Cue approach to depth perception
The approach to explaining depth perception that identifies information in the retinal image, and also information provided by aiming and focusing the eyes on an object that is correlated with depth in the scene. Some of the depth cues that have been identified are overlap, relative height, relative size, atmospheric perspective, convergence, and accommodation.
cue approach to depth perception
-ask what information in the retinal image enables us to perceive depth in the scene -e.g. occlusion
What are the 3 major type of cues in cue approach to depth perception?
1. Oculomotor. Cues based on our ability to sense the position of our eyes and the tension in our eye muscles. 2. Monocular. Cues that work with one eye. 3. Binocular. Cues that depend on two eyes.
is a signal, or cue, that one object is in front of another. the object that is partially covered must be at a greater distance than the object that is covering it. According to cue theory, we learn the connection between this cue and depth through our previous experience with the environment.
oculomotor cues are created by (2)
a. convergence: inward movement of the eyes that occurs when we look at a nearby object
b. accomodation: the change in the shape of
the lends that occurs when we focus on objects at various distances
The idea behind these oculomotor cues
we can feel the inward movement of the eyes that occurs when the eyes converge to look at nearby objects, and we feel the tightening of eye muscles that change the shape of the lens to focus on a nearby object.
Monocular cues work with only one eye. They include accommodation, which we have described under oculomotor cues; pictorial cues, which is depth information that can be depicted in a two-dimensional picture; and movement based cues, which are based on depth information created by movement.
Pictorial cues are sources of depth information that can be depicted in a picture, such as the illustrations in this book or the image on the retina (Goldstein, 2001). -occlusion -relative height -relative size -perspective convergence -familiar size -atomspheric perspective -texture gradient -shadows
pictorial monocular cues - Occlusion occurs when one object hides or partially hides another from view. The partially hidden object is seen as being farther away - occlusion does not provide information about an object’s absolute distance; it only indicates relative distance.
pictorial monocular cues - objects that are below the horizon and have their bases higher in the field of view are usually seen as being more distant. - When objects are above the horizon, like the clouds, being lower in the field of view indicates more distance - connection between an observer's gaze and distance. Looking straight out at an ojbect high in the visual field, near the horizon, indicates greater depth than looking down, as you would for an object lower in the visual field
pictorial monocular cue when two objects are of equal size, the one that is farther away will take up less of your field of view than the one that is closer. This cue depends, to some extent, on a person’s knowledge of physical sizes
pictorial monocular cue When parallel lines extend out from an observer, they are perceived as converging— becoming closer together—as distance increases. This perceptual coming-together of parallel lines is called perspective convergence
pictorial monocular cue judge distance based on our prior knowledge of the sizes of objects. Epstein (1965) shows that under certain conditions, our knowledge of an object’s size influences our perception of that object’s distance.
Epstein (1965): familiar size
The stimuli in Epstein’s experiment were equal-sized photographs of a dime, a quarter, and a half-dollar, which were positioned the same distance from an observer. By placing these photographs in a darkened room, illuminating them with a spot of light, and having subjects view them with one eye, Epstein created the illusion that these pictures were real coins. When the observers judged the distance of each of the coin photographs, they estimated that the dime was closest, the quarter was farther than the dime and the half dollar was the farthest of them all The observers' judgements were influenced by their knowledge of the sizes of the real dimes, quarters and half-dollars. This result did not occur, however, when observers viewed the scene with both eyes, because the use of eyes provided information indicating the coins were at the same distance. The cue of familiar size is therefore most effective when other information about depth is absent
monocular pictorial cue - occurs when more distant objects appear less sharp and often have a slight blue tint. The farther away an object is, the more air and particles (dust, water droplets, airborne pollution) we have to look through, making objects that are farther away look less sharp and bluer than close objects. -"calibrated' to locations, so more difficulty estimating distances in the clearer air
monocular pictorial cue -Elements that are equally spaced in a scene appear to be more closely packed as distance increases -according to the cue of relative size, more distant objects take up less of our field of view. This is exactly what happens to the faraway elements in the texture gradient.
pictorial monocular cue - Shadows that are associated with objects can provide information regarding the locations of these objects - Shadows also enhance the three-dimensionality of objects.
Motion-produced monocular cues
All of the cues we have described so far work if the observer is stationary. If, however, we decide to take a walk, new cues emerge that further enhance our perception of depth. two different motion-produced cues: (1) motionparallax and (2) deletion and accretion.
two different motion-produced cues
(1) motion parallax and (2) deletion and accretion.
motion-produced monocular cue -occurs when, as we move, nearby objects appear to glide rapidly past us, but more distant objects appear to move more slowly -The image of the far object travels a much smaller distance across the retina, so it appears to move more slowly as the observer moves. -one of the most important sources of depth information for many animals. - the information provided by motion parallax has also been used to enable human-designed mechanical robots to determine how far they are from obstacles as they navigate through the environment -motion parallax is widely used to create an impression of depth in cartoons and video games
objects are covered or uncovered as we move relative to them
Covering an object is Deletion Uncovering an object is Accretion
ex: left hand in front of right with eyes closed, moving head back an forth exposes and covers the right hand
Range of effectiveness of different depth cues
these cues work over different distances, some only at close range (accommodation and convergence), some at close and medium ranges (motion parallax), some at long range (atmospheric perspective), and some at the whole range of depth perception (occlusion and relative size;
Binocular depth information
information—the differences in the images received by our two eyes. Because our eyes view the world from positions that are about 6 cm apart in the average adult, this difference in the viewpoint of the two eyes creates the cue of binocular disparity.
Difference in the position of the same image on 2 retinas enables you to construct a 3-D perception from 2-D images on your retinas
Corresponding retinal points
The points on each retina that would overlap if one retina were slid on top of the other.
- for any given dist. there is a set of points that fall on corresponding retinal points and the locus of all these points is called horopter
- single vision occurs for all points in the horopter and all points should appear equidistant from the observer
the location of objects that have zero disparity (same distance from point of fixation)
two points, one on each retina, that would not overlap if the retinas were slid onto each other. also called disparate points
The visual angle between the images of an object on the two retinas. When images of an object fall on corresponding points, the angle of disparity is zero. When images fall on noncorresponding points, the angle of disparity indicates the degree of noncorrespondence.
The visual angle between the images of an object on the two retinas. When images of an object fall on corresponding points, the angle of disparity is zero. When images fall on noncorresponding points, the angle of disparity indicates the degree of noncorrespondence.
Why is absolute disparity important?
it provides information about the distances of objects. The amount of absolute disparity indicates how far an object is from the horopter. Greater disparity is associated with greater distance from the horopter.
How many fixations a seconds does a person make when scanning a scene?
3 fixation per second; every new fixation establishes a new horopter, this means that the absolute disparities for every object in a scene have to be constantly recalculated
the difference in absolute disparities of two elements in the
visual scene
- Relative depth of objects
Connecting disparity information and the perception of depth
-stereopsis: the impression of depth that results from information provided by binocular disparity
One of the binocular cues; it is based on the small discrepancy in the retinal images in each eye when viewing a visual scene (binocular disparity)
A device for simultaneously presenting one image to one eye and another image to the other eye. These can be used to present dichoptic stimuli for stereopsis and binocular rivalry -Charles Wheatstone (1802-1875)
What is another way to separate the images other than using a stereoscope?
create the left and right images from polarized light—light waves that vibrate in only one direction. One image is polarized so its vibration is vertical, and the other is polarized so its vibration is horizontal. Viewing the film through polarizing lenses, which let vertically polarized light into one eye and horizontally polarized light into the other eye, creates the disparity that results in three-dimensional perception.
Bela Julesz (1971): stimulus called random-dot stereogram; advantage over stereopsis?
-stereopsis did not provide that disparity creates a perception of depth because images also contain potential depth cues, such as occlusion and relative height, which could contribute to our perception of depth -By creating stereoscopic images of random-dot patterns, Julesz showed that observers can perceive depth in displays that contain no depth information other than disparity
random-dot stereoscope; how they were created and the effect
These patterns were constructed by first generating two identical random-dot patterns on a computer and then shifting a square-shaped section of the dots one or more units to the side. - The effect of shifting one section of the pattern in this way is to create disparity. When the two patterns are presented simultaneously to the left and the right eyes in a stereoscope, observers perceive a small square floating above the background
Psychophysical experiments, particularly those using Julesz’s random-dot stereograms, show that retinal disparity creates what?
correspondence problem - stated as a question
How does the visual system match the parts of the images in the left and right eyes that correspond to one another? in order for the visual system to calculate this disparity, it must compare the images of the cactus on the left and right retinas and the images of the window on the left and right retinas. This is the correspondence problem. How does the visual system match up the images in the two eyes?
Possible answer to the correspondence problem
A possible answer to this question is that the visual system may match the images on the left and right retinas on the basis of the specific features of the objects. For example, the upper-left window pane on the left could be matched with the upper-left pane on the right, and so on. Explained in this way, the solution seems simple: Most things in the world are quite discriminable from one another, so it is easy to match an image on the left retina with the image of the same thing on the right retina.
Many animals have excellent depth perception
Cats leap on their prey; monkeys swing from one branch to the next; a male housefly follows a flying female, maintaining a constant distance of about 10 cm; and a frog accurately jumps across a chasm
There is no doubt that many animals are able to judge distances in their environment, but what depth information do they use?
A survey of mechanisms used by different animals reveals that animals use the entire range of cues described in this chapter. Some animals use many cues, and others rely on just one or two.
To make use of binocular disparity, an animal must have eyes that have overlapping visual fields
-animals such as cats, monkeys and humans have frontal eyes which result in overlapping fields of view, can use disparity to perceive depth -animals with lateral eyes, such as the rabbit, do not have overlapping visual fields and therefore cannot use disparity to perceive depth. In sacrificing binocular disparity, animals with lateral eyes gain a wider field of view - something that is extremely important for animals that need to constantly be on the lookout for predatos e.g. pidgeon
The pigeon is an example of an animal with lateral eyes
that are placed so the visual fields of the left and right eyes overlap only in a 35-degree area surrounding the pigeon’s beak. This overlapping area, however, happens to be exactly where pieces of grain would be located when the pigeon is pecking at them, and psychophysical experiments have shown that the pigeon does have a small area of binocular depth perception right in front of its beak
Movement parallax is probably insects’ most important method of judging distance, and they use it in a number of different ways
e.g. the locust uses a “peering” response—moving its body from side to side to create movement of its head—as it observes potential prey. T. S. Collett (1978) measured a locust’s “peering amplitude”—the distance of this side-to-side sway—as it observed prey at different distances, and found that the locust swayed more when targets were farther away. Since more distant objects move less across the retina than nearer objects for a given amount of observer movement, a larger sway would be needed to cause the image of a far object to move the same distance across the retina as the image of a near object. The locust may therefore be judging distance by noting how much sway is needed to cause the image to move a certain distance across its retina
The above examples show how depth can be determined from different sources of information in light. But bats, some of which are blind to light, use a form of energy we usually associate with sound to sense depth
Sonar, which stands for sound navigation and ranging, works by sending out pulses of sound and using information contained in the echoes of this sound to determine the location of objects. Donald Griffin (1944) coined the term echolocation to describe the biological sonar system used by bats to avoid objects in the dark.
Bats emit pulsed sounds that are far above the upper limit of human hearing, and they sense objects’ distances by noting the interval between when they send out the pulse and when they receive the echo. Since they use sound echoes to sense objects, they can avoid obstacles even when it is totally dark. the timing of these echoes provides the information the bat needs to locate objects in its environment
The physiology of depth perception
Most of the research on the physiology of depth perception has concentrated on looking for neurons that signal information about binocular disparity. But neurons have also been found that signal the depth indicated by pictorial depth cues.
Physiology of depth perception: neurons that respond to pictorial depth
Tsutsui et al., (2002, 2005) studied the physiology of neurons that respond to the depth indicated by texture gradients by having monkeys match stimuli. The records below the texture gradient patterns are the responses of a neuron in an area in the parietal cortex that had been associated with depth perception in other studies. This neuron does not fire to the right-slanting gradient, or to a flat pattern, but does fire to the left-slanting gradient. Thus, this neuron fires to a display in which depth is indicated by the pictorial depth cues of texture gradients. This neuron also responds when depth is indicated by disparity, so it is tuned to respond to depth whether it is determined by pictorial depth cues or by binocular disparity
what is the case when binocular neurons respond at the highest freq? when does the number of binocular cells that respond to binocular disparity inc? what are the main lobes that process stereopsis?
1. when the same image occupies slightly diff areas of the retina in the two eyes
2. inc the further along into the cortex the stimulation travels, higher cortical processing is important for perceiving stereopsis
3. occipital and parietal lobes
what is the horopter? how do objects appear in the horopter? how are objects seen when close up?
1. surface on the circle, it is diff than muller circle but is often used interchangeably
2. as a single object when seen w/ both eyes
3. seen as two ojbects this double vision is known as diplopia
neurons that respond to binocular disparity
The first research on these neurons described neurons in the striate cortex (V1) that responded to absolute disparity. These neurons are called binocular depth cells or disparity-selective cells. A given cell responds best when stimuli presented to the left and right eyes create a specific amount of absolute disparity. Further research has shown that there are also neurons higher up in the visual system that respond to relative disparity. Examine disparity tuning curvr
connecting binocular depth cells to depth perception
Just because disparity-selective neurons fire best to a specific angle of disparity doesn’t prove that these neurons have anything to do with depth perception. -Blake and Hirsch (1975) demonstrated this connection by doing a selective rearing experiment that resulted in the elimination of binocular neurons. - Another technique that has been used to demonstrate a link between neural responding and depth perception is microstimulation - DeAngelis et al., (1998) trained a monkey to indicate the depth created by presenting images with different absolute disparities to the left and right eyes. - brain-imaging experiments on humans show that a number of different areas are activated by stimuli that create binocular disparity - experiments on monkeys have determined that neurons sensitive to absolute disparity are found in the primary visual receiving area, and neurons sensitive to relative disparity are found higher in the visual system, in the temporal lobe and other areas - depth perception involves a number of stages of processing that begins in the primary visual cortex and extends to many different areas in both the ventral and dorsal streams
-demonstrated connection between disparity and behaviour by doing a selective rearing experiment that resulted in the elimination of binocular neurons. They reared cats so that their vision was alternated between the left and right eyes every other day during the first 6 months of their lives. After this 6 month period of presenting stimuli to just one eye at a time, Blake and Hirsch recorded from neurons in the cat's cortex and found that (1) these cats had few binocular neurons and (2) they were not able to use binocular disparity to perceive depth. Thus, eliminating binocular neurons eliminates stereopsis and confirms what everyone suspected all along - that disparity-selective neurons are responsible for stereopsis
Another technique that has been used to demonstrate a link between neural responding and depth perception is microstimulation
Microstimulation is achieved by inserting a small electrode into the cortex and passing an electrical charge through the electrode to activate the neurons near the electrode (M. R. Cohen & Newsome, 2004). Neurons that are sensitive to the same disparities tend to be organized in clusters, so stimulating one of these clusters activates a group of neurons that respond best to a specific disparity.
Gregory DeAngelis and coworkers (1998) trained a monkey to indicate the depth created by presenting images with different absolute disparities to the left and right eyes
Presumably, the monkey perceived depth because the disparate images on the monkey’s retina activated disparityselective neurons in the cortex.
Gregory DeAngelis and coworkers (1998): what happened when microstimulation was used to activate a different group of disparity-selective neurons?
DeAngelis and coworkers stimulated disparity-selective neurons that were tuned to a disparity different from what was indicated by the images on the retina. When they did this, the monkey shifted its depth judgment toward the disparity signaled by the stimulated neurons
Are depth and size perception related?
What is a real life example of how distance and size perception are interrelated
Whiteout—one of the most treacherous weather conditions possible for flying—can arise quickly and unexpectedly. Frank flew helicopter across Antarctic waves, thought he saw a big vehicle far away, actually a grenade, A small box seen close up can, in the absence of accurate information about its distance, be misperceived as a large truck seen from far away
What classic experiment demonstrated the idea that we can misperceive size when accurate depth information is not present
A. H. Holway and Edwin Boring (1941): the holway and boring experiment
Testing how the visual angle of an object relates to the observer's perception of distance. Without depth cues size estimation is based on visual angle.
describe the holway and boring experiment
Observers in Holway and Boring’s experiment sat at the intersection of two hallways and saw a luminous test circle when looking down the right hallway and a luminous comparison circle when looking down the left hallway. The comparison circle was always 10 feet from the observer, but the test circles were presented at distances ranging from 10 feet to 120 feet. Task: adjust the diameter of the comparison circle on the left to match their perception of the size of the test circle on the right: cast same retina
the angle of an object relative to the observer’s eye
What does the visual angle do?
The visual angle tells us how large the object will be on the back of the eye. There are 360 degrees around the entire circumference of the eyeball, and an object with a visual angle of 1 degree would take up 1/360 of this circumference— about 0.3 mm in an average-sized adult eye
Main result of Holway and Boring (1941)
Eliminating depth information made it more difficult to judge the physical sizes of the circles. Without depth information, the perception of size was determined not by the actual size of an object but by the size of the object’s image on the observer’s retina. Because all of the test circles in Holway and Boring’s experiment had the same retinal size, they were judged to be about the same size once depth information was eliminated.
visual angle when depth information is eliminated (red lines). An example of size perception that is determined by visual angle is our perception
of the sizes of the sun and the moon, which, due to a cosmic coincidence, have the same visual angle. The fact that they have identical visual angles becomes most obvious during an eclipse of the sun.
the tendency to perceive the size of an object as constant despite changes in its retinal image (when we change distance)
Size Constancy as a Calculation
size–distance scaling that takes an object’s distance into account: S = K (R*D), where S is object's perceived size, K is constant, R is the size of the retinal image, and D is the perceived distance of the object.
According to the size–distance equation, as a person walks away from you, what happens to R, D and S?
the size of the person’s image on your retina (R) gets smaller, but your perception of the person’s distance (D) gets larger. These two changes balance each other, and the net result is that you perceive the person’s size (S) as remaining constant.
This relationship between the apparent distance of an afterimage and its perceived size is known as
The farther away an afterimage appears, the larger it will seem. This result follows from our size–distance scaling equation, S= R*D. The size of the bleached area of pigment on the retina (R) always stays the same, so that increasing the afterimage’s distance (D) increases the magnitude of R*D. We therefore perceive the size of the afterimage (S) as larger when it is viewed against the far wall.
Other information for size perception
One source of info for size perception is relative size. -relationship between objects and texture information on the ground. According to Gibron (195), we perceive two cylinders resting on a texture gradient as the same size because the bases of both cylinders cover the same number nof units on the gradient indicates that the bases of the two cylinders are the same size
-The Mueller-Lyer illusion -the ponzo illusion -the ames room -the moon illusion
Muller-Lyer Illusion- Classic Illusions
lines of equal length appear unequal because of the orientation of the arrow marks at the end
Take away: we don't really know why we see the Muller-Lyer illusion, we just know we do and that it may have cultural influences
Why does the Müller-Lyer display cause a misperception of size? Richard Gregory (1966) explains the illusion on the basis of a mechanism he calls
misapplied size constancy scaling.
Misapplied Size Constancy Scaling
A principle stating that when mechanisms that help maintain size constancy in the three-dimensional world are applied to two-dimensional pictures, an illusion of size sometimes results.
Challenge to Gregory's misapplied size constancy scaling
e.g. figures like the dumbbells, which contain no obvious perspective or depth, still result in an illuson. DeLucia and Hochberg have shown that the Müller-Lyer illusion occurs for a 3d display
R.H. Day has proposed what theory to explain the Muller-Lyer illusion
states that our perception of line length depends on two cues: (1) the actual length of the vertical lines, and (2) the overall length of the figure. According to Day, these two conflicting cues are integrated to form a compromise perception of length.Thus, although Gregory believes that depth information is involved in determining illusions, Day rejects this idea and says that cues for length are what is important
The Ponzo illusion: describe
In the Ponzo (or railroad track) illusion, both animals are the same size on the page and have the same visual angle, but the one on top appears longer.
The Ponzo illusion: explain in terms of Gregory's misapplied scaling explanation
the top animal appears larger because of depth information provided by the converging railroad tracks that make the top animal appear farther away. Thus, just as in the Müller-Lyer illusion, the scaling mechanism corrects for this apparently increased depth (even though there really isn’t any, because the illusion is on a flat page), and we perceive the top animal to be larger
Describe the Ames illusion
The Ames room causes two people of equal size to appear very different in size. The reason for this erroneous perception of size lies in the construction of the room. The shapes of the wall and the windows at the rear of the room make it look like a normal rectangular room when viewed from a particular observation point. however the Ames room is actually shaped so that the left corner of the room is almost twice as far from the observer as the right corner
Ames room: explain using distance scaling equation
We can understand why this occurs by returning to our size–distance scaling equation, S =R* D. Because the perceived distance (D) is the same for the two women, but the size of the retinal image (R) is smaller for the woman on the left, her perceived size (S) is smaller.
Explanation for Ames room based on relative size
The relative size explanation states that our perception of the size of the two women is determined by how they fill the distance between the bottom and top of the room. Because the woman on the right fills the entire space and the woman on the left occupies only a little of it, we perceive the woman on the right as taller
Describe the moon illusion
You may have noticed that when the moon is on the horizon, it appears much larger than when it is higher in the sky. This enlargement of the horizon moon compared to the elevated moon CONSTANT visual angle agreeance
Explanation for moon illusion: apparent distance theory
According to apparent distance theory, the moon on the horizon appears more distant because it is viewed across the filled space of the terrain, which contains depth information; but when the moon is higher in the sky, it appears less distant because it is viewed through empty space, which contains little depth information.
The key to the moon illusion, according to apparent distance theory, is that
both the horizon and the elevated moons have the same visual angle, but because the horizon moon is seen against the horizon, which appears farther than the zenith sky, it appears larger. Follows from size-distance scaling equation, S=R*D
Kaufman and Rock (1962) have done a number of experiments that support the apparent distance theory. Describe on of their experiments
they showed that when the horizon moon was viewed over the terrain, which made it seem farther away, it appeared 1.3 times larger than the elevated moon; however, when the terrain was masked off so that the horizon moon was viewed through a hole in a sheet of cardboard, the illusion vanished
Another theory of the moon illusion (other than apparent distance theory) is what?
angular size contrast theory
angular contrast theory - the moon illusion
states that the moon appears smaller when it is surrounded by larger objects. Thus, when the moon is elevated, the large expanse of sky surrounding it makes it appear smaller. However, when the moon is on the horizon, less sky surrounds it, so it appears larger
Is there an agreed upon theory for the moon illusion?
Even though scientists have been proposing theories to explain the moon illusion for hundreds of years, there is still no agreement on an explanation (Hershenson, 1989). Apparently a number of factors are involved, name 3 of them
atmospheric perspective (looking through haze on the horizon can increase size perception), color (redness increases perceived size), and oculomotor factors (convergence of the eyes, which tends to occur when we look toward the horizon and can cause an increase in perceived size)
Participants made distance judgements with or without a backpack -Those with backpacks increased their estimates (without walking)
had participants throw balls to targets ranging from 4 to 10 meters away. After they had thrown either a light ball or a heavy ball, participants estimated the distances to the targets. -distance estimates were larger after throwing the heavy ball -expectiation also influences distance judgements