Module 7,8 & 9 Flashcards

(43 cards)

1
Q

State Weber’s Law and define the Weber fraction.

A

Weber’s Law holds that the ratio of the DL (ΔS) to the standard stimulus (S) is constant (K):

Δ
𝑆
𝑆
=
𝐾
S
ΔS

=K
The Weber fraction K is this constant ratio (e.g., 0.02 for weight discrimination).

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is Stevens’s Power Law and its formula?

A

It describes perceived magnitude (P) as a power function of stimulus intensity (S):

𝑃 =𝐾*𝑆^𝑛

where K is a constant and n is the exponent.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What does the exponent (n) signify in Stevens’s Power Law?

A

n < 1: Compresses perception (e.g., brightness)

n ≈ 1: Linear scaling (e.g., electric shock)

n > 1: Expands perception (e.g., perceived heaviness)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What is a Receiver Operating Characteristic (ROC) curve?

A

It’s a plot of hit rate versus false‐alarm rate across different decision criteria, showing the trade‐off between sensitivity and specificity.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

How does the shape of an ROC curve reflect sensitivity, and what does it mean if two observers’ points lie on the same curve?

A

The curve’s shape encodes an observer’s sensitivity independently of their criterion. If two observers’ hit/FA points all lie on the same ROC curve, they have equal sensitivity.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

How do criterion shifts affect the ROC curve and an observer’s sensitivity?

A

Shifting the decision criterion (e.g., via payoffs or instructions) moves the operating point along the ROC curve.

Sensitivity (area/shape of the curve) does not change with criterion shifts.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What does sensitivity (d′) represent in Signal Detection Theory, and how does it affect the ROC curve?

A

d′ is the distance between the peaks of the noise (N) and signal+noise (S + N) distributions.

Larger d′: Distributions are farther apart → ROC curve bows sharply toward the upper‐left (high sensitivity).

Smaller d′: More overlap → ROC curve lies closer to the diagonal (low sensitivity).

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

How does the steepness of the psychometric function slope relate to the JND?

A

The psychometric function is a curve that relates a measure of perceptual experience (e.g., whether a person detects a stimulus) to the intensity of a physical stimulus.

Steeper slope → smaller JND (higher discriminability)

Shallower slope → larger JND (lower discriminability)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What effect does bias (criterion shifts) have on ROC performance?

A

Bias moves the operating point along the existing ROC curve, changing hit and false‐alarm rates but not sensitivity.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Opponent Processing

A

After the initial cone-based encoding, visual signals are fed into antagonistic neural channels that compare cone outputs in pairs.
- These opponent channels typically include:
- A red–green channel, which contrasts signals from “long-wavelength” cones against “medium-wavelength” cones
- A blue–yellow channel, which pits “short-wavelength” cone signals against the combined output of the other two cone types
- A luminance channel, summing inputs to encode brightness rather than hue

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What role do retinal ganglion cells (RGCs) play?

A

They act as opponent neurons, combining excitatory (+) and inhibitory (–) cone inputs.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Which circuit is “+S – (M + L)”?

A

The blue–yellow opponent channel excited by S, inhibited by M+L.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Which circuit is “+(M + L) – S”?

A

The yellow–blue opponent channel excited by M+L, inhibited by S.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What perceptual pair does “+L – M” correspond to?

A

Red–green (excited by L/yellow-reddish, inhibited by M/greenish).

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What perceptual pair does “+M – L” correspond to?

A

Green–red (excited by M/greenish-yellowish, inhibited by L/yellowish-reddish).

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What does “uniform center–surround receptive field” mean for these RGCs?

A

They respond best when their entire RF sees a uniform patch of preferred (or opponent) color.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

A +L–M RGC sees a uniform green field—what happens?

A

It is suppressed (firing rate drops), because M (green) inhibits that cell.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

What is a single-opponent neuron in V1? What stimulus best drives single-opponent neurons?

A

A cell with a center–surround RF where the center is driven by one cone type and the surround by its opponent. A uniform patch of their preferred hue (average color over the RF).Encode the average hue within a region (detect uniform color fields).

19
Q

What makes a neuron “double-opponent”? What stimulus best drives double-opponent neurons?

A

Both its center and its surround compare two cone inputs with opposite signs (e.g. center +L/–M, surround +M/–L).A sharp border between two opponent hues (chromatic edge).Detect local hue transitions (chromatic contours) even without luminance contrast.

20
Q

What is “discounting the illuminant”?

A

The brain estimates and removes the light source’s effect to recover the object’s true color.

21
Q

What is chromatic adaptation?

A

Cones adjust sensitivity after exposure to dominant light, helping maintain stable color perception.

22
Q

Dichromacy (Partial Color Blindness)

A

Mechanism: Missing exactly one cone type → two remaining receptor classes.
- Types & Frequencies:
- Protanopia (no L-cones): ~1% of males, 0.02% of females.
- Deuteranopia (no M-cones): ~1% of males, 0.01% of females.
- Tritanopia (no S-cones): ~0.002% of population.

23
Q

What are oculomotor depth cues?

A

Convergence (eye rotation) and accommodation (lens shape changes).

24
Q

Retinal Image Cues What are some key visual cues the brain uses to perceive depth from 2D retinal images?

A

Binocular disparity (differences between eyes),

partial occlusion (closer objects block others),

motion parallax (nearby objects move faster across the retina),

visual angle (smaller angle = farther object),

familiar size (known objects help judge distance), and

linear perspective (parallel lines appear to converge with distance).

25
Types of Retinal Disparity & Horopter
Horopter: The imaginary 3D surface in space where objects project to corresponding retinal points in both eyes (e.g., both foveas) → seen as single and aligned. Zero Disparity: Object is on the horopter → no vergence change needed. Crossed Disparity: Object is closer than the horopter → eyes must increase convergence to fixate. Uncrossed Disparity: Object is farther than the horopter → eyes must decrease convergence to fixate.
26
What is the correspondence problem in depth perception?
The challenge of matching features from the left and right eye images so the brain can compute binocular disparity and perceive depth.
27
What are the two main hypotheses for solving the correspondence problem?
Recognition-First: Recognize objects in 2D first, then match. Matching-First: Match simple image features first, then recognize objects. (most plausible)
28
What is a random dot stereogram (RDS) and what does it prove?
An image with identical random dots in both eyes except a small shifted region. It proves depth can be perceived before object recognition.
29
What are Marr & Poggio’s two constraints for matching features?
Uniqueness – Each feature matches only one feature in the other eye. Continuity – Nearby features usually have similar depth.
30
What do binocular cells respond to?
They fire when both eyes view the same object with a specific disparity (zero, crossed, or uncrossed).
31
Where are disparity-tuned cells found in the brain?
Early areas (V1–V3): Start depth processing. Dorsal stream: Guides action (e.g., area MT). Ventral stream: Helps with object recognition (e.g., V4).
32
What is Bayesian cue integration in depth perception?
The brain weights each cue by reliability and combines them with prior knowledge to compute the most likely 3D layout.
33
What does Emmert’s Law demonstrate?
Size–distance invariance: an afterimage appears larger on a farther surface even though its retinal size is fixed.
34
What is shape constancy?
Perceived shape of an object stays the same despite changes in the retinal image caused by orientation.
35
What is shape–slant invariance?
Perceived shape depends on estimated slant; correcting for slant (via disparity, shading, texture) gives correct perceived shape.
36
What is forced perspective?
An illusion where near and far objects are deliberately aligned to appear at the same depth, tricking size perception
37
What causes the Ponzo Illusion?
Linear perspective (e.g., converging lines) makes identical objects appear different in size due to implied depth.
38
How does the Ames Room create a size illusion?
A trapezoidal room appears rectangular from one viewpoint. It manipulates depth cues, making equal-height people appear different in size.
39
What explains the Moon Illusion?
The Moon looks larger on the horizon because it's perceived as farther away. By size–distance invariance, its perceived size increases.
40
What causes the Tabletop Illusion?
Identical tabletops are seen with different slants, leading to shape–slant corrections that alter perceived shape.
41
Which brain areas respond to pictorial depth cues?
Parietal cortex neurons respond to slant defined by texture gradients and binocular disparity; other neurons respond to motion parallax.
42
Where are disparity-selective cells found and what do they encode?
V1: Encodes absolute disparity. Temporal lobe: Encodes relative disparity, which stays constant across fixations.
43
What is the size–distance scaling equation?
S = R × D S = perceived size R = retinal image size D = perceived distance