M31, brown or blue?

[ollypenrice]

Qualia on form as ever!

To pick up on Hume's "Sounds, colours, heat and cold...are not qualities in objects, but perceptions in the mind," I'd want to add that we know know more about the mechanical relationship between qualities in objects and our perceptioons of those objects in the mind. We know why object x appears red (in white light against a grey background!) and why it emits a sound we call middle c. This lends a degree of repeatability and predictability and objectivity to our perceptions, as it seems to me. This is why I don't find myself able to subscribe to a totally relativistic view of colour.

Olly

PS, Steve, pull yourself together, man: we've had much harder than this from Qualia!!!!!

[Qualia]

Imagine we know why object x appears red for we understand that – ‘object x appears red for it absorbs given wavelengths and reflects others’.

The strict empiricist will say that this experience or statement or observation is independent from any beliefs held by the observer, it is objective but is that correct?

It seems, for example, a statement riddled with beliefs: inferences from prior experience; theory about the apparatus required to derive such insights; the physical theories upon which the instruments and concepts are based; other theories that determine the effect (if any) of wavelengths and so on; so it seems we are still far from the empiricist’s ‘basic’ proposition.

It would be difficult to imagine a scientist going about their research without bias, without presuppositions and theoretical grounding in place. All scientific observations are guided by theory, so the empiricist’s statement of an unbiased observer merely recording events doesn’t bear close inspection.

If we understand the term ‘objective’ as a kind of transcendent gaze from nowhere, in which the subject is distanced from and set above the object of inquiry, then such objectivity must be impossible. If we understand objective as meaning the same results for all, how do we judge the competence of an experimenter or theory other than by whether he or she or it obtains exactly the "correct" result? If we mean observable as an experiment that copies all the features of an experiment which a theory already determines are relevant, we find ourselves repeating the experiment only as an example of the theory. There would be no way of advancing, no way of pushing science on towards new discoveries. Science in these senses, then, may not be strictly objective and by comitting to it such a stance may be condemning it to the realms of dogma.

None of this I see as an attack on science per se, for science is where we get so much of our knowledge of the world and of us, but we can also ask of it how and why. I think there is also an important moral reason for taking this stance, for by acknowledging that our observations and truth-claims are not gained from some disembodied gaze to which we can say, ‘well, it is just like that, or, that is just the way things are’ and then proceed to wash our hands of them, we can take responsibility of our truth-claims and of the consequences that may arise from them.

[laser_jock99]

Imagine we know why object x appears red for we understand that – ‘object x appears red for it absorbs given wavelengths and reflects others’.

The strict empiricist will say that this experience or statement or observation is independent from any beliefs held by the observer, it is objective but is that correct?

It seems, for example, a statement riddled with beliefs: inferences from prior experience; theory about the apparatus required to derive such insights; the physical theories upon which the instruments and concepts are based; other theories that determine the effect (if any) of wavelengths and so on; so it seems we are still far from the empiricist’s ‘basic’ proposition.

It would be difficult to imagine a scientist going about their research without bias, without presuppositions and theoretical grounding in place. All scientific observations are guided by theory, so the empiricist’s statement of an unbiased observer merely recording events doesn’t bear close inspection.

If we understand the term ‘objective’ as a kind of transcendent gaze from nowhere, in which the subject is distanced from and set above the object of inquiry, then such objectivity must be impossible. If we understand objective as meaning the same results for all, how do we judge the competence of an experimenter or theory other than by whether he or she or it obtains exactly the "correct" result? If we mean observable as an experiment that copies all the features of an experiment which a theory already determines are relevant, we find ourselves repeating the experiment only as an example of the theory. There would be no way of advancing, no way of pushing science on towards new discoveries. Science in these senses, then, may not be strictly objective and by comitting to it such a stance may be condemning it to the realms of dogma.

None of this I see as an attack on science per se, for science is where we get so much of our knowledge of the world and of us, but we can also ask of it how and why. I think there is also an important moral reason for taking this stance, for by acknowledging that our observations and truth-claims are not gained from some disembodied gaze to which we can say, ‘well, it is just like that, or, that is just the way things are’ and then proceed to wash our hands of them, we can take responsibility of our truth-claims and of the consequences that may arise from them.

Erm...you get the colour of a photo 'correct' by knowing the spectral response of the sensor used- it's an absolute measurment.

[Qualia]

Erm...you get the colour of a photo 'correct' by knowing the spectral response of the sensor used- it's an absolute measurment.

but isn't that knowing already-always couched in some number of coressponding theories? Moreover, by claiming an absolute isn't one deriding the spirit of science and its claim to knowledge via induction, inference, falsification, etc?

[Josh Wilson]

Where is IanL when we need him?!?

[IanL]

Where is IanL when we need him?!?

I'm not going there, philosophy is not something I know much about! I'll stand by my original assertion that colour is created by the eye & brain in response to a external stimuli. The fact that the external stimuli (light of various wavelengths) is something that can be measured empirically does not mean that one individual's perception of colour will be exactly the same as another individual's perception of colour for the same object or scene.

First let's be clear what I am not proposing:

- For one thing there are a variety of colour-blindness conditions, probably not relevant to this discussion though since they are (generally) well understood in terms of the physical/genetic changes that cause them and can reasonably be excluded as atypical.

- A previous poster did state that individuals from different cultural backgrounds have different ways of describing colours. This is certainly true. Here is an excerpt from a paper which is well worth reading as a background to this whole subject (http://www.imaging.o...eporter20_1.pdf):

"..though small in number, there are cultures today which have only two basic color terms. The Jalé in Highland New Guinea have only two basic color terms: hóló, which could be glossed as “brilliant,” and sing, which could be glossed as “dull.” The term hóló appears to encompass the colors that we call white and yellow, and likely covers the lighter ranges of blue and green; the term sing names black and red and likely the darker ranges of blue, green, purple, and brown. Among the Tangma of New Guinea, the two basic color terms are mola and muli. The term mola encompasses the “brilliant” colors white, red, and yellow, and muli encompasses the “dull” colors black, green, and blue. Along with Jalé, three other Damian family languages (New Guinea) have been found to be two color term systems also. These limited systems do not reflect a physical impairment of color vision across the population, but appear to be the product of adaptive need (or lack of need). In simple terms, their cultures and environments have not provided significant pressures which would warrant the distinctions between certain colors."

Whilst there is clearly an issue of 'lost in translation' between cultures, I am not saying that typical individuals from the same cultural background would describe objects differently in the same scene. For example, one person would not describe an apple in a still life as 'red' and another person describe it as 'blue'.

- I am not saying that colour perception and wavelength are unrelated, Our eye-brain system has an 'auto-calibration' system of some sort (colour constancy - see below for more). Each retina is different and the relative number of the three different types of cone cell can vary between individuals by as much as a factor of 40. Nonetheless, when asked to tune a block of colour to be as yellow as possible, individuals would always select a wavelength that was the same of very close to that chosen by others. More here (http://www.scienceda...51026082313.htm), including an interesting application of adaptive optics.

So what am I saying?:

- I have shown you some (admittedly extreme) examples that prove that that the overall composition of the scene affects your perception of the colour of individual objects in that scene. That was taken that I was saying that one person would therefore call a tomato red and another would call it blue, which was absolutely not my point.

- Colour is perceived in context. If the context is a single block of colour (as in the experiment noted above), then there does appear to be an 'absolute' relationship between a single wavelength of light and a colour (I do not know if the same follows for a spectrum of light though, that is not what the above experiment tested as far as I can tell). Conversely if the colour is in the context of a scene, then there is not an absolute relationship between the colour of an object and its wavelength/spectrum.

- Your perception of its colour is affected by the contents of the whole scene, both due to the contrast between objects emitting different wavelengths (blue and yellow, and red and green) and also brightness contrast between objects. There is a long, but helpful, summary of the science of colour perception here (http://webvision.med...n/color-vision/). One of the key points is this:

"The schemes of the previous figures neglect a problem called “color constancy”. We see colors as unchanged even when there are large changes in the spectral properties of an illuminant. The colors in a scene illuminated by fluorescent lights, which generate much short wavelength energy or by tungsten filament lights, which generate much long wavelength energy are not significantly altered by such changes. In other words we should see things “bluer” in fluorescent light and redder in tungsten light but we don’t. This constancy of colors despite changes in illumination intrigued Edwin Land, the founder of the Polaroid Corporation, who spent years investigating this phenomenon and demonstrating significant global aspects of color vision. Local objects can reflect identical spectral components from their surfaces but will appear of different color because of the influence of the entire scene. He proposed a model in which the signals from each cone mechanism are normalized over the entire visual scene before being compared with each other locally to generate the perception of the color of an object scene. "

- So due to colour constancy, we would see objects that are emitting different spectra as being the 'same colour' because of their context in scene, their spectra and contrast relative to other objects in the scene. This is a physical phenomena due to to the way the cells in your eye and brain operate, and appears to be 'hard wired' and therefore subject to measurement and calibration (in simple scenes like the optical illusions we saw in my earlier post at least).

- Perhaps less importantly, expectations influence your perception of colour. You know your car is blue, so even if it is illuminated under an orange sodium street lamp, you believe it as blue even though it is actually some dull shade of orange. You know a strawberry is red, so you believe it as red under any lighting conditions. You know Orion is red because ... hold on, you've never seen a red Orion nebula. You have seen lots of pictures of the Orion nebula that people have balanced to be some shade of red. So you believe it is red, but exactly what shade of red? It's a matter of consensus, one person creates an image and balances it one way, another does something a bit different, third person sees both images and shoots somewhere in the middle, and so on we go. This season, black is the new red darling!

- The issue is that you're working 'backwards' by trying to calibrate your image data to produce a natural looking scene by colour balancing it, but you have no reference other than prior images and wavelengths (and wavelength is not colour). Here is an interesting paper on the problems of colour balancing digital images (http://elynxsdk.free...nce problem.pdf). Now I'm not going to pretend that I understand the several pages of equations but I would ask you to take a look at the three images in figure four on the final page. It shows the same scenes processed by using different colour balancing algorithms.

To my eye some of the results look very unnatural, but equally there are some examples where the same scene looks different from another scene, but it would be very hard to say that one or the other was 'more natural' or a truer representation of the original scene. This is especially the case since we haven't actually seen the original scene with our own eyes. We face the same issue with our astronomical targets (well if any of you have a scope that can show Orion in full technicolour through an eyepiece, I'm coming round to your house for a look!)

So back to my original assertion; the colour balance of an astronomical image is a matter of taste and personal preference. There is no absolute basis for saying that one colour balance of an astronomical image is 'right' and by inference that all others are less 'right'. That is not to say there is no 'wrong'. I am not stating that Ha isn't red and so you can colour it blue if you like, but I am absolutely saying that for a given scene you might choose to colour it more magenta or more orange than red and produce an image that is perfectly 'natural' looking.

[laser_jock99]

The issue is that you're working 'backwards' by trying to calibrate your image data to produce a natural looking scene by colour balancing it, but you have no reference other than prior images and wavelengths (and wavelength is not colour). Here is an interesting paper on the problems of colour balancing digital images (http://elynxsdk.free...nce problem.pdf). Now I'm not going to pretend that I understand the several pages of equations but I would ask you to take a look at the three images in figure four on the final page. It shows the same scenes processed by using different colour balancing algorithms

Wavelength is a fixed physical constant as is your camera's response to it. In practical terms this is one reason why it's best to work from RAW images with DSLRs since basic colour data is not lost by the camara applying 'auto white balance' or some other unwanted algorithm. At least you can start work with a clean set of data. .

On the subject of white balance I tend to leave my cameras set to 'Sunny Day' for astro work so at least the .jpeg sets are consistent. I have cocked up a number of timelapse movies by accidentally leaving the white balance on auto with result the colour changes from shot-to-shot.

[IanL]

Yes absolutely agree that if you are shooting jpeg (or using them for preview purposes) then a manually set daylight white balance is the best move. No way an auto-white balance is going to get a long exposure astro image right. RAW is always the way to go though, since you can disregard the white balance setting on the camera and balance it yourself (it is recorded in the file if you do want to use it, or you can pick a different white balance in software if you just want a quick squint at what is in the RAW).

[ollypenrice]

Ian, your long post (I won't quote all of it on ecological grounds!) was excellent. I greatly enjoyed it. I'm far from expert on this, something I know because my father was an academic in perception theory and I could follow only a small percentage of his output on the subject!

Groups of people whose language describes colour very differently should certainly give us pause. The spectral response of the eye is one thing, but an enormous amount of human perception is filtered through language. Good point.

Olly

[laser_jock99]

The spectral response of the eye is one thing, but an enormous amount of human perception is filtered through language. Good point.

Olly

I expect the spectral response of everyones eyes is similar (we have the same cells)- it's our brains that are in question!

[IanL]

I expect the spectral response of everyones eyes is similar (we have the same cells)- it's our brains that are in question!

Interesting, that isn't actually the case. One of the papers I linked above did an experiment to measure the make up of the three different types of cone cell in the human eye. Before this, it wasn't really possible to tell the arrangement of the cells as the pigments degrade rapidly and so cutting open a spare eye didn't help. They borrowed from Astronomy and build an adaptive optics system to shoot a laser in to individual cells in test subjects' eyes and measured the reflected light to determine the arrangement of cells.

Turns out that the ratios of the three different types of cell can vary between individuals by as much as a factor of 40, so whilst the spectral response of an individual cell is roughly the same between individuals, the spectral response of the whole eye can be very different. Nonetheless, this does not affect colour perception. Again as noted above, the colour constancy phenomenon seems to indicate that the eye-brain system has some kind of 'auto-white balance' mechanism whereby it compensates for this physiological difference and we all see discrete colours as been the same, or very nearly so.