Perception and Attention
Information Processing Model
 models human thought like its a computer
Attention
Resources
Sensory
Register
Response
Selection
Perception
Thought
Decision
Making
Working
Memory
Long Term
Memory
feedback
Response
Execution
Controller
Sensors
Feature
Detectors,
Pattern
Recognition
Output
Routines
AI System
RAM
Disk storage
feedback
Output,
Motor
actions
Perception
Processing
Response
Attention
Resources
Sensory
Register
Response
Selection
Perception
Thought
Decision
Making
Working
Memory
Long Term
Memory
feedback
Response
Execution
Perception
 compares incoming sensory data to stored
knowledge
 reduces from many pieces of data to
meaningful units
 Three aspects, feature analysis (bottom up
processing), unitization, and top down
processing
Feature Analysis
 Analysis of the raw features of an event

Colour, size, shape, loudness
A
A AA
Unitization
 Matching sets of features with long term
memory to create units

E.g. reading words in a familiar language vs an
unfamiliar language
 things are built up hierarchically
 feature-> letter -> word ->sentence
A AA
A Apple
unitisation: creating a unit out of features
Objects
 Can also be broken down into features, and
conversely unitized
Design Implications of Feature
Analysis
 Speed and accuracy greatest for most often
seen fonts (use a common font)
 For single words (labels) use all caps. STOP
 For sentences use upper and lower case,
NOT ALL CAPS ITS HARD TO READ
 Use print not script.
Design Implications of Feature
Analysis
 Minimize abbreviations and use complete
words when possible (min abb.)
 If you have to abb. trunc. don’t abvt.
 Leavespacesbetweenwords.
Top Down Processing
 Uses the context of the situation to resolve the
image
 Occurs simultaneously with bottom up processing
 resolves ambiguous situations in the absence of
clear physical features
 your expectation of what you’ll see affects what
you’ll see
Exercise: Feature Analysis and
Top Down Processing in
Conflict
GREEN
RED
YELLOW
BLACK
BLUE
PURPLE
BLACK
BLUE
RED
BLACK
GREEN
PURPLE
YELLOW
BLUE
RED
GREEN
BLACK
GREEN
PURPLE
YELLOW
GREEN
RED
BLUE
BLUE
BLACK
BLUE
RED
BLACK
GREEN
PURPLE
YELLOW
PURPLE
RED
YELLOW
BLACK
GREEN
YELLOW
GREEN
PURPLE
BLACK
XXX
X
XXXX
XXXXX
XXXXX
XXXXXX
XX
XXX
XXXX
XX
XXX
XXXXXX
X
XXXX
XXXXXX
XXXX
XX
XXX
XXXXX
X
333
1
4444
55555
55555
666666
22
333
4444
22
333
666666
1
4444
666666
4444
22
333
55555
1
555
6
2222
1111
44444
222222
33
666
1111
44
11
444444
2
3333
444
66666
666
555555
33333
5
Attention
 focuses resources
 like a spotlight
 filters out what you don’t need to know
 shifts:


multi-tasking when driving in a straightforward
situation (drive, radio, friend)
focus when in heavy traffic, or accident seen
Two Kinds of Attention
 Selective attention

Focusing on the environment with the goal of
extracting certain information
 Divided attention

Processing two sources of information at once
Divided Attention: Resource
Demands
 Difficult tasks reduce ability to divide
attention
 “Resource theory” (Kahneman, 1974)



Mental resources are shared by tasks
Mental resources are finite
Two tasks share resources
Divided Attention: Structural
Similarity
 People have multiple pools of resources




Auditory vs. visual senses
Spatial vs. verbal cognitive demands
Stages of processing (memory vs. responding)
Visual channel (focal vs. ambient)
 Tasks share resources within a pool
 “Multiple resource theory”
Divided Attention: Task
Confusion
 More similar tasks will cause confusion


Baseball scores and math
Auditory background and words
 How does this explanation fit with multiple
resource theory?
Divided Attention: Task
Management
 Users choose a primary task vs. secondary
task
 Strategies


Doing secondary tasks when primary task allows
Success relies on switching appropriately
Automatic and Controlled
Processing
 when tasks are new they take alot of attention (e.g.
learning to drive)
 Controlled processing: Effortful cognitive processes
that require attention to initiate and sustain them
 With practice become automatic
 Automatic: can be run without cognitive demand on
attentional resources
How do things become
automatic?
 High degree of consistency
 Extremely strong mappings between world
and required action
World
Action
STOP
Design Implications of
Attention
 use strong mappings to make things
automatic
 exploit visual and auditory information to
share attention on different resource pools
 make most critical things most salient
(visible) to capture attention - attention is
competitive!
 SPOT exercise