W10: Concepts & Categorisation Flashcards
categorisation and cognition
Understanding of categorisation
Categorisation is the engine of cognition
Fundamental ideas for understanding categorisation:
1) Importance of categorisation
2) Generalization
3) Selective attention
What is categorisation?
Ability
The ability to form equivalence classes (treat similarly) of discriminable (distinguishable) entities
Categorisation is largely automatic
Reasons?
Survival:
Provides a basis of deciding what constitutes appropriate action
Facial categorisation: warm and competence
simple organisms categorize:
E. coli: cat nutrient source
Female Mouse: unfamiliar male mouse
Benefits of categorisation
4
1) Provides a means for identification
2) Reduces the complexity of the environment:
World Color Survey
3) Allows organisation of knowledge
4) Allows for generalization
Categorisation Induction:
meaning
Induction: Generalizing from the particular to the general
Generalization is sensitive to the elements in the category
4 key empirical effects that help us understand generalization via induction
and their effect on generalization (how to get greater gen)
1) Typicality of Instances (more strongly relate to category) 2) Typicality of category (more typical category members) 3) Category Size (greater to more specific (smaller) categories) 4) Category Variability (greater when the examples are more variable)
Development of categorisation:
2 studies related to habituation
1) Habituation in Infants
Focusing more on “novel” stimuli during successive exposures
Fantz (1964)
2) Category Habituation: After a familiarization phase (keep showing animal from same cat), infant in general fixate on new cat images compared to horse CATS -> 60.1% ZEBRAS -> 62.1% GIRAFFES -> 57.2% Eimas & Quinn (1994)
Development of Categorisation
Infants VS adults
Rudimentary categorisation is evident in infancy:
3-4 Month Old Infants have increased looking times for novel categories
There is amazing flexibility in categorisation:
Adults can acquire exceedingly complex, non-perceptual categories; children can not
Children’s categories are focused on perceptual grouping
Children seems to attend to all features.
Abstract categories based on:
3
1) unobservable attributes (e.g., love, doubt, thought)
2) relational concepts (e.g., enemy or barrier)
3) rules (e.g., island, uncle, or acceleration)
What may have contributed to the dif from adults and infants?
Children’s categories are focused on perceptual grouping
And the ability to represent things that aren’t perceptually available.
Selective attention:
Factor which determines the influence or weight of a stimulus dimension or attribute on categorisation
Categories which require you to pay attention to more dimensions are harder to learn (how does selective attention helps here)
Different type of categories: Type I (Easy) Type II (mediam) Type III-IV: (difficult) Type 6: (Extremely difficult)
Type I: single dimension matters:
Color OR shape…
Type II: Two dimensions matter:
Color AND shape
Type III-IV: Rule + exceptions:
Learning a rule + exception is more difficult than learning a “stable” rule even if that rule requires more than one dimension
Type VI: Three dimensions relevant
No rule can be used to simplify the learning
Need hard learning (memorising)
Category Learning Task
Shepard, Hovland & Jenkins (1961)
S37-44
P were presented 2 categories and they are to sort them to cat (cat A or B?)
Instant feedbacks were given
Results:
Type I: Very fast learning (after 4 blocks). Very easy
Type II: slower (after 10 blocks). more difficult
Type III-VI: much slower (~ 16 blocks). difficult
Type 6: very difficult to learn (still .1 error after 16 blocks)
CONCLUSION:
category learning difficulty depends on how many different dimensions are used to define the category
What is the consequence of selective attention?
2
Selective attention has limited capacity:
cannot attend to everything
attending to one ==not attending to other aspects
Children have a hard time focusing attention on one important thing:
children seem to use all of the features
Different dimensions of Categorisation for Selective attention
Handel & Imai (1972)
2 dimensions:
given 3 objects (same shape). ask them to sort them into cat: color OR size
sort on the basis of a single dimension:
SIZE
sort on the basis of similarity across both size and colour:
2D
Categorisation between adults and children
Selective attention
Smith (1989)
S47 - S55
Conditions
2 dimensions:
3 object
color/size
Condition: 3
Discriminable (similarity increased between objects)
Standard (same as Handel & Imai (1972))
Extreme (similarity decreased)
S49
Looked through the whole lifespan.
Categorisation between adults and children
Smith (1989)
S47 - S55
Results
Adults use a one dimensional grouping (mainly size)
2 and 3 year olds:
Standard: tend to group based on similarity (2D grping)
Discriminable: all in one grp > similarity
Extreme: All dif cat ~= similarity
4-yr & 5-yr olds:
continue to group by similarity when objects are less discriminable (and standard),
Use a one dimensional-grouping when the objects are far apart (Extreme)
8-yr olds:
use a one dimensional-grouping unless the objects are very similar (discriminable)
Categorisation between adults and children
Smith (1989)
S47 - S55
conclusions
Children very different from adults
can’t resist the urge to look at other aspects
3 to 4 yr olds-> discriminate finer quantities starts to develop
Adults sort stimuli on the basis of a single individual dimensions of those stimuli:
Utilising their ability of selective attention to simplify the world
This ability takes time to develop:
Children tend to sort on the basis of overall similarity across all dimensions
Different dimensions of Categorisation
Selective attention
Rehder & Hoffman (2005)
Eye-tracking reflects attention:
Similar to Shepard, Hovland & Jenkins (1961)
Eye-tracking movement showed that as blocks increases:
Type I: Fixated on 1D (6)
Type II: Fixated on 2D (10)
Type 3,4,5: slowly goes to 3D
All start ard 2D,3D then slowly fixate on respective Ds.
Real word example:
Why is identifying recycling bins easy?
But recycling itself…
Identifying Bins is like a Type I problem
recycling is difficult (type 6) need to memorise….
Memorising the features of categories is difficult but knowing the reason why different categories exists makes them easier to learn
Causal Models
Murphy & Medin (1985)
categorization should be seen as a process of explaining why an exemplar belongs with the rest of a category
Knowing the cause underlying the category provides an additional “deeper” dimension or feature that can be used to understand the category
Real word example:
What is the best way to understand the dif cat. of pre- and post- renaissance paintings?
Hockney’s Hypothesis: Camera Obscura
the painters from this period may have used optics to achieve the photorealistic quality of their paintings. In particular, he hypothesized that painters from this latter period may have used a camera obscura in order to render the more true to life images
Categorisation and Causal Reasoning
Defining the relevant features which separate the categories
Allows for the development of theories regarding the cause of those differences
Categorisation and Causal Reasoning Example: Camera Obscura & Causal reasoning S65 66, 67, 68 good and faults
Good model as it was falsifiable.
To-be-explained:
Difference between representational painting to photorealism
Theory:
Introduction of optics
Hypothesis:
Hockney’s Hypothesis: Camera Obscura
Proves and faults: provides quite a general explanation for a general behaviour BUT can be done without optics light-colour issues
Categorisation and Causal Reasoning Example: comparative mirror technique VS Camera Obscura
To-be-explained:
How did Vermeer paint the photorealistic painting The Music Lesson?
Hypothesis:
Optics but specifically the comparative mirror technique
Comparative Mirror is better than the Camera Obscura hypothesis because it explains how one could get accurate detail and colour matching
Comparative Mirror Hypothesis ->
A Causal Model
The simulation makes the comparative mirror theory a model:
It becomes a set of PROCEDURE, an ALGORITHM, for producing PREDICTIONS
produces data which look to an observer like the actual data we want to explain
Explains several well-known facts about Vermeer’s painting
Causal Theory of Categorisation
Surface differences between categories are often assumed to be caused by some underlying deep difference
(Even if not understood)
- drive science and can lead to prejudice
Causal theories can drive generalisation
Simplify our understanding of the word
