CH 10: Depth Perception

Depth Perception

Determining how far away an object is

Cues:

• Pick out individual pieces of info

• Pieces correlated with depth

Occlusion: one object partially hides another is perceived as closer

Oculomotor Cues

senses the position & tension in the eye muscles

two: 

• convergence

• accomodation

Convergence

inward movement of the eyes

inversely correlated with depth (closer it is = more inward movement)

Ex: looking at person and coffee cup (looking at coffee = eyes come in)

Accomodation

Eye changes lens shape to focus.

Far objects (>20 ft):

Light rays = parallel

Ciliary muscles relax → lens flattens

Near objects (<20 ft):

Light rays = spread out

Ciliary muscles contract → lens curves more

Monocular Cues

Depth cues from one eye

Pictorial cues: depth in 2D images (like paintings/photos)

Movement cues: depth from motion

Oculomotor cues can also count

Pictorial Cues

Depth info shown in still images

Occlusion

One object blocks another

Partially hidden object = farther away

Gives relative, not exact, depth

Monocular Pictorial Cues

Relative Height: higher in the field of view = further away

Familiar & Relative Size

Use known object size to judge distance

Farther = smaller retinal image

If a large object looks small, it’s seen as far away

Judge distance using known object size

Epstein (1965):

Participants judged coin distance

Larger coins → seen as farther

Effect only with one eye (no binocular cues)

Perspective Convergence (linear perspective)

Parallel lines appear to meet as they go into the distance (get further away)

Used by Renaissance artists to show depth

Atmospheric Perspective

Farther objects look bluer or hazier

Caused by light scattering from air and dust

More distance = more scattering, so objects lose contrast and color clarity

Texture Gradient

On a textured surface, patterns or details appear closer together as they get farther away

At great distances, texture details can blend or disappear

Shadows

Show an object’s location and depth

Position of the shadow helps us judge where the object is in space

Provide relative position cues

Motion Produced Cues

visual indicators produced by motion

two cues:

• motion parallax

• deletion & accertion

Motion Parallax

When you move, close objects seem to move faster, and far objects seem to move slower — like when you drive past a cornfield.

Deletion & Accertion

From side-to-side (lateral) movement

When one object covers another → Deletion

When the hidden object is uncovered → Accretion

Works like occlusion + motion parallax to show depth

Depth Cue Summary

Each cue gives a “guess” about depth

Alone, they’re not always accurate

Some work better at certain distances

Some give relative (comparison) depth, not absolute distance

Binocular Depth Perception

Each eye gets a slightly different image

Brain compares the differences to judge depth

Demonstrated by:

• Finger test (close vs. far)

• Hand on face illusion

Closer finger

moves further in visual field

moves further on retina

Stereoscopic Vision

Each eye has a slightly different view of the world

Brain compares these views to create stereoscopic (3D) depth

Depth comes from differences between the eyes’ images

Stabismus

The eyes don’t line up correctly (cross-eyed or wall-eyed).
Visual system ignores one eye’s input to prevent double vision, so depth is judged using only one eye’s (monocular) cues.

Standard Movies/Films

use only one-eye (monocular) depth cues

3D movies

Filmed with two cameras; glasses let each eye see a different image to create depth; combines everything to create one image

Binocular Disparity

retinal images of an object fall on disparate points on the retinas

basis of stereoscopic depth perception

Corresponding Retinal Points

Same matching spots on each retina that line up when looking at an object; these send signals to the same place in the brain’s visual area (V1).

Retinal Disparity Example: Julie & Tree

• Julie: Image falls on fovea → you’re looking directly at her.

• Horopter (dashed line): Imaginary line where images fall on corresponding retinal points → objects look single.

• Tree: Falls on corresponding points left of fovea → seen as single, but off to the side of Julie.

Horopter

• Imaginary line through what you’re looking at.

• Objects on it → fall on corresponding points → seen as single.

• Objects off it → fall on noncorresponding points → cause retinal disparity.

• The brain uses this disparity to judge depth (in front/behind fixation).

Noncorresponding points

points that would overlap

i.e different parts of the visual field

Absolute Disparity

the degree to which an object’s image deviates from falling on corresponding points in the two retinas. It tells you how much the object is in front of or behind the point you’re fixating on. Larger disparity means the object is farther away from the fixation plane. (ex: 2L and 3R = 5 absolute disparity)

Angle of Disparity

the visual angle between the locations of an object’s images on the two retinas.

When an object’s images fall on corresponding points, the angle of disparity is 0°.

The greater the angle, the more the images fall on noncorresponding points, and the more depth difference you perceive.

The farther an object is from the horopter (in either direction), the greater the angle of disparity — meaning it appears deeper or closer in space.

(ex: catching a fly ball and knowing when to close glove)

Crossed Disparity (in front of the Horopter)

Objects that are closer than the horopter create crossed disparity because their images fall on opposite sides of the fovea in each eye.

Left eye: image lands to the right of the fovea.

Right eye: image lands to the left of the fovea.
This “crossing” refers to the image positions on the retinas — not an actual flipping or switching of the eyes’ views.

(BB8 and R2D2)

Uncrossed Disparity (behind the Horopter)

Objects that are farther away than the horopter create uncrossed disparity because their images fall on the same side of the fovea in each eye.

Left eye: image lands to the left of the fovea.

Right eye: image lands to the right of the fovea.
In this case, the images stay left with left and right with right — they do not cross.

No longer on corresponding points

Crossed vs Uncrossed Disparity

These indicate whether objects are closer to or farther from the point you’re fixating on.

They provide information about an object’s relative position in depth (in front of or behind).

Stereopsis

The perception or impression of depth created by retinal (binocular) disparity — the slight differences between the images in each eye.
Earlier studies showed how disparity information changes between the eyes, but additional proof was needed to show that this difference actually causes depth perception rather than just changes in the retinal image.

Binocular Disparity: Stereotest

In the Wirt Circles test, you push down the “raised” circles that appear closer.

In the Fly test, you try to “pinch” the fly’s wings that seem to pop out in 3D.

The special glasses create binocular disparity, similar to how 3D movie glasses work, allowing each eye to see a slightly different image and produce a sense of depth.

Correspondence Problem

Refers to how the visual system determines which parts of the two retinal images correspond to the same object in the world.
To find the angle of disparity, the brain must match images from the slightly different viewpoints of each eye.
Exactly how this matching happens is not fully understood.

Problems with Matching Features

If images are too complex or lack clear features, it’s hard for the visual system to find matching points between the two eyes.

Random Dot Stereograms

Made of random dot patterns with no distinct features to match.
They show that depth perception (stereopsis) can occur without recognizable objects—the brain can match patterns purely based on disparity, not identifiable shapes.

Disparity-Selective Cells (binocular depth cells)

Neurons in V1 that fire when the two eyes’ images have a specific retinal disparity.
Each cell is tuned to a particular disparity, helping the brain judge whether objects are near or far.

Disparity Tuning Curve

a graph showing how active (how much the cell fires) a neuron is for different disparities.

Selective Rearing Experiments (disparity)

Cats were raised with only one eye open each day, so their eyes never worked together.
After 6 months, they had few binocular neurons and poor depth perception on tasks
This shows that disparity-selective cells are needed for normal depth perception.

DeAngelis et al. (1998) - binocular depth perception

Monkeys were trained to judge depth using random dot patterns.
When researchers stimulated disparity-selective cells, the monkeys’ depth judgments changed — showing these cells directly influence depth perception.

Disparity-Selective Cells

Found not only in V1, but also in MT, V2, V3, and parts of the temporal cortex.
Minini et al. (2010): fMRI studies showed disparity responses in these higher visual areas too.

Binocular Cues (predators)

Requires Frontal Eyes/overlapping fields of vision

• humans, cats, dogs, bears, and more predators

Binocular Cues (prey)

Lateral eyes

• eyes on the side of the head

• deer, horses, rabbits, etc.

• much less overlap in visual field

• trade off overall detection vs accuracy of depth

Size & Depth perception

Your brain judges true size by using distance.
If distance cues are unclear, size judgments become inaccurate.

For example, in a whiteout, a helicopter pilot can’t see depth or distance clearly, so it’s hard to tell how close or far (and how big) objects are — which can be dangerous.

Visual Angle

How big something appears on your retina (in degrees).
Closer/larger objects → larger visual angle.
Farther/smaller objects → smaller visual angle.
Same-size objects at different distances look different in size; different-size objects at different distances can look the same if their visual angles match.

At the same distance (perceiving size)

the bigger one will take up a larger visual angle — meaning it looks bigger on your retina.

Holway & Boring (1941)

Tested how distance cues affect size perception by having participants match a variable circle to a fixed-size circle at different distances.

Viewing Conditions (removing cues one by one):

• Monocular viewing → removes retinal disparity.

• Peep hole → removes motion parallax.

• Dim lighting → removes pictorial cues (e.g., linear perspective).

Findings:

• With full depth cues, size judgments were accurate.

• As cues were removed, judgments got worse; people relied on retinal image size, so farther circles looked smaller.

Conclusion:
Size perception depends on accurate distance perception.

Size Constancy

means that we see an object as being the same size, even when it’s closer or farther away.

For example:
When a friend walks away from you, their image on your retina gets smaller, but you still know they’re the same height — your brain adjusts for the change in distance.

Size-Distance Scaling (Holway & Boring)

our brain keeps size constancy by combining:

• Visual angle (how big the image is on your retina)

• Perceived distance

If something looks small but is far away, your brain interprets it as actually large.
Holway & Boring showed that when distance cues are removed, this scaling breaks down and people rely only on retinal size.

Size-Distance scaling Equation

s = perceived size

k = constant

r = retinal image size

d = perceived distance

s = k(RxD)

Red Dot Demo

create an after image of a red dot 

look at both a near and far object 

retinal image stays constant

• perceived size changes with the change in perceived distance

Emmert’s Law

Your afterimage grows or shrinks depending on how far away the surface is, because your brain scales size based on perceived distance.

• match size - distance scaling equation

Optical Illusions & Depth/Size Cues

Many optical illusions trick your brain by messing with the same cues you use to judge size and distance.
But not every illusion works the same way — some rely heavily on depth cues, others on size cues, and some on both.

MuLler-Lyer Illusion

You see two lines that are actually the same length, but the fins at the ends make one look longer or shorter.

Outward fins ( >—< ) → looks longer

Inward fins ( <—> ) → looks shorter

Gregory Explanation (1966)

Your brain applies size constancy scaling as if the lines were 3D corners.

• Outward fins ( >—< ) → interpreted as an inside corner → seen as farther → brain scales up → looks longer.

• Inward fins ( <—> ) → interpreted as an outside corner → seen as closer → brain scales down → looks shorter.

Even though the lines are the same length, depth-based scaling makes them appear different.

Dumbbell

no depth cues

3D Fins

positioned so interior vs. exterior corners are equidistant

can be replicated with books on your desk

• i.e. distance x = distance y

Conflict Cues Theory

Day (1989/1990) argued the Müller-Lyer illusion isn’t mainly about depth—it’s about your brain trying to combine two conflicting cues:

Actual line length
Your brain knows the two lines themselves are the same length.

Overall figure length
The fins (or circles) change the global size of the whole shape, so one figure looks larger overall than the other.

Fins/Dumbbell circles 

Your brain tries to merge these two cues—actual line length and the global size of the whole figure—and the bigger-looking figure wins, making its line appear longer even though they’re identical.

Ponzo Illusion (railroad track illusion)

Two identical objects placed between converging lines appear different in size.
The converging lines act as depth cues, making the upper object seem farther away.
Your brain applies size–distance scaling, so the “farther” object looks larger, even though both are the same physical size.

Ponzo Illusion works directly on distance cues:

linear perspective convince your brain that one object is farther away, and your size–distance scaling system overcompensates, making the “farther” object look larger even though both are identical.

Ames Room

• A specially constructed room viewed through a peep hole.

• Designed to look rectangular, but one corner is actually farther and the other closer.

• Floor patterns and windows are distorted to trick linear perspective.

Effect:
Your brain assumes both corners are the same distance, so size–distance scaling fails.

• Far corner → smaller retinal image → person looks tiny

• Near corner → larger retinal image → person looks huge

The hidden distance difference makes accurate size perception impossible.

Harvest Moon (moon illusion)

looks huge on the horizon even though its visual angle doesn’t change.
Your brain thinks the horizon Moon is farther away (because of depth cues like trees/buildings), so it scales up its perceived size.

Super Moon (moon illusion)

actually physically closer, so its retinal image really is larger, but we only perceive a small size increase.

Apparent Distance Theory (moon illusion)

• Horizon Moon: Seen with distance cues (trees, buildings, landscape) → brain interprets it as farther away.

• Sky Moon: Surrounded by empty space → brain interprets it as closer.

• Since both Moons have the same visual angle, a greater perceived distance makes the horizon Moon appear larger due to size–distance scaling.

Angular Size Contrast Theory (moon illusion)

• The Moon looks larger on the horizon because it’s compared to a smaller, compressed-looking portion of the sky.

• High in the sky, the surrounding sky looks huge and expansive, so the same Moon appears smaller by comparison.

• The effect comes from contrast with the surrounding sky, not any real change in the Moon’s size.