Indirect measures of perception (3)
- Magnitude estimation and production studies
give observers standard stimulus with a given value and ask them to give a corresponding value for their perception of a new stimulus (relative to the standard one) OR adjust a new stimulus until it appeared a certain level relative to the standard
SS. Stevens (1957) - differences between stimuli conformed to general pattern [subjective level of sensation = a constant x stimulus’ physical intensity^constant specific to stimulus] - Matching
participants match appearance of two stimuli in two different conditions - compare known standard to options and pick which it matches - Detection and discrimination tasks
looking at sensitivity to small differences between stimuli
absolute threshold = minimum detectable stimulus (50% of time detected)
difference threshold = minimum detectable change in stimulus
lower threshold –> less stimulation to perceive (more sensitive)
Weber (1830s) - Weber’s law = [min. detectable intensity change = background intensity x constant] - a linear relationship between difference threshold and intensity
Light/dark adaptation:
what is it?
evidence for it?
- Human visual system operates at large light range - due to pupil size changes and mechanisms in the retina
- adaptation allows for differentiation between small changes in luminescence and background
- a form of ‘gain control’ - increases sensitivity when light is low and decreases sensitivity when it is high
Hecht (1937) - adapted participants to light room then put them in the dark
- red flashes (>690nm) - sensitivity increased 100x then got no better
- violet flashes (<480nm) sensitivity increased 100x then again by 10,000x
- evidence for 2 systems:
phototopic = light-adapted - chromatic, high acuity, cones, 550nm most sensitive
scotopic = dark-adapted - achromatic, poor acuity, rods, 505nm most sensitive
How do humans perceive colour?
- main theories (+ evidence to support)
- cones
- pattern coding
Dichromatic = 2 cone types, confuse frequencies that elicit equal responses in white light (mammalian) Trichromatic = 3 cone types (L/M/S)
Young (1807) - suggested colour vision is trichromatic
- evidence from metameric matching - match appearance of any single wavelength to a primary colour mix
- metamer = mix of 2 or 3 lights that look the same colour at a single wavelength but differ spectrally
- 2 primary colours cannot match all wavelengths, 3 can
Brown + Wald (1966) - physiological support for trichromacy using microspectrophotometry - shining thin monochromatic beam through individual receptors + examining absorption
- peak absorptions cluster in cones as: 440nm, 545nm, 565nm
Hering (1878) - opponent process theory
- 4 primaries in 2 opponent relationships (RvG, BvY)
- early colour vision = trichromatic, later = conforms to conscious colour perception (seems to be mix of 4)
De Valois et al., (1966) - physiological evidence for 4 primary colours arranged in 2 opponent pairs
- monkey lateral geniculate nucleus
- colour opponent cells (R+G, B+Y)
- both trichromatic + colour-opponent correct (tri at receptoral stage, 4-primaries in post-receptoral)
- S = blue, M = green, M+L = yellow, L=red
L+S cones alone confuse certain wavelengths in white light - M allows to distinguish
Pattern coding = reflects relative responses of the 3 types of cone - post-receptoral coding looks at imbalance in LvM or SvL+M
Contrast perception
- how does it work
- centre-surround antagonism (CSA)
- illusions
Photoreceptors have receptive fields - regions of VF where light stimulation makes them respond (opposite effects in peripheral vs centre) -
- ganglion cells also respond to centre vs peripheral contrast
CSA = tendency for stimulation of centre of cells’ RF to have opposite effect to that elicited by stimulation of peripheral RF
- Hartline + Graham (1932) - CSA in limulus demonstrated
- Kuffler (1953) - CSA in cat retinal ganglion cells
- ‘on-centre’ cells = stimulate centre increases firing rate, surround decreases firing rate
- ‘off-centre’ cells = opposite
Illusions tell us about mechanisms:
- simultaneous brightness contrast - gradient background (seems lighter on a darker background because comparison)
- Troxler fading - focus on dot + background pastel colours fade because there are no sharp edges
Principles of neural coding (3)
- Neurons are preferentially activated by particular features –> specific for certain features (wavelength/orientation) - see Hubel + Wiesel (1962) - cat LGN + orientation
- Adaptive independence –> visual system can adapt to different features (light level/orientation/movement) - different mechanisms tuned to different features can be adapted independently
- afterimages = clear images on black background
- aftereffects = adaptation alters perception of 2nd stimulus - Univariance –> cell responses varies along one dimension (firing rate increases or decreases based on stimulus)- response of one cell is ambiguous so need to combine responses (pattern coding)
Perception of orientation
- how does it work?
- tilt orientation aftereffect
- Hubel + Wiesel (1977)
Ganglion cells’ RF respond equally to edge/line of any orientation as long as centre=surround light
TOAE:
- adapt neuron to tilted orientation then test a different one
- unadapted - collective responses give rise to our perception (cells tuned to different orientations - will fire more if orientation tuned to is presented)
- adapted - decreases sensitivity for that orientation so perception will be that it’ll tilt the opposite way
- adapting has largest effect if stimuli presented are 10-15 degrees apart (orientation units in visual system tuned to1 0-15 degrees apart may mutually inhibit each other)
Hubel + Wiesel (1977) - V1 in macaque - hypercolumns - neighbouring regions tend to code orientations 10-15 degrees apart
Heuristics
- what is it?
- Gestalt principles
- how does it help for depth
- how does it help for motion
- how does it help for brightness?
- pattern coding
- spatial antagonism
- visual input often ambiguous - perception isn’t because it makes used of heuristics (rules of thumb) to disambiguate
- perception involves making assumptions because through observation alone, we cannot make accurate guesses of the nature of visual stimuli (Goldstein, 2013)
Gestalt principles provide a nice summary of heuristic rules: proximity, similarity and common fate being characteristics by which objects are grouped together (Bruce, Green & Georgeson, 2003).
- Common fate: Johansson (1973) attached lights to the joints of an actor who wore dark clothes and filmed his movement in a dark room, so only the lights were visible. If the actor was stationary, the lights were perceived as a random collection of points; however, when they were moving, observers perceived a walking human figure.
- Additional principles such as good continuation (a preference for smooth continuity over abrupt changes), closure and relative size (that the smaller of two areas will be seen as a figure against a background) are also useful in disambiguating information when visual cues fail to do so (Bruce, Green & Georgeson, 2003).
Depth:
- binocular disparity
- if binocular disparity not enough, need to look
at other heuristics e.g. lighting from above/pictorial clues (Snowden, R. Thompson, P. & Troscianko, T.,2012).
Motion:
- inertia (assumed to continue in same direction) –> Sekuler, Sekuler & Lau (1997)
- rigidity (two things moving will be perceived to stay the same distance from each other) –> Johansson (1964)
Brightness:
- bright dark object vs dark light object
- use spatial antagonism (Hering, 1878 - visual vs surround) - ganglion cells good at it
- Wallach (1948) - evidence for luminance ratio
Spatial frequency
- what it it?
- evidence for it?
SF = number of cycles per degree of a visual angle (how often sinusoidal components of a stimulus repeat per unit of distance) –> edges appear at different scales (angles, sharpness) and one index of scale is SF - luminance varying cyclically
- High SF = luminance varies quickly across space
- Low SF = luminance varies slowly across space
- contrast sensitivity function (SF on X, contrast sensitivity on Y) - sensitivity greater for certain SFs
Campbell + Robson (1968) - adapted observers to a SF
- led to decreased sensitivity in specific range of SF, not ALL –> shows there are a few different channels in visual system tuned to different SF
Blakemore + Campbell (1969) - visual cortex neurons in cat - responded optimally to sinusoidal SF gratings
De Valois et al., (1982) - macaque V1 cells tuned to variety of SF channels
Depth perception
- how do we turn 2D image –> 3D perception?
- Binocular disparity –> L + R retinal images differ - disparity of visual angle - compared + fused
- objects at certain distance focus on retina in each eye at corresponding points
- ellipse formed by locations at the same distance (horopter) - object on here have no disparity
- region surrounding horopter = Panum’s area (within this, images fuse; outside this, double images - diplopic)
- Objects further than horopter = displaced leftwards in L eye wrt R, uncrossed disparity (need to uncross eye to fixate)
- Object closer than horopter = displaced rightwards in L eye wrt R, crossed disparity - Motion parallax - same eye, different times
- objects closer = appear to move faster in opposite direction to you
- objects further = appear to move slower in same direction - Pictorial cues
- relative size of objects - appear bigger = closer
- perspective - convergence of parallel lines
- shadows/lighting - show separation in space (Kersten et al., 1997)
Motion perception
- simple case
- adaptation
- mechanism
Simple case - Reichardt (1969) - motion in flies
- delay of one of the inputs in cells to ensure that if object move across one receptor then another at the right speed - summed excitation, cells fire
- BUT: one direction and one speed
Adaptation: Aristotle’s waterfall illusion
- adapted to constant movement in one direction (30s) then focus on static stimulus - looks like it’s moving the other way (adaptive independence, pattern coding)
Mechanism: Sutherland (1961)
- motion perception = average of motion signals for cells tuned to different directions
- spatial pattern coding –> cells with RF in same location may code for different directions of motion
- static = motion detectors signal all directions moving equally
Interocular transfer
- what is it?
- where does it occur?
- examples of it
IOT = effect of stimulus from one eye perceived in the other
Occurs in…
- pre-cortical stage –> retina, LGN, geniculostriate pathway - monocular (cells receive inputs from one eye)- aftereffects only in adapted eye
- cortical stage –> V1, extrastriate cortex - binocular - aftereffects similar in adapted and unadapted eyes
Examples:
- Paradiso, Shimojo, Nakayama (1989) - IOT of tilt aftereffects:
subjective contours = 92% aftereffect same in both eyes (binocular coding)
real contours = 46% aftereffect same in both eyes (substantial monocular coding)
- Nishida, Ashida, Sato (1994) - IOT motion aftereffect with flickering:
can be up to 100% in unadapted eye with flickering stimulus rather than static - seems to isolate high-level binocular mechanisms
Perceptual constancy
- what is it?
- size constancy
- lightness constancy (including spatial antagonism)
Perceptual constancy = vision adjusts perception according to current conditions
- identical retinal inputs can appear very different
- e.g. same object in different light/angle/distance
Size constancy - needs to adjust for distance (further = smaller retinal image)
- Emmert’s law –> perceived size of object scaled up according to distance from observer
- Ames room illusion (assume normal room so people look different sizes when really room is different shape)
- Titchener illusion (circles), Ponzo illusion (train tracks)- misapplied size constancy
Lightness constancy - retinal input = illumination x lightness
- lightness = how reflective surface is
- light dark-object = dark light-object –> ambiguous
- Helmholtz - we break down into illuminant + lightness components - if well lit, reduce estimated lightness; if poorly lit, increase estimated lightness
- Spatial antagonism –> compare to background illumination
- Hering (1878) - respond to contrast between light reflected from centre vs surround - ganglion cells give accurate comparison of centre + surround
- Wallach (1948) - luminance ratio = apparent lightness depends on ratio light reflected by object to light reflected by immediate background (same ratio = same lightness perceived)
BUT:
- Gelb effect (1929) - based on context - if placed next to white object it seems grey, if alone it seems white (depends on 3D interpretation of scene too)
- Gilchrist (1980) - depth perception can influence light perception –> if appears to be in dark room, judged as relatively light, if appears to be in light room, judged as relatively dark
Filling in
- what it is?
- evidence for it (perceptual + neural)
We have a blind spot (monocular) where there are no photoreceptors and yet vision fills in colour/texture information
Perceptual evidence:
- Ramachandran + Gregory (1991) - aftereffect of filled in twinkle –> patch filled in using centre and surround information
- Davis + Driver (1994) - modal completion - detect parially-camoflaged objects by perceiving illusory contours –> continued perception of a rectangle even if it is camouflaged with background
Neural evidence:
- Fiorani et al, (1991) - found neurons appearing to respond to perceptual info being filled in - V1 neurons corresponding to blind spot fire in response to orientation of line overlapping blind spot
- De Weerd et al., (1995) - macaque V2 + V3 - humans + monkeys showed same stimuli, humans indicate when box disappeared from visual - recorded monkey neurons at this point - little activity to begin with then activity increases just as square perceived to disappear
The visual brain:
- Hubel + Wiesel (1962)
- Wong-Riley (1979)
- Extra-striate visual areas
H+W (1962)
- V1 in cats - cortical surface cells near each other have similar RF locations, prefer same orientations + code same eye
- all cells in a hypercolumn - similar RF location
- neighbouring hypercolumns - code neighbouring regions of space, differ in orientation by 10-15 deg.
W-R (1979)
- stained cortex - saw patterns of cytochrome oxidase blobs within hypercolumns
- cells WITHIN blobs = blob cells - code colour info mainly
- cells BETWEEN blobs = interblob cells - respond to high spatial frequencies and colour edges
Other:
- simple cells = size, orientation, position of stimulus
- complex cells = moving stimuli, direction of motion
- hypercomplex cells = corners/line ends moving
V2: codes visual surfaces, beyond retinal input (perceptual filling in)
V4: codes sophisticated colour responses, shows evidence for colour constancy (Zeki, 1978)
- Hadjikhani et al., 1998 –> V8 is crucial for high-level colour processing (Zeki replies –> V8 is part of V4)
V5: high-level motion processing (global motion of object)
- Beckers + Zeki (1995) –> TMS pulse to V5 causes motion perception to drop to chance
- von Zihl, Crammon + Mai (1983) –> LM akinetopsia after V5 damage - lost motion perception - saw world as stills
Conscious vision
- what it is?
- change blindness
- parallel vision
We think we consciously process visual info but we don’t - only know the product not the processes used to produce it
Change blindness = if fixated on one aspect of a scene and it changes, we don’t notice
- Simons + Levin (1998) - the door study - experimenter changes, 50% don’t notice
- change in stimulus results in perceptual transient - usually heightens awareness (not if not attended to or if everywhere)
Parallel vision –>processing information all at once
- Treisman + Gelade (1980) - visual search paradigm –>
parallel search for low level visual features (colour + orientation) - pop out, no increase in RT as set increases; high level features processes serially, conscious search - RT increases as set increases
- Ramachandran (1988) - shading cues/convexity are hard to differentiate, need to do visual search through all features
