Page 1
Baobab Health, March 2014Harry Hochheiser, [email protected]
Introduction to Human-Computer Interaction – Cognitive Issues
Attribution-ShareAlike CC BY-SA
Harry Hochheiser University of Pittsburgh Department of Biomedical Informatics [email protected] !+1 410 648 9300
Page 2
Baobab Health, March 2014Harry Hochheiser, [email protected]
What is usability?
" What does it mean to be usable? " Which factors influence usability? " How can we make an interface usable? " Is interface design an art or a science?
!" Goal – develop some intuitions
Page 3
Baobab Health, March 2014Harry Hochheiser, [email protected]
What does it mean to be “usable”?
" Find something usable " Software, hardware, etc.
" Why is it usable? What do you mean when you say that? " Something unusable?
" What doesn't work? What's hard?
Page 4
Baobab Health, March 2014Harry Hochheiser, [email protected]
Usability Successes
Page 5
Baobab Health, March 2014Harry Hochheiser, [email protected]
Usability Failures
Page 6
Baobab Health, March 2014Harry Hochheiser, [email protected]
A Usability FailureThe light switch in my office…
Page 7
Baobab Health, March 2014Harry Hochheiser, [email protected]
A Usability FailureControl to turn off light sensor
Page 8
Baobab Health, March 2014Harry Hochheiser, [email protected]
A Usability Failure
Controls??
How might we improve upon this?
Page 9
Baobab Health, March 2014Harry Hochheiser, [email protected]
Dimensions of Usability
"Efficiency "Learnability "Memorability "Error Handling/Prevention "Satisfaction
Page 10
Baobab Health, March 2014Harry Hochheiser, [email protected]
Efficiency
" Can the system be used to complete the specified task? " How much time does it take to complete a task? " How many operations? " How much movement?
Page 11
Baobab Health, March 2014Harry Hochheiser, [email protected]
Learnability
" How quickly can a novice learn to use tool? " What help/assistance is given? " Consistency with similar system and/or convention?
Page 12
Baobab Health, March 2014Harry Hochheiser, [email protected]
Memorability
" Retention of proficiency over time? " Reminders and cues " Recognition vs. Recall " Minimizing cognitive load
Page 13
Baobab Health, March 2014Harry Hochheiser, [email protected]
Error Prevention/Handling
"What is the error rate? "Slips vs. mistakes "Preventing errors "Responding to errors
Page 14
Baobab Health, March 2014Harry Hochheiser, [email protected]
Satisfaction
" Do users feel that the tool was usable? " Do they want to use it more? " Is it aesthetically appealing?
" Don Norman: “Attractive things work better”
Page 15
Baobab Health, March 2014Harry Hochheiser, [email protected]
Tradeoffs/challenges
" Efficiency vs. Learnability " Expert interfaces
" Learnability vs. power/expressive control? " Power tools may be hard to learn
" Satisfaction vs. learnability " Hidden interactions on tablets
" Efficiency vs. Errors?
Page 16
Baobab Health, March 2014Harry Hochheiser, [email protected]
Human Abilities
Motor Abilities: How we can manipulate our world
Perception: What we take in from surroundings
Cognition: What we know
!
Capabilities influence theories, models, guidelines, etc.
leading to interfaces based on realistic understanding of what people can do and how they can work.
not how we think they should work.
Page 17
Baobab Health, March 2014Harry Hochheiser, [email protected]
Motor abilities• How quickly can we move around?
• Cost of various actions
• mouse vs. keyboard?
• cut down on number of steps required?
Fitts's law: time between two targets is proportional to distance + inverse of size of targets:
a – start time, b - speed of device, d- distance, w – target widths
Page 18
Baobab Health, March 2014Harry Hochheiser, [email protected]
BAOBAB JAMA video..
BAOBAB Screen shots.
Page 19
Baobab Health, March 2014Harry Hochheiser, [email protected]
Perception
Visual
Tactile
Auditory
Taste/Smell
rarely used in HCI
Page 20
Baobab Health, March 2014Harry Hochheiser, [email protected]
Visual PerceptionGood use of Contrast/Color help
except when they don't – color blindness
Which is easiest to read and why?
What is the time?
What is the time?
What is the time?
!What is the time?
!What is the time?
What is the time?
Page 21
Baobab Health, March 2014Harry Hochheiser, [email protected]
show baobab web site.
Page 22
Baobab Health, March 2014Harry Hochheiser, [email protected]
Pre-attentive Processing
• Limited set of visual properties are processed pre-attentivel without need for focusing attention • < 200 - 250ms qualifies as pre-attentive !
• Important for design: • what can be perceived immediately: things pop out • what properties are good discriminators
• Color, shape • what can mislead viewers !
• Combinations may not be pre-attentive
Page 23
Baobab Health, March 2014Harry Hochheiser, [email protected]
Find the red circle
Page 24
Baobab Health, March 2014Harry Hochheiser, [email protected]
Find the red circle
Page 25
Baobab Health, March 2014Harry Hochheiser, [email protected]
Not Pre-attentive: conjunction of shape and color
Page 26
Baobab Health, March 2014Harry Hochheiser, [email protected]
• Pattern perception driven by
• proximity
• similarity
• smooth/good continuation
Perceptual Grouping - Gestalt
Page 27
Baobab Health, March 2014Harry Hochheiser, [email protected]
Grouping Rosenholtz, et al. 2009
Page 28
Baobab Health, March 2014Harry Hochheiser, [email protected]
Grouping Rosenholtz, et al. 2009
Page 29
Baobab Health, March 2014Harry Hochheiser, [email protected]
Cognitive concerns
Attention
Memory
long-term
short-term
Language: reading, speaking, listening
Problem solving: planning, reasoning & decision-making
Learning
Page 30
Baobab Health, March 2014Harry Hochheiser, [email protected]
Attention We don't see -or “attend to” most of the world around us
attention -filtering out the noise
without attention, we'd couldn't function
how do we capture attention
“retinal attributes” intensity, marking, size, fonts, color..
change
• movement
• lights
• sound
• etc..
Cost of capturing attention: distraction
Page 31
Baobab Health, March 2014Harry Hochheiser, [email protected]
Question: When to claim attention?
Want to notify users of problems
Each claim on attention has a cost
context-shift takes time, loses focus
When should an interface interrupt the user?
Page 32
Baobab Health, March 2014Harry Hochheiser, [email protected]
Change Blindness http://www.youtube.com/watch?v=Ahg6qcgoay4
Page 33
Baobab Health, March 2014Harry Hochheiser, [email protected]
Short-term Memory
temporary recall
small - 7 +/-2 “chunks” • George Miller – remembering strings of numbers
rapid access, rapid decay
repeat to send to long-term
sometimes used in interaction design.
not necessarily appropriate.
• interfaces are about recognition
• STM is about recall
Page 34
Baobab Health, March 2014Harry Hochheiser, [email protected]
Long-term memory
Huge capacity
Slow access, decay
episodic: events and experiences
semantic: structured records of facts, concepts, skills.
Page 35
Baobab Health, March 2014Harry Hochheiser, [email protected]
Examples of ways that interfaces can work with memory constraints?
• Don't overtax short-term
• excel copying
• Help build long-term
• consistency
• (don’t force people to remember what’s been entered -)
• vb screenshot.
Page 36
Baobab Health, March 2014Harry Hochheiser, [email protected]
Conceptual models" Mental constructions that describe how something works
" or how you think it works
Good models help with predictability
key concept in HCI
To build good conceptual model
must understand the way users work and approach problems
otherwise, you get interfaces that reflect models from the designer, organization, etc., that don't reflect how users work.
Deep vs. shallow
deep – understanding of mechanisms - “white box”
shallow – how to use something - “black box”
Page 37
Baobab Health, March 2014Harry Hochheiser, [email protected]
Whose conceptual model? Don Norman, Design of Everyday Things
Page 38
Baobab Health, March 2014Harry Hochheiser, [email protected]
Norman's Refrigerator Thermostat Don Norman, Design of Everyday Things
Page 39
Baobab Health, March 2014Harry Hochheiser, [email protected]
Norman's refrigerator thermostatMental Models
It's been > 15 years since I first read this book, and I'm still confused by this model.
.. and my fridge is still not correctly adjusted.
Page 40
Baobab Health, March 2014Harry Hochheiser, [email protected]
Examples of mental model mismatches?
Page 41
Baobab Health, March 2014Harry Hochheiser, [email protected]
Language: reading, speaking, listening
• We're good at these things
• but, they require attention
• May distract from other tasks
• don't like to read manuals
• try listening to the news while you work on computer
• Speech recognition: thinking about your voice controls may interfere with language tasks (i.e, writing)
Page 42
Baobab Health, March 2014Harry Hochheiser, [email protected]
Problem solving: planning, reasoning & decision-making
• When solving the problem of how to accomplish a task with the computer..
• System should help figure out next tasks
• logical design – visibility and mapping
• When trying to use the computer to solve a problem
• System should get out of the way
• Not distract from focus on problem
• interfaces that disappear
• No one wants to just “use the computer” - they want to accomplish goals
Page 43
Baobab Health, March 2014Harry Hochheiser, [email protected]
The Psychopathology of Everyday ThingsDon Norman – Design of Everyday Things, Ch. 1
Favorite quote: “if a door handle needs a sign, then its design is probably faulty”
“When simple things need pictures, labels, or instructions, the design has failed”
Thinking about how design relates to human psychology
ideally, design should be consistent with how we think
Learn from failures
Principles all applicable to computer interfaces
Page 44
Baobab Health, March 2014Harry Hochheiser, [email protected]
Norman's Seven Stages of Actions
Form goal Form intention Specify action execute action perceive state interpret state evaluate outcome !!Cycles evaluation – mismatch between system output and user expectations
Page 45
Baobab Health, March 2014Harry Hochheiser, [email protected]
Gulfs of Execution and Evaluation
User Goals System Capabilities
Gulf of Execution - interface design
Intentions Action Specification
Interface Mechanism
Gulf of Evaluation - information design
Interface Display
InterpretationEvaluation
Page 46
Baobab Health, March 2014Harry Hochheiser, [email protected]
• Goal: Enter a new prescription
• Initial intention: How do I decide where to start?
• selecting a medication…
• Action Specification: Once I know first step, what do I do?
• Interface Mechanism: which buttons do I press, in which order?
Gulf of Execution
Page 47
Baobab Health, March 2014Harry Hochheiser, [email protected]
• After execution
• Interface display: What happened/changed in response to my action?
• Interpretation: What does it mean?
• Evaluation
• did the right thing happen?
• am I closer to my larger goal?
• do I need to change my goal or my choice of tactics?
!
Gulf of Evaluation
Page 48
Baobab Health, March 2014Harry Hochheiser, [email protected]
Bridging Gulfs
Execution clarity of action
Transparent interfaces with affordances
Wizards
Help/Tutorials
Evaluation Feedback
Explanation
Error Messages
Undo
Page 49
Baobab Health, March 2014Harry Hochheiser, [email protected]
Affordances
“actionable properties between
the world and an actor (a person or animal).” (Gibson, quoted by Nielsen http://www.jnd.org/dn.mss/affordances_and.html)
!“Perceived affordance” - design that communicates possibility of an action to the user.
!
What makes an affordance? Examples?
Page 50
Baobab Health, March 2014Harry Hochheiser, [email protected]
Models
Response time - how long it takes from initiation to display of results
Think time – time before starting next action
As response time decreases, approach simpler model
Interface can help reduce planning time?
How to measure think time?
Page 51
Baobab Health, March 2014Harry Hochheiser, [email protected]
• Predictive Models of task completion times
• Motion
• Button press
• Think, etc.
• CogTool: www.cogtool.com
Keystroke-Level Models & Goals, Operators, Methods, Selection Rules
Page 52
Baobab Health, March 2014Harry Hochheiser, [email protected]
Cogtool and ANC
Page 53
Baobab Health, March 2014Harry Hochheiser, [email protected]
Cognitive Load, Errors, and Response Time
More effort spent planning and waiting increases cognitive load and short-term memory - more errors
But, errors are more expensive
vicious cycle
Interruptions can be expensive
derail thinking/flow
Page 54
Baobab Health, March 2014Harry Hochheiser, [email protected]
Mappings
Correspondence between what you want to do
what appears to be possible
You want to go forwards/backwards, but controls are only up and down
visibility indicates mapping between intended actions and operations.
Light switches in order of where the lights are?
Good mappings reduce cognitive overload
leverage previous knowledge
Page 55
Baobab Health, March 2014Harry Hochheiser, [email protected]
Mapping Successes and/or Failures?
Refrigerator controls -
2 controls implies 2 systems to regulate
!
Others?
Page 56
Baobab Health, March 2014Harry Hochheiser, [email protected]
Visibility• “correct parts must be visible, and they must convey the
correct message”.
• “natural signals”
• What's natural in a computer interface? • what's intuitive?
• What does visibility look like?
• Is strict visibility crucial?
• Weaker notion: parts that aren't immediately visible should be findable in appropriate places?
• “Whenever the number of possible actions exceeds the number of controls, there is apt to be difficulty.”
Page 57
Baobab Health, March 2014Harry Hochheiser, [email protected]
• Interfaces are languages
Levels of Visibility Lexical-Syntactic-Semantic-Conceptual Foley & van Dam, as interpreted by Myers
(http://www.cs.cmu.edu/~bam/uicourse/830spring13/830-09-softorg.pptx)
Letters/Words Sentences Paragraphs Ideas
Lexical Syntactic Semantic Conceptual
Keys, Buttons, Sequences Tasks Domain Concepts
Page 58
Baobab Health, March 2014Harry Hochheiser, [email protected]
• Spelling and composition of “tokens”
• “add” vs. “append” vs “^a” vs
• Location of controls
• “Key-stroke”
• Make items visible
• Distinguish between items that can and cannot be acted upon
Lexical Visibility Myers http://www.cs.cmu.edu/~bam/uicourse/830spring13/830-09-softorg.pptx
Page 59
Baobab Health, March 2014Harry Hochheiser, [email protected]
• Ordering of inputs
• Legal a tokens must sensible order in?
Syntactic Visibility
• Tokens must be used in a legal, sensible order.
• Visual cues can improve visibility
• Gray out inactive/inappropriate items
• Grouping and spacing
Page 60
Baobab Health, March 2014Harry Hochheiser, [email protected]
• System functionality
• High-level tasks
• Patient creation
• Patient lookup
• data retrieval
• How do sequences of actions complete a task?
• Task sequence and completion indicators
• Wizards
• Titles/screen layout
Semantic Visibility
Page 61
Baobab Health, March 2014Harry Hochheiser, [email protected]
• Conceptual (definition from Foley & Van Dam text, 1st edition)
• key application concepts that must be understood by user
• User model
• Objects and classes of objects
• Relationships
• Operations
• Example: text editor
• objects = characters, files, paragraphs
• relationships = files contain paragraphs contain chars
• operations = insert, delete, etc.
Conceptual Visibility Myers http://www.cs.cmu.edu/~bam/uicourse/830spring13/830-09-softorg.pptx
Page 62
Baobab Health, March 2014Harry Hochheiser, [email protected]
Feedback
Action without impact is potentially problematic.
as is reducing complex set of actions to one piece of
information (done/failed)
or, lack of feedback on long-running actions
Don't know if your action has taken place or not
so, you do it again
Therac 25
Page 63
Baobab Health, March 2014Harry Hochheiser, [email protected]
Errors, Mistakes, Slips
Slips – you know what to do, but you do the wrong thing
!
Mistake – incorrect mental model
!
Examples ? Slips? Mistakes?
Page 64
Baobab Health, March 2014Harry Hochheiser, [email protected]
How to tell the difference?
"Mistakes are based on difficulty in underlying mental model "But, mental models aren't visible "Must elicit model from user "“think-aloud” protocols –User describes actions, goals, ideas, as they do the work.
" Statements that suggest a mental model might be inconsistent with system model → mistakes
Page 65
Baobab Health, March 2014Harry Hochheiser, [email protected]
System Capabilities
Execution, Evaluation, Mistakes, Slips
User Goals
Gulf of Execution - interface design
Intentions Action Specification
Interface Mechanism
Gulf of Evaluation - information design
Interface Display
InterpretationEvaluation
SlipsMistakes
SlipsMistakes
Page 66
Baobab Health, March 2014Harry Hochheiser, [email protected]
How to redesign to avoid slips?
Page 67
Baobab Health, March 2014Harry Hochheiser, [email protected]
How to redesign to avoid slips?
Clarity in labeling?
!Clear affordances
Mapping
Visibility
Feedback..
!Undo!
Page 68
Baobab Health, March 2014Harry Hochheiser, [email protected]
Redesign to avoid mistakes?
Page 69
Baobab Health, March 2014Harry Hochheiser, [email protected]
Redesign to avoid mistakes?
Preferred approaches:
Revise the system model to meet the user model
Revise work and information flow.
!Acceptable:
Increase visibility, feedback, mappings of the system
model
!Not mutually exclusive!
Page 70
Baobab Health, March 2014Harry Hochheiser, [email protected]
Slips and Mistakes in Clinical Informatics
Why so important?
!
Factors that distinguish clinical informatics from other settings?
!
“To Err is Human” 1999 – Institute of Medicine
Page 71
Baobab Health, March 2014Harry Hochheiser, [email protected]
A cognitive taxonomy of medical errors Zhang, et al. 2004
System Hierarchy !U = f(u,c,t,s)
U → usability
u → user
c → context
t → task
s → system
Page 72
Baobab Health, March 2014Harry Hochheiser, [email protected]
A cognitive taxonomy of medical errors Zhang, et al. 2004