The Structure of Knowledge

1
The Structure of Knowledge
2
Characteristics of KnowledgeThe storage,
integration, and organization of information in
memory

• Long term memory includes not only specific past
  events but also general knowledge about the world
  (e.g. semantic memory).
• General knowledge is abstract in that it is not
  constrained by a particular context (e.g.
  knowledge of driving rules).
• Knowledge is not bound by a particular time and
  place of occurrence although it may include that
  (again knowledge of driving rules).
• What we can deduce from the example used above is
  that knowledge of even a single concept consists
  of a wide range of information! It is different
  from information derived from the senses!

3
An Example of the extent of General (semantic)
Knowledge

• Knowledge can be represented (viewed) in several
  ways, through lexical relationships,
  propositional relationships, through visual
  images, or as neurological components

4
Expert Performance

• One way to study general knowledge is to study
  the performance of people who are experts in a
  subject.
• By experts (or high knowledge individuals), we
  refer to highly developed or exceptional skills
  abilities, not general intelligence.
• Studies of expert knowledge grew out of the field
  of artificial intelligence, the attempt to design
  computer systems (sometimes called expert
  systems) that could mimic what a human expert
  knows.
• The best know of these types of systems is Big
  Blue, IBM corporations chess playing computer.
• Cognitive psychologists have now studied a wide
  variety of experts, including those who play
  chess, Go, or baseball, and those who are expert
  in complex fields like radiology, painting, or
  architecture.

5
Expert Performance

• From these studies, several factors concerning
  experts have emerged (Glazer Chi, 1988)
• -Experts excel mainly in their own domains (e.g.
  experts in mental calculations are not likely to
  excel to medical diagnosis)
• -Experts perceive large meaningful patterns in
  their own domain (e.g. chess masters or x-ray
  diagnosticians are able to see more meaningful
  patterns within their own specialties than
  non-specialists) (
• -Experts are fast (e.g. expert programmers and
  typists are able to work much faster within their
  own specialty than non-specialists.
• -Experts seems to utilize STM and LTM more
  effectively. It seems that they have superior
  memories, but it may just be that they use it
  more effectively.
• -Experts see and represent a problem in their
  domain at a deeper level than novices. When
  experts are asked to sort and analyze problems,
  they tend to deal with deep issues rather than
  superficial ones (e.g. physics professors vs.
  novices Chi, Feltovich, Glazer, 1982)
• -Experts spend a great deal of time analyzing a
  problem qualitatively, They tend to look at a
  problem from several different angles before
  plunging into a solution.
• -Experts have self monitoring skills. They seem
  to be aware of their errors and are able to make
  in course corrections

6
Studies of Expert Performance
7
Expert Performance of Artists Eye Tracking

• Eye tracking studies have been performed on
  expert artists as they draw a picture based on a
  model, this information is compared to novice
  artists who attempt the same task (Miall
  Tchalenko, 2001).
• This study used an eye camera, a visual scene
  camera, a motion detector which marked hand
  movements.
• This allowed the researchers to examine the eye
  fixations, eye movements, and hand movements, and
  the subjects.
• They discovered that the expert artist had eye
  fixations nearly twice the time of the novice, he
  used special fixations different from his
  ordinary looking pattern, and he would build
  details point by point rather than holistically.

8
Expert Performance fMRI (Solso, 2001)

• Neuroscience evidence also suggests that experts
  spend less effort processing information than do
  novices.
• These fMRI scans were taken from an expert and
  several novice artists as they drew a portrait.
  Both types of individuals show activation in the
  right parietal lobe (an area know to be involved
  during facial recognition), however the expert
  shows far less activation than does the novice.
• This suggests that experts are able to do more
  than a novice with less neural activity,
  indicating more efficient processing of
  information.

9
A final note on experts Memory

• Either experts have an entirely different memory
  system than most of us or they are using
  knowledge stored in LTM to expand their working
  memory capacity.
• Chase Ericsson (1981) have proposed three
  principals that propose ways in which experts
  exploit their LTM to perform unusual tasks
• The mnemonic encoding principal-Asserts that
  experts encode information in terms of a large
  existing knowledge base (e.g. using existing
  knowledge to chunk new information).
• Retrieval Structure Principal-Experts use their
  knowledge of a subject to develop abstract,
  highly specialized mechanisms for systematically
  encoding and retrieving meaningful patterns from
  LTM This allows them to anticipate informational
  needs for a familiar task, and to store info in a
  format that will facilitate its retrieval.
• The speed up principal-Practice increases the
  speed with which experts recognize and encode
  patterns. This means they can also retrieve it
  faster than a novice.

10
Theories of Knowledge Structure Theories of
Knowledge Structures propose general types of
structures that represent information.

• Clustering Model (Bower et al, 1970).
• Associative Network Models (Collins Loftus
  Anderson, 1983)

11
Clustering Models A Conceptual Hierarchy

• Clustering Models propose that concepts tend to
  be organized in clusters of groups that are
  related within a hierarchical structure.
• Evidence for this model came from free recall
  tests which indicated that categorically similar
  words tended to be recalled together

12
Associative Networks Models of Knowledge

• Associative network models propose that
  everything regarding a concept is related
  together (Collins Quillian, 1969).
• A major concept is represented by each node
  (group) of information (e.g. bird), nodes are
  tied to together (Anderson et al, 1983).
• Activation of one concept can spread to other
  related concepts (e.g. a bird has feathers
  feet) (Collins Loftus, 1975).
• Concepts can overlap with some unrelated concepts
  (e.g. a pig also has feet) causing minor levels
  of activation for that concept.

13
Cognitive Evidence for associative Networks

• Reaction times taken to questions about knowledge
  (Collins Quillian, 1969).

14
Neurological Evidence for associative Networks
Knowledge networks are stored by modality

• Neuroimaging studies have shown that physical,
  phonological, and semantic codes of words
  activate very different areas of the brain.
• Encoding just the physical features (e.g. the
  color of an object), leads to less activation
  than encoding the phonological features (e.g.
  sounds), which leads to less activation than
  semantic elaboration (e.g. meaning).
• Repetition of the same item leads to increases in
  brain activity as well (e.g. increasing the
  semantic elaboration)

15
More Neurological Evidence for hierarchical
associative Networks

• Brain damage can selectively effect the knowledge
  systems
• People with visual agnosia (the inability to
  recognize objects) due to occipital-temporal
  damage, can describe the characteristics of an
  object but they cannot identify the object
  itself.
• In some cases they can recognize certain
  categories of objects (like non-living objects)
  while recognizing other categories (Living
  objects)
• This again implies that different types of
  knowledge are organized in hierarchical
  conceptual form based on there different
  characteristics.

16
Comprehension and Knowledge The Central Theme

• One interesting note about the relationship
  between comprehension and knowledge is that
  lengthy material is not recalled verbatim.
• Comprehension is the process of extracting the
  general meaning of a communication and discarding
  the details.
• However, comprehension is also a process of
  adding information to the communication not
  included in the original event.
• We tend to summarize information, capture the
  essential meaning of the information of the
  information or the gist.
• In other words we capture the central theme and
  attempt to organize information around that
  theme.
• The theme is the general topic of the material,
  and a theme will guide both comprehension and
  memory for material.
• However, the theme will often lead you to infer
  information, not present in the actually
  presented information
• The process of inferring information based on
  the main theme is important, in fact it is
  essential to connect or organize ideas expressed
  in a passage.

17
Frederick Bartletts Experiment (1932)

• The war of the Ghosts was designed to
  demonstrate the reconstructive nature of memory

18
Bartletts Experiment

• Individual sentences are not always remembered
  verbatim, but the ideas expressed are integrated
  into a representation, e.g. the general idea
• This can lead to alterations in memory that
  increase over time and with repeated attempts at
  recollection
• Does this have implications for serious issues
  like eyewitness memory for crimes, or the memory
  for traumatic events like abuse that have
  occurred in someone's past???