View
219
Download
0
Tags:
Embed Size (px)
Citation preview
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
MAPS - LifeLine(Distributed Support System)
Stefan Carmien11/4/2003
L3D
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Challenge -
• Individuals with cognitive disabilities are often unable to live on their own because of deficiencies in memory, attention, and executive functionalities.
• Computationally enhanced prompting systems can provide a bridge to independence.
•Unacceptably high abandonment rate (50% - 70%) of expensive (~.5-4 K$) Assistive Technology (AT)
• Difficult (re) configuration contributes to abandonment
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Research Objectives• Lower abandonment
– Dual user interface design• Provide safety net functionality
– Panic button– Caregiver monitoring
• Context aware cognitive assistance– Dynamic prompt generation – Effective error detection and correction
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
MAPS (Memory Aiding Prompting
System)
• Hand held prompter – to display plans
• Caregiver script editor– To create, modify and share plans
• Interface to LifeLine– Safety net/dynamic context aware prompting
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Plans, Errors? What are you talking about?
• What is a plan/script?– A task of the right size broken down into small atomic
chunks– Made into a script (of visual and verbal prompts) that can
be followed• What is an error or error state ?
– A detected deviation from the plan– Sensors provide input– Boolean tests of expected states
• Example: <bus GPS and your GPS are the same and distance between correct stop and bus is getting larger >
• How do you correct a detected error?– User templates provide ‘lookup table’ written in script data
structure
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
MAPS Prompter
MAPS Prompter (user interface)
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Caregiver Script Editor
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Demo here
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Conceptual Framework• User modeling
– Templates support static user modeling in configuration– Templates for error trapping/correction
• Information ecologies – Users with cognitive disabilities– Caregivers– System components
• Distributed Cognition– Dynamic prompt generation– Sensor & table driven
• Situated Action– Expect & plan for failed plans (driven by breakdowns)
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Tool Design
SituatedAction
Time
EnvisionedUse
ActualContext
MetaDesign, Plans and Situated Action
ClientCaregiver
Key
Designer Caregiver
Plans
EnvisionedContext
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
LifeLine prototype
Andrew Gorman’s LifeLine project
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Conceptual Design
Information ecology supporting caregivers and users with cognitive disabilities
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Distributed Support System
• Distributed– Prompting (MAPS) is inherently distributed
cognition– Knowledge to maintain system stability is
distributed between: user/caregiver/maps/lifeline/sensors/database
• Support– Caregiver is supported by ensuring safety– User activity is supported to afford
independence and inclusion
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Demo 2 here
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
User Modeling & Error Trapping / Correction
• Reusable rules• Templates for tasks & segments of
tasks• Error
– Trapping (how do I recognize an error for this person)
– Correction (how do I correct an error for this person)
– Embedded in script database records (schema)
• Caregiver combinatorial explosion dilemma:
(User type) X (task segment) X ( |prompts in this segment| )
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
MAPS – Error Detection & Correction
• Template based dynamic error detection & correction• Lifeline monitors the running script, detects error states & contacts caregiver as needed
Tom(more cognitive
resources)
Anne(less cognitive
resources)MissedBus Stop !
Tell the Bus driver
your problem
Wait on bus for
Caregiver
Aler
t Car
egiver
Summ
on
Caregive
r
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Context Sensitive Prompting
• Dynamic prompting – Loop till event
• wait for your bus
– Fork • employment (if this then do this….. else do that)
• Error (trapping and correction)– Sensors– User type – Task segment
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
MAPS – Context Awareness and Dynamic Prompting
•Dynamic prompt generation•Blocking for event driven prompt
This is not your bus
Wait here for
your bus
Here is your bus,
get on it.
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Meta Design• Abandonment strongly influenced by lack
of a tool to reconfigure/add content– Empowering and enabling caregivers to be
designers of scripts– Leverage existing skills
• Tool needs to provide just enough support to make script generation easy but not so much that it itself needs a configuration tool
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Meta Design II• User Modeling
– User model– Task and task segment error trapping/correction
• Critiquing – Verbiage– Image attributes
• Sharing scripts – Scripts as templates– Community building
• MAPS odd twist on MetaDesign – Phased/tiered metadesign
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Shifting gears now….
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Tools for LivingTools for Learning
• Initial insight from AT world– Enabling learning – Enabling daily life
• Examples– Eyeglasses– Learning to read / use maps & schedules
• Grey areas and argument able issues– Hand calculators– Scanners/readers & dyslexia
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Tools for living
• You require this tool to accomplish this task - otherwise you cannot
• The tool rarely changes or is abandoned (in the sense of ‘graduate’)
• It is specifically tailored to you • Your use of it does not typically
change over time
External, Fitting You, & Permanent
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Tools for Learning
• Is your skill in the tools domain significantly greater (such that you no longer need the tool to perform meaningful work) as a result of using/experiencing this tool?
• Are you different as a result of using this tool?
• Is there a method of having this tool disappear
Internalized, Changing You, & Scaffolded
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Design Implications
• What would you do to make a TfLea– Low entrance cost – Tool adapts over time (Scaffolding)
• What would you do to make a TfLiv– Use of the tool will not change over time– Need for good personalized fit at the
beginning– robust (crutches, eyeglasses)
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Arguments against this idea
• Too amorphous
• Not useful (Just YetAnotherDumbAcronym)
i.e. yada yada yada
• Not really news (other ideas like this)
• What do you think?
CLever: Building Cognitive LeversCLever: Building Cognitive Levers to help people help themselves
Center for LifeLong Learning & DesignCenter for LifeLong Learning & Design
University of Colorado at Boulder
Thanks