed to (1)-(3) by the passage I quoted in the previous post.
I don't have a source for (4), but I am not sure what externally existing object is variably colored in a perceptual situation involving a perceiver and a uniformly blue sphere.
(5) follows because the perception is either correct (and therefore not of any external object) or not. If it is incorrect, it may as well be a perception of the sphere.
(6) comes from the veridicality of perception on Reid's picture (he goes to some lengths to argue that the sense do not deceive us), and (7) comes from the fact that Reid is insistent that the objects of perception are external objects (and their qualities).

So, this is the puzzle. Reid can avoid the puzzle when it comes to visible and linear distances, for instance, because there are two different (but related) qualities he can invoke. But there is no such distinction available when it comes to color.


Keith said...

What exactly do you mean here by 'original perceptions'? If it the sort of experience had by a child, for example, or non-acquired perceptions, then I'm not sure that Reid need be committed to 6 or 7. Isn't it only the sort of perceptions that a mature adult observer would experience - in this case acquired perceptions - that need to be veridical?

Also, I'm not sure that no suitable distinction can be found for colour as for other properties. There's nothing incompatible between the claim that the sphere appears uniformly blue (actual colour) whilst parts of it appear to be different shades of blue when viewed from here under the current lighting conditions (apparent colour, this being a perfectly objective relational property involving the object, observer, and lighting conditions, just as visual figure is, for Reid, a relational property and the 'immediate object' of visual perception).

Lewis Powell said...


First off, thanks for the questions/suggestions. I want to preface my reply by mentioning that I am probably a lot more open to revising my position on some of the things I am about to say than my tone suggests (I have a lot of "Reid's must say X about this"-type comments where it might be more appropriate to say something like "I thought Reid was committed to X, but I could be wrong", so please my responses accordingly.

A few things on what I think the "original"/"acquired" distinction is supposed to be tracked.

1) The distinction between original and acquired perceptions in Reid as I am using it here, approximates (I believe) the Berkelayan distinction between "proper" objects of perception (in a given modality) and that which is "suggested" to the mind by proper perceptions in light of experience. I don't know if this helps, or how strong the parallel is.

2) Reid thinks that each sense modality is originally constituted so as to provide perceptions of a certain set of qualities, so that, for instance, in a mind completely unexposed to experience, the qualities perceived by vision only include: angular distance, visible figure, color (there may be one that I am leaving off). Now, crucially, Reid is committed to these perceptions being veridical (he argues at length that there are "no fallacies of the senses").


Lewis Powell said...

And, it is easy to see how angular distance perception will be veridical on Reid's view, since it is a real (but relational) property of a pair of objects that they have such-and-such angular distance.
In fact, insofar as there may be error in perception, the best candidate for how this could be is in the case of acquired perception (since one might be exposed to a prejudicial set of experiences, generating faulty non-inferential perceptual judgments), but it is not clear that Reid would want to say that such cases are genuinely acquired perception, rather than that the individuals involved failed to achieve acquired perception.

On your second point, here are some (potential) disanalogies between a two-color-qualities approach and a two-distance/shape/etc.-qualities approach:
a) in distance/shape/etc., the acquired perceptions of one modality are the original perceptions of another, not so with a color/color approach.
b) in distance/shape/etc., the quality originally perceived by sight is uncontroversially mind independent, not so with a color/color approach,
c) Reid is explicit that angular distance is a distinct quality from linear distance, and does not say, for instance, simply that distance is originally perceived by vision.

So, consider a two-types-of-color-quality approach (which may be the move Reid _should_ have adopted, but I am not sure I see any significant textual support that he consciously embraced the view that there are two types of color qualities). We'll call them 2d and 3d color qualities for lack better terms:
First: what is the basis for our acquired perception of 3d color qualities: in the case of angular/linear distance, we get experiences generating an association between the original perceptions of original sense modalities underwriting the "improvement of the senses" as Reid terms it, but in the case of 2d/3d color qualities, it is hard to see what could possibly ground the new perceptions (unless, for instance, an object's possessing the 3d color quality is simply constituted by its possessing the right array of 2d color qualities).


Lewis Powell said...

Second: I can understand how angular distance is mind independent, insofar as smaller angular distance corresponds to narrowness of some stripe (it is harder to kick a ball between two poles from off to one side than from dead-on, this is in part a function of the angular distance between the two poles from the point where the ball is kicked). I have more difficulty understanding what real quality of an object its 2d color is relative to a point. This is because we also need to supply lighting conditions (fair enough), and an actual observer (or the constitution of the visual faculties of such an observer). I'm not sure if Reid's treatment of color as a secondary quality is getting in the way here, or if, rather, I am overlooking some way in which color being a secondary quality _helps_ Reid explain this.
Third: Reid doesn't ever talk about there being different types of color qualities. He simply lists color as an original perception of vision, and describes how our acquired perception of 3d shape effects whether we see a gradual variation of color, or a uniform color. Now this might be because both types of color qualities are perceived only* by sight or something like that, but at any rate, the way Reid describes the case makes it harder to read him as adopting such a distinction.

I think it might be the case that the right thing to say here is "ok, so Reid doesn't have that 2d/3d view, but he _should_ have". And that might be so, but I'm not sure he could simply adopt that position without having to give up some further commitments of his.

*This can't be quite right, since it may be possible for me to have acquired perceptions of colors from odors (say, if a given species of flower has a particularly distinctive smell and color, I may be able to perceive by smell the color of a particular flower (for instance, if I am blindfolded). It also can't be that both are originally perceived by sight, since, ex hypothesi, 3d color qualities are not original perceptions.

Joshua said...


I am not sure if I am reading the premise correctly, but it seems to me that (1) is mistaken. First, is it okay if I read the premise as follows:

1) A perceiver observing a uniformly blue sphere would perceive something uniformly blue only if that perceiver sees the sphere as a 3-dimensional object.

I am trying to figure out why someone might believe (1) and my guess is that someone might believe it if he/she is reading it as follows:

1*) A perceiver observing a uniformly blue sphere would perceive a uniformly blue sphere only if one sees the sphere as a 3-dimensional object.

But, then it looks like there is room for two different interpretations of this premise. On one interpretation, the content of the perception includes the sphericality of the sphere. On the other the content of the perception does not include the sphericality of the sphere. I'll try to distinguish the two as follows:

(1.1*) For any x and any y, if x is an observer and y is a uniformly blue sphere, then x perceives that y is both a sphere and uniformly blue only if x sees the y as a 3-dimensional object.

(1.2*) For any x and any y, if x is an observer and y is a uniformly blue sphere, then x perceives that y is uniformly blue only if x sees the y as a 3-dimensional object.

On the first of these, the content of a the observers perception includes something about the object being a sphere and on the second it does not. It seems to me that (1.1*) is fairly plausible whereas (1.2*) seems false. However, if we interpret premise (2) so that something validly follows from it and premise (1.1*), then we get the following:

(2*) for any x and y , if x is an original perceiver and y is a uniformly blue sphere, then x does not see y as a 3-dimensional object.

But, from (1.1) and (2) it only follows that:

(3*) So, for any x and y , if x is an original perceiver and y is a uniformly blue sphere, then x does not perceives that y is both a sphere and uniformly blue.

But, this shouldn't be troubling. After all, he is an original perceiver. So, presumably one thing he fails to perceive is the sphericality of the sphere.

Lewis Powell said...


First, let me reproduce the relevant portions of Reid's text, indicating which elements I think support Premise 1 by emboldening them:

"Thus, if a sphere of one uniform color be set before me, I perceive evidently by my eye its spherical figure, and its three dimensions. All the world will acknowledge, that by sight only, without touching it, I may be certain that it is a sphere; yet it is no less certain, that, by the original power of sight, I could not perceive it to be a sphere, and to have three dimensions. The eye originally could perceive only two dimensions, and a gradual variation of colour on the different sides of the object.
It is experience that teaches me that the variation of colour is an effect of spherical convexity, and of the distribution of light and shade. But so rapid is the progress of the thought, from the effect to the cause, that we attend only to the last, and can hardly be persuaded that we do not immediately see the three dimensions of the sphere.
Nay, it may be observed, that, in this case, the acquired perception in a manner effaces the original one; for the sphere is seen to be of one uniform color, though originally there would have appeared a gradual variation of color: But that apparent variation, we learn to interpret as the effect of light and shade falling upon a sphere of one uniform color."

This strikes me as a statement that the original color perception is of a gradient of blues, while the acquired color perception is of a uniform blue.

In a separate comment, I will consider your more specific points.

Lewis Powell said...

To your first question: Yes, it is okay to read premise (1) as you indicate (replacing "one" from my statement with "that perceiver". That was a simple typo on my part.

On your second point (re: formulation 1*), Reid seems to have in mind that there is, originally perceived, a gradient of blue, and that when one comes to see the shape as spherical (a process which occurs through associating the tangible shape with the visible shape, permitting one to visually perceive the tangible shape) one then comes to see a uniform color. Since Reid takes the two shifts in perception (shape appearance and color appearance) to go hand-in-hand, it is difficult to say whether he is better understood as endorsing a view on which there is a single perception with both sphericality and uniform blueness in its content.

However, I think it is clear that Reid would endorse 1.2*. I think that, even if Reid should be interpreted as endorsing 1.1*, 1.2* would also be attributable to him on the grounds that (on this 1.1* intepretation of Reid) no circumstance in which one sees the sphere as uniformly blue is one in which one fails to see the sphere as a sphere. In other words, 1.2* is either the view itself, or is a consequence of Reid's commitment to 1.1* (holding fixed the constitution of Human minds).

On the 1.2* reading of the premise, the argument gets the stronger conclusion.

Lewis Powell said...

Here is another way I could have framed the argument, where "shape-experienced perceiver" designates a perceiver who has experienced the association between visual and tangible shape enough to have acquired visual perception of tangible shape:

According to Reid, this is true:
1) The color appearances of an original perceiver presented with a uniformly blue sphere differ from and conflict with the color appearances of a shape-experienced perceiver presented with a uniformly blue sphere.

But, it is also clear that:
2) If the color appearances of an original perceiver conflict with the color appearances of a shape-experienced perceiver, either the color appearances of the original perceiver do not produce color perception, or the color appearances of the shape-experienced perceiver do not produce color perception.

Here is why we should accept 2: A pair of conflicting perceptions could not both be veridical (we can either fix the time and location, ambient conditions, etc. to ensure this, or define "conflicting" in such a way that it is trivial). Since, for Reid, all perception is veridical, this ensures that if two perceivers have conflicting color appearances, at most one of those appearances can produce perception.* appearances can produce perception.

Given 2 however, it will follow (for Reid) that either color is not originally perceived by vision (since the original perceiver does not have color perceptions), _or_ that we do not acquire perceptions of the color of three dimensional objects (since the shape-experienced perceiver does not perceive the uniform blue of the sphere).

I take it that Keith's suggestion(s) were a) to offer a challenge to (1); suggesting that the color appearances do not really conflict (the same way 2d and 3d shape appearances don't conflict), and b) to 'bite the bullet' on the conclusion, maintaining that we need not interpret Reid as treating color appearances as giving rise to perceptions (in original perceivers).

*This should probably say "at most one of those appearances can produce a perception with content conforming to the appearance".

Joshua said...

Oh I think I see how the puzzle works. I guess didn't realize that perceptions were supposed to be veridical. Might we state it as a set of inconsistent statements as follows:

A) There are some original perceivers and there are some experiences perceivers of a particular uniformly blue sphere.

B) Any experienced perceiver of a uniformly blue sphere will perceive that it is uniformly blue.

C) Any original perceiver of a uniformly blue sphere will perceive that it is variable in color.

D) All perceptions are veridical.

E) No sphere is both uniformly blue and variable in color

Since these five claims are inconsistent one must be false. But, this is a problem for Reid because he seems to accept them all. Would that be an acceptable way to state the problem?

Lewis Powell said...

Yeah, I think that is a fair way to put things.

Keith said...

I still don't see that there is any more of a problem for Reid here than there is with the issue of distance or real (as opposed to apparent) figure.

The original 'perceiver' experiences colour sensations of varying intensity corresponding to objective relations between the properties of the perceived object, the lighting, viewpoint, and so on. In simple cases, such as a uniformly coloured and illuminated flat surface, they are able to perceive colour veridically. To the extent that they judge a uniformly coloured sphere to be of varying colour they are in error since there is no such uniformly blue object. Hence, no such object is perceived, since perception is always veridical.

The experienced perceiver, on the other hand, is able to judge on the basis of the same colour sensations, along with a perception of the three-dimensional shape of the object, that the sphere is uniformly coloured, and so has a veridical perception. There is no conflict between perceptions because only the experienced subject perceives the object's true colour.

This diagnosis commits Reid to saying that the original 'perceiver' does not perceive the colour of the sphere, which is unsurprising since they judge incorrectly and are in any case unable to perceive the object's real figure. This is incompatible with premise 5, which should be rejected, but (so far as I can remember) in line with the rest of Reid's account, including the passage quoted above.

I'd be interested to know what textual support you can find for premise 5. Reid does credit Berkeley with noticing that the word 'colour' is ambiguous between the real colour of objects and the sensation of colour, which I think gives him all the resources he needs to respond to this puzzle.

Lewis Powell said...

You write:
"This diagnosis commits Reid to saying that the original 'perceiver' does not perceive the colour of the sphere, which is unsurprising since they judge incorrectly and are in any case unable to perceive the object's real figure."
However, on Reid's analysis of the figure case, there is original perception of visible figure, and no perceptual error. Original perceivers correctly perceive the quality of the object that is its visible figure (from their current point of observation).

Second, your diagnosis has Reid denying that (the quality of) color is an original perception of vision, but typical cases of acquired perception for Reid involve one sense modality acquiring the ability to perceive the objects of another. But on your diagnosis, color is exclusively an acquired perception whenever it is perceived.

Third, I think you are confusing the sensation/perception distinction (which is indeed important for Reid) with the original/acquired perception distinction. Sensations, for Reid, are mental occurrences which are causally involved with perception, but which are distinct from perception itself. Insofar as sensation occur, they are either reflexively directed or objectless (this is a debate among Reid scholars because of his ambiguous phrasing of the position). Perception, on the other hand, always has an object external to the perceivers mind as its object, and is always veridical, even in the case of original perceivers. Reid, in his objection to Hume's table argument about perceiving real magnitudes, is very clear that the visual perception of visible figure is a genuine perception of a relational property of the external object.

Later today I will try to find some more textual support for the points you regard as controversial.

Lewis Powell said...

Keith, here are some passages to cite support for the relevant claims I attributed to Reid (all page/line references are to the 2002 Brookes edition):

The discussion (reproduced in my first blog post on this subject) which gives rise to my attribution of (1) and (2) to Reid, is p. 236, lines 21-39.

I attribute (4) to Reid on the basis of the same passage, insofar as he describes the sphere in the case he describes as uniformly colored (not as both uniformly and variably colored).

On page 237 (lines 9-15), Reid writes: "In perception, whether original or acquired, there is something which may be called the sign, and something which is signified to us, or brought to our knowledge by the sign.
In original perception, the signs are the various sensations which are produced by the impressions made upon our organs. The things signified are the objects perceived in consequence of those sensations, by the original constitution of our nature."

This strikes me as distinguishing the color sensations from the color perceptions (Reid prefers to use terms like "color" to refer to the qualities of objects, not our sensations). Since sensations are the signs of the qualities perceived, this would mean that in original visual perception, the color-sensations are signs of some perceived quality. In other words.

In the very next lines, Reid explains that, in acquired perception, the sign may be either sensations or something else originally perceived. So, in the shape case, there are sensations which produce perceptions of visual shape (i.e. an original perception of a real-but-relational quality of objects), and when one acquires the ability to visually perceive tangible shape by sight, either the sensations or the original perception (i.e. visual shape) is the sign of the acquired perception (i.e. tangible shape).

In Essay II, chapter 22 ("Of the fallacy of the senses"), p. 244, line 16-22, Reid indicates that some judgments called "fallacies of the senses" are the result of inferring incorrectly from the testimony of the senses (Reid's example is a counterfeit coin: the mistake is not that your senses tell you it is a real coin and it is not, the mistake is that your senses tell you it is such-and-such shape, size, etc. and you conclude that it is a real coin. Your senses told you truths, and you inferred a falsehood). I mention this passage since here Reid lists color among the things perceived, lending support to including color among the original objects of visual perception, after all, it is not as though color is an original object of some other sense modality).

See also the discussion of color perception coming again on p. 247 (line 29) through 248 (line 7).