Scene perception

Question 1

What is a scene?

Answer 1

A semantically coherent view of real-world environment comprising background elements and multiple discrete objects arranged in a spatially licensed manner

Question 2

Compare and contrast advantages and disadvantages of naturalistic and artificial scenes

Answer 2

ARTIFICIAL - Precise control so clear conclusions but may not be fully representative of natural behaviour
NATURAL - Normal behaviour but more cluttered and complex so harder to control variables

Question 3

What 4 variables need to be controlled for in experimental stimuli?

Answer 3

VISUAL - Colour, luminance, contrast etc
TACTILE - Surfaces/textures, contours
SOCIAL - Crowding (visual complexity)
SHERLOCK - Familiarity and predictability

ALL NEED TO BE CONTROLLED FOR TO ALLOW FOR ACCURATE CONCLUSIONS

Question 4

What did Biederman find in his scene perception experiments on scene coherence?

Answer 4

Participants shown normal image, or same image jumbled into 6 sections with only one target section left in its original place
Shown for 1 second then masked –> shown cue for target location and asked what object was there
There was better object discrimination for the normal scene

Question 5

What was a criticism of Biederman’s work?

Answer 5

The jumbled scenes were more visually complex so any differences in discrimination could be due to this rather than actual visual processing (variables such as contours not controlled)

Question 6

What are the 5 LAWS OF COHERENCE?

Answer 6

SYNTACTIC (based on laws of physics):
1) SUPPORT i.e. obeys gravity 
2) INTERPOSITION i.e. foreground should occlude background 
SEMANTIC (learned by experience)
3) PROBABILITY 
4)SIZE 
5) POSITION

Question 7

What determines where our eyes move?

Answer 7

Depends on information from peripheral vision e.g. cues from visually salient stimuli

Question 8

What determines how many fixations are made in an area?

Answer 8

Both top-down and bottom-up processing e.g. when we recognise something as a face, and informative quality such as when someone is talking

Question 9

What are 2 examples of early eye movement studies?

Answer 9

BUSWELL - used old paintings found that fixations on areas of specific informational value
YARBUS - same image but different instructions –> instructions changed eye scan paths

Suggest that our eyes gather a general gist of a scene and then focus on regions of informational value

Question 10

What was Loftus and Mackworth’s experiment for semantic informativeness (i.e. predictability) ?

Answer 10

Line drawings
Target with high or low semantic informativeness
Greater fixation density found on semantically informative (e.g. unpredicted) regions and through use of extrafoveal vision any semantically inconsistent objects were fixated earlier

Question 11

What 3 possible reasons can explain why other studies have not replicated Loftus’s findings?

Answer 11

Possible confounding effect of visual complexity
Very low scene complexity i.e. not naturalistic
Highly incongruous semantically informative objects

Question 12

What did Reingold et al test in their experiments?

Answer 12

The effect of expertise

Question 13

What is the gaze-contingent window technique?

Answer 13

Use different size experimental “windows” when observing a chess board - the windows restrict covert attention for determination of next fixation
Larger windows give more access to the perceptually degraded peripheral info

Question 14

What is Perceptual Span?

Answer 14

The smallest window size that has no effect on performance
Amount of visual field we can process in one fixation - experts found to need larger window i.e. they take in more visual field info in one fixation

Question 15

What is the change blindness technique?

Answer 15

Flicker between 2 images and see if spot differences i.e. ability to discriminate stimuli within visual field when no visual cue

Question 16

What were Reingold et al’s findings?

Answer 16

Random configurations - expertise had no effect
Normal configuration - experts had greater visual span and faster reaction times for detection of differences i.e. when a scene has meaning for us we process more of it in a single fixation
Good example is learner vs experienced drivers

Question 17

What are Global Image Features and what are the 5 types?

Answer 17

Help us in perceiving the gist of a scene - can be perceived rapidly and holistically

1) Degree of naturalness e.g. textured zones, contours
2) Degree of openness e.g. visible horizon line
3) Degree of roughness e.g. rougher has more small elements and appears more complex
4) Degree of expansion e.g. convergence of parallel lines
5) Colour e.g. green suggests foresty

Question 18

How does use of global image features indicate use of top-down processing?

Answer 18

Past experiences play a role in determining associations made e.g. with colours
Make use of regularities in the environment without much conscious awareness (result from experience-dependent plasticity)

Question 19

What are 2 examples of physical regularities?

Answer 19

GESTALT LAW OF GOOD CONTINUATION - when one object partially covers another we still see one whole object rather than 2 parts
Shape from shading - assume light source above so a pattern higher up on something indicates bulging out while lighter at the bottom indicates indentation

Question 20

How did the “multiple personalities of a blob” experiment illustrate our use of semantic regularities?

Answer 20

We use our knowledge of a scene to help us to perceive an ambiguous object

Question 21

What is the role of inference in scene perception?

Answer 21

We use our knowledge of physical and semantic regularities to infer what is present in a scene (remember Helmholtz’s unconscious inferences based off idea of subjective idealism)
Correlates with the LIKELIHOOD PRINCIPLE which is when we predict from experience what an object should be based on context

Question 22

How have psychologists quantified Helmholtz’s ideas about inference?

Answer 22

Using BAYESIAN INFERENCE i.e. statistically technique that assesses probabilities and expresses likelihood in mathematical formulas