Pure Green

Colours are physical properties of external objects. One such colour is Pure Green: the shade of green that looks not at all yellowish or blueish. However, if people are asked to identify the shade of green that looks not at all yellowish or blueish, they come up with (slighly) different shades: what looks pure green to me looks slighly blueish to you; what looks pure green to you looks slightly yellowish to me. What shall we make of this?

We could claim that one of the groups is simply right about Pure Green and the other wrong, even though there is no way to find out which is which. That is incredible.

Alternatively, we could say that colours are relational properties: something is pure green only relative to an observer. That's not entirely incredible, but also suspicious. It seems to confuse the way we pick out a colour with what we thereby pick out.

Compare two thermometers A and B of which whenever A displays a number n, B displays n + 23. It would be silly to conclude that temperatures are relations to thermometers.

Is that because, in this case, (at least) one of the thermometers simply gets the temperature wrong? Not necessarily. We could also say that they measure temperature in different units: A in degrees Celsius and B in degrees Celsius + 23.

I think we should treat the case of Pure Green in the same way: what I call "Pure Green" is a physical property, not a relation to observers. It is the very same property you call "slightly blueish green". That we give them different names doesn't prove that they are different properties. It merely indicates that we pick out the property differently.

In my childhood, we had a hoover with a distinctive greenish colour we called "hoover green" ("staubsaugergrÃ¼n"). Other people perhaps used that same term for a different shade of green. This doesn't show that Hoover Green is not a real colour -- that it is a relation between various shades of green and people or hoovers.

Suppose hoovers were available in all shades of green, but never in red or blue or yellow, and "hoover green" became a common term that everyone used for the colour of their (or their household's) hoover. People would apply "hoover green" to different colours, but it would be ordinary colours they applied it to. Of course, "hoover green" wouldn't be a terribly useful term. If I told you that I painted my wall hoover green, you wouldn't know what shade of green I painted it unless you knew the colour of my hoover.

"pure green", I think, works like "hoover green", except that people wrongly assume that everybody's hoover has (almost) the same colour. We all use "pure green" for the shade of green our visual system processes in such a way that it looks not at all yellowish or blueish to us, and most of us wrongly believe that this is the same shade for us all. Well, in fact, I suppose "pure green" is hardly used at all; I should really speak of "the shade of green that looks not at all yellowish or blueish". This is also hardly used, but we have determinate intentions and expectations towards its use; for instance, members of our community are disposed to believe that a certain wall has a shade of green that looks (to them, and to all ordinary people) not at all yellowish or blueish if a trustworthy person tells them, "the wall has a shade of green not at all yellowish or blueish". The error is in the brackets: most people, I suspect, assume that what looks pure green to them looks pure green to most others as well.

Is this error built into the definition of "pure green"? That is, once we notice that people classify slightly different colours as pure green, should we conclude that there is no such colour as Pure Green at all? Or should we conclude that "pure green" is an indexical colour term like "hoover green"? Well, since "pure green" is kinda technical, we're free to choose, but I think the second choice is more natural. "pure green" thereby becomes less useful than we might have hoped: if I tell you that I painted my wall pure green, you don't know what precise shade of green I painted it unless you know what colour looks pure green to me.

(I should add that I think "pure green" is a very special case. Most other colour terms, like "Brunswick green" and "#00a913" and "green", are not defined by how they look to us, hence there's no danger that they might turn out to be uninformative indexicals.)

(I should also add that I am not at all familiar with the literature on colour. I've only read "Color Realism and Color Science" (PDF) by Byrne and Hilbert and the replies to it (also in the PDF); B&H argue for the incredible thesis that most people are wrong about Pure Green, while some of the replies argue that one should instead construe colours as relations. The solution I favour doesn't come up, but I'd be surprised if it is new. Maybe it faces serious objections I fail to see?)

# on 10 June 2006, 17:50

Hi Wo,
I'm not sure why you say that colors are physical properties of the external objects. Maybe I don't understand your intention or the context of analysis.
When we talk about colors we usualy talk about objects which don't emit, but reflect light. And it is this reflected light which gives the color of the object. So, those objects which reflect light (as opposed to emmiting light), don't have any color by themselves. So speaking physically the color they reflect will depend on their reflection/absorbtion qualities, but also on the structure of the light that is shined on them.
Maybe we should try to be more precise. We could probably speak either of things which are e.g. Pure Green, given that pure white light is shined on the thing, so, that from the whole spectrum it reflects just Pure Green; or about objects which emit particular frequency of light.
But then the light can be in any continuum of frequencies, and none of those frequencies has anything in itself which makes it green, blue or red. Wikipedia says that there are three kind of cones (neurons photoreceptors) in the eye ...
From http://en.wikipedia.org/wiki/Cone_cell :
"The color yellow, for example, is perceived when the yellow-green receptor is stimulated slightly more than the blue-green receptor, and the color red is perceived when the yellow-green receptor is stimulated significantly more than the blue-green receptor. Similarly, blue and blue-like hues are perceived when the blue-violet receptor is stimulated more than the other two."

So, if those photoreceptors react differently in different people (as it seems they do), then they would judge different colors as being Pure Green.

# on 10 June 2006, 19:38

Hi Tanasije,

yes, I took it for granted that colours are, at least prima facie, physical properties of external objects. The paper I linked to contains an elaborate argument for this view, so I'll be brief here:

I agree that it is the reflected light which gives the colour of the object. But why does it follow that "so, those objects which reflect light don't have any color by themselves"? The reflected light also "gives" (tells us about) the shape of the object. Would you conclude that objects don't have any shape by themselves?

I also agree that from a physical point of view, there is nothing very special about the light or the surface properties we perceive as having different colours. (This is different for shapes.) Arguably (due to metamers), what we perceive as green is a gerrymandered mess of a surface property. Nevertheless, it's still a surface property.

In the end, I don't have a good argument for why flowers, leaves and rivers (themselves) have colours. It just sounds utterly obvious to me, and I've never heard a convincing reason to deny it.

# on 11 June 2006, 19:39

Hi Wo,

I glanced over the paper, and I understand the context in which you are talking now...
I think that is position that color is a objective property of the surface of the object, more precisely a function (or set of functions) over its spectral-reflectance(spectral-reflectance being "the proportion of incident light the object is disposed to reflect at each wavelength in the visible spectrum.").
As spectral-reflectance is objective property of the surface, so any mathematical function, or set of mathematical functions on it, will give us another objective property of the object.
If I understood the paper, they intend to show that there is such function, or set of functions based on the objective spectral-reflectance of the surface of the object, in which case the the color we are looking at, in conditions necessary for perceiving color, would present us with the objective color, as defined by the function (or set of function) over the reflectance property. The functions or set of functions, are needed to give account for metamers, similarity of colors; so that in the end as result we have objective values for the color that the surface has.

But, it might be argued that in order for the properties of the surface to be mapped to phenomenal color, we need different functions for different people. And even that is not the case, we can imagine the case in which it is (e.g. cones which react differently). And this I guess is connected to what you are saying... that phenomenal colors (as Pure Green) in case of different people will be connected with different reflectance properties of the surface.
But it seems that in such case it would be confusing to say that same-objective-color is pure-green for one person, and bluish-green for another. In my opinion we could say that same reflectance properties of the surface would be pure-green for one person, and bluish-green for another (as in the case of thermometer), and I would avoid to use objective-color term. As if we loose the objective way to map from the reflective properties to non-subjective colors, it would be hard to say why would we speak of "colors" (i.e. a specific set of functions over the reflective properties, which were taken just to give account for the issues connected to the issues with phenomenal colors - similarity, metamers), and not about the reflective properties themselves.

# on 11 June 2006, 20:55

I'm not sure I understood all of this. In the case of the thermometers, would you also say that they measure different ("subjective") temperatures corresponding to the same objective level of molecular energy? Isn't it far more natural to say that they measure the same temperature, but merely display the value differently?

I find the notion of phenomenal colour rather more confusing than objective colour terms. What are phenomenal colours supposed to be? Apparently not properties of physical objects. So are they properties of experiences? But experiences don't have colours (or at least not of the right sort). I think it is better to just speak of different kinds of colour experiences on the phenomenal side: there's a Pure Green experience, a yellow experience, etc. For the most part, colour terms are not defined in terms of those experiences, I believe: when we teach children the meaning of "green", we show them green objects, and don't care about their experiences. So it doesn't matter if there is no unique map between objective colours (reflectances or whatever) and colour experiences ("phenomenal colours"). "Pure green" is a very special case in that it is really defined in terms of colour experiences, which is why here differences in experiences do show up.

Another comparison: there is a certain range of skin temperature we feel as neither too cold nor too hot, something around 25-35?C, I guess. Let's introduce a name for it: "mild". Now mild is an objective temperature, not an experience or anything like that (after all, I *defined* it as a (determinable) objective temperature). But the definition is indexical: the temperature *I* properly call "mild" might differ from the temperature *you* properly call "mild".

# on 12 June 2006, 02:59

I'm also little confused, so to avoid misunderstanding i will try to summarize what I understand as main point in the paper, and what I undersand as your proposal. Please correct me if I got something wrong.

The objects in our experience appear to have certain color.
We can call the color that objects appear to have - phenomenal color.
The color that objects do have, would be physical color.

The proper physicalist account of color, as I understood the paper, would need to show how the following is possible:
(*)Things can have the color they appear to have.
The relation there is relation of identity, the color that thing appear to have, in the given circumpstances would be identical to the color the thing have.

But in the pure-green case we have situation in which:
1.the object appears to be pure-green to person A. Let's assume (*) to be the case, so if person A says "that thing is pure-green", he will be right.
2.the object appears to be bluish-green to person B. Let's assume (*) to be the case, so if person B says "that thing is bluish-green", she will be right.

If we assume all those things. the result is that we have object which is both pure-green, and bluish-green, which seems to me as contradiction.

I understand that you are proposing to make it matter of definition - that we agree color to mean some objective property of the objects (reflectance or function of reflectance)... Then, there is no choice but to agree, that things objectively have certain color (by this definition).
But in this case, seems we will end up with in principle separate two concepts of color: physical color and phenomenal color. Things would not have colors that they appear to have (i.e. no pure-green objects in themselves), but colors they appear to have will stand in some relation with the color they have. The relation between phenomenal color and physical color (defined in some way) will not be that of identity, but would be probably more simmilar to relation between the temperature of the object and the length of mercury in thermometer. The length of mercury may represent the temperature, but it is not the temperature. It is in some kind of causal relation with the temperature.

# on 15 June 2006, 18:51

That sounds right, except that I still wouldn't speak of "phenomenal colour": we wouldn't say that the length of the mercury is a special kind of *temperature*. It is something that represents temperature, just as colour experiences represent colours. When they misrepresent, we get a case where the colour an object seems to have (is represented as having) differs from the colour it actually has. But this is not what happens with Pure Green, I believe. (Byrne and Hilbert think it is.) Rather, what we have here is a many-one map between colour experiences and colours. As you say, the relation between those is not identity (an experience is never a colour), but similar to the relation between temperature and the length of mercury in the thermometer.

# on 22 June 2006, 13:09

I struggled with this paper a little while ago. As I understood the authors, one thing they would really like to capture is how veridical perception is possible.
Yet it seems to me (as it does to many people doing vision science), that the process of perception is divided into three stages: refraction of light (1), excitation of the retina (2) and finally (3) processing be it retinal or post-retinal. The phenomenon that different perceivers may perceive different hues is then usually explained by differences in the composition of their sensor (the retina) and/or differences in processing.
So if I should take a stance on the subjectivity/objectivity distinction or the absolut/relational distinction I'd definitely opt for a distinction of levels. The physical color of an object is the way it refracts light, while the phenomenal color is that what we perceive, make judgements about and so forth. Sometimes I feel the whole problem arises only because philosophers still strive to get hold of the physical stimulus which, I suggest, they may never achieve ...

# on 26 June 2006, 15:55

Hi Bert, I don't think anyone would disagree with that three-stage picture of perception. But I'm not sure why you say that "what we perceive, make judgements about and so forth" is not physical colour. The fact that perception of X requires a causal chain linking X and our conscious experience (a rather trivial fact, once noticed) doesn't entail that we don't perceive X after all, but some mysterious other thing, "phenomenal X". At any rate, if I call a wall "red" and then notice that it's only illuminated by red light, I'd say my earlier claim was wrong: the wall isn't actually red. So what I make judgements about is not just how things appear to me.

# on 28 June 2006, 11:00

Hi Wo, it is one thing not to disagree, another to overlook that it might be problematic to hold two (or more) views that are hard to maintain at the same time.
Naturally phenomenal color is dependent on physical color. Call this relation causal if you like. But what also has to be accounted for is that parts of subject's visual systems can be different. And how are we supposed to factor that into our causal explanation, which for now just is a relation between an object X in the world and a conscious experience of a thing X. Once thought this way I do not find the "trivial fact" trivial any more. Where are the possible differences in subjects? And that there obviously are such differences seems to be more than empirically plausible compare for the case of pure green Munsell Chips.
I want to catch as much of our intuitions as possible yet, it simple seems to me that some aspects of those intuitions need to be revised. We talk about colors as properties of objects and rightly so, because the objects appear colored to us. (But have you ever wondered why a white wall in natural light doesn't look bluish to you even so the incident light is in fact bluish? Or why white things in a room that is lighted by neon light don't appear greenish to you?) Furthermore we have to face the fact that our ability discriminate two different hues is far better than our ability to regognize a hue. Our color concepts simply do not account for such subtle differences between say red_12 and red_87. Yet, there might be differences. And now we face a problem either we rescue our ideas about veridical perception and judgements about our perceptions or we admit that our claim might be to strong. And I simply have a strong hunch that the latter is the case. Hope this made my thought clearer...