Object recognition, categorisation and achieving object constancy

HideShow resource information
  • Created by: The Shrew
  • Created on: 21-01-16 15:19
Generalising
Achieving invariance across different views and body positions
1 of 41
Selectivity and discrimination
see differences in individuals and generalise to a wider group
2 of 41
Invariant features theory
speed of detecting if a picture is the same or rotated- time cost= view sensitivity/ no cost= view invariance
3 of 41
Invariant features predict
view invariance
4 of 41
No invariant features leads to
view sensitivity
5 of 41
There are probably no invariant features that
uniquely discriminate all everyday objects and are visible under all viewing conditions
6 of 41
Wittgenstein
No commonalities between games
7 of 41
Template theory
Inputs of views are stored in 2D- lots of templates for each object- no transformation or decomposition of input
8 of 41
Large LTM demands but
less online processing= faster
9 of 41
Hierarchy of cortical neurones
respond to increasingly complex info, start with simple features
10 of 41
Demonstrated
electrophysiologically
11 of 41
Top of hierarchy
Grandmother cells respond to only specific and complex objects
12 of 41
Evidence of specific face neurones in
monkey inferotemporal cortex
13 of 41
fMRI studies
different regions in brain for faces, stick figures, body parts
14 of 41
If stimuli less socially important
little evidence for templates
15 of 41
Templates bad
too inflexible/ too much storage
16 of 41
Multiple views and transformations
Input-> transform and normalise-> match to templates-> identify
17 of 41
Upright
easies/ 180 degree plane rotation= hardest
18 of 41
Deciding if letter inverted or not
Easiest when upright
19 of 41
Fewer stores needed
but higher computational demands at transformation stage
20 of 41
Structural descriptions
Biederman- visual systems extracts non-accidental features which are invariant across different views of an object- not sufficiently unique to identify object on their own but several together can be used to infer which geons are present in an objec
21 of 41
Geon structure- internal reconstruction in 3D
produced by geons and spatial relations
22 of 41
Computationally insensitive
too fast?
23 of 41
Palmer et al
canonical view- view that is prefered
24 of 41
Lawson et al
problems recognising unusual views of objects
25 of 41
Bent wire stimulus/ wire frame/ blocks
hard to detect from different angles- supports multiple view theory and provides evidence against structural description theory
26 of 41
Biederman and Gerhardstein
Canonical views preferred because they are more useful and informative rather than because they match template
27 of 41
Results with novel objects
don't tell us much about recognising familiar, everyday objects- novel objects similar to each other
28 of 41
Priming tasks
Objects named twice in 2 blocks- name objects faster second time
29 of 41
Block 1= 820ms
Block 2= same exemplar= 610ms/ 135 degree rotation= 610ms/ different exemplar= 680ms/ different exemplar at 135 degree rotation= 710ms
30 of 41
Lawson and Humphreys
Significantly more priming when object primed by identical view of that object compared to if it was primed by a different view
31 of 41
Use all four
Depends on stimuli, complexity, familiarity
32 of 41
Object categorisation
Rosch et al
33 of 41
Subordinate level
Specific- extra work for little extra info
34 of 41
Basic level
Input- generally preferred level
35 of 41
Superordinate
little visual or semantic info
36 of 41
Basic level preferred
Cognitive economy
37 of 41
Jolicoeur et al
people name typical members of category faster at basic level
38 of 41
Unusual members of categories
subordinate level rather than basic- maximise cognitive economy
39 of 41
Rosch et al
experts more likely to use subordinate levels/ children more likely to use superordinate
40 of 41
Johnson and Mervis
Fish and bird experts
41 of 41

Other cards in this set

Card 2

Front

Selectivity and discrimination

Back

see differences in individuals and generalise to a wider group

Card 3

Front

Invariant features theory

Back

Preview of the front of card 3

Card 4

Front

Invariant features predict

Back

Preview of the front of card 4

Card 5

Front

No invariant features leads to

Back

Preview of the front of card 5
View more cards

Comments

No comments have yet been made

Similar Psychology resources:

See all Psychology resources »See all Cognitive Psychology resources »