Flashcards in Chapter 5 Perceiving Color Pg. 153 Deck (69):
Portion of the electromagnetic spectrum in the range of about 400-700nm; within this range, people with normal vision perceive differences in wavelength as differences in color
Spectral power distribution (SPD)
Intensity (power) of a light at each wavelength in the visible spectrum
Light that consists of more than one wavelength
*White light = heterochromatic light that contains wavelengths from across the entire visible spectrum and has no really dominant wavelengths
Light that consists of only one wavelength-> vertical spike
Proportion of light that a surface reflects at each wavelength
*Reflect and absorb light
Quality usually referred to as "color"- that is blue, green, yellow, red and so on, the perceptual characteristic most closely associated with the wavelength of light
*Wavelength, vary wavelength peak
*Strongest among CHROM processes (RGBY)
Vividness (or purity or richness) of a hue
*Purity, vary spectral purity
*Strength of CHROM vs. ACHROM
*Amount of light
*Vary total energy intensity
*Strength of ACHROM white vs. black
2-D depiction in which hue varies around the circumference and saturation varies along any radius
only be created by mixing together two or more wavelengths
3-D depiction in which hue varies around the circumference, saturation varies, along any radius and brightness varies vertically
*Brightnesses increasing as you move up
*Decrease (shrinking) radius -> smaller range -> either increase or decrease brightness from mid level so colors get very dim or very bright , they become less vivid
Subtractive color mixture
Mixture of different-colored substances, called "subtractive" because the light reflected from the mixture has certain wavelength subtracted (absorbed) by each substance in mixture
*Color printing, inkjet printers
Additive color mixture
Mixture of different-colored lights, called "additive" because the perceive color of the mixture is the result of adding together all the wavelengths in all the lights in the mixture
Pairs of colors that combine in equal proportion to yield a shade of gray
Any three colors that can be combined in different proportions to produce a range of other colors (magenta, cyan, yellow/red, green, blue)
*TVs and computer monitors
Trichromatic color representation
Light evokes different responses from three different types of cone photoreceptors in retina
Opponent color representation
Responses from the cones are combined and processed by a subset of retinal ganglion cells and by color-selective neurons in the brain
*4 basic colors can be divided into 2 pairs of complementary colors: red, green, blue, yellow
Metameric color-matching experiments
Whether the right mixture of 3 monochromatic primary colors is perceived as identical in color to some other monochromatic light
*Observers adjusted amounts of three wavelengths in a comparison field to match a test field of one wavelength
*420nm, 560nm, 640nm
Any 2 stimuli that are physically different but are perceived as identical
*Adjust intensities so that the additive color mixture in comparison patch= color as test patch-> metameric color match
Spectral sensitivity function
Probability that a cone's photopigment will absorb a photon of light of any given wavelength
Principle of univariance
With regard to cones, the principle that absorption of a photon of light of any given wavelength
*The strength of response generated by a cone when it traduces light depends only on the amount of light transduced, no on the wavelength of the light
*A M-cone's response o dim 543nm light and to a bright 450nm light could be identical, with right choice of intensities
If you had only one type of cone (or only rods)
A person with normal vision will perceive two lights of these wavelengths as the same if their intensities are equal
*Under equal illumination, green might look brighter than red/blue because the relative sensitivity of rods is higher in green than red/blue-but they will not look different in color
If you had only two types of cones
A person with 2 types or cones cannot adjust the intensity of a single arbitrary comparison light to match the color of a test light with different wavelength
*People with 2 types of cones match a monochromatic test light of any wavelength if they have a mixture of 2 monochromatic comparison lights to work with instead of just 1
*Only 2 primary colors needed to match any other color
Physiological evidence for Trichromacy
Retinal densitometry: produces high-resolution images of retina -> mosaic of 3 types of cones
Photocurrent measurements: directly measure an individual cone's response to light
Experimental technique in which the person cancels out any perception of a particular color in test light by adding light of complementary color
*Adding blue to cancel out yellowness
*Adding red to cancel out greeness
*"Unique blue" at zero between red and green and no yellow
*"Perceive green" between "unique blue" and "unique yellow"
*No trully unique red
*Amount of a physical light needed to cancel a complementary hue percept is a measure of the strength of that original complementary hue percept
Physiological evidence for opponency
*Introspection/ hue cancellation
*Measurements of neurons in lateral geniculate nucleus that also responded to color in opponent fashion
*Neural circuits supporting red-green, blue-yellow
*3 types of cones' nerve impulse-> bipolar, horizontal + amacrine -> excitatory+ inhibitory inputs to retinal ganglion cells -> 4 different types of "opponent color circuits"
*S= Short wavelength/ bluish-greenish; M=Medium wavelength/greenish-yellowish; L=Long wavelength/yellowish-reddish
*+S-ML circuit: RGC fires above baseline in response to short-wavelength light and below baseline rate in response to medium and long wavelength; respond oppositely to blue and yellow
*+ML-S circuit: Respond oppositely to blue and yellow
*+L-M Circuit: Respond oppositely to red and green
*+M-L Circuit: Respond oppositely to red and green
*Color-selective neurons: RGCs, LGN cells, cortical cells have RFs that produce more elaborate patterns of response
*V1: single opponent center-surround RF
*Carry info about wavelength of light within uniformly colored regions of visual scene but don't provide much info about color edges, locations, where adjacent regions are illuminated by different wavelengths
*Double opponent center-surround RF: COLOR EDGES
Photopigment molecule's loss of ability to absorb light for a period after undergoing photisomerization
a kind of photopigment bleaching results from exposure to relatively intense light consisting of a narrow range of wavelengths-> color afterimages
*Not sensitive to M-cone, and look at white surface, L-cone is stronger than M-cone, +L-M -> red (opposed from green)
Tendency to see a surface as having the same color under illumination by lights with very different spectral power distributions
*Our perception of color of surfaces corresponds to estimated reflectance of each surface, not to SPD of reflected light
*Fail if the illuminating light consists of only a narrow range of wavelengths
*Fail if just one surface is seen against a black and empty background; no way for system to know whether distribution of wavelengths is due to reflectance of a colored surface illuminated by a perfectly white light or perfectly white surface illuminated by colored light
Theory: Visual system automatically determines number of each wavelength reflected from all surfaces on average -> estimate of SPD f illuminant -> discounting illuminant
See a surface as having same lightness under illumination by very different amounts of light
*Ratio principle: perceived lightness of a region is based on relative amount reflected from region and is surround
See everything in shades of gray
Must rely on rod vision all the time
*No color; hypersensitive to light, low acuity
Less frequent than rod monochromy; have both rods and cones, but only one type of cone, which can be either S-, M- or L-cones
*Lack color vision
*Different wavelengths- different shades of gray
More common-but rare overalls; just one of three types of cone is missing
*can discriminate between colors that a rod monochromat would see as equivalent
*confuse some colors that a trichromat could tell apart
*Person who needs only two wavelengths to match any color, confuses many hues seen by normal trichromats (has only 2 cone pigments)
Ishihara color vision test
a test using configurations of multicolored disks with embedded symbols; the symbols can be seen by people with normal color vision but not by people with particular color vision deficiencies
Loss of vision caused by brain damage
Additive color mixtures
Digital color video displays
Use additive mixtures of 3 primary colors at each location on the screen to produce nearly the full gamut of colors you can see
*Taking advantage of ability of human eye-> distinguish dots that are sufficiently small and close together
picture elements, 3 subelements -> emit the light of one of three primary colors- red, green, blue
Digital color printing
*Applying tiny droplets of different color inks to a material
*Cyan, magenta, yellow, black
*Location and thickness -> light absorbs
*Increase thickness, increase photo absorb
*Subtractive effects when color on top of others
Ability to see differences between lights of different wavelengths
What does a single photoreceptor type encode?
One receptor type cannot lead to color vision because:
*Absorption of a photon causes the same effect in the cone, no matter what the wavelength is, because every absorbed photon causes the same isomerization of the photopigment molecule
*This is termed the principle of univariance
*There is no "memory" for the wavelength of an absorbed photon
*Any two wavelengths can cause the same overall response by adjusting their relative intensities
*Thus, there is no way to discriminate changes of wavelength from changes of intensity and, therefore, no color vision (rod vision)
One non-spectrally opponent retinal output
Parasol ganglion cells
*Sums L&M in both center & surround, both - and + center types
*Could this be the black/white perceptual opponent process
Two spectrally-opponent retinal outputs
Midget ganglion cells:
*Opposing M and L cones (M-L, L-M)
*Could this be the red/green perceptual opponent process?
Small bistratified ganglion cells:
*S cones opposed by L+ M
*Could this be the blue/yellow perceptual opponent process?
What does cortex need to do to produce our actual opponent-hue perceptions?
1. Add S-cone input to red-green opponent hue response, to produce short-wavelength red.
2. Shift cone weights in opponent processes so that bbalance points fall at observed unique-hue wavelength
3. Create a way of representing the thousands of binary hues we see.
V1 response to color
*Response in all hue directions of color circle
*Reflects combination of signals from paravocellular and koniocellular pathways
*Not biased perceptual unique-hue directions
*Could be part of basis for representation of full range of binary hues
*Show achromatic (black-white) response
How is mutual exclusivity of opponent processing maintained in cortex?
*Easy to see how LGN opponent cells can signal different hues by an increase or decrease from their baseline firing rate
*Cortical cells have near-zero baseline firing rate
*Cortical circuits create cells that only increase their firing rate to a narrow range of wavelengths
*One way this could be accomplished is if cortical circuits create cells that only respond to increases (not decreases) of firing of LGN-like cells.
Color processing in the Cortex
*There is no single module for color perception
-Cortical cells in V1, V2, V4 and beyond show differential response to wavelengths
-These cells usually also respond to forms and orientations
-Cortical cells that respond to color may also respond to brightness variations
-Some evidence of focal centers for color processing-blobs in V1, thin stripes in V2, globs past V4-but still controversial
What does Xiao study reveal about color in V2?
*There may be clusters ("modules") of color sensitive cells in or near thin stripes that display narrow color tuning and are arranged in a spatial pattern reflecting differences in hue.
*Responses of cortical color cells are one-directional, not opponent.
*The color patches indicate a dominant collective response of many cells but patches may overlap and show response to many different hues.
*Neither optical imaging nor the the averaged electrophysiological recordings presented here tell what individual cells are doing (similar to comparison of fMRI and single-cell physiology).
*No evidence of special status of unique hues.
Conway study shows:
*fMRI shows there are clumps ("globs") of color-processing cells in V4 and PIT (posterior infratemporal cortex)
*Single-unit electrophysiology shows that many cells in these areas have narrow hue tuning
*Preferred axes are most commonly in general direction of unique-hue axes, not LGN axes
*Globs may be first level we've seen in cortical processing that shows special status of unique hues
*Person who needs three wavelengths to match any color (has 3 normal cone pigments)
*Needs three wavelengths in different proportions than normal trichromat and has reduced color discrimination ability (has 2 normal and one shifted cone pigment)
*Need 3 wavelengths to match a 4th (thus, trichromatic), but in different proportions than normal trichromat
*Has reduced color discrimination ability, can range from mild to severe deficiency
*Modern color-vision genetics has revealed they have 2 normal and 1 shifted cone pigment
*In deuteranomaly (most common of all inherited color deficiencies), M-cone pigment is shifted toward L-cone pigment In protanomaly, L-cone pigment is shifted toward M-cone pigment.
*Result for both is that an L-M chromatic mechanism varies much less with wavelength change than for normals
*Person who needs only one wavelength to match any color, no color vision (has only 1 or 0 cone pigments). Cone monochromats typically have only S cones. Rod monochromats have only rods.
Sex-linked, "red-green" dichromacies
*Most common types; mostly males affected
*No red-green color discrimination (confuses color in normals' green/yellow/red range and in normals' aqua/blue/violet range)
*Deuteranope, missing cone M, remaining cones L and S
Protanope, missing cone L, remaining cones M and S
Autosomal, "blue-yellow" dichromacy
*Rare; males and females eqally affected
*No-blue-yellow color discrimination (confuses colors in normals' blue/aqua/green range and in normals' orange/yellow/red)
*Tritanope, miss cone S, remaining cones L and M
*Only rods and no functioning cones
*True color blindness: Ability to perceive only in white, gray and black tones
*Poor visual acuity
*Eyes very sensitive to bright light
*Have S-cones and rods only
*True monochromat at bright and dim light levels when only S-cones or only rods function
*At intermediate light levels, rods and S-cones can work together to permit wavelength discrimination
Basic human genetics
*23 pairs of chromosomes (2 autosomal pairs, 1 sex pair)
*Women have XX, men have XY (sex chromosomes)
*L and M genes on X chromosome (sex linked)
*S gene on autosome (not sex linked)
Action of genes on photopigment molecules
*Genes determine which amino acids end up in the protein part of the photopigment molecule (opsin)
*Different amino acids have different effects on the pigment's peak sensitivity, shifting it to longer or shorter wavelengths
*18 amino-acid dimorphisms in LM pigment molecule (one of two specific amino acids at each location) identified by number in the sequence of the opsin protein
Spectral proximity principle
*Separation of anomalous and normal pigment determines severity of color vision impairment
How can the brain set a consistent red-green balance across individuals with 30:1 differences in L:M ratios?
*We think that, over time, the brain "averages" every day's light exposure and adjusts its sensitivity so that there is a net balance of red and green. A similar sort of "normalization" is possible for blue-yellow but has not been demonstrated.
*Evidence for normalization comes from long-term adaptation studies with subjects exposed to red or green light for several hours a day
The job of the cortex: color and lightness constancy
*Cortex needs to transform information about the wavelength distribution of light coming from an object (as represented by the pattern of cone excitations) into perceptions of color/lightness of that object.
*The aim is to take the best inference about the state of the world. In this case, real objects don't change under different illumination, so cortex assumes color/lightness stay constant.
*The cortex will use anything it can get its hands on to make its inferences about the state of the world.
Why Color Constancy?
*In the real world, our views of objects and people always changing (amount and spectrum of illumination) yet the objects/people themselves don't change
*Visual system has evolved to provide representations of objects, not of actual pattern of light falling on the retina.
What are the factors (cues) that influence the interpretation of color or lightness made by the visual system?
*Shadows and perceived illumination gradients
*Depth and size
*Perceptual grouping (Gestalt principles)
*Edges and junctions (local configurations)