Class Notes (1,100,000)
CA (650,000)
Western (60,000)
PSYCH (8,000)

Psychology 2135A/B Lecture Notes - Gestalt Psychology, Ames Room, Change Blindness

Course Code
Robert Brown

of 3
Lecture 3 - Object Perception
- pattern recognition is more general
- recognize objects with pattern recognition
Why is object recognition difficult?
- image on retina does not constrain the possibilities; not enough information
- human faces probably single most important visual stimulus
- whether or not we can correctly recognize important differences
- must recognize things to engage in right behaviour
- must find visual cues
- somehow know objects do not change with different orientations
- somehow able to generate name of objects
- do not need to see whole head to recognize/distinguish faces
The stimulus on the receptors is ambiguous
- inverse projection problem
- image on retina can be caused by an infinite number of objects
- objects can be hidden or blurred
- common, but typically do not hinder environment
Objects look different from different viewpoints
- viewpoint invariance: ability to recognize object regardless of viewpoint
Two kinds of theories
- some of how we perceive things is in our culture, and some is in our evolution
The stimulus is sufficient
Structuralist Approach
- there is enough information in the stimulus to explain our perceptions
- percept of object is sum of combining elements
- cannot explain apparent movement or illusory contours since observer sees things not present in
- apparent movement occurs when we see something moving across when it is not really moving
due to the small amount of time between stimulus
- bi-stable: two different perceptions; can flip between two, can stay with one
Direct Perception
- 3 principles:
1. all information required for object perception is available in environment
2. perception is immediate; no need for inferences
3. perception is for action; perception guides action and action generates new percepts
- as we move, we get different perceptions of same objects and learn that they do not change
- a problem is that stimulus available at the retina can be same with different sizes and distances
- different angles create different shapes on the retina
- response is that you can tell difference when you move
- movement provides new information that resolves uncertainty
- Ames room goes against this hypothesis
- we bring assumptions that may distort perception
The stimulus is insufficient
Gestalt theory
- not enough information in stimulus to explain perception
- nervous systems add information, based on experience, to what is available to senses
- perception guided by principles of organization of stimuli (same thing as previous note)
- whole differs from sum of parts; not build from sensations, but result of perceptual organization
- perceptual organization involves grouping of elements in a visual stimulus to produce a larger
object; grouping process requires knowledge of the world
- principles:
- good form (Pragnanz): every stimulus is seen as simply as possible
- similarity: similar things grouped together
- good continuation: lines seen as smoothest path
- proximity: things that are near each other are grouped
- meaningfulness or familiarity: form groups that are familiar or meaningful
Unconscious Inference - Bottom-Up Processes
- data-driven or stimulus driven
- stimulus determines final percept
- does not use contextual information
- four classes: template, prototype, feature analysis, recognition by components
- template: match image to template, pattern stored in memory; when match found, stimulus
- problems: many templates for different orientations, how do we learn new objects,
template for every colour/texture?
- prototype: do not need exact match, just similar
- problems: how similar is similar enough
- feature analysis: objects have sets of features that are stored in memory, match features; small
set of features should allow perception of large number of objects
- problems: does not explain how spatial relations are processed i.e. which feature goes
where and what is connected to what; nobody able to say what the features are, no principle to
specify what a feature is; "little templates", so some problems from template theory
- recognition-by-components theory: made by Biederman; objects are recognized by volumetric
features called geons (geometric ions)
- 36 geons that combine to make all 3D objects
- geons include cylinders, rectangular solids, and pyramids
- includes specification of spatial relationships
- Properties of Geons:
- each geon has unique non-accidental properties; properties of edges in the retinal image
that correspond to edges of objects in 3D environment
- view-invariant properties: aspects of the object that remain visible from different
viewpoints; some viewpoints that cannot distinguish geons
- discriminability - ability to distinguish geons from one another, which depends upon
- principle of componential recovery - ability to recognize an object if we can identify its
- obscure geons create impossible recovery of objects
Unconscious Inference - Top-Down Processes
- context contributes to our ability to recognize
- theory-driven or conceptually driven
- knowledge, expectations influence perception, based on experience
- inferences speed up visual processing - but can be a source of misperceptions i.e. context
effects, change blindness
- context determines how we see ambiguous stimulus
- once we see something, we cannot un-see it; cannot prevent process, just take output
- objects that fit context easier to identify in brief flashes
- knowledge of world influence perceptual processing of scene
- change blindness: unable to see large changes; process very little in the world; what we notice
is what matters to us