Memory

1
Memory
SENSORY STORE
WORKING MEMORY
LONG-TERM MEMORY
2
A little experiment in memory

• Courtesy of NASA Ames Cognition Laboratory
  (http//human-factors.arc.nasa.gov/cognition/tutor
  ials/ModelOf/memory5.html)
• Step 1 take out a blank sheet of paper and put
  List 1 on the top. Then put your pencil/pen
  down.
• Step 2 listen to the list of words carefully.
• Step 3 after the entire list is finished, you
  will be instructed to write down as many of the
  words as you can remember.
• Step 4 check your list against the one I show
  you and write the number correct at the top of
  the page.
• Repeat steps 1 4 with List 2 and List 3.

3
Results from an earlier experiment
http//human-factors.arc.nasa.gov/cognition/tutori
als/ModelOf/memory5.html
4
Impact of memory on system design ...

• Power
• Vast store of knowledge
• Limitations
• Forgetting
• Limited working memory
• Attention

5
Just the facts about memory ...

• Three subsystems of memory
• Short-term sensory store
• Working memory (short-term memory) WM/STM
• Long-term memory - LTM
• These subsystems differ in several ways
• Capacity
• Sensory store __________________________________
• WM is ______________________________
• (the "magic number" 7 plus or minus 2)
• LTM __________________________

6
Just the facts about memory (cont.)

• Differences in memory subsystems (cont.)
• Duration
• Sensory store ____________________________________
  _
• WM _____________________________________________
• LTM _____________________________
• Codes
• Sensory store ____________________
• WM ____________________________
• LTM ____________________________

7
How it works (or doesnt) ...

• Working Memory (WM)
• A model (from Baddeley)

8
WM How it works (or doesnt) ...

• Restrictions
• Capacity - 7 2 items of information.
• Time - 7 - 70 second half-life
• Some solutions ...
• Increase capacity by chunking
• Create meaningful sequence already present in LTM
• Experiments
• Subject could recall gt 20 binary digits by coding
  into octal (0101111? 57)
• Subject could recall gt 80 digits by coding into
  running times (353431653 ? 3 min, 53.4 sec mile
  3 hr, 16 min, 53 sec marathon)
• Chess masters recall board with great accuracy
  "chunk" into strategic patterns

9
WM How it works (or doesnt) ...

• Examples of everyday chunking
• Parsing - break up into chunks
• phone numbers, social security numbers
• Reading musical staffs ("Every Good Boy Does
  Fine")
• Medical school mnemonics
• Songs constraints of rhythm, rhyme
• "We Didn't Start the Fire"
• "Joseph and the Amazing Technicolor Dreamcoat"
• Preamble to the US Constitution
• Other approaches to handling WM limitations
• Minimize load
• Visual echoes
• Exploit different codes

10
How it works (or doesnt) ...

• Long-term memory (LTM)
• Types
• Semantic memory - general knowledge
• Event memory
• Episodic - an event in the past
• Prospective - remember to do something
• Basic mechanisms
• Storage - through active rehearsal, involvement,
  or link to an existing memory.
• Alternatively - everything gets in
• Retrieval - depends on
• item strength
• number and strength of associations to other items

11
LTM How it works (or doesnt) ...

• Organization of information in LTM
• Most-used information is semantic
• retrieval depends on semantic associations
• good design builds / uses appropriate semantic
  associations
• The network of semantic associations around
  specific topics are schemas
• Schemas involving sequences of activities are
  scripts
• Schemas concerning how equipment and systems work
  are mental models

12
LTM How it works (or doesnt) ...

• What it means for design
• Encourage regular use of info
• Standardize
• Design information to be remembered
• Provide memory aids

13
Memory versus knowledge in the world

• When do you not need to remember something?
• (Why do you not need to remember what a penny
  looks like?)
• When the knowledge is already "in the world"!
• (Because you only need to recognize a penny - and
  nothing else looks like it.)

14
Knowledge in the worldfrom Norman, D.A. The
Design of Everyday Things, (formerly "P.O.E.T.")
1988. New York Currency/ Doubleday.)

• Affordances
• Constraints
• Mappings
• Conceptual Models
• Visible Structure
• Reveals
• 1. affordances
• 2. constraints
• 3. mappings

15
Affordance

• "refers to perceived or actual properties of the
  thing, primarily those fundamental properties
  that determine just how the thing could possibly
  be used. (Norman, pg. 9)
• Affordances of objects e.g., chairs, tables,
  cups
• Affordances of materials e.g., glass, wood
• Affordances of controls How are things
  operated?

16
Examples ...
17
Constraints

• Those aspects of a device or material that limit
  its perceived possible uses.
• Physical size, shape, possibilities for
  movement, etc.
• Semantic meaning of the situation
• related to the notion of conceptual models
• Cultural defined by tradition, meaning within
  the culture (e.g., the color red, triangular
  shape)
• Logical placement of controls, direction of
  movement, etc.
• related to mappings

18
Examples ...

• Physical constraints
• Semantic constraints
• Cultural constraints
• Logical constraints

19
Conceptual Models

• Our understanding of the way things work, how
  things are put together, cause effect, etc.
• Depends on the visibility of the system
  structure, the timing of the feedback, and
  consistency of cause/effect relationships
• Builds a framework for storing knowledge about a
  system or device in the head.
• Used to develop explanations, recreate forgotten
  knowledge, and make predictions.

20
Mappings

• Making the connection between how things work and
  how we think they work.
• Some examples (stay tuned - more in the display
  design lesson!)
• Principle of Pictorial Realism Displayed
  quantities should correspond to the human's
  internal model of these quantities.
• Congruence The linear motion of a control and
  display should be along the same axis and the
  rotational motion of a control and display should
  be in the same direction.
• Principle of the Moving Part The direction of
  movement of an indicator on a display should be
  compatible with the direction of movement of an
  operator's internal representation of the
  variable whose change is indicated.
• Spatial compatibility The spatial arrangement of
  displays should be preserved in the controls.

21
Your turn

• Recall the question regarding Benjamin Franklin
  given to you as homework last week.
• List a few of the things youve thought of that
  Mr. Franklin would be able to figure out in
  your apartment/home.
• Describe how Mr. Franklin is able to figure these
  things out in terms of the affordances,
  constraints, mappings, and visible structure.
• Use the following table to help organize your
  answer.

22
(No Transcript)
23
ATTENTION!!!
ATTENTION RESOURCES
From page 147 of Wickens et al.
24
ATTENTION!!!

• A "flexible, sharable, processing resource of
  limited availability".
• Our ability to attend to several things at once
  (time-sharing) depends on
• Controlled vs automatic processing
• Skill
• Which resource(s) required
• Attention tasks can be divided into 4
  categories ...

25
1. Selective Attention

• "requires the monitoring of several channels
  (sources) of information to perform a single
  task.
• Example scanning cockpit instruments
• Limitations
• As the number of channels of information
  increases, performance declines (even when the
  overall signal rate is the same).
• Can select inappropriate aspect(s) of the
  environment to process.
• "Cognitive tunnel vision" in complex environments
  with many displays, especially under stress.
  (Example 1972 Eastern Airlines crash in the
  Everglades).
• Errors associated with Selective Attention are
  generally the result of an intentional, but
  unwise choice.

26
Selective Attention

• Design Guidelines
• Place frequently sampled displays together.
• Place sequentially sampled displays together.
• Use external aids/reminders to help people
  remember when the display was last sampled.

27
2. Focused Attention

• Requires attending to one source of information
  at the exclusion of all others
• Examples
• Trying to study while someone else is talking on
  the phone
• Trying to enter numerical data into Excel while
  others are discussing basketball scores and
  stats.
• Limitations
• Impossible to ignore a visual stimulus within 1
  degree of visual angle of the visual information
  you are interested in.
• Auditory stimuli sufficiently loud with respect
  to the signal you are interested in, and/or
  similar to it, can interfere with the signal.
• Errors associated with focused attention are
  generally unintentional, driven by the
  environment.

28
Focused Attention

• Design Guidelines Parallel vs serial processing
• Parallel processing is helpful when
• two tightly coupled tasks are performed
  simultaneously (e.g., control roll and pitch of
  aircraft)
• two or more information sources imply common
  action (redundancy gain)
• Parallel processing is harmful when
• similar aspects of different stimuli must be
  processed (resource competition)
• two or more stimuli imply different actions
• e.g., a batter distracted by a moth

29
3. Sustained Attention

• "the ability of observers to maintain attention
  and remain alert over prolonged periods of time."
• Example Security guard watching monitor for
  intruders.
• Limitations
• Vigilance decrement - a decline in the speed and
  accuracy of signal detection with time on the
  task (found more in the laboratory than in real
  world tasks).

30
Sustained Attention

• Design Guidelines
• Appropriate work-rest schedules and task
  variation.
• Increase the conspicuity of the signal.
• Reduce uncertainty as to when and where.
• Training.

31
4. Divided Attention

• "two or more separate tasks must be performed at
  the same time, and attention must be paid to
  both.
• Example Driving and talking to a passenger.
• Limitations
• Time-sharing ...

32
The Resource Metaphor of Attention

• Time-sharing (or doing two tasks simultaneously)
  is difficult because we have limited attention
  resources.
• The Performance-Resource Function (PRF)

33
The Performance Operating Characteristic (POC)
34
Limitations of the "Single-Resource" Theory of
Attention

• Difficulty insensitivity
• In some experiments it has been shown that making
  one time-shared task more difficult has no effect
  on the performance of the other.
• Perfect time-sharing
• Structural alteration effects
• In some experiments it has been shown that
  altering the structure (but NOT the difficulty)
  of one task affects performance on the other.
• Example Manual vs vocal responses to a tone
  discrimination task while tracking.

35
Multiple-Resource Theory

• Instead of one "pool" of resources, there are
  several different capacities of resources
• Codes spatial or verbal
• Modalities visual or auditory
• Stages of processing early (encoding/central
  processing) or late (responding)
• The more resources are shared, the more tasks
  will interfere.

36
Multiple-Resource Theory

• To the extent that tasks demand separate rather
  than common resources
• Time-sharing will be more efficient
• Difficulty insensitivity will be observed
• The POC will be more "boxy"

37
Limitation of Multiple Resource Theory

• The three proposed dimensions (stages, codes,
  modalities) do not account for all experimental
  findings. For example
• Tasks with different rhythmic requirements are
  hard to time-share.
• Control dynamics affect the efficiency of
  time-sharing a manual tracking task with another
  task.

38
Implications Design Recommendations

• Since spatial and verbal codes draw upon separate
  resources, time-sharing manual and verbal
  responses is highly efficient (assuming that the
  manual response is spatial in nature and that the
  vocal response is verbal). Example
• pilots fly the airplane (spatial, manual task)
  and simultaneously talk to air traffic control
  (verbal, vocal task).
• This example also demonstrates different
  modalities (visual and auditory) which also draw
  from separate resources
• therefore
• Design systems to support a mix of manual and
  vocal responses for time-shared tasks.

39
Multiple Resource Theory

• The effect of training
• Training can make tasks data limited rather than
  resource limited
• Data limited tasks can coexist more easily than
  resource-limited
• Reasoning behind part-task training paradigms
• People can also be trained to timeshare tasks
  more efficiently
• Rapid switching between tasks
• True multi-tasking