Cognitive Psychology - Knowledge Representation

Front 1

Tulving (1972, 1983) identified two seperate and distinct long term memory systems, these are...

Back 1

1. Episodic
2. Semantic

Front 2

The memory system that holds information about events in which you somehow participated is called

Back 2

Episodic Memory

Front 3

The memory system that holds information about general knowledge is called

Back 3

Semantic Memory

Front 4

Knowledge representation refers to how knowledge is _______ and the _______ that allow us to access and use this knowledge.

Back 4

represented; processes

Front 5

What are the two models of cognitive architecture?

Back 5

Computational and Connectionist

Front 6

Cognitive architecture refers to

Back 6

the organisation of cognitive structures and the processes that work on them.

Front 7

The computational model or ________ ________ approach of cognitive architecture are those in which _____ _________ processes operate over complex _______ ________ ________.

Back 7

information processing;
rule governed;
symbolic mental representations.

Front 8

The connectionist model or _____ ______ approach of cognitive architecture are those in which _________ is captured in patterns of ______ across a complex network of interconnected, ______ _____.

Back 8

neural processing;
knowledge;
activation;
simple units.

Front 9

Computational models of cognitive architecture are characterised by the following 3 things:

Back 9

1. symbolic representation - abstract representations
2. local representation - reside in specific memory
3. serial processing - sequential

Front 10

There are many types of knowledge that we retain, in general a distinction is made between two basic knowledge types these are:

Back 10

1. Declarative Knowledge
2. Procedural Knowledge

Front 11

Sternberg suggests that researchers differentiate between ________ based models and ____________ based models.

Back 11

perception; meaning

Front 12

Declarative Knowledge is knowledge about _____ such as facts, information and ideas and can be easily made explicit, ______ and described in _______, pictures or _______.

Back 12

"what";
recalled
words;
symbols.

Front 13

Procedural knowledge represents our knowledge of _____ we do things, and includes our abilities to perform both _____ and ______ tasks.

Back 13

"how";
motor;
cognitive.

Front 14

We studied 3 symbolic representation of declarative knowledge, these are

Back 14

1. Concepts;
2. Propositions; and
3. Schemas.

Front 15

Concepts are...

Back 15

fundamental units of thought

Front 16

Concepts help _____ the multitude of objects, events and _______ in our physical and ______ world, categorising them by similarity of their _____.

Back 16

organise;
relations;
mental;
referent.

Front 17

Concepts allow us _____ _ ______ by not having to represent each instance of the object eg a ball

Back 17

economy of representation

Front 18

Concepts embody knowledge about objects that is not _______ _______, eg snake knowledge

Back 18

perceptually apparent

Front 19

What are the two models of how conceptual knowledge is represented that we study?

Back 19

Semantic networks & feature comparison models

Front 20

A netowrk consists of ______ and _____.

Back 20

nodes and links

Front 21

Semantic networks describe the organisation of concepts in terms of a

Back 21

network of associated concepts.

Front 22

A nodes corresponds to a _____ or _____. The links designate the _______ between two _____.

Back 22

concept;
idea;
relationship;
nodes.

Front 23

Information is retrieved from the semantic network via the process of

Back 23

Spreading activation

Front 24

The semantic network approach has been successful in accounting for a number of findings such as the

Back 24

typicality effect

Front 25

The phenomenon in which experimental subjects are faster to respond to typical instances (robin is a bird) of a concepts that they are to atypical instances (penguin) is called the

Back 25

typicality effect

Front 26

Spreading activation refers to the

Back 26

excitation of one node in a semantic network by the excitation of another node to which it is connected.

Front 27

Declarative memory is _______ and consciously ________ whereas procedural memory is _______ represented and ___ _______ accessible.

Back 27

explicitly;
accessible;
implicitly;
not consciously.

Front 28

Anderson ACT theory the interaction between...

Back 28

working, declarative and procedural memory.

Front 29

Working memory is that part of ______ memory that is highly ______ at any one time.

Back 29

Declarative;
Activated;

Front 30

Proposition representation theories assert that knowledge is representations comprise ______, which are the _____ ___ of knowledge about which one can make true or false judgements.

Back 30

propositions;
smallest units.

Front 31

Anderson model of cognitive architecture, the adaptive control of thought or _____ assumes the basic level of representation for events are ________.

Back 31

ACT;
Propostions.

Front 32

Computational, rule-based systems use the logic of _______ calculus to compute meaning from a _______ structure, and make ________.

Back 32

predicate;
propositional;
inferences.

Front 33

Propositional networks extract and represent meaning, while letting go of the

Back 33

peripheral structure

Front 34

In propositional representations a given fact is expressed in terms of a ________ which specifies the relationship between the associated concepts.

Back 34

predicate

Front 35

The representation of more complex information is thought to be achieved through the organisation of ideas according to a general knowledge structure called a

Back 35

schema

Front 36

A schema can be viewed as a structure with several slots in which to

Back 36

organise relevant information

Front 37

Special kinds of schemas called ______ are thought to represent knowledge about routine everyday situations, by organising everyday information into ordered ________ of ______.

Back 37

scripts;
Sequences of events.

Front 38

Schemas and scripts are important in providing organisational structure for complex knowledge but the recall of information based on general knowledge structures can lead to

Back 38

memory distortions and incorrect inferences.

Front 39

Bower (1979) showed that people remember ______ and may construct ____ ______ based on schemas and scripts. This has implications for the accuracy of _____ ______ ______.

Back 39

selectively;
False memories;
eye witness accuracy.

Front 40

In contrast to computational/rule-based models, which were based on the information processing view of the way computers process information, the connectionist approach takes its inspirations from the way the

Back 40

brain processes information.

Front 41

Connectionist models represent knowledge as a

Back 41

pattern of activation across the network.

Front 42

In connectionist models cognitive processing occurs in ______ not ______.

Back 42

parellel;
serial.

Front 43

The 4 basic principles/properties in the connectionist model are:

Back 43

1. Neural network structure
2. Distributed representations
3. Spreading activation
4. Parallel Processing

Front 44

Connectionist networks can have single or _____ layers. A single layer has an ______ unit connected to the _____ unit. In complex networks there can be _______units which act as an intermediary between input and output units.

Back 44

multiple;
input:
output;
hidden.

Front 45

In connectionist models distributed processing refers to the sum of all ______ units distributed across the ______ ______ of the network, where units represent ______-______ only.

Back 45

active;
output layer;
micro-features.

Front 46

In spreading activation the connection strength is determined by the

Back 46

statistical association between two units and the activation rule.

Front 47

Connection units can be ______ or ______ and vary between ___ and ___.

Back 47

excitatory;
inhibitory;
-1;
+ 1.

Front 48

A connection weight of 0 means

Back 48

no activation between the connected units.

Front 49

Statistical association happens when the output unit becomes

Back 49

strongly associated with the input unit over time

Front 50

The connectionist approach to cognitive architecture was inspired by the brain metaphor and neural processing but this approach was

Back 50

not intended to be biologically faithful to the way the brain actually processes information.

Front 51

In computational models, cognition involves the manipulation of complex symbolic representations Such as declarative eg. _____, _____ & ____, and procedural eg ____ & _____.

Back 51

concepts, propositions & schemas;
conditions & actions.

Front 52

McClellend described connectionism as "When an event is experienced, it creates a

Back 52

pattern of activity over a set of processing units that is then considered a representation of the event"

Front 53

The Transfer function is output =

Back 53

sum of input times weight.

Front 54

The mental process in the the computational model are _______ in nature and are limited by capacity constraints eg _____ ______

Back 54

algorithmic;
serial processing.

Front 55

Knowledge representation in computational models features:
1. complex ________ representations
2. _______ representations
3. Constructed form combinations of _________ structures
4. are stored at _____ _______ in memory

Back 55

1. mental
2. symbolic
3. propositional
4.specific loacations

Front 56

Whch approach to cognitive architecture is regarded as better at accounting for knowledge represention.

Back 56

Experts do not agree whether thecomputational or connectionist models are better.

Front 57

4 of the main issues used to debate the relative merits of connectionist versus computational models are:

Back 57

1. Learning
2. Retrieval
3. Structure of mental representations
4. Constraint satisfaction

Front 58

In the computational model the goal of learning is to

Back 58

formulate explicit rules (proposition, productions) and requires innate knowledge.

Front 59

In the connectionist model the goal of learning is to find the right connection weights so that the correct pattern of activation is returned

Back 59

connection weights so that the correct pattern of activation is returned

Front 60

In Anderson's ACT model what are the three stages of procedural learning?

Back 60

1. Cognitive Stage - learning explicit rules for procedures
2. Associative Stage - Strengthening associations through practice
3. Autonomous Stage - the new rules become implicit and automatic

Front 61

The law of contiguity refers to the idea that

Back 61

two events that occur closely together in time or space become associated with one another.

Front 62

In connectionism the Hebb learning rule is when unit A and B are simultaneously activated this

Back 62

increases the strength of connection between them in proportion with their joint activation.

Front 63

In connectionism the Delta learning rule is superior to the Hebb rule as it accounts for ______.

Back 63

feedback

Front 64

In Delta learning in successive presentations the size of the error is _______, and the pattern of activation more closely resembles the ____pattern of activation.

Back 64

reduced;
target.

Front 65

Content addressability refers to the

Back 65

natural account of certain attributes in human memory.

Front 66

In a distributed system, target information can be retrieved by providing a _____ of the items _______. In contrast _________ representational system requires the knowledge of ______ __ _____ which requires a ______ search.

Back 66

subset;
features;
where to look;
serial.

Front 67

A criticsim of the connectionism approach is that ideas are simply adding together individual units and do not account for ______ which governs the way ______ are combined in certain rule determined ways. ie Jack chases butterflies

Back 67

syntax;
elements.

Front 68

Cognitive models need to allow for situations where making the correct response depends on

Back 68

several conditions being met at the same time or parallel processing.