Concepts and
Conceptual Structure
L.P. Alampay
Cognitive Psychology
Our concepts reflect how we
divide the world into
classes or categories.
To categorize is to treat two
or more distinct entities as
the SAME or EQUIVALENT.
The world can be
partitioned in a
limitless variety of
ways, yet people find
only a miniscule subset
of possible
classifications to be
Functions of Concepts
1. Cognitive economy
By partitioning the world into
categories, we decrease the
amount of information we
must perceive, learn,
remember, communicate, and
reason about
Functions of Concepts
Enable us to infer beyond
the given information
Can infer non-perceptible
attributes about an object
(even if unfamiliar) once we
are able to classify it
Provide us with expectations
to guide our behaviors
QuickTimeª and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Functions of Concepts
1. Can be combined to form more
complex concepts and thoughts
Our understanding of complex
ideas hinges on understanding
of constituent concepts
Ang buhay ko ay parang
telenovela.
A dog is man’s best friend.
The structure &
content of conceptual
knowledge
WHAT is being represented?
Theories of concepts should explain
the coherence of categories, and
why some categories seem natural
(or privileged) and others are not.
the class of things that weigh less than a
ton and do not have stripes…
the class of things that have feathers and
can fly…
View 1: Feature-based
or Similarity-based
views
Our conceptual knowledge
consists of the properties,
features, or attributes of
the instances of the concept
or the members of the
category
We put things together
because they are SIMILAR.
Feature-based views:
Classical
Concepts are defined by sets of
individually necessary and
collectively sufficient
features.
- Means that every instance of
the concept must possess ALL
features
- And that objects that have all
those features MUST be
classified as instances of that
Feature-based views:
Classical
Bachelor = male + adult +
unmarried
QuickTimeª and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Triangle = closed form + three
sides + interior angles sum
Feature-based views:
Classical
What are the defining features
of the following concepts?
Cup
Chair
Vegetable
Problems with the
Classical View
Failure to specify defining
properties
- There seems to be no necessary
and sufficient features to
define most everyday concepts
- Unclear cases regarding
category membership
- Disagreement over whether
objects are instances of
certain categories
- Am I friend or lover ???
Problems with the
Classical View
3. Typicality effects
- People can think about objects
in terms of being more “typical”
or better/worse instances of a
concept
- Measured by speed of
categorization, latency of
memory retrieval, recognition
tasks
- e.g., Malt & Smith, 1984
Feature-based views:
Probabilistic
Concepts are organized around a
set of features that are only
characteristic or typical
among category members.
- Category membership is thus
fuzzy or ill-defined
Feature-based views:
Probabilistic
Family Resemblance principle
The more typical instances of a
category are those that share
more features with other
members of the category
Feature-based views:
Probabilistic
Concept represented as
Prototype
An abstracted, MOST TYPICAL
instance that possesses the most
characteristic features
Concept represented as
Exemplars
BUT! Typicality effects do not
rule out classical views of
conceptual structure.
Typicality effects were
demonstrated EVEN IF subjects
were able to state the defining
properties of concepts
(Armstrong, Gleitman, &
Gleitman, 1983)
odd number
even number
Feature-based views:
Dual-code
Representation?
We code for two representations of
a concept:
A prototype (i.e. most typical)
- May be used for “quick and dirty”
identification purposes
- A core consisting of properties
that are necessary for
determining category membership
- Defines boundaries, used for
reasoning and making important
judgements
Feature-based views:
Dual-code
Representation?
Landau, 1984: Will the real
grandmother please stand up?
- Adults use core definition to
pick out the grandmother
- Children use the prototypical
definition to identify the
grandmother
Feature-based views:
Dual-code
Representation?
Keil & Batterman (1984)
A smelly, mean old man with a
gun in his pocket who came to
your house and takes your TV
because your parents didn’t
want it anymore and told him
he could have it.
A very friendly, cheerful woman
who gave you a hug, but then
disconnected your toilet bowl
and took it away without
permission and no intention to
Problems with
Probabilistic Views
Feature similarity does not
explain categorization.
• What counts as a feature?
(Without
constraints, any two things can
be
arbitrarily similar or
dissimilar!)
• What features are “essential”,
“weighted”? (The importance or
Problems with
Probabilistic Views
Feature similarity does not
explain categorization
(continued)
• Goal-derived categories are
not based on similarity to a
prototype/exemplar
• Thus, need to specify the
principles that determine WHAT
are to count as relevant
features!
Problems with
Probabilistic Views
2. The sum of the features is
not the whole concept.
- Concepts are not simply a sum of
independent features…
- …rather, the structure of the
concept is held together by the
features PLUS the (functional,
causal) relations binding the
features together
Features of a car
Has wheels
Has an engine
Has four doors
Runs on fuel
Has a roof
“Today I saw a car with orange
wheels.”
“Today I saw a car with square
wheels.”
“Today I saw a “Today I saw a
car with orange car with square
wheels.” wheels”
The owner is a It cannot
hippie move
The rest of It was not
the car is intended to
brightly function
painted normally
Car functions It may be a
normally sculpture
intended to
elicit
View 2: Knowledge or
Theory-based views
Our knowledge of concepts and
categories consist of the
features or attributes of the
instances of the
concept/members of the
category…
PLUS
…the underlying explanatory
principles that determine the
relevant features, their
View 2: Knowledge or
Theory-based views
1. Explanations or theories
provide the “cognitive glue”
that binds features together
in some meaningful
relationship.
Context
Goals
Higher-order knowledge
memories
Cell Lion
phone Tiger
Keys Cat
Wallet
Penguin
Laptop
Parrot
Chicken
View 2: Knowledge or
Theory-based views
2. Our knowledge/theories
dictate which features are
central and explains why
instances of a category tend
to go together.
straight bananas vs straight
boomerangs
soft knives vs soft diamonds
polka-dot fire hydrants vs
polka-dot yield sign
orange wheels vs square wheels
Psychological Essentialism (or the
Essentialist Heuristic): People
act as if things have “essences”
that make them the thing they
View 2: Knowledge or
Theory-based views
3. Our knowledge/theories
provide the link between
features of a concept and the
“core” or “essence” of the
concept
- human perceptual system is
biased towards discerning the
features of objects that are
likely to lead to the core of
the objects
