6. depth perception

What is the inverse problem in vision?

It's the idea that any 2D image on the retina can come from many different 3D scenes — the brain must figure out which one is correct.

What are the four main sources of depth information?

1) Binocular cues
2) Motion cues
3) Pictorial cues
4) Oculomotor cues

What is binocular disparity?

The slight difference in images between the two eyes, used by the brain to judge depth (also called stereo vision).

What is motion parallax?

When objects move across the retina at different speeds depending on their distance — closer objects move faster than farther ones.

List 5 examples of pictorial depth cues.

Texture, elevation, relative size, perspective, shading, and occlusion.

What is convergence?

The inward turning of the eyes when focusing on a close object.

What is accommodation?

The lens in the eye thickens (gets more curved) to focus on near objects.

What is the main challenge with interpreting 2D retinal images?

They are ambiguous and can represent many possible 3D scenarios.

How do we overcome ambiguity in perspective?

We assume lines in the world are usually parallel and use past knowledge (top-down processing).

Why can shading be ambiguous?

The same shading pattern can come from a convex shape lit from above, a concave shape lit from below, or a flat surface

What assumption helps solve shading ambiguity?

The brain assumes light comes from above.

What is an erroneous assumption in perception?

A mistaken belief that leads to perceptual errors, like visual illusions.

How does the Ames Room demonstrate an erroneous assumption?

We wrongly assume the room's walls are rectangular and lines are parallel, causing a distorted perception of size.

Why do we integrate multiple depth cues?

To reduce ambiguity, improve reliability, and resolve conflicts between cues.

What is compromise in depth perception?

When conflicting cues (e.g., motion and texture) are blended by the brain into an in-between perception.

Describe Young et al. (1993)’s experiment.

Participants saw a cylinder with motion (suggesting it was stretched) and texture (suggesting it was flattened). The final percept was a blend depending on cue reliability.

What happens when motion is less smooth in Young et al.’s study?

Texture becomes the more reliable cue and dominates perception.

What is dominance in cue integration?

When two cues conflict strongly, the brain may ignore one and rely entirely on the other (often the more valid one).

Give an example of dominance leading to illusion.

Using elevation on a scene without a ground plane may mislead the brain, causing illusion

What is interaction in depth cue integration?

When one ambiguous cue is clarified by another reliable one — e.g., binocular disparity disambiguating texture.

What is the “order of integration”?

1) Interaction stage: Reliable cues help clarify ambiguous ones
2) Compromise stage: If ambiguity remains, brain blends cues for a final percept