Lecture 6 - Chapter 7 Colour Vision Stimulus (what is colour?) The neural code for colour Theories of colour vision Disorders of colour vision Colour constancy Lightness constancy
Functions of colour vision: Detection of objects - ie. Picking berries Discrimination of objects - ie. Making sure you pick ripe berries.
Other Functions of colour vision: Grouping of objects - Gestalt grouping Recognition of familiar objects Tanaka & Presnell, 1999
Describing colour experience We can describe all the colours we can discriminate using just four words : red, yellow, green and blue. If presented with many colours and asked to describe them, they can’t do it if one of the 4 colours is omitted. So, red, yellow, blue and green are considered to be basic colours. The basic colours can be arranged in a circle, and the ordering matches the ordering of the colours in the visible spectrum.
Colour circle
Visible spectrum We can discriminate up to 200 hues across the length of the spectrum. More colours can be made by changing saturation, the amount of white. If you have more white in a colour, it is less saturated, and if you take white away it becomes more saturated.
Saturation more white is added as you move to the centre of the circle. Hue = chromatic colour, it changes as you move around the circle. There are about 1 million discriminable colours
Wavelength and colour perception
What makes an object appear coloured? Selective Reflection Reflectance is the percentage of light falling on an object that is reflected from it. Selective reflection is the reflection of different amounts of different wavelengths.
Selective Reflectance - Glacial ice
Selective transmission Some objects are not ‘solid’ and allow light to pass through. Some of these selectively transmit certain wavelengths. For example, translucent objects (clear liquids) transmit light
Types of Colour… Chromatic - blue, red, green etc. these are seen when an object selectively reflects or transmits some wavelengths. Achromatic - white, black, shades of grey. These are seen when all wavelengths are reflected or transmitted equally.
Theories of colour vision: 1. Trichromatic Theory (Young & Helmholtz) Colour matching experiments Any colour can be matched using 3 wavelengths
Colour vision depends on 3 receptors with different spectral sensitivities. The pattern of activity across 3 receptors codes colour
Physiological Support for Young/Helmholtz 1. Identification of three types of cone pigments (70 yrs later)
Physiological support for Young/Helmholtz 2. Can predict colour perception based on pattern of activity in cones.
Colour mixing Additive - Add lights Subtractive - Mixing paint
Mixing light - Additive mixing
Mixing paints - Subtractive mixing Blue pigment only reflects S + M + Yellow pigment reflects M and some L = All that is reflected by the combination is M We can only perceive wavelengths that reach the eye
Metamers If two combinations of wavelengths produce the same neural pattern, the observer will experience the same percept.
Are three receptors necessary? Colour vision is possible with only 2 types of receptors, but not with just one type. If you only had one type of receptor, then it would respond to light, but changing the wavelength would not change it’s response. One receptor type couldn’t signal the difference in wavelength, it would only signal a difference in intensity. This is called the principle of univariance. When there is only one receptor/pigment you can make any two wavelengths create the same response, by adjusting their intensities.
Principle of Univariance
Principle of Univariance
Principle of Univariance
Trichromats People with three cone types are called trichromats. People with only two cone types are called dichromats. Dichromats see colour, but not the full range experienced by trichromats. Monochromats have only one cone type (or no cones). There is a condition called unilateral dichromat, where only the cones in one eye are deficient. This allows the person to make comparisons that aren’t typically possible.
Colour Deficiencies Monocromatism - can match any colour using one other Extremely rare condition, usually lacking any cones. These individuals are truly colourblind, and see in shades of grey. This has implications for their vision during the daytime (sensitivity to glare). If they only have Rods, then their resolution is poor.
Dichromatism Dichromats usually experience colour. There are three main types of dichromatism: Protanopia - perceive blues and yellows, with a neutral point (grey) around 492nm. Long wavelength cones missing. Deuteranopia - similar to protanopia, but the neutral point is at 498 nm. Missing medium wavelength cones. Tritanopia - can perceive green and red with a neutral point around 570nm. Likely missing short wavelength cones.
Colour deficiencies
Colour deficiencies
Triteranopia Protanopia Deuteranopia original
Red-green colour deficient people will see only spots in the 4 images on the right
Colour normal will see a 5, r-g deficient will see a 2
Theories of colour vision continued: 2. Opponent-Process Theory (Hering) Based on phenomenological observations of colour pairings Afterimages Simultaneous colour contrast Colour descriptions Demo
Afterimages
Simultaneous contrast
Hering’s theory: W+ B- Y+ B- R+ G- Three mechanisms Each responds in opposite ways to different wavelengths or intensities of light. W+ B- Y+ B- R+ G-
Physiological Support for Hering R. DeValois found colour opponent cells in LGN These cells were organized: R+G- / R-G+, Y+B- / B+Y Many of these opponent cells were later identified in monkey cortex.
Opponent cells in LGN
Both They just describe different levels of processing Which model is correct Both They just describe different levels of processing
Trichromatic and opponent processing combined
Colour vision in the cortex Is there a special area for colour in the cortex? Electrophysiology and Neuropsychology suggested ‘yes’ (see description of cerebral achromatopsia). Recent single-unit recording and fMRI studies say ‘no’.
Colour vision in the cortex Wavelength sensitive neurons have been found throughout the cortex, so it now appears that it isn’t a ‘module’ as once believed. Instead there is distributed processing. But there are two issues: Where is wavelength information processed? Where is the perception of colour determined? Some people (e.g. M.S.) can use wavelength information, but do not experience colour.
Colours and changing illumination Colours retain their general appearance, in spite of changes in lighting. This is called colour constancy. Note that it isn’t perfect, but pretty good, particularly given the range of lighting conditions we experience.
Effect of surroundings We see colour relative to other coloured objects nearby. Land proposed that the colour-sensitive cells in the cortex calculated the ratio of nearby colours. If this ratio is constant then the colours look the same even though lighting is different.
The main way that colour constancy is achieved is by taking into account the illumination. The textbook calls this ‘chromatic adaptation’, but that isn’t a complete description. Just recall that: if the visual system can see either the source of illumination or its effect on nearby colours, it can keep colour perception constant by discounting the effect of the illumination.
Lightness Constancy Occurs when our perception of achromatic colour is not influenced by illumination. Albedo = percentage of light reflected from an object (grey = 10-70%, white = 80-90%, black =5%). Note: it’s not the total amount of light, but the percentage determines our perception. So if total light increases, the perceived lightness remains constant.
Relative reflectance Because the relative amount of light reflected from the two locations is the same, they are perceived to have the same lightness under both conditions.
How is lightness constancy achieved? The system takes into account the ratio of lightness across regions of an image (similar to Land’s colour constancy explanation). This ratio principle was used in the checkerboard example - but only works under even lighting. 2. Under uneven lighting conditions more complex operations have to occur. The visual system must segregate the scene into its components, based on lighting…
Lightness constancy (uneven lighting) The brain must distinguish between reflectance edges and illumination edges. Reflectance edges occur when the surface material changes (for example painted areas) causing changes in the amount of reflected light. Illumination edges occur when the lighting changes, for example in the case of shadows.
Information in shadows How do we interpret shadows? Why aren’t they seen as changes in the surface colour? 1. Meaningful shape 2. Edge information (penumbra)
Discounting illumination… The white squares inside the shadow are the same grey as the DARK squares outside the shadow…
Illusions from mis-applying lightness constancy Adelson The brain cannot always discount shadows…
Assumptions (heuristics) about illumination Pigment vs. Shading
Equal perceived lightness due to illumination Unequal perceived lightness due to unequal perceived illumination
Exactly the same Intensity of white Haze Illusion
His web page where this article can be downloaded: http://www-bcs.mit.edu/people/adelson/papers.html You are not required to read this extra material, it is for your own interest…
Perceptual organization and lightness constancy The disks are identical in the two images…
What is colour? Is colour an inherent property of objects? We cannot measure the sensation in the way that we can measure other qualities.
What is colour? If not, are colours created by the visual system? There is nothing inherently ‘red’ about long wavelengths of light. One way to think of this is that colour is the brain’s way of letting us know what wavelengths of light are present. This is not a phenomenon restricted to colour, it applies to many of the senses (ie. Taste, smell…)
Sections to omit Chromatic Adaptation pp 157-158 Memory and colour pp 158-159 Note that there is a small section on page 158 on the effect of surroundings that is included in the lecture.
Next week Chapter 8…