Color Vision - Science topic

Most of the stimuli in the color blindness test may be due to being printed in ink.

Emad, it has been few years since you raised this question. I am working on X-rite version of hue test. Did you publish your work? How may I get a PDF?

This is great, thank you Jasleen Kaur Jolly

This might help:

While there is a lot of debate regarding the issue, there have been studies (I forget the article name) which indicated that women possess more cones (or variety in the cones i.e. polymorphism, to be more specific) in their retina, thereby attributing to recognizing a wider colour spectrum. This has widely been attributed to the division of labour among our ancestors wherein, historically early-women being gatherer's needed to identify edible/ripe fruits whereas the men served as hunters and needed to spot movement. While I am not sure how credible these theories are, they do sound plausible.

Murray et.al's 2012 paper is an interesting read on the subject of male/female colour vision comparison... doi:https://doi.org/10.1167/12.1.18

Thank you guys! I will consider both answers on the next time I run a colour vision model!

Cheers!

Dear Amaranath,

I am sorry, but unfortunately this is not my area of expertise. Maybe the following papers can help you a bit with your question:

Joselevitch C. Human retinal circuitry and physiology. Psychol Neurosci 2008;1(2):141-165. http://www.scielo.br/pdf/pn/v1n2/v1n2a08.pdf

Asari H, Meister M. Divergence of visual channels in the inner retina. Nat Neurosci 2012;15(11):1581-9. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3717330/pdf/nihms488915.pdf

Hoon M, Okawa H, Della Santina L, Wong RO. Functional architecture of the retina: development and disease. Prog Retin Eye Res 2014;42:44-84. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4134977/pdf/nihms-609782.pdf

Cruz-Martín A, El-Danaf RN, Osakada F, Sriram B, Dhande OS, Nguyen PL, Callaway EM, Ghosh A, Huberman AD. A dedicated circuit links direction-selective retinal ganglion cells to the primary visual cortex. Nature 2014;507(7492):358-61. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4143386/pdf/nihms552751.pdf

Neitz J, Neitz M. Evolution of the circuitry for conscious color vision in primates. Eye (Lond) 2017;31(2):286-300. https://www.nature.com/articles/eye2016257.pdf

Have a good research day, Martin

Jasleen, thank you for your answer. I will take a closer look on the article you recommended. :) Have a nice day!

Blind people have no visual imagery in their dreams, if born blind. Dreams do process past memories; in a dream, different temporal memory layers can be mixed over into 'one film'. We do not dream with our eyes, it is a brain function in sleep as the brain is a non-stop organ (you could put a light bulb on it).The existence of pre-cognitive dreams, which is portrayed in prophetic literature, e.g. Joseph in Egypt, cannot be ruled out. A healthy sleep cycle (chronobiology) and dreamimg are closely connected, in medical terms, to 'free' the memory from non-necessary psycho-logical ballast of life experiences.

Hi Pierpaolo,

in general, when it come to colour spaces your choice depends on what you want to test. Gabor patches are usually used to control cone-opponent low-level mechanisms (second-stage mechanisms). For this reason, colour is typically modulated along the axis of cone-opponent spaces, such as DKL and, less common for this purpose, MacLeod-Boynton. However, you can also use Gabor patches to investigate paradigms that require the control of colour appearance (i.e. how colours appear to the observer), such as the use of Gabor patches and the control of salience in attention research. In this case, you would rather use colour appearance models, such as CIELUV, CIELAB, CIECAM02. I add a little overview on colour contrast, where Gabors are modulated along the cone-opponent axes.

I hope this is of anz help. All the best,

Christoph

A good solution would be to specify each image as a set of R,G,B quantified pixels and characterise the resulting populations in terms of their spectral Hue, Chroma (or purity) and Lightness. The key of course is a computer model that correctly maps RGB pixel values onto HCL perceptual colour values or the less intuitive CIE L*a*b* colour space co-ordinates. When the task is simply to quantify the difference between images, a simple solution is available via Adobe Photoshop and ICC (International Colour Council) colour profiling. I describe an earlier and perhaps better approach in detail in two of my RG posts namely “Measuring the contribution of texture to colour appearance” 1998, and ”Computer systems for the specification and communication of colour” 2009. In these papers the quantified R,G,B difference, the overall model and the HCL perceptual difference co-ordinates used may all be fully verified step by step, by reference to the internationally agreed CIE Standard Observer colorimetric model.

Does color vision testing in the primary school screening pick up amblyopia ? I do not think so.

However, color vision testing in the school vision screening is important to pick young children with congenital color vision defect to help them adapt to their environment and in later years select appropriate career. Infact teachers and parents are to be educated on this in India. At our institute, as part of all our school vision screening we have included color vision screening

Although your question is several years old, I would be curious to see how you dealt with this problem. I can only imagine using saturation as a variable since a dark red and a light red would not really be equal in luminance, but you could mix pure red with gray of the same luminance to de-saturate the red and obtain (perhaps) a lower salience.

The opponent color system requires less energy and keeps the head cool. The achromatic channel is fast (enemy attacking from behind) and high resolution (for edge detection), while the chromatic channels can be slow and low resolution. To further save energy, when the chromatic content does not change, the retina propagates only information about the edges and the color on each side of the edge (it does NOT work like a camcorder) by inhibiting the chromatic response from that region. This adaptation process is slow, so when the region's color changes, it takes a while for the inhibition to be rescinded, thus we perceive an afterimage of the complementary color.

Well I would say if you compare an old desktop version to a new handheld you might find comparable results. But comparing new equipment the difference will still be there. Handheld instruments generally has to sacrifice stability and accuracy to reduce size and weight. Knowing the error tolerance in your experiments is therefor important.  Try to send the vendor a test spectrum and ask them if they can give you the uncertainty on a key parameter if it had been measured with their instrument. If they can't give you that - they might not know what they are doing.

Anders

Hi,

 As a first step, you basically need to calculate how much the spectrum you measured excites the 3 human photoreceptors. In principle, you need the spectral sensitivities of the 3 human photoreceptors, L, M, and S (long-, middle, and short-wavelength). However, depending on what you want to do with your colour specifications, you can also directly calculate the Tristimulus Values (XYZ). XYZ allow you to directly calculate chromaticity coordinates (xyY), which are widely used to communicate colour specifications.

If you want to obtain LMS values you need to get spectral sensitivities (cone fundamentals). If you want to get XYZ you need colour-matching functions (CMF). The CMF are – in principle - a linear transformation of LMS for the purpose that one dimension (Y) corresponds to luminance (i.e. low-level lightness). For further details see e.g. here http://cvrl.ioo.ucl.ac.uk/database/text/intros/introcones_cmfs.htm. You will find cone fundamentals and CMF here http://cvrl.ioo.ucl.ac.uk/cmfs.htm and here http://cvrl.ioo.ucl.ac.uk/cones.htm.

Both consist of 3 columns with the contribution of each wavelength to the cone excitation (LMS) or XYZ, respectively. You just need to calculate the sum over all wavelengths (i.e. the integral) to obtain how much your spectrum excites each of the cones / each of XYZ. This can be done by an inner matrix product (= sum of products). So, in principle you need to do this:

SPECTRA(:,1:3)'*CMF(:,1:3)

where the rows of each matrix (SPECTRA & CMF) are the weights for each wavelength.

 However, you also need to make sure that your spectrum has data for the same wavelengths as the CMF/cone fundamental. If not, you will need to interpolate your spectrum for the respective wavelengths of the CMFs (either linearly or by cubic spline). Moreover, in order to get standardized XYZ you also need to multiply by 683 (=maximum photopic luminous efficacy) and by your wavelength resolution (the wavelength-interval between each data point, e.g. 300:5:800nm à interval = 5nm). Once your CMF and spectrum are matched for wavelengths, the complete formula looks like that:

XYZ = SPECTRA(:,1:3)'*CMF(:,1:3)*683*Interval;

Then you obtain xyY by normalization (x = X./(X+Y+Z); y = Y./(X+Y+Z); Y=Y).

A final complication: All I said above requires your spectra being specified in spectral radiance (w/sr/m2), as measured by a spectroradiometer. However, if you measured photon catches by a spectrophotometer you first need to calculate spectral radiance in order to get XYZ.

All the best,

Christoph

Attached.

This is a little old because I no longer teach. But I think it is still in use. It is very stepwise with a small spoon, as students need!!. But once in the swing. I find it works well. A couple of minutes for a dichromat. Maybe 10 for a good anomalous trichromat.

The key is also that they adjust the brightness overtime and we check that they are seeing the colour difference accurately not a brightness difference.

Happy reading

Best wishes

Stephen

There are 3 aspects of this issue: 1- For Monitor/Display, 2-For Color Sensor, 3- Color mixing

1- For Monitor/Display: Code Value (eg. 0 - 255) will decide the color through complicated conversions of this system dependently. I think that you do not mean this aspect. But if you want to know it, please read: https://en.wikipedia.org/wiki/Bayer_filter

2- For Color Sensor: in the color sensor there are filters and a photodiode. The filter will capture the desired spectra such as R(lambda), G(lambda), B(lambda) and then the photodiode will convert these optical energies into the desired electrical signals such as current and votage. I think you do not mean this aspect.

3- Color Mixing: I guess that you want to mean this aspect. If it is correct, actually it depends on your color light sources. Principally, I can say as follows:

a- R light source has chromaticity xr,yr und 'brightness' YR (you mean Light Intensity).

b- G light source has chromaticity xg,yg und 'brightness' YG (you mean Light Intensity).

c- B light source has chromaticity xb,yb und 'brightness' YB (you mean Light Intensity).

Then you can determine the factors mr=yr/Yr, mg=yg/Yg, mb=yb/YB and msum =mr+mg+mb. And the mixed point will has chromaticity as

xsum=(xr*mr+xg*mg+xb*mb)/msum

ysum=(yr*mr+yg*mg+yb*mb)/msum

Ysum=YR+YG+YB

There is a matrix to convert from x-sum, y-sum, Y-sum into RGB-Value (0-255). You can read it here: http://www.brucelindbloom.com/index.html?Eqn_RGB_XYZ_Matrix.html

Finally, you always have the relationships YR,YG,YB=f(electrical signal), mixing principle and matrix. They help you to solve the problem. Additionally, you should read about adaptation CMC2000, because the conversion depends on the reference light source. Currently I am too busy to be able to give a more careful answer and but hope, this quick answer can help you.

Thank you!

Dear Luis,

I have only done research on humans. In vision research on human color perception, perception of gray shades is a separate field.

First step (for us) is to distinguish between above mentioned

(1) brightness, for example shades coming from your computer screen (i.e. emitted light) and

(2) lightness, the shades you see on paper (i.e. reflected light, a composite of light source intensity and the surface reflectance).

(http://www.cs.tau.ac.il/~hezy/Vision%20Seminar/Lightness%20Perception%20and%20Lightness%20Illusions.htm)

Visual system of humans provides you with different results when asked for lightness and brightness judgment. However, the differences does not come from retinal responses but from visual cortex, such as Brodmann area 17 (V1) (http://www.pnas.org/content/98/15/8827.full)   in charge of lightness constancy – the fact that you see the same shade despite the change in illumination (http://www.sciencedirect.com/science/article/pii/S0960982207013826). Obviously there is no brightness constancy as that is the percept of illumination itself (i.e. emitted light).

The reason for my long intro is because Stevens reports his exponents from research much like mine (12 students, behavioural exp, perception judgments).

I am not sure how generalizable is this result to sheep retina, given the differences mentioned in this fantastic response by Ian Ashdown.

Alternatively you can use Bevil R. Conway : Neural mechanism of Color Vision...., Springer Science+Business Media 2002, ISBN 978-1-4419-5291-2

Dear Muhammad,

Are you asking the difference between paper vs. software tests or between online vs. offline software tests? The most important point is if they are validated or not. Paper tests of Ishihara Tests and/or Farnsworth-Munsell Test are validated and require specific viewing conditions. If you do not keep the viewing conditions, this may result in misdiagnosing or over-diagnosing. If you want to use software versions, the key point is validation - most of the online tests are not validated.

Ethambutol has a toxic effect on the retinal ganglion cells.  That damage often can manifest as a blue-yellow loss early on but as the damage progresses one can also observe red-green defects.  e.g. see Polak BC, Leys M, van Lith GH. Blue-yellow colour vision changes as early symptoms of ethambutol oculotoxicity. Ophthalmologica. 1985;191(4):223-6.

The mechanism is excitotoxic damage.  See Heng JE, Vorwerk CK, Lessell E, Zurakowski D, Levin LA, Dreyer EB. Ethambutol is toxic to retinal ganglion cells via an excitotoxic pathway. Invest Ophthalmol Vis Sci. 1999; 40(1):190-6.

There are two valid answers here. The first point is that you need to be specific on the type of RGB values in question. If they are just monitor drive values the calculation is specifically for that screen and it has no wider significance. The basic properties of the  r,g plane of the  monitor colour space are however sufficient to reveal the R:G:B ratio that defines your Hue. The problem of course is where is your b value  coming from?. Given that r+g+b=1, then r =1 represents pure red gun colour etc. It follows that pure B must be at r=g=0 i.e.at the origin in the r,g diagram. likewise the b value is calculated by b = 1 - r - g and the ratio r:g:b then specifies your Hue.

The second answer is device independent. There is a widely recognized  colour space used in imaging known as sRGB space where the R,G and B hues have an unambiguous device independent definition and the same derivation applies. The ultimately precise r:g:b ratio for any given screen colour may also be back calculated from measured CIE XYZ co-ordinates by matrix multiplication.

We have worked a bit with the hue-100 test for determination of color discrimination. The subject is simply asked to arrange small colored patches according to their hue. Note that the choise of illuminant is quite important. Just as the choise of monitor is important for these kinds of test on a computer or tablet.

Regards

Anders

Hi Daniel,

A point in the LMS space represents the photoreceptor response to a given light stimuli. It means that the response depends on the spectral sensitivities of the cones, which may vary from observer to observer.  In other words there is not "the" LMS space, but different possible LMS spaces associated with a given set of cone sensitivities.  The book of Mark Fairchild is good reference to find out how to compute the LMS values from XYZ and from spectral data, given a set of cone sensitivities. You may check also "Color Science: Concepts and Methods, Quantitative Data and Formulae" by Wyszecki and Stiles, or "Digital color imaging handbook, Chp1",  or other  colorimetry/color science book.

Check also the spectral sensitivities for the L  and M cones you may notice that they are very close, so it is expected that the L and M values are correlated (but not the same). Besides that, the photoreceptor response is just the first step in the process of perceiving color, where the chromatic adaptation mentioned by Volker plays an important role, with an opponent process stage where color is encoded in achromatic, Blue/Yellow, and Red/Green channels.

Finally, be careful how you visualize the final result of your experiment, specially make sure that the variations you're generating are inside the sRGB gamut,  and also keep in mid the transformation RGB to sRGB.

You should read the many papers published by G.H. Jacobs who, along with numerous colleagues, has performed behavioral tests on many different primate species (squirrel monkey, spider monkey, owl monkey, lemur). The methods sections of his papers give detailed descriptions of how the testing was performed. Also see his book entitled Comparative Color Vision.

You can use the nomogram found in:

Han DP and Thompson HS (1983) Nomograms for the Assessment of Farnsworth-Munsell 100-Hue Test Scores. Am Journal of Ophthalmology 95: 622–625

to determine if the values you obtained fall outside the normal ranges from their study.

There are a number of other articles that describe unilateral congenital loss:

Cox J. (1961) Unilateral color deficiency, congenital and acquired. J. opt. Soc. Am. 51, 992-999.

Sloan L. L. and Wollach L. (1948) A case of unilateral deuteranopia. J. opt. Soc. Am. 38, 502-509.

Graham C. H. and Hsia Y. (1959) Studies of color blindness: a unilaterally dichromatic subject. Proc. Nat. Acad. Sci. 15. 96-99.

MacLeod D. I. A. and Lennie P. (1974) A unilateral defect resembling deuteranopia. Mod.Prob.Ophthal.13, 130-134.

thanks Terrace Waggoner and Sentilvel.

hope it will be helpful to me.

"I have grayscale images of same scene but with different intensity values. I want to join them in such a way that difference in intensities can be observed through color."

The following MatLab code may be helpful for this problem:

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

% image1 and image2 are two matrices containing grayscale images

imageDiff = image1-image2;

h=surf(imageDiff);

set(h,'edgecolor','none');

axis off

view(0,90)

%you can use various predefined colormaps defined in MatLab or create your

%own

colormap(jet)

% set color scale if you don't like the MatLab default

% caxis([cmin cmax])

% shows color scale

colorbar

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

Dela E (CIE Lab):

0,0 … 0,5 no color difference

0,5 … 1,0 difference only perceivable for experienced observers

1,0 … 2,0 minimal color difference

2,0 … 4,0 perceivable color difference

4,0 … 5,0 significant color difference

larger than 5 different color

Literature e.g.,

R. F. Witzel, R. W. Burnham, and J. W. Onley. Threshold and suprathreshold perceptual color differences. J. Optical Society of America, 63:615{625, 1973. 14

William David Wright. A re-determination of the trichromatic coeffients of the spectral colours. Transactions of the Optical Society, 30:141{164, 1928. 8

G. Wyszecki and G. H. Fielder. New color-matching ellipses. J. Optical Society of America, 61(9):1135{1152, 1971.

Is this actually a solvable problem ?? You are going from 3 to 4 channels. The information of the 4th channel is nowhere in your RGB ( only in assumptions about source materials ). This means to me that the only "new" information you get is something you yourself just added implicitly.

To me, this problem is close to: can I create a color a human sees from the 2color signal most other mammals have.

Or, at the extreme end: how do I get a color image from a b&w image.

Thank you!

Dear Roy, we implemented several experimental protocols in order to detect early signs of multiple sclerosis. For this purpose we used pC high-quality monitors in order to provide color visual stimuli to the patients of the Neurological Dept of the Umberto I° University Hospital of Roma Sapienza. At your request I can send you more details (giancarlo.filligoi@uniroma1.it ). See also.

FILLIGOI G.C., CAPITANIO L., ACCORNERO N. "Chromatic perception test: a computer based approach" Proc. IEEE/EMBS Int. Conf., Amsterdam (The Netherlands), 31 Oct - 3 Nov 1996

N.ACCORNERO, S. RINALDUZZI, M. FILIPPI, E. MILLEFIORINI, G.C. FILLIGOI, L. CAPITANIO "Computerized color perimetry in multiple sclerosis" Multiple Sclerosis, 1998, 4, 79-84

Many thanks Carl, this was really useful!

I will check the bibliography that you suggest!

Thank you!

I am not convinced that the curves have been "calculated".

Otherwise, they would specify the various responses of the cones at various wavelength for human beings.

It seems to me that the curves have been manually tuned to fit our perceptions.

I am not sure that I will be any help with this but I will have try because I am interested in the general idea of a tetrahedral colour space for birds and would like to ask a question at the end. I think you may have a very tricky problem! If you have transformed the values into x,y, and z co-ordinates presumably you can derive values for e.g. u/v=0, s=0, m=0, l=max. which would be an 'ideal vertex'. A possible set up for would be

u/v: x=1, y=0, z=0

s: x=-1, y=0, z=0

m: x=0, y=sqroot 2, z=1

l: x=0, y=sqroot 2, z=-1

However, although it is possible that pure u/v and pure l excitation can occur using light (rather than say electrodes) almost certainly no spectrum will give pure s or m excitation so the bird's colour space will be within the tetrahedron above but quite a lot smaller in two axes. How much smaller should be possible to work out from all the sensitivity curves but would I guess be a slog.

You might want to 'restretch' the reduced space to make it a tetrahedron again for display purposes and this could be considered legitimate since there is no 'metric' to subjective colour. Whether it would be worth the effort of recalibrating all your values I don't know. You might want a dedicated programme to calculate all that!

My next thought is that presumably the birds could tell the spiders apart on the basis of shade as well as hue, in which case don't you need a four dimensional space anyway? As I understand it a tetrahedral space will only cover the hues, all normalised for relative intensity? The human ability to distinguish by colour would I think be based on a three dimensional space.

And finally my question is whether we know that birds actually use their four receptors to generate a four dimensional hue/shade space, as the receptors would theoretically allow them. Is it possible that the pathways that extract hue and shade 'throw away' a dimension and just construct a triangular space like ours, but spanning a wider range up to UV? Since the bird's brain is so much smaller than ours it may not have space for analysing in four dimensions and may simply use the increased number of receptors as a short cut to getting more colours with less processing. They might even use the four receptors to create a linear hue sequence using a system of analysis closer to the cochlear, where pitch is assigned to wherever maximum excitation is. OK, we perceive harmonies, but most of us are not that good at discriminating them. I don't know the literature here so may be way off tack but I would love to know if this has been established somehow.

Codons 180, 277, and 285, exon 3 on q arm of X-chromosome seems to be important, Asenjo, A.B., Rim, J. and Oprian, D.D., “Molecular determinants of human red/green color discrimination,” Neuron 12, 1131-1138 (1994).

Sharpe, L.T., Stockman, A., Jägle, H. and Nathans, J., “Opsin genes, cone photopigments, color vision, and color blindness,” in Color Vision: From Genes to Perception, Gegenfurtner, K. R., and Sharpe, L.T., eds. (Cambridge University Press, 1999), 3–51.

Yokoyama, S. and Radlwimmer, F.B., “The ‘Five-Sites’ Rule and the evolution of red and green color vision in mammals,” Molecular Biology and Evolution 15, 560–567 (1998).

particularly codon 180…..

Wasserman, L.M., Szeszel, M.K. and Jameson, K.A., “Long-Range Polymerase Chain Reaction Analysis for Specifying Photopigment Opsin Gene Polymorphisms,” Technical Report Series # MBS 09-07. Institute for Mathematical Behavioral Sciences University of California at Irvine, Irvine, CA, USA (2009). Available on-line at http://www.imbs.uci.edu/tr/abs/2009/mbs_09-07.pdf

Also see: Jameson KA, Highnote S M., Wasserman LM Richer color experience in observers with multiple photopigment opsin genes. Psychonomic Bulletin & Review (2001), 8, 244-261

and http://aris.ss.uci.edu/~kjameson/Jameson2009GlimpsejournalTetrachromacyAndSupplement.pdf

So because of this overlap in the olfaction and visual transaction pathways, shouldn't these two cues that are likely to interact, be studied simultaneously?

Do you mean ambient light during night-time? Twilight? Moonlight? You will need an spectroradiometer, that has an absolute spectral irradiance calibration. What do you use for measuring the spectral reflectance? If you are using an array spectrometer like Ocean Optics, Avantes, StellarNet or similar, it might be possible to measure spectral irradiance with it, if it is sensitive enough, you get a cosine diffuser for it, and get it calibrated. In the range 300-800 nm an array spectrometer should be good enough if it is sensitive enough, but you will need to check with a UV blocking filter that stray light near 300 nm is not a problem under your conditions. If you do not need fine spectral resolution then a wide slit will help increase the signal. If you need to measure extremely low levels of light, then you might need an array spectrometer with a cooled array, like those used for measuring fluorescence. (A cooled array, has lower electrical noise, so it allows the use of longer integration times.) You need to give us some idea of the irradiance level you want to measure... that would let us to give you a better answer.

Of course, if you will be measuring very low light levels, you will need to be doubly careful when measuring the spectra: you will need to avoid shading the sensor, as usual, but also to be careful to avoid that the laptop screen, or any other light source you use affects your measurements.

Psychology of Colors

There have been other proposed roles, in addition to the various functional roles that Quirin mentioned above, including facilitation of an optimum mosaic distribution (to presumably improve photon catch) and changes during development, see e.g. http://jcs.biologists.org/content/s3-98/42/189.full.pdf+html

These are references you have likely seen but good overviews of some of the proposed functional roles of the cone types including some experimental studies:

http://www.vision-in-cichlids.com/why_double_cones.htm

Cheers-cjb

There are quite a few:

Intraindividual comparison of color contrast sensitivity in patients with clear and blue-light-filtering intraocular lenses. Schmidinger G, Menapace R, Pieh S. J Cataract Refract Surg. 2008 May;34(5):769-73. doi: 10.1016/j.jcrs.2007.12.034.

Color discrimination by patients with different types of light-filtering intraocular lenses. Ao M, Chen X, Huang C, Li X, Hou Z, Chen X, Zhang C, Wang W. J Cataract Refract Surg. 2010 Mar;36(3):389-95. doi: 10.1016/j.jcrs.2009.09.038.

Intraindividual comparison of color perception and contrast sensitivity with and without a blue light-filtering intraocular lens. Eberhard R, Roberti P, Prünte C. Eur J Ophthalmol. 2009 Mar-Apr;19(2):235-9.

Color perception with AcrySof natural and AcrySof single-piece intraocular lenses under photopic and mesopic conditions. Cionni RJ, Tsai JH. J Cataract Refract Surg. 2006 Feb;32(2):236-42.

Blue-light filtering intraocular lens in patients with diabetes: contrast sensitivity and chromatic discrimination. Rodríguez-Galietero A, Montés-Micó R, Muñoz G, Albarrán-Diego C. J Cataract Refract Surg. 2005 Nov;31(11):2088-92.

Effect of a yellow intraocular lens on scotopic vision, glare disability, and blue color perception. Muftuoglu O, Karel F, Duman R. J Cataract Refract Surg. 2007 Apr;33(4):658-66.

Comparison of contrast sensitivity and color discrimination after clear and yellow intraocular lens implantation. Rodríguez-Galietero A, Montés-Micó R, Muñoz G, Albarrán-Diego C. J Cataract Refract Surg. 2005 Sep;31(9):1736-40.

Comparison of color perception after tinted blue light-filtering and clear ultraviolet-filtering intraocular lens implantation. Khokhar SK, Jindal A, Agarwal T, Panda A. J Cataract Refract Surg. 2011 Sep;37(9):1598-604. doi: 10.1016/j.jcrs.2011.03.044. Epub 2011 Jul 12

just to name a few...

One reason might be how much the colour saturation of the bleached pulp contains after bleaching with different chemicals. The objective of bleaching is to get white pulp without colour saturation.