Saturday, 19 July 2014

Colour Perception And The Impossibility Of Ever Painting Things Exactly As They Are.



When I was sixteen I had my first serious opportunity to do an oil painting en plein air as it is sometimes referred to, harking back to the French impressionist practice of painting in the open. I had an easel and primed board set in front of a huge field of oilseed rape and the choice of a handsome selection of colours. I also had plenty of time, it being the summer holiday, and my plan was to build up the painting gradually as each successive layer dried. Everything went well enough until I began working on the intense yellow of the oilseed rape. I could mix the colour very convincingly on the palette and for a while there seemed to be little difference between my yellow and the yellow of the field. But as the lighting conditions changed the difference between painting and field was unmistakable. This was a source of enormous frustration to me since it seemed that either my skills of colour mixing were defective or else the paints were sub-standard. Actually the paints were professional quality oils so it seemed unlikely that they were at fault but what I didn't realise then but is clear to me now is that the paints were indeed defective. All paints are. The reason we don't complain to Windsor and Newton or Rowney is because our perceptual powers are also systematically limited in ways that we do not notice because these flaws are shared and are therefore the perceptual norm. Pictorial representations such as oil paintings or photographs would be impossible without these shared perceptual limitations because pulverised pigments or chemical dyes—even when expertly mixed—would almost never seem to have the same properties as the things they represent. Their chemical structure and therefore their spectral properties under varying lighting conditions would simply be too easily discriminable for use as colour substitutes. If our perceptual capacities were faultless, no two different things would ever seem to be the same.

The issue can be better understood in the case of my struggles with oil painting. The reason the yellow of my painting and the field seemed to be the same for a short time was because under those particular lighting conditions I was incapable of discriminating between them. Other humans would have the same difficulty whereas other perceivers with different perceptual skills probably wouldn't have nearly the same trouble. Tetrachromats who—like many birds, fish and amphibians—possess an extra set of colour-sensitive cones on their retinas, would be unlikely to make the same mistake. When the lighting conditions over the field changed, the differing spectral properties of the paint and oilseed rape absorbed and reflected light in ways that were more easily discriminated (for me with my trichromatic vision) so it was obvious that the paint wasn't the same shade as the oilseed rape. Any hope I had of mixing a perfect spectral match of the crop was doomed before I had even begun. Perhaps if I had known this at the time I wouldn't have mistakenly concluded that the fault was due to a lack of skill on my part, but rather to inherited weaknesses which make all the wonders of pictorial representation possible in the first place.



16 comments:

Tor said...


The slaughterhouse sequence from Predator 2 on YouTube

Jim Hamlyn said...

Thanks Tor, I was thinking of Predator too: This is the scene from the original film where the predator can't see Arni because he's covered in cold mud which blocks the infra red.

Drawstillwater said...

Yes, you can't literally match things with paint.

Jim Hamlyn said...

Exactly. You can simulate them though. Which means to say that under specific circumstances of presentation our species is susceptible to an incapacity to discriminate (sensorily - not intellectually) between two things that in the relevant respect are objectively different.

Helge said...

Thank you for the interesting article, Jim! I have also been wondering about my inability to capture the luminous quality of rapeseed fields on film - one of the things I only recently realised is, how much that luminosity depends on plants converting UV light into visible light, thereby exceeding the brightness of any paint or photo paper.

Jim Hamlyn said...

That's right Helge, I'm pretty sure oilseed fluoresces which I didn't know at 16 either. Paintings and photographs are terrible at mimicking reflective surfaces too. When have you ever seen a picture of a mirror that actually reflected the same as the mirror! I'm joking of course but in fact that's exactly the point I'm making (stretched to absurdity). On a different note, fluorescence is interesting from another point of view too because it's caused by something called down-conversion. I wonder if any materials down-convert visible spectrum to lower energy forms of visible (or invisible light) and how would we know?

Helge said...

You could maybe test it with infrared photography - it also helps to know which range of the IR-R-G-B-UV-spectrum the insects that pollinate a specific plant are sensitive to.

Jim Hamlyn said...

Yes, though I'm not sure that what I was trying to say was clear. I'll give it another go. If we could talk to a bee and ask it how fluorescent colours look it would probably deny that they look strangely bright. The bee might instead say that they are just bright colours. Or, to put it another way, say there is a material that absorbs visible blue and down-converts it to green. How would that materiel look? Would it look fluorescent? How could it? It would just look green? And there would be no net gain in luminance overall so it could't appear to fluoresce. The only reason fluorescing materials appear bright is exactly as you say: because they give off more visible energy than they appear to receive. If I'm right, then it must be possible to find a material that down-converts visible light to lower energy visible light. So for example it must be possible to find a material that looks red or green under blue light. Sadly I can't find any information about such a thing online.

Felix K said...

I think you may be a bit hard on artists' paint manufacturers, Jim, just because these colours, when they are applied to a frontally presented canvas and seen at a moderate distance, don't match the colours on the surfaces of more distantly seen and often obliquely or otherwise unfavourably presented things.

You are obviously right to blame our shared weaknesses of acuity in sensory discrimination for the fact that the blue colour on a canvas successfully simulates for us the colour of a distant hill (whatever colour that hill may really be) in a way that probably wouldn't work for eagles. But I think that artists' colour manufactures ought to be congratulated rather than blamed for catching on to this, and adjusting their product accordingly.

Jim Hamlyn said...

True, paint manufacturers should be congratulated on their acumen for exploiting (or catering to) our frequent inability to discriminate between the colours that can be mixed with what they sell and the things we see under changing light. Their products compliment our flaws perfectly don't they? Though not with fluorescing things like oilseed. Daler Rowney and Windsor and Newton don't make fluorescent oil colours.

Drawstillwater said...

Jim, I still am a bit confused.

You are saying that ‘all the wonders of pictorial representation’ are only possible for us humans to enjoy because our perceptual capacities are faulty?

And it is because all humans posess these perceptual faults that we are able to recognise (& accept as ‘correct’) the representation of e.g. a cornfield by Van Gogh?

If its impossible for humans to match ’reality’ in paint then what exactly are we 'recognising’ ?

Did Van Gogh just manage to paint the ‘moment’ when the field actually was that colour?

Jim Hamlyn said...

You are saying that ‘all the wonders of pictorial representation’ are only possible for us humans to enjoy because our perceptual capacities are faulty?

Yes. But terms like "faulty" and "failure" tend to make people baulk — as if our sensory systems were only ever very mildly or very occasionally faulty. It's often more palatable to speak of systematic "limitations" in various ways and in various circumstances. Remember though that whilst we're susceptible to these weaknesses we also have other senses and capacities to confirm or disconfirm our discriminations and if we're language users we also have the advantage of concepts to help distinguish one thing from another.

And it is because all humans posess these perceptual faults that we are able to recognise (& accept as ‘correct’) the representation of e.g. a cornfield by Van Gogh?

Sort of, yes. Accepting something as correct is something we do because we can make judgements by virtue of our language skills. The point I'm making is more fundamental than our language though (hence optical illusions and their imperviousness to conceptual knowledge). But broadly yes, it's because we're all often incapable of discriminating on a sensory level (not a conceptual level) between two things that are actually different in the relevant respect/s.

If its impossible for humans to match ’reality’ in paint then what exactly are we 'recognising’ ?

Recognition is when the sensory setup—let's just say brain for simplicity—responds in the same way to one thing as to another; i.e. cannot discriminate (sensorily but not conceptually) between them.

Did Van Gogh just manage to paint the ‘moment’ when the field actually was that colour?

"Actually that colour"? What colour was the field actually at that moment? Let's say that we could measure a sheaf of Van Gogh's actual straw and determine that it was beige123. Later in the day, during the sunset say, would the colour of the corn "actually" have changed? No, it was still beige123. Corn has what's known as "colour constancy". Obviously it gradually changes over the seasons, but throughout each day (and night for that matter) its colour remains relatively unchanged.
So then you might want to say that the "apparent" colour changed due to the different incident light at sunset. We could measure this colour of course and give it a spectral specification in terms of hue saturation and brightness. Perhaps we could also mix a paint with exactly the same spectral specification under those specific lighting conditions, simulating the colour perfectly, say Orange234. However, actually matching, not just beige123 but the full spectral characteristics of the corn, under all possible lighting conditions would be vastly more difficult. But all of this is academic. We don't need to try to match the exact spectral characteristics of the corn because actually painting the corn as beige123 when you want it to look like it's illuminated by evening light will completely ruin your chances of a convincing result. Use orange234 and Bob's your uncle!

One hurdle I run into arguing this issue is that many theorists want to get down to the nitty gritty of visual stimuli - as if seeing corn or whatever were a case of seeing visual stimuli. We don't see visual stimuli, we see things. We can't see light, we can only see things illuminated by light or giving off light (i.e. lightbulbs or the sun). So when we see a field of corn in the sunset, what we perceive is beige123 corn in orange light, not corn that changes colour moment-to-moment and is just at this moment orange234 and soon to change to black for the hours of darkness. When we simulate the colour of things we are not making an objective claim about their actual colour, we are exploiting the fact that under many circumstances our perceptual capacities are limited in an incredibly useful way.

Drawstillwater said...

The nitty gritty hurdle can’t be ignored if you want to change people’s minds about a deeply held conviction.

Can you explain what are the visual stimuli which stimulate us visually?

Are they responses in the retina or the brain…to light (what else?) …& then our (unconscious, or thoughtful based on past experience, or mitigated by inherited ‘weakness’) interpretation to ‘make sense’ of what we are looking at?

Jim Hamlyn said...

When theorists speak of visual stimuli they are trying to describe the phenomena involved at a 'scientific' level but this strategy fails in the same way that the strategy of attempting to explain consciousness by talking of electrical processes in the brain. The level of description is inappropriate to the issue we are trying to solve. Water and H2O are two levels of description of precisely the same thing, but these levels of description have different uses and limitations depending on what we wish to understand. It's like attempting to explain the taste of food by describing chemical formulae. Of course food has chemical formulae but our sense of taste is best understood by describing what we are capable of discriminating, not by describing chemical interactions.

Drawstillwater said...

"Our sense of taste is best understood by describing what we are capable of discriminating, not by describing chemical interactions."

Sweet salt, sour, bitter etc…..Yes I can understand that

So what would be the equivalent visual ‘phenomena’ explained in equally simple terms?

Jim Hamlyn said...

Sure, we say things like: "It looks red," "It looks beige123," "It looks circular," "It looks ten feet high" etc.

Note also that we can say either something "looks red" because it is red or we can also say that a white thing looks red in red light.

With each of the senses, we talk comparatively in these ways all of the time. Sometimes we are describing objective properties of objects and substances and sometimes we're describing simulating cases. Do all things that taste lemony share objective qualities with lemons? Probably not. But the fact that some things aren't objectively the same as others hasn't stopped us simulating chicken flavour, or onion flavour etc. by using synthetic formulations.

Audition is perhaps an easier case. We can simulate a loud sound in the distance with a more proximate quieter sound. We can also simulate the song of a bird even though the setup of our mouthparts is completely different than a bird's. We might Match their song in respect of its length or number of trills or several other attributes but in some respects we can also exploit the fact that our discriminatory capacities are imperfect. And by imperfect here I mean, by way of example, that if audition were perfect, distant things would never sound quieter than proximate things.

So, to return to the theorist who wants to interrogate the stimuli involved in some sensory modality. They will argue that the wavelengths etc. involved are what make something objective. Sure, just as H2O is water. But what about audio volume for instance? Is volume an objective attribute? If the volume is 100db at source and 50db a certain distance from source then which is the objective volume? We don't need to know. What we want to know is what the sound is of, not the spectral characteristics of the sound. We want to know if it really is the sound of someone prowling around in the basement or if it's just the boiler playing up again.

We evolved to sense things and events, not raw stimuli. Of course stimuli are involved, but even though we have developed sophisticated ways of interrogating the phenomena that mediate our access to the world it is only very recently that there has been any advantage whatsoever in knowing about these photons, wavelengths, energetic molecular excitations etc. Light enters our eyes and stimulates our rods and cones, but we don't see the light, we see the world.

Post a Comment