Roschelle: Designing for Conversations

1
Roschelle Designing for Conversations

• CptS 538Spring, 2006
• Christopher Hundhausen

2
Who is Jeremy Roschelle?

• One of the pioneers of the computer-supported
  collaborative learning community
• Got Ph.D. from Stanford under Andrew DiSessa
• Studied the use of visualization in collaborative
  science learning
• Comes from Situated action tradition used
  conversation analysis in his studies
• Current work investigates how to scale up
  educational technologies such as SimCalc
• Nice bio can be found at http//ctl.sri.com/people
  /displayPerson.jsp?Nickjroschelle

3
Where Did This Paper Come From?

• Designing for Conversations is one of my
  all-time favorite papers, and a key source of
  inspiration for my dissertation research.
  Unfortunately, it was presented at a conference,
  and never officially published.
• (Closest Ive found J. Roschelle (1994)
  Designing for cognitive communication Epistemic
  fidelity or mediating collaborative inquiry? The
  Arachnet Electronic Journal on Virtual Culture 2.
• I acquired this paper from my advisor, Sarah
  Douglas, when I was a Ph.D. student at the
  University of Oregon.

4
The Key Contrast

• Epistemic Fidelity A visualization is
  educationally valuable because it accurately
  portrays an experts mental model of a phenomenon
  (denotational correspondence)
• Symbolic Mediation A visualization is
  educationally valuable because it provides
  resources for managing the uncertainty of meaning
  in conversations that inevitably occurs

Conventional Representation of Velocity
Acceleration
Alternative Representation of Velocity
Acceleration
5
Evidence of Symbolic Mediation

• Hal I still I dont understand what its
  doing.
• Gerry OK, I think. See it acceleration pulls
  it velocity down.
• Hal Mm, hmm.
• Gerry So it acceleration makes this velocity
  shorter.
• Hal Ohright.
• Gerry cause thatll slow it down.

6
Roschelles Thesis

• An ability to benefit from an epistemic
  correspondence in a visualization depends on
  viewer.
• Learners often do not have the knowledge to
  appreciate and benefit from an epistemic
  correspondence.
• As designers, we need to be sensitive to this,
  and to design for symbolic mediation, which may
  often be in contradiction to epistemic fidelity.

7
Differing Roles that Visualization Can Play in
Education

• System Type Role
• Intelligent Tutoring System Teacher/Coach
• Page Turner Textbook
• Simulated laboratory Lab Bench
• Microworld Blackboard
• Roschelles EM system falls squarely in the
  microworld category. Notice the different
  learning theories that underlie each role.

8
Case Study Envisioning Machine

• Roschelles thinking on the EF vs. SM distinction
  evolved out of the iterative design of the
  Envisioning Machine
• The Envisioning Machine
• a simulated microworld for learning Newtonian
  physics
• Represents velocity and acceleration as vectors
  on a particle in an ideal world
• Vectors can be directly manipulated
• Particle can be set in motion, allowing viewer to
  see an animation of the particle moving through
  space, and a trace of its path

9
Case Study Methodology

• Design an EM version based on a set of design
  principles
• Implement EM version
• Videotape pairs of participants using EM version
• Transcribe and analyze interaction for evidence
  of learning
• Iterate

10
EM-1 Case Study

• Design Rationale Epistemic Fidelity!
• Velocity and acceleration represented as single
  vectors, with acceleration at top of velocity
  vector
• Real-time registration of velocity vector values
  an experts mental model would register such
  changes
• Contradiction in EF principle uncovered
• Representing an abstract concept requires
  arbitrary design choices that could be read in
  to
• Conventions adopted for this purpose. But
  newcomers arent privy to them!

11
EM-1 Case Study (cont.)

• Observations
• Expert participants dug it!
• Novice participants (newcomers) had trouble
• Didnt know how to benefit from the display
• Didnt know where and when to look Focused on
  moving particle instead of changing length of
  velocity vector
• Developed mental model in which acceleration was
  first order instead of second order

12
EM-2 Case Study

• Design Rationale Correct shortcomings of EM-1
• Add tools to help newcomers interpret the display
  more accurately (see Figure 4, p. 13)
• Grid and clock allow one to measure change of
  position
• Velocity space view
• Numerical representations of velocity and
  acceleration

13
EM-2 Case Study (cont.)

• Observations
• New features overwhelmed newcomers!
• Numerical representations were misleading,
  because students focused on the numbers instead
  of the arrows
• Decision was made to turn control over to a
  tutor, who directed learners attention to
  appropriate displays
• After 3 hours with tutor, students could
  articulate proper understandings of acceleration
  and velocity
• However, Jeremy was interested in less tutor
  intervention

14
EM-3 Case Study

• Design Rationale Develop environment to support
  30-60 minute activity for pairs of learners.
• Minimal tutor intervention
• Assist learners in developing correct
  interpretations of velocity and acceleration

15
EM-3 Case Study (cont.)

• Design principles
• Minimalism Support the minimum necessary
  features
• Persistence Past trajectories should be visible
  while learners are adjusting vectors for the next
  attempt
• Selective redundancy Redundant representation
  of foreground objects such as speed, which is
  represented as trace dots, a vector, and an
  animation

16
EM-3 Case Study (cont.)

• Design principles (cont.)
• Direct manipulation for Communication
• Show past vector settings while vectors are being
  adjusted, so that adjustments can be discussed
• Support forwards and backwards execution of
  animation
• Challenge activity structure
• Give learners progressively more difficult
  challenges in which they need to match particle
  motion to a goal trajectory.

17
EM-3 Case Study (cont.)

• Observations
• Success!
• Visualization mediated conversations in which
  students were able to build scientific
  perspectives
• However, students still had misconceptions like
  those with EM-1. A coach is still needed.
• Lots of implementation required for all three
  design iterations. Dedicate most of your time to
  designing activities, not software!

18
Three Analytical Perspectives

• 1. Participating in communities
• Learning takes place by participating more fully
  in a community of practice (where have we heard
  that before)?
• A newcomers perspective is gradually transformed
  into a more expert one
• Mediational tools such as visualizations are
  indispensable in this process for negotiating
  meaning
• Microwords provide valuable scaffolding that
  enables learners to participate in authentic
  activities without undue expense or danger

19
Three Analytical Perspectives (cont.)

• 2. Engaging in Activities
• Learners engage in coordinated work
• Such work should accord with authentic practices
• Visualization helps them get that work done
• 3. Negotiating meaning in Conversations
• The conduit metaphor vs. collaboratively-constru
  cted meanings
• Visualizations are resources for establishing
  common ground
• The graphical elements of the EM-3 display enter
  into the composition, design, and structuring of
  students interaction, thereby facilitating
  meaningful conversations about motion (p. 22).

20
Discussion Questions

• 1. What are the implications of epistemic
  fidelity for the design of visualizations? How
  do those implications contrast with those of
  symbolic mediation?
• 2. What is activity fidelity? Why is it
  important for learning?
• 3. Based on your own experience, can you cite an
  example in which a violation of epistemic
  fidelity in a visual representation increased
  the value of that visual representation as a
  mediational resource?
• 4. Describe an internal conflict in the principle
  of epistemic fidelity that can lead to
  undesirable interpretations of visualizations.
• 5. Roschelle suggests that epistemic fidelity and
  symbolic mediation perspectives are not
  completely at odds. What is his view, and do you
  agree?
• 6. To what extent do Roschelles design
  principles (minimalism, persistence, selective
  redundancy, direct manipulation for
  communication, and challenge activity
  structure) jive with Hundhausens Communicative
  Dimensions? Are there any notable similarities?
  Do they even share the same goals?
• 7. In your opinion, what visualization design
  process has the best chance of leading to an
  effective visualization, both with respect to
  symbolic mediation, and with respect to the
  cognitive efficiency concerns we considered
  earlier in the course?