6. Depth Perception

What is depth perception?

Ability to perceive in three-dimensions

Why is depth perception important?

3D perception is vital for interacting with the world and recognising objects

What is the inverse problem?

Any retinal image is consistent with infinitely many possible configurations of the world

What are 4 sources of 3D information?

1. Binocular 2. Motion 3. Pictorial 4. Oculomotor

What are binocular cues?

Binocular disparity (stereo vision), Two eyes get different views of the world so two retinal images are slightly different, Brain puts two images together to get one image

What are motion cues?

1. Motion parallax (due to self-motion). When we move it creates movement in the retinal image. Things further away move slower on the retina 2. Kinetic depth (KDE) due to object motion which creates movement on the retina

What are pictorial cues?

1. Texture 2. Elevation 3. Relative size 4. Perspective 5. Shading 6. Occlusion

Texture (pictorial cues)?

Gradient gets smaller and more dense as it gets further away

Elevation (pictorial cues)

Objects higher up in an image are further away

Relative size (pictorial cues)

Objects that are bigger in an image are closer

Perspective (pictorial cues)

As things get further away, lines converge

Shading (pictorial cues)

Patterns of light and dark tells us about orientation and surface

Occlusion (pictorial cues)

One object occluding our view means it is closer to us

What are occulomotor cues?

Information to do what the positioning of the eyes and the muscles that control the eyes : 1. Convergence- greater convergence when looking at a close object 2. Accommodation- lens in the eye changes shape to focus the image on the retina. Len is fatter when looking at something closer to us and thinner when further away. Ciliary muscles control the shape of the lens

What makes depth perception ambiguous what assumption can we make about it?

A 2D retinal image could have been produced by different lines. We must use prior knowledge to interpret an image e.g. lines in the world tend to be parallel. This a type of top-down processing

What makes shading ambiguous what assumption can we make about it?

A retinal image could have been created by different convex/concave objects. We use prior knowledge to interpret an image e.g. assume light comes from above

What makes elevation ambiguous and what assumption can we make about it?

Retinal image could have been created by objects sitting/ floating. We use prior knowledge to interpret an image e.g. assume objects rest on a ground plane

What is wrong with assumptions?

Assumptions in some cases are not valid and can lead to perceptual errors (illusions)

What is the Ames room erroneous assumption?

assume all lines are parallel or at right-angles so leads to misperception, person on the right is closer

What is multi-cue perception?

Real world scenes have multiple cues present and cues must be integrated to achieve a single unified percept

What does integration help to overcome problems of?

1. reliability 2. ambiguity 3. conflict

What are three types of multi-cue integration?

1. Compromise 2. Dominance 3. Interaction

What is compromise (multi-cue integration)?

When two sources of depth information are conflicting, the brain finds a compromise between the two

What is Young et al's (1993) study on compromise?

Ppts viewed computer generated cylinder defined by 2 cues; texture and motion, Ppts asked if the cylinder looks flattened, circular or stretched

What were the findings of Young et al's (1993) study on compromise?

when 2 cues are conflicting the perceived shape is a compromise between the 2 cues- when texture is made less regular/ reliable perceived shape is biased towards motion cue- when motion is made less smooth/ reliable perceived shape is biased towards texture cue

What do the findings of Young et al's (1993) study on compromise show?

When 2 cues are conflicting the brain will try to average them. The final percept of shape will be biased towards the most reliable cue

What is dominance (multi-cue integration)?

When two cues define different shapes or depths, the brain may choose to ignore one in preference for the other. Usually the cue with valid assumptions will dominate. When an invalid cue dominates an illusion is created

What is interaction (multi-cue integration)?

Some cues are ambiguous e.g. texture and shading but other cues can disambiguate them

What is an example of interaction?

Texture ambiguity, Image could be convex or concave, Binocular disparity can be used to disambiguate

What is Landy et al's (1995) model of cue integration

When we are faced with lots of different cues we go through interaction process first (disambiguate cues) and compromise afterwards to combine information together