Birds Can Discriminate Human Metameric Colors

Table of contents

• Introduction
• Materials and Methods
• Results
• Discussion

Introduction

Two surfaces that are matched in color, for a given observer under given illumination, are metameric if their spectral reflectances differ from each other within the visible spectral range. The metameric color match can be broken by the change of either the illumination (illumination metamerism) or the observer (observer metamerism). The question is if the birds color matching functions differ significantly from human ones to get high color mismatch for the ordinary painted surfaces.

In their ordinary life humans are hardly concerned for metamerism and intuitively take the specimens identical in color for identical in general. So, the main problem of the experiments, after the metameric specimens were painted, was to overcome this "intuition" in experimenters themselves. It looks as if we had more problems with metamerism than our birds.

To recognize close colors by memory is evidently more difficult for humans than to detect the difference in color between the surfaces presented side by side. When looking at our metameric paints one can hardly suppose that it is ever possible to remember the difference between the specimens to recognize them by color. Therefore, it was decided to supplement the task of color recognition with the easier task of form discrimination and to make something like pseudoisochromatic plates for the birds, with metameric colors presented side by side. For example, to discriminate orientations of the grid pattern made from alternating stripes of metameric colors. Moreover, in the form discrimination experiments it is possible to further facilitate the training procedure by using more distinct grids, and to use the metameric grids only in the test.

Contrary to the expectation, it turned out that it was more difficult to train the birds to discriminate grid orientation than to recognize colors metameric for human subjects, while the last task took as much time as simple color discrimination.

Materials and Methods

Subjects
The experiments were conducted in the mixed woods of the Moscow and Vladimir districts on breeding pairs of hollow-dwelling small passerine birds: pied flycatcher Muscicapa hypoleuca (three nests), great tit Parus major (two nests) and redstart Phoenicurus phoenicurus (one nest). We used the ability of hollow-dwelling birds to find their nest with nestlings by the features of the surroundings that could be changed in the experiment.

Procedure and Apparatus
In spring, nesting-boxes were hung up on trees where the birds nested. When the nestlings hatched, the nesting-box with the nest was gradually lowered to a height convenient for experiments. Meanwhile, the birds were accustomed to the presence of the experimenters. After that the birds were trained to find the true entrance to the nest by special marks as conditioning stimuli - see Procedure in Pictures for details (87K).
Experiments were conducted during day-light hours (with breaks) over a period of a few days.

Metameric paints
Two sets, each of 10 color papers were painted with mixtures of different blue and yellow mineral paints. According to their compositions these sets will be referred to as FePb and CoCd. Despite the clear difference between these sets in spectral reflectances (see figure), all papers had close green colors for human observer under daylight illuminations.
To eliminate the influence of ultra-violet light both types of papers were made equally "non-reflective" in this end of the spectrum (about 6% reflectance for wavelengths below 380 nm).

These sets are metameric for humans under daylight illumination in the sense of
1) the light, sent to the eye from all papers, causes close calculated quantum catches by each cone type of the human retina, sensitive to long- (L), medium- (M) and short-wavelength (S);
2) in the CIE standard colorimetric system all papers have close tristimulus values, their colors occupying a compact locus in the color space;
3) when observing the papers under daylight illumination humans failed to identify the set by color of the paper;
4) human observers could select several pairs of papers from the sets that matched in color.

Stimuli
The stimuli were 10-12 cm squares with a central hole for the entrance. For the color recognition experiments stimuli were cut from the different painted papers. For the form discrimination experiments stimuli were grid patterns made from alternating 1cm wide stripes of two different metameric colors. As far as the experimenters failed to agree about exactly which specimens of painted papers matched in color, each stripe was composed of small patches (5x5 mm) of different papers from the corresponding set. This made the stripes invisible to all observers.
Stimuli with clear grid pattern (for humans) were made from gray/green, blue/green and black/white pairs of color.

Illumination
When attached to the entrances, the stimuli were in equal lighting conditions. The experiments were made with such nests and in such periods of day when the front panel of the experimental box was illuminated by diffuse daylight with spectra close to the Standard Illuminant D65.

Training and Testing
A non-trained bird usually shows one or another position habit in choosing the entrance. It practically do not looks at the stimuli and flies straight into the closest entrance from the wood. Usually, one can recognize that the bird has learned the task by changes in its very behavior. Now, it first settled on the perch and examines the entrances with the attached stimuli, and only from there makes the choice.
Once it become clear that a bird had learned to fly to the necessary stimulus despite the changes in its position, the last training session(s) was carried out to score the bird's final performance. The pair of stimuli for this session was replaced with a new one (made from the same papers) to exclude cues other than coloration.
Testing was conducted with new pairs of stimuli, sometimes different from the initial ones in color. Several consecutive test sessions used to be made with various stimuli to get more reliable results.

Statistics
Birds feed nestlings in a very irregular manner. Also many environmental factors affected the duration of experimental sessions. So, the number of trials in sessions varied. Accumulated scores were analysed with the binomial test for each session of a certain length.

Results

• Metameric colors as seen by human observer
  
  • Observer metamerism
    • Appreciable individual differences in matches of our FePb and CoCd colors were revealed in humans. The largest difference of opinions appeared between students and professors. This discrepancy can be accounted for more yellow ocular media in elder observers. The accord turned out to be better when the young observers examined the metameric specimens through yellow glasses.
    • Most people rejected the matches calculated for the CIE Standard Colorimetric Observer and the CIE Standard Illuminant D65, the calculations being in closer agreement with the matches made by young people.
    • Color matches of FePb and CoCd papers proved to be broken down by change in viewing distance. The property was discovered accidentally during the experiments with birds when the appearance of the attached metameric stimuli changed while moving away from the experimental box. This color shift could be accounted for the shift in the visual angle and coincided in direction in the color space with change of the observer from the CIE 1931 Standard Colorimetric Observer (2^o) to the CIE 1964 Supplementary Standard Colorimetric Observer (10^o).
  • Illuminant metamerism
    Metameric matches of our FePb and CoCd paints proved to be very sensitive to changes in the spectral content of the illumination. Usually it was sufficient to turn the papers away from the window to the tinted wall to see mismatch. With changes of illumination from daylight to incandescent light, the colors of metameric sets FePb and CoCd underwent significant changes: the first ones became more saturated green, while the last ones lost saturation and even turned slightly lilac.
    Calculated color coordinates of our papers under the two illuminants are plotted on the CIE uniform chromaticity diagram and are shown in figure at the left. As it can be seen from an enlarged insert, under daylight illumination (D65) both FePb papers (blue points) and CoCd papers (red points) have chromaticities intermixed within the limited area. After change of the illumination (arrow) colors of the sets are parted and occupy separate areas, papers CoCd being less saturated than FePb (lie closer to the illuminant color, indicated by the cross A).
    The picture below shows how our stimuli (with grid patterns invisible for human observer under daylight illumination) look under incandescent light.

• Metameric colors as seen by camera and scanner
  "Systems of color vision" in cameras and scanners are different from that in humans. So, being photographed, metameric specimens (matched in color for human observers) take clearly different hues on photographs and slides or at the screen in case of digital cameras.
  An example of snapshot made with a Creative Video Blaster WebCam Go under daylight illumination is shown above. This camera posesses an appreciable sensitivity in the far red end. As a result, it reproduces CoCd stripes much lighter than FePb ones - see reflectance curves in Materials and Methods.

  Besides, scanners use their own light source. So, in this case we meet both observer metamerism and illuminant metamerism. In other words, the color rendering of both camera and scanner were wrong in the case of our color papers. This feature of scanned images of the metameric paints created troubles in preparing this poster, for we failed to escape the process of scanning. The examples of raw scanned metameric grids (without subsequent tonal correction in graphical editors) are given below.
  

  

• Metameric colors as seen by birds
  At different nests the experiments were organized in a variety of ways. Moreover, birds differ in their ability to learn and display different behavior in experiments. So, experimental results for the different birds are described separately.
  • In the form discrimination experiments, four birds showed the ability to discriminate vertical metameric grids from horizontal ones. However for some birds the discrimination by orientation (both metameric grids and grids made from some other colors) proved difficult.
    See Individual Results in Form Discrimination experiments for details (26K).

  • In color recognition experiments, two flycatchers and one great tit successfully learned to discriminate papers of the FePb set from the matched in color papers of the CoCd set.
    See Individual Results in Color Recognition experiments for details (30K).

Discussion

Different lights that look similar for human observer are not a rare case. Metameric lights are necessary tool in colorimetry. We all have met with an illuminant metamerism when trying to match clothes by artificial light at shops. Now it appears that it is not so difficult to make high metameric colors from ordinary mineral paints.

Color mismatch for such metameric pairs permits to detect

• small changes in illuminant spectra owing to illuminant metamerism, or
• small individual differences in color matching functions owing to observer metamerism.

Acknowledgements

Authors are indebted to cohorts of unnamed students from Orekhovo-Zuevo Pedagogical Institute participating in experiments as observers during their practical works in zoology and to Vitaliy Vodyanov for photos.

Literature

4. V.V.Maximov, E.N.Derim-Oglu. Birds can discriminate human metameric colours. Perception, 25, Suppl., 98, 1996.

Back to IITP Laboratory 8 Activity Page

Back to IITP Electronic Posters Page

Back to IITP Projects Page ( IITP russian Projects Page )

Last Update: 20 February, 2001