colour reading

• Luminance


  Difference in lighting; e.g. black-and-white TV
• Colour conveys information beyond luminance
• Colour vision has evolved over millions of years through evolutionary pressures and natural selection
• Nature of light


  • Light is packaged into photons; the quantity of photons determines luminance/intensity
  • Light is a type of electromagnetic radiation occupying a tiny part of the spectrum
  • Spans a fraction of a nm - kms (gamma rays to radio waves)
  • Visible spectrum (to us) ranges from violet to red
• Seeing light


  • Snakes see infrared light to detect warm objects in the dark (we can't see infrared because we have warm blood)
  • UV light is visible to birds and insects for mate choice + detecting patterns in flowers (we can't see UV because it damages our long-living eyes; it's partially filtered out)
• ROYGBIV includes indigo because Newton wanted to show that there's 7 basic colours (like 7 basic musical notes; he was first to show light was a mix of different colours)
• The colour isn't out there; it's in your brain
• Monochromatic/1 cone vision


  To see, you need to detect light ⇒ photoreceptor
• Most of the Earth's surface reflects light around 550–560 nm, as chlorophyll reflects most light at that wavelength
• Principle of univariance


  Problem of monochromatic vision; if a lower WVL was twice as intense, it would register the same as the peak WVL; you can't differentiate between intensity and WVL because your photoreceptor can only change its response in magnitude vs stimulus is different in 2 ways
• Monobloc mixer tap


  Ratio of red : green tells us colour; adding the output of R + G cones gives us luminance
• We knew about trichromacy before being able to record individual cones from colour-matching experiments
• Colour-matching experiments


  1. Participant mixes different lights to match the test light

  2. For one light: only the intensity/luminance can be changed

  3. Two lights with different wavelengths: test light is sometimes matched

  4. Three lights: many different lights can be made
• The basis of colour TV: intensity of red, green and blue lights (these three colours correspond to retinal cone types) is changed to produce different colours
• Metamer


  Different combinations of wavelengths that appear the same despite physically being different (meaning we're technically colour-blind; non-discriminance; great for colour TV because you only need 3 lights instead of a light for every colour)
• Colour addition vs. subtraction

  • Adding lights gives you white/lighter colours
  • Adding those colours but in paint form gives you dark colours
  • Paint absorbs light
  • Subtractive primaries absorb a wavelength (e.g. yellow paint absorbs blue, magenta absorbs green, cyan absorbs red)
  • Continuing: yellow: -B primary/ magenta: -G primary/ cyan: - R primary
  • Mixing them together absorbs more and more light ⇒ black
• Dichromatic/2 cone vision


  • Colour vision is useful if 2 cone types for very different WVLs (e.g. 550 nm - 420 nm/yellow-blue)
  • Colour difference tells you orientation (up/down) + helps differentiation (e.g. between vegetation)
  • Simple two-cone systems are effective enough to be widespread (especially in mammals; dogs, cats, bulls, etc.)
  • Introducing a second cone type would mean getting rid of existing cones ⇒ acuity compromisation (more cones = fine detail better perceived because more cells to compare inputs)
  • This second cone should be sparsely distributed on the retina; responds to coarse detail (no blue foveal cones for acuity + short WVLs are always out of focus in the eye)
  • Chromatic aberration→why blue/purple neon signs are always a bit fuzzy; common defect in most visual systems (low blue cone distribution to preserve foveal acuity; short WVLs don't register clearly)
  • Second/shorter WVL cone type wouldn't be in fovea ⇒ no foveal colour vision BUT luminance/intensity information + coarse colour info overlay allows differentiation between blue and yellow light
• Trichromatic/3 cone vision


  • Before adding the 3rd cone, identify what tasks are obscured without this cone (e.g. finding red/yellow fruit from a background of dappled/mixed luminance foliage + detecting sexual readiness from faces or other parts in primates)
  • Red-green vision in primates has the advantage of obscuring the dappling (figure out that the foliage is all the same despite luminance difference) + makes the fruit stand out
  • 30 - 40 m.y.a: yellow cone split into blue and red cones for sensitivity to short, medium and long WVLs
  • Red and green cones are similar in sensitivity
  • Trichromatic vision is rare in the animal kingdom; only really seen in humans and other primates (but can have +3; birds, shrimp, etc.)
  • ~10% cones are blue with red cone majority; cone distribution in human retinas (lower acuity where S cones are)
• Colour opponency (cone activity comparison)

  • Comparing the ratio of 2 cones regardless of luminance would give the same colour in a range of conditions (e.g. cones responding proportionally the same across luminance)
  • Opponent coding of colour: Theory that color vision is based on the coding of pairs of opponent colors: red-green, blue-yellow, and black-white
  • Cone/colour signal ratios: luminance signal (R + G), red : green signal = R output ÷ G output (or vice versa; red : green ratio), blue-yellow signal = luminance signal ÷ B output
  • If luminance/intensity changes, colour stays constant (increased activity signals increased luminance but not colour if cone activity ratio is the same)
  • If WVL is changed, colour changes but not luminance
  • This encodes colour by comparing activity across cone types; dual cone activity causes opposite changes in neural activity (R:G increased by +R and decreased by +G)
  • Adaptation to a colour swings channel sensitivity to opponent colour
  • Ewald Hering: 4 'psychological pure primary colours'; red + green (no Y/B), yellow + blue (no R/G) ⇒ supported by the fact there's four opponent signals in the colour aftereffect
• Colour-opponent cells


  • Cells that oppose cone activity (decrease activity as red cones are excited + vice versa for green; G+/R- cell)
  • These can be found in macaque LGNs in the parvocellular layers
  • Koniocellular pathway: same thing with blue-yellow vision system
• Evidence for two-colour vision systems: genes & recordings from cells in LGN + beyond
• Blue-others


  Shared with dogs and cats (they don't need red-green system; see below)
• Red-green


  Found in primates + compares red and green cones for detecting fruit
• How we know the red–green system was developed for finding fruit

  1. Sumner & Mollon, 2000 + Lovell et al., 2005 found a monkey in tropical rainforests that likes a certain type of yellow fruit, measured WVL range of fruit + surrounding leaves, measured WVL sensitivity of red and green cones in monkey's retina, tuning of cone sensitivity curve was best for differentiating 'fruit' signal from 'leaf' signal; small changes in the cone response curves would make differentiation harder

  2. Lovell et al., 2005 found red-green system worked across luminance levels (e.g. shady/sunny days, weather changes, etc.)
• Colour blindness


  • Protanopia→protanopes have no red/long-wave cones
  • Deuteranopia→deuteranopes have no green/medium-wave cones
  • Tritanopia→tritanopes have no blue/short-wave cones (super rare)
  • Anomalous trichromats→trichromats with a genetic anomaly affecting a cone type's sensitivity have abnormal colour vision
  • Deuteranomaly→green cone's peak sensitivity is closer to red than normal (e.g. in a yellow colour-matching experiment, they would make the colour match with less red)
  • Protanomaly→red cone's peak sensitivity is closer to green than normal (e.g. in a yellow colour-matching experiment, they would make the colour match with more red than most people)
  • Genetically inherited + gender differences (more common in men)
  • Rod monochromacy→total lack of cones; incidence rate of 0.0003% in the population
  • Colour vision can be affected by diseases (e.g. diabetes) and other factors (e.g. doing drugs)
  • Dichromats are less susceptible to certain camouflages ⇒ they were recruited in WWII to see past camouflage
  • Asian/African monkeys are trichromats (with some colour-blind members); male American monkeys only have the primitive blue-yellow system, with half of females being trichromats
• Cortical processes


  • 3 cones that give us 3 different images of the world
  • Above images used to generate three different signals: luminance, red-green differences and blue-yellow differences
  • These retinal + LGN signals go to V1
  • Blobs→specialized regions in V1 that process stimulus colour/WVL surrounded by non-colour-processing cells; info is passed onto prestriate cortex ⇒ V4
  • V4 cells are selective to certain colours; some cells respond to stimulus colour than WVL
• Colour constancy


  • To know the 'true' colour of an object, you need to disregard lighting ⇒ brain looks at different objects all at once ⇒ e.g. if lighting is red, everything appears red ⇒ system disregards red
  • Things remain the same colour even across WVL changes because they are e.g. bluer/redder/greener than other objects
  • Edwin Land, 1983: lit a Mondrian pattern with red, green and blue light ⇒ a single patch changes colour over WVLs; doesn't if part of Mondrian pattern = = = demonstrates colour constancy, where an isolated patch of colour changes with wavelength but remains constant when part of a pattern (because there's nothing to compare against)
  • Single cells also respond to Mondrian experiment; WVL changes impacts V1 but not V4 (differentiates between surface colour and light wavelength)
• Cerebral achromatopsia


  • Loss of colour vision after damage to extrastriate cortex, even with retaining their cones (luminance differences detectable)
  • V4 lesion study (Cowey and Heywood, 1995) reveals this area is not solely responsible for colour vision
  • V8 is a small area adjacent to V4; activity could be tested with colour gratings (adjusted for luminance so that activity corresponds only to colour; compare response to luminance grating)
  • V1-3, 8 excited by colour grating but not V4
  • V8 is active even without actual present colour info; evidenced by activity + V4 inactivity during colour aftereffect
  • Consensus is roughly divided between V4 and more popularly V8