HCI 3

1
HCI 3 4

• Cognitive Models
• and
• Perception and Representation

2
Don Norman Interview

• If accidents are caused by humans, maybe its not
  the humans, maybe things arent designed
  appropriately for the humans.
• You dont blame an electronic circuit for burning
  itself up (you blame the designer?)
• Either
• design the system to avoid the kinds of errors
  humans make
• make the system tolerant of those errors.
• Coffee cup over flaps putting things in the
  world rather than in your head

3
Models play two roles

• Generative
• Contribute during the design process
• Evaluative
• Determine, after the fact, whether the design
  meets the intended criteria

4
Models of human behaviorA short history

• Human information processor
• Human information processor extended
• Multi-store model of human memory
• Model human processor
• Computational vs Connectionist approach
• Human actors
• Distributed cognition

5
Human information processor

• Assumes
• Information is unidirectional
• Sequential
• Each step takes time depending on complexity
• Steps
• Input (stimulus) is encoded
• Encoded stimulus compared with stored stimuli
• One response is selected from the possibilities
• That response is executed

6
Human information processor extended

• Adds to the original model
• Attention
• Memory
• Attention and Memory are a limited (and therefore
  limiting) resource
• Therefore attention and memory mediate (and
  therefore affect) the encoding, comparing,
  selection and execution

7
Multi-store model of human memory

• The extended model includes memory in the picture
  of human performance this breaks that memory
  into smaller, distinguishable pieces for closer
  scrutiny.
• Sensory store (tied to the senses themselves, NOT
  in the brain)
• Short term store (severely limited in size and
  duration, in the brain)
• Long term memory (in the brain)

8
Stu Cards Model Human Processor

• Modeled as three systems, each with memory and
  processing
• Perceptual system
• Motor system
• Cognitive system
• Enabled predictions(?) of expected performance
• Lead to GOMS predictive analysis (keystrokes)

9
Stu Cards Model Human ProcessorThe Problems

• It models performance as a series of processing
  steps (is that appropriate?)
• Too focused on one person, one task
• An overly simplistic view of human behavior
  (ignores environment other people)

10
Beyond the Model Human ProcessorFocus on

• How knowledge is represented
• How mental models are used in HCI
• How users learn and become experienced on systems
• How interface metaphors help to match users
  expectations (and how they dont!)
• How a persons mentally-held conceptual model
  affects behavior

11
Computational vs Connectionist approach

• Computational approach uses the computer as a
  metaphor for how the brain works
• Connectionist approach rejects the computer
  metaphor of the brain in favor of the neural
  network approach the brain metaphor cognitive
  processes are characterized as activation of
  nodes and connections between them

12
Human actors

• A change from human factors to human actors
• A change in focus
• from passive, depersonalized
• to active, controlling
• Person as autonomous agent (what an idea!) able
  to coordinate and regulate behavior
• Not a passive element in a human machine system

13
Distributed cognitionEd HutchinsCognition in
the Wild

• To explain human behavior you have to look beyond
  the individual human and the individual task
• A collection of actors, technology, setting and
  the interrelations to one another The Functional
  System
• Human behavior is situated in an environment

14
Distributed cognition

• Used to analyze coordination components in the
  functional system
• Looks at information and how it propagates
  through the system
• Looks at how it transforms between the different
  representational states found in the functional
  system

15
Distributed Cognition

• Situation awareness (Norman) is the silent and
  inter-subjective communication that is shared
  among a group
• Not explicit, rather through glancing and
  inadvertent overhearing
• Sounds like periphery?

16
Distributed Cognition

• Two main concerns
• Map out how the different representational states
  are coordinated across
• Time
• Location
• Objects
• Analyze and explain breakdowns
• Can it be used to avoid the design of systems
  that break down?

17
Distributed CognitionFinally

• Provide a method of analysis and a theoretical
  framework that enables researchers to
  conceptualize cognition in the real world
• in a way that can inform system design.

18
Perception and Representation

• Competing theories of perception
• Constructivist
• View is constructed from stuff in the environment
  and from previously stored knowledge
• Ecological
• Perception is a process of picking up
  information from the environment, no construction
  or elaboration needed.
• The notion of affordances plays into this.

19
Constructivist approach

• Perception involves the intervention of
  representations and memories
• Not a replica or copy rather a model that is
  constructed by the visual system through
  transforming, enhancing, distorting and
  discarding information.

20
Constructivist approach (cont)

• Models provide us with a more constant view of
  the world they reduce the amount of stuff we
  have to perceive and carry with us
• Buildings are seen as stable even though they
  float around on the retinas surface
• People are seen as being a consistent size even
  though the image on the retina is smaller when
  seen at a distance

21
Context and Gestalt

• THE CAT example the context of the ambiguous
  words used in conjunction with our prior
  knowledge allow us to disambiguate the phrase.
• Object/foreground/background image decomposition
  (partitioning) is an aspect of the constructivist
  process

22
Gestalt laws of organization

• Proximity - objects form groups.
• Similarity - same shape/color? Same class.
• Closure - completion of missing parts
• Continuity - organization into lines
• Symmetry - bounding by symmetrical borders
  segments the contents into a figure

23
Ecological approach

• Perception is direct and does not need to be
  constructed
• Users intentionally engage in activities that
  cause the necessary information to become
  apparent we explore objects in the environment
• Tied to the idea of the affordances of systems,
  objects and events.
• If affordances are obvious we know how to interact

24
Affordances A quandry...

• Understanding affordances can be beneficial when
  designing direct manipulation interfaces
• But The idea of affordances derive from real 3D
  objects in the world, not necessarily 2D
  representations of 3D objects
• How direct is direct manipulation when applied to
  2D interfaces?

25
Graphical representation

• Wireframe - computationally less expensive, small
  in size, good for showing interior workings,
  lower fidelity
• Solid-object - computationally more expensive,
  larger in size, can disambiguate internal and
  external surfaces, higher fidelity.

26
Monocular depth cues

• Size
• Interposition (occlusion)
• Contrast, clarity, brightness, atmosphere
• Shadow
• Texture
• Motion parallax (moving images)

27
Graphical codingLeast direct to most direct
mapping

• Abstract code (filenames)
• Abstract shape to represent different objects
• Color to represent options
• Different sized representation indicates
  different sized files
• Bar charts to show numerical data trends
• Different line widths represent different pen
  size
• Picture of waste bin to represent delete
• Paper file folder to represent a file

28
Icons

• Believed to reduce complexity
• What happens if there are many options? It is
  difficult to design icons that are small but
  distinct.

29
The End!
30
(No Transcript)