3.4 Foundations of Cognitive Brain Research
All documents for this subject (12)
Seller
Follow
evelienrodenburg
Reviews received
Content preview
1
Problem 2 - Mental representation
Learning goals:
- What are mental representations and how are they formed?
Markman & Dietrich (2000) - Extending the classical view of representation
Introduction:
Since the late 1950s, models of cognition have been dominated by representational approaches.
- These models posit (veronderstellen) some kind of internal mechanism for storing and
manipulating data as well as processes that act on representations to carry out intelligent
behaviors.
The early predictions that we would soon (now) have autonomous robots and intelligent computers
have not yet come to pass.
- Researchers from a variety of perspectives have suggested that the standard
representational assumptions made by cognitive models are to blame for this lack of
progress.
This article:
- Sketches the classical view of representation that is widely employed in cognitive models
- Four recent approaches to cognitive modeling are examined
- perceptual symbol systems
- situated action
- embodied cognition
- dynamical systems
- The authors suggest that each of the four alternative approaches has something important to
offer, but can’t replace the classical view.
- Discussion of ways to reconcile the classical view with these alternatives.
The classical view of representation:
All approaches to representation share five key assumptions:
1. Representations are mediating states of intelligent systems that carry information
a. There are mediating states that are internal to the cognitive systems.
b. In order for something to qualify as a mediating state, it must meet four conditions:
i. There must be some representing world.
1. The representing world consists of the elements that serve as the
representations.
a. Features that make up an object.
ii. There must be some represented world.
1. The represented world is the information that is being represented.
a. The object itself
iii. There is a set of representing relationships that determine how elements in
the representing world come to stand for elements in the represented world.
iv. There are processes that use the information in the representing world.
v. Example: Tversky’s contrast model of similarity assumed that objects (the
represented world) are represented by sets of features (the representing
world). Each feature is a symbol that stands for a particular property of the
object, such as the color blue (the representing relationships). Pairs of sets
representing two objects are compared by finding the intersection of the sets
(a process that acts on the representations). The features in the intersection
are the commonalities of the pair, and the features that are not in the
intersection are the differences of the pair.
, 2
2. Cognitive systems require some enduring representations
a. Some representations are enduring states of the system. Agents must use their
experience as a guide. Thus: they have internal states that endure longer than the
states in the represented world that gave rise to them,
b. Example: an object representation can contain a particular feature (e.g. Blue)
regardless of whether that property is currently accessible in the environment.
3. Cognitive systems have some symbols in them
a. Some representations are symbols.
b. Symbols have to central qualities:
i. Their relationship to the represented world is arbitrary (willekeurig)
ii. They are discrete packets of information
c. Symbols are necessary for referring to specific values or properties in the represented
world, and mirror the observation that languages consist of words that permit people
to fix common reference.
d. In the contrast model (Tversky) features are symbols in the representing world.
4. Some representations are tied to particular perceptual systems but others are amodal
a. Representational elements exist at a variety of levels of abstraction. Some
representations correspond directly to aspects of perceptual experience. Other
representations are more interpreted and refer to abstracts concepts (e.g. truth or
justice), which are quite removed from perceptual experience.
b. In the contrast model (Tversky) there is no necessary connection between the
features that describe an object and the perceptual information.
5. Many cognitive functions can be modeled without regard to the particular sensor and effector
systems of the cognitive agent
a. Cognitive agent = organisms & intelligent machines
b. Some cognitive models need not be concerned with perceptual and motor
representations. Some representations in the cognitive system are sheltered from the
body of the cognitive agent.
c. The contrast model (Tversky) makes no assumptions about the nature of the
perceptual or motor systems of cognitive agents.
Alternatives to the classical view
The four alternative approaches to representation have all taken umbrage (=aanstoot) with at least
one of the core assumptions of the classical view.
Perceptual symbol system
Cognitive processing is flexible → People are able to recognize when a new situation is like one they
have experienced before, but they are also good at handling
deviations (=afwijkingen) from normal situations.
The classical approach to representation assumes that flexibility requires abstraction
→ By abstracting away (=onttrekken) from the perceptual details of specific situations, the
commonalities (=gemeenschappelijkheden) across situations can be preserved. The classical
approach typically assumes that there are abstract amodal representations that play an important
role in cognitive processes.
Amodal representations are not as flexible as they were initially assumed to be.
- Schank et al.: Suggested ways to represent abstract scripts and schemas to enable an agent
to comprehend new events. These systems had difficulty dealing with the potential variations
of simple events, and therefore can’t explain the human flexibility in situations.
, 3
→ Later work had to posit (=veronderstellen) both abstract and specific representations in
order to account for human-like flexibility in dealing with the variations on events.
Symbolic models have also had difficulty accounting for differences in the way a property manifest
itself across items. The same spatial preposition can describe many subtly different situations.
- The English preposition ‘in’ normally means that one object is contained inside another, but
an apple can be ‘in’ a bowl, even when it is stacked on other apples such that it rises above
the top of the bowl → it is difficult to account for this ambiguity using traditional symbolic
models.
Current research suggests that flexibility in cognitive processing arises from the storage and use of
specific episodes in memory and their perceptual content.
- Barsalou’s perceptual symbol system approach: the perceptual system is used to simulate
objects and events. The connection between perception and language in this case is
accomplished using principles derived from cognitive grammar.
- Cognitive grammar: attempts to account for
grammatical phenomena using
representations and processes that are
continuous with those used by other
cognitive processes. Grammar facilitates
the construction of representations using
both perceptual and attentional processes.
- For example: the representation of
the prepositions ‘above’ and ‘below’
involves setting up locations in a
semantic space, and then focusing
attention on one of the objects.
Evidence suggests that cognitive science should eschew (=mijden) amodal representations.
- Schyns et al.: most amodal theories assume a fixed feature set. Categorization models then
use these features to predict the category to which an instance belongs. Models can calculate
the similarity to a prototype, or to various known exemplars, ore they can form rules to
describe the categories. People often learn new features, even new perceptual features
→ to understand categorization, we need to further understand processes of perceptual
feature creation.
Schyns & Rodet: taught people perceptual
categories each consisting of unfamiliar shapes.
What people considered to be the basic
perceptual components of the categories
depended on the order in which they were
exposed to the categories. For example: they
might see some items that contained the feature
XY as one of its components. If this was the first
category they learned, they treated XY as one
unit. But, if they first saw some categories with
component X, when they later saw XY, they
divided this shape up into X and Y → the set of perceptual features used to construct the categories is
learned.
The benefits of buying summaries with Stuvia:
Guaranteed quality through customer reviews
Stuvia customers have reviewed more than 700,000 summaries. This how you know that you are buying the best documents.
Quick and easy check-out
You can quickly pay through credit card or Stuvia-credit for the summaries. There is no membership needed.
Focus on what matters
Your fellow students write the study notes themselves, which is why the documents are always reliable and up-to-date. This ensures you quickly get to the core!
Frequently asked questions
What do I get when I buy this document?
You get a PDF, available immediately after your purchase. The purchased document is accessible anytime, anywhere and indefinitely through your profile.
Satisfaction guarantee: how does it work?
Our satisfaction guarantee ensures that you always find a study document that suits you well. You fill out a form, and our customer service team takes care of the rest.
Who am I buying these notes from?
Stuvia is a marketplace, so you are not buying this document from us, but from seller evelienrodenburg. Stuvia facilitates payment to the seller.
Will I be stuck with a subscription?
No, you only buy these notes for $3.75. You're not tied to anything after your purchase.