Color Perception

  • Vocabulary:

    • Light: Part of the electromagnetic spectrum between 400 and 700 nm

    • Pure white: The ‘color’ that contains all wavelengths, demonstrated via prisms

    • Prism: Used by Newton, when white light was shined onto this, light was refracted into short and long wavelengths, splitting the light into a rainbow. Also done in reverse, light was bent back together and formed while light again

    • Newton: Used a prism to refract white light into a rainbow, said “For the Rays to speak properly are not colored. In them there is nothing else than a certain power and disposition to stir up a sensation of this for the color”, meaning there is no color in the light, and that color is completely perceptual. There is no physical properties of color

    • S-cones: Detect blue light

    • M-cones: Detect green light

    • L-cones: Detect red light

    • Univariance problem: When only one cone is active, could be as a response to two different colors

    • Univariance: Measurement only varies along one dimension, if more than one environmental factor can result in a change in activation a system cannot distinguish between possible causes

    • Trichromatic theory: The activity of all three cones is used and compared to actually determine what color light is being perceived, since activity level depends on luminance this measure is comparative, activation is in ratios (division, not subtraction at low levels). The photoreceptor layer displays this

    • Protanopia: Red/green colorblindness, have short and medium cones, long cones act like medium comes. Can see blue well, but lots of yellowing of greens and reds

    • Deuteranopia: Medium cones behave like long cones, very similar to protanopia

    • Tritanopia: Blue-yellow colorblindness, medium cones and long cones work properly, short don’t

    • Monochromatopia: Two cones don’t work properly, vision is monochrome

    • Color spectra: The distribution of wavelengths striking our eye

    • Metamers: Produce the same perceived color by many different color spectra, elicit the same cone responses so there will be no perceived difference in color

    • Color mixing experiment: The hypothesis is that all seen colors are some kind of combination of activity of blue, green, and red sensors. Light is shown of one color from one lamp, then light from 3 (blue, green, and red) light are shown to a single spot (combined), participants asked to adjust intensity of the 3 color lights to match the single light, found that every color could be made with the 3 lamps. Was shown that using only two lamps (dichromatic theory) wouldn’t make all light colors

    • Pigments: Absorb some wavelengths more than others, such as those in blue paint absorb shorter wavelengths better

    • Additive color mixing: Mixture of different color lights (red, green, blue), subtracts the number of colors absorbed and adds the number reflected

    • Subtractive color mixing: Mixing pigments together so that different pigments absorb different wavelengths (cyan, yellow, magenta), subtracts the number of colors that are reflected and adds the number absorbed

    • Opponent process theory: States that there are 4 primary colors (red, green, blue, yellow), with every color being along a continuum between red/green, blue/yellow, and black/white. Can be visualized on a color wheel. The ganglion cell layer displays this

    • Color opponent neurons: There are ganglion cells that are inhibitory like on-center off-center to red/green and yellow/blue, some wavelengths start as excitatory signals and go through cells that flip them to inhibitory before refeeding into cells. Example red will excite it and green will inhibit it

    • Dual process theory: Both trichromatic theory and opponent processing theory are correct, trichromatic cones feed into opponent process brain mechanisms

    • Basic color words: Describe colors, need to be common, not object/substances, and not compound (blueish-green) words

    • English basic color words: Red, orange, yellow, green, blue, purple, pink, black, brown, white, gray

    • Color constancy: Seeing a surface as the same color, irrespective of the color of light it is in

  • Marr’s Levels applied to basic color vision:

    • Computational Theory: The goal is to perceive the wavelengths from reflective properties of objects and stimuli in a surrounding environment

    • Algorithm Representation: Average activity of wavelength sensors (3 of them for RGB)

    • Hardware Implementation: The photoreceptors (cones) detect various wavelength, the CNS interprets activation levels

  • A certain cone will activate for many different wavelengths, and can produce a bell-shaped kind of graph

  • Evolutionarily, there used to only be two cones, short and long. Recent evolution evolved the medium come from the long cone, making it so that the medium and the long share similar activation levels. Humans and old-world primates have 3 cones, new-world primates only have 2. Now visual experience is more rich and descriptive. With only 2, some colors wouldn’t be able to be distinguished in the way they can with 3. This is what happens with color blindness.

  • Men tend to get colorblindness more frequently than women as it is an X-linked recessive trait (males more likely to get recessive copy, females need two). Can only occur in one eye sometimes. Can be evaluated by ability to distinguish between colors, like red and green (think about those dot things with numbers)

  • Challenges to Trichromatic theory:

    • Why don’t colors appear to be a combination of red/green or blue/yellow if there are colors that are blue/red, red/yellow, blue/green, green/yellow?

  • Opponent process theory explains color afterimages (residual color is the opposite of what was seen, across on a wheel/scale). It also explains why yellow doesn’t look half red and half green (on a separate scale/dimension/spectrum, on the red/green it is half way)

  • Internal experiences of color between people are likely similar, as if they were randomly shuffled the relationships between color (red is like orange) wouldn’t be preserved. This, however, means that perception of color can be totally shifted around (reverse, blue=yellow and purple=orange). Also, with fully saturated colors blue is ‘darker’ than yellow and has more contrast with while, meaning people cannot have yellow/blue swapped. This doesn’t apply to red/green as full saturation has similar contrast with red and green, meaning inversion could occur for those colors. This may be occurring with those who have both protanopia and deuteranopia (L act like M and M act like L), for males 1.3% and 1.2%, or 16/10,000 for both, though it has never once been observed

  • Cultures with only two color words have light/dark, if there is a third it tends to be red, the 4th is more variable but tends to be yellow, and after that it is highly variable. For example, the Dani use Mola for light/warm colors and mili for dark/cool colors

  • How the color of a surface is perceived is based on the light coming into our cones which varies based on the chemical properties of the surface pigment absorbing/reflecting, and spectral properties or which wavelengths reach our eyes and cones based on that. O=S*I, where O is observed color, S is surface reflectance, and I is the illuminant. If you only have O, the univariance problem emerges because infinite possibilities could multiply to that value. For purposes of color constancy, we care more about the color of the surface than the color of the light, because knowing that allows us to tell objects apart and determine properties about it (like if a fruit is rotten)

  • The brain keeps color constancy even under different lights by observing that if many objects are the same skewed color in one direction, assuming it is due to the light and not the objects. Familiarity with a location will also allow for previous knowledge that things are skewed. For a new space, if you know you are wearing a blue shirt and it looks green in a new space, you can infer that the light in the area is yellow. This is easier to assume then that your shift has spontaneously changed colors. There are assumptions about the light source and about surfaces that won’t spontaneously change

  • The brain makes out shadows by detecting areas of similar color but lower luminance grouped in an area, though if there is a shift in color it might be seen as a new object

  • People perceive the same color categories regardless of language/culture. Evidence by Dani (warm dark and cool light), Dani don’t see particular colors as more similar than English speakers. Green/blue boundary for English people was also seen by Dani, despite them using the same word.

  • To create a wiring diagram with blue on and yellow off

    • (S)(M)(L)

    • |       \   /

    • \       (-)

    •  \      / 

    • (B+/Y-)