Promoting Transfer through Case-Based Reasoning: Rituals and Practices in Learning by Design

1
Promoting Transfer through Case-Based Reasoning
Rituals and Practices in Learning by Design
Classrooms

• Janet L. Kolodner
• College of Computing
• Georgia Institute of Technology
• Atlanta, GA, USA
• (and the Learning by Design project team)

2
Learning by Design

• A project-based inquiry approach to science
  education for middle school
• Students learn science concepts and practices in
  the context of attempting to achieve design
  challenges.
• Highly collaborative
• A variety of practices and scaffolding tools are
  embedded in the approach to promote the kinds of
  experiences and reflection that promote transfer.

3
Science Practices

• Understanding a problem and what might need to be
  investigated
• Investigation with a purpose experimentation,
  modeling, learning from cases, library lookup,
  ...
• Informed decision making, reporting on and
  justifying conclusions
• Iteration towards understanding
• Explaining scientifically
• Teamwork, collaboration across teams, giving
  credit

4
Our Units

• Physical Science
• Apollo 13 introduction to practices of design
  and science
• Vehicles in Motion motion and forces
• Machines that Help simple machines and
  mechanical advantage
• Earth Science
• Digging In -- launcher unit
• Managing Erosion erosion and accretion
• Tunneling through Georgia geology, rocks and
  minerals, rock formations, underground water
• Looking for a publisher

5
Novel Features

• Ritualized classroom activities matched to
  science practices
• Design diary pages matched to activities provide
  scaffolding for performance and reflection
• Software scaffolding matched to activities and
  presentations promotes summary and interpretation
• Launcher units introduce practices and culture
• Orchestration such that students need each
  others' results
• Lots of presentations to promote good kinds of
  reflection
• Highly iterative to promote explanation and
  iterative refinement of conceptions

6
LBD's Community Rituals

• Gallery walk -- scaffolds explanation
• Pin-up session -- scaffolds justification
• Results presentation -- scaffolds justification
  and data interpretation
• Design rules of thumb generation-- scaffolds data
  interpretation
• Messing About and Whiteboarding -- scaffold
  question asking

7
Novel Features

• Ritualized classroom activities matched to
  science practices
• Design diary pages matched to activities provide
  scaffolding for performance and reflection
• Software scaffolding matched to activities and
  presentations promotes summary and interpretation
• Launcher units introduce practices and culture
• Orchestration such that students need each
  others results
• Lots of presentations to promote good kinds of
  reflection
• Highly iterative to promote explanation and
  iterative refinement of conceptions

8
Two Design Diary Pages
9
SMILE

• Design discussions
• Investigation planning and presentation
• Design plans and decisions (pin-up)
• Design experiences (gallery walks)
• Summary authoring
• Goals, plans, results, science used
• Two most important iterations
• Lessons learned and how they might be used

10
Our Experiment

• Your question
• Your hypothesis
• How you will test it
• Your procedure
• Variable you will vary
• Values you will give it
• Properties you will hold constant
• Data
• Interpretation of results
• Rule of thumb

• Scaffolding is of three kinds
• structuring with a leading question or topic
• hints
• examples

11
Describe a design iteration

• Describe your design iteration. Be sure to tell
  us what is different this time and why.
• Describe what you expected to happen when you
  implemented this design.
• Describe scientifically the positive aspects of
  this design. Tell why these aspects were
  positive, using scientific vocabulary. Make sure
  to specify the constraints you took into account
  and the criteria your design achieved.
• Describe scientifically the negative aspects of
  this design. Tell why these aspects were
  negative, using scientific vocabulary.
• Describe the parts of the challenge that were not
  solved.
• What, if anything, needed to be changed?
• Anything else?

12
Lessons Learned

• Choose one of the things you learned and describe
  it. Be as specific as possible. Explain so that
  someone who doesn't yet know what you know could
  understand.
• Describe the experience when you learned it.
• Describe another situation when you have
  experienced this.
• Explain why what you learned might be important
  to someone else. When and how might they use it?
• Describe a situation in the future in which you
  might use what you have learned again.

13
Novel Features

• Ritualized classroom activities matched to
  science practices
• Design diary pages matched to activities provide
  scaffolding for performance and reflection
• Software scaffolding matched to activities and
  presentations promotes summary and interpretation
• Launcher units introduce practices and culture
• Orchestration such that students need each
  others results
• Lots of presentations to promote good kinds of
  reflection
• Highly iterative to promote explanation and
  iterative refinement of conceptions

14
Vehicles in Motion

• Design and build a vehicle that can propel itself
  over several hills and beyond the farther the
  better
• Coaster Car Challenge
• Friction and keeping things going
• Balloon Car Challenge
• Getting and keeping things going
• Rubber-band and Falling Weight Challenge
• Comparing different kinds of propulsion
• Put it all together

15
LBD's Sequencing

• Pose design challenge.
• "Messing about" leads to question posing.
  (Messing About)
• Investigation following scientific methodology.
  (My Experiment SMILE)

• Balloon-car challenge
• W/balloon engines
• Size of balloons?
• Length of straw?
• Diameter of straw?
• Double balloon?
• Double engine?
• Each group chooses a question and designs and
  runs an experiment

16
From Group Work to Class Discussion

• Sharing results
• Drawing out design rules of thumb (My Rules of
  Thumb)

• Why were the results of that run so different?
• Maybe you didn't blow up the balloons the same
  every time.
• Two engines are better than one
• ...

17
Getting to the Science

• Design planning (SMILE)
• Pin-up session (Pin-up Notes)
• Construction and testing (Testing my Design)
• Gallery Walk (SMILE Gallery Walk Notes)
• Need for science (My Rules of Thumb)

• Let's use two engines and double the balloons in
  each because
• When we did that, the wheels spun out. We don't
  know why.
• Read text pages about ...

18
Summary Pulling it all together

• Iterative refinement
• Final gallery walk
• Product history (SMILE)
• Application problems and scenarios
• Lessons learned (SMILE)

• Individual and group writeups
• About science, science practice, collaboration,
  ...

19
More on Sequencing

• Iteration towards better solutions provides
  opportunities for iteration towards better
  understanding.
• Sharing experimental results, design ideas, and
  design experiences promotes focus on
  investigative methodologies.
• Design diary pages and software provide
  scaffolding for doing and reflection in the
  context of ritualized activities.
• Multiple opportunities for students to engage in
  and learn science process, communication,
  collaboration, planning, reflection, design in
  the process of learning and applying science
  content.

20
(No Transcript)
21
How did we get to all this?
22
LBD's Foundations

• Case-based reasoning's model of learning from
  experience (Kolodner, Schank, Hammond, )
• Problem-Based Learning's model of the classroom
  (Barrows, )
• Communities of Learners (Brown, Campione),
  Constructionism (Papert, Harel, Kafai, ),
  Cognitive Apprenticeship (Collins, Brown, ),
  architecture studio, Decision-Based Design
  (Mistree, ),
• Transfer literature (as in How People Learn)

23
Transfer

• The ability to apply something learned in one
  situation in another situation that wasnt
  directly targeted in the learning
• Example 1 sliding up the driveway (reusing
  knowledge)
• Example 2 we did a little pin-up
  (participating in practices)

24
Near and Far Transfer (Low-Road and High-Road)

• Near
• Measuring water during baking gt measuring
  liquid in a beaker or pipette in chemistry class
• Really far (between two contexts)
• Texture of shark skin makes sharks aerodynamic
  gt maybe same texture in bathing costumes will
  make people more aerodynamic

25
What How People Learn Tells Us About Transfer

• Initial learning is necessary for transfer, and a
  considerable amount is known about the kinds of
  learning experiences that support transfer.
• Knowledge that is overly contextualized can
  reduce transfer abstract representations of
  knowledge can help promote transfer.
• Transfer is best viewed as an active, dynamic
  process rather than a passive end-product of a
  particular set of learning experiences.
• All new learning involves transfer based on
  previous learning, and this fact has important
  implications for the design of instruction that
  helps students learn.

26
What Does This Tell us About Designing Classroom
Practices?

• Help learners generalize across individual
  experiences and examples.
• Help learners use what they know.
• Help learners develop the ability to transfer.
• Help them practice transferring
• Make reasoning explicit
• Give them experience over a variety of situations
• Here is where Case-Based Reasoning comes into the
  picture.

27
A Case-Based Reasoner is Constantly Engaged in
Transfer

• It applies lessons learned in old situations to
  new ones.
• Sometimes it succeeds in its endeavors sometimes
  it fails.
• When it fails, it attempts to explain what was
  responsible for the failure and updates its
  memory accordingly.
• It learns by adding new cases, re-encoding and
  re-indexing old cases, and abstracting out
  generalizations.

28
Case-Based Reasoning

• Developed as a way to allow the computer to solve
  complex problems and learn from its experiences.
• The computer remembers and applies the lessons
  learned in one situation to another, storing its
  experiences in its case library.
• Much like people ...

29
Our computer models

• we give our computer programs memories to store
  their experiences
• we have our programs keep track of their
  experiences in memory, intepret their experiences
  so as to extract lessons that could be learned
  from them and the conditions of applicability for
  those lessons, and encode those experiences based
  on those lessons learned and their conditions of
  applicability
• we give our programs means of retrieving
  applicable cases from their memories, judging
  which of several potential cases might be most
  applicable in a new situation, and merging,
  adapting, and applying the lessons learned in new
  situations
• we provide our computer programs with feedback on
  their decisions, helping them explain mistakes
  and/or poor predictions, and helping them revise
  memorys encodings and interpretations as those
  explanations suggest and
• we have our programs notice similarities and
  general rules and draw abstractions to use for
  more sophisticated encoding.

30
CBR defines transfer as spontaneously reusing
some past experience productively.

• Three steps involved in reuse
• remembering (access)
• deciding on applicability
• application
• Getting to productive transfer is a
  developmental process
• practice and explanation
• articulation of lessons learned
• iterative and reflective application of lessons
  learned

31
Our computer implementations provide insights
into the processes involved in "getting to
productive transfer."

• Remembering
• Application
• Learning

32
Remembering (the indexing problem)

• Depends on the quality of three processes
• Interpretation at encoding How well and how
  completely the reasoner interpreted the old
  situation
• Interpretation at retrieval (situation
  assessment) How well and how completely the
  reasoner interprets a new situation
• Matching How good the reasoner's partial
  matching capabilities are

33
Interpretation -- the better one encodes, the
more chance there is of noticing relevant
similarity

• Making of connections
• Extraction of lessons learned
• Making conditions of applicability of those
  lessons explicit
• Allows indexing (labeling according to most
  important complexes of descriptors)

34
Implications

• The better a reasoner can extract lessons and
  conditions of applicability from a situation, the
  better (s)he will see connections between new and
  old situations.
• Help learners interpret their experiences to
  extract what can be learned from them anticipate
  the kinds of situations in which lessons might be
  applied and abstract across a variety of
  experiences to extract general principles.

35
Application

• With a description of a problem situation and its
  solution, an old solution can be repeated.
• If, in addition, the reasoner knows whether the
  solution succeeded when applied, there is a basis
  for deciding whether or not to reuse the old
  situation.
• If, in addition, the reasoner knows what happened
  as a result of applying the solution, more
  reasoned judgments are possible (e.g., does that
  old result make sense in my new situation?).
• If, in addition, factors responsible for the
  result (success or failure) are known, judgment
  of applicability is possible.

36
Implications

• If we help people embellish their understanding
  of what happened in ways that include rich
  connections between goals, actions, and results,
  and if we help them explain the factors
  responsible for what happened, ability to judge
  applicability of whats remembered and to apply
  it will be more likely.

37
What we do in LBD (preview)

• That students need each others results provides
  reasons for presentations.
• The need to present coherently encourages rich
  interpretation.
• The need to give advice to others encourages
  anticipation.
• Design diary pages and software provide
  scaffolding for that interpretation and reminders
  to anticipate use.
• The need to understand applicability of the
  advice of peers encourages active listening,
  questioning of peers, and the drawing of lessons
  from presentations.

38
Learning

• A case-based reasoner learns when it extends its
  knowledge by
• incorporating new experiences into memory in ways
  that are consistent with what is already in
  memory,
• re-encoding old experiences to more accurately
  reflect what one can learn from them and their
  applicability, and
• abstracting out generalizations from experiences.

39
What is required for learning?

• failure and explanation
• useful feedback from the world what happens
  when I try it out?
• the want and ability to explain
• iteration towards better and better explanations
• the want to iterate towards better and better
  explanations
• All of this requires that learners have goals
  they are trying to achieve and that they want to
  achieve and that they have the ability to make
  predictions.

40
How do we help students learn?

• we ask them to achieve engaging goals that can be
  achieved in ways that provide feedback and that
  require several iterations
• we help them keep track of their experiences in
  memory, intepret their experiences so as to
  extract lessons that could be learned from them
  and the conditions of applicability for those
  lessons, and encode those experiences based on
  those lessons learned and their conditions of
  applicability
• we give them practice retrieving applicable cases
  from their memories, judging which of several
  potential cases might be most applicable in a new
  situation, and merging, adapting, and applying
  the lessons learned in new situations
• we make sure they get feedback on their
  decisions, help them explain mistakes and/or poor
  predictions, and help them revise memorys
  encodings and interpretations as those
  explanations suggest and
• we have them notice similarities and general
  rules and draw out abstractions to use for more
  sophisticated encoding.

41
(No Transcript)
42
Novel Features

• Ritualized classroom activities matched to
  science practices
• Design diary pages matched to activities provide
  scaffolding for performance and reflection
• Software scaffolding matched to activities and
  presentations promotes summary and interpretation
• Launcher units introduce practices and culture
• Orchestration such that students need each
  others results
• Lots of presentations to promote good kinds of
  reflection
• Highly iterative to promote explanation and
  iterative refinement of conceptions

43
LBD's Community Rituals

• Gallery walk -- scaffolds explanation
• Pin-up session -- scaffolds justification
• Results presentation -- scaffolds justification
  and data interpretation
• Design rules of thumb generation-- scaffolds data
  interpretation
• Messing About and Whiteboarding -- scaffold
  question asking

44
What makes the rituals work

• A systematic way of carrying out some important
  skill set that
• systematizes practices to make them methodical
  promotes habits
• situates practices in several contexts promoting
  adaptability
• engages students in public practice as
  collaborators affording noticing, asking,
  discussion, productive reflection

45
Show and Tell Rituals

• Gallery walks (explanation)
• Pin-up sessions (justification)
• Results presentation (experimental method)
• Ritualized public ways of participating in
  science practices
• Well-articulated expectations
• Repeatedly practiced and publicly discussed

46
LBDs Connections to CBR

• The design challenge gives kids goals both
  learning goals and performance goals.
• Messing about and experimentation with rules of
  thumb as results helps kids make predictions.They
  articulate predictions in pin ups.
• They discuss results and get help explaining in
  gallery walks, with focus on justification and
  use of evidence.
• Gallery walks and pin-ups give them the chance to
  vicariously experience and explain a wide range
  of applications of science content and practices
  of scientists.

47
Where's the CBR?

• Students reason using cases
• Students collect lots of cases
• They are helped to index their cases well
• They expect, try, fail, explain, try again
  iteratively moving toward better solutions and
  better understanding at the same time.
• They experience and interpret feedback from the
  real world as they run what they design and
  build.

48
What How People Learn Tells Us About Transfer

• Initial learning is necessary for transfer, and a
  considerable amount is known about the kinds of
  learning experiences that support transfer.
• Knowledge that is overly contextualized can
  reduce transfer abstract representations of
  knowledge can help promote transfer.
• Transfer is best viewed as an active, dynamic
  process rather than a passive end-product of a
  particular set of learning experiences.
• All new learning involves transfer based on
  previous learning, and this fact has important
  implications for the design of instruction that
  helps students learn.

49
How is transfer promoted?

• Motivating activity keeps students attention
• Reflection on their experiences in ways that
  promote abstraction from experience, explanation
  of results, comparing and contrasting,
  understanding conditions of applicability
• Repeated use of concepts repeated practice of
  skills experience with concepts and skills over
  a variety of circumstances
• Reasons to reflect on their experiences (they
  have to explain to others others need their
  results)
• Help with reflecting on their experiences, help
  with remembering, help with judging
  applicability, help with application, help with
  explanation

50
Evidence of Transfer

• Reuse of content
• bookstand remindings while doing bridge
• triangle, truss shape, distribution of weight
• new bridge in neighborhood what kind of bridge
  is it?
• Water skiing
• Enculturation into community practices
• pin up episode
• community support for doing and learning
• gallery walks
• planfulness in post-tests
• work together in post-tests
• science fair projects

51
More evidence

• Students initiate their own use of these
  practices
• the pin-up session story
• creating rules of thumb
• messing about during interviews and performance
  assessments
• whiteboards science fair projects
• Students adapt the rituals to later needs
• e.g., when teacher stops calling gallery walks

52
Evidence from Performance Assessment
53
The Big Design Question How do you get from
what the learning literature says to classroom
design?

• Find an approach to practice that almost matches
  what you're aiming for
• ours was Problem-Based Learning
• Adapt it to your constraints
• Iterate