Designing Interfaces that Help Students Think

1
Designing Interfaces that Help Students Think
Sharon Oviatt oviatt_at_incaadesigns.org
HCSNet, March 2008
2
Whats Wrong with Modern Interfaces Are We
Productively Digital Yet?
Leonardo da Vincis Codices

• We have become highly technical, but
  computing has not permeated many
  important areas of our lives- Math education
  still hardcopy pencil paper
• Current interfaces
• Support mechanical tasks well, but not complex
  tasks focused problem
  solving
• Not optimally flexible, collaborative mobile
  yet
• Limit users expressivity- linguistic numeric
  input better supported than symbolic
  diagrammatic representations
• Generate many interruptions, distractions
  excess cognitive load
• How can we design new interfaces that stimulate
  thinking on paper? (like productive
  fluency in da Vincis codices)

3
Tom Friedmans Age of Interruption the
Malady of Modernity

• Chronic multitasking continuous partial
  attention, induced by cell phones, email,
  internet, handhelds other devices
• Is the Age of Interruption shrinking our
  attention spans, ability to think synthetically
  stay attuned to our world?
• Tom marvels at Gilbert his rainforest guide who
• carried no devices did not suffer from
  continuous partial attention. Just the opposite.
  He heard every chirpor crackle in the forest and
  would stop us in our tracks immediately and
  identify what birdit was. He also had incredible
  vision and never missed a spiders webHe was
  totally disconnected from the Web, but totally in
  touch with the incredible web of life around
  him.
• 
  -Friedman, Lima Peru,
  2006

4
Whos Designing for Gilbert the Rest of Us?

• Whats human-centered design?
• Modeling users natural behavior (including
  constraints on attention, learning, performance
  communication) so interfaces are more intuitive,
  easier to learn use, and freer of errors
• Designing interfaces that minimize users
  cognitive load, so mental resources are
  available for primary tasks
  staying attuned to the world (especially when
  mobile)
• Supporting existing work practice communication
  patterns
• Societal Impact Substantial improvement in
  commercial viability of real-world applications
  (e.g., education)

5
Is Technology Helping to Conquer the Persistent
Achievement Gap?

• During the past 8 years
• Gap in WASL math achievement scores between
  white black 7th graders has increased 13
• While gap in basic computer internet access
  between white black Americans has narrowed 6
  
• Simple availability of computing hardware
  digital tools cannot alone reduce the achievement
  gap if
• Interfaces arent equally usable accessible by
  all groups
• Content isnt culturally relevant meaningful
• Lower-performing students have weaker
  meta-cognitive skills, including knowledge of
  when how to use digital tools

6
Human-Centered Design Principles Directions
for Educational Interfaces

• Promising directions for educational interfaces
  include new pen-based, tangible multimodal
  interfaces that
• Preserve students existing work practice
  (including collaborations), rather than
  attempting to change them
• Support students flexible communication patterns
  expressive range, including linguistic,
  numeric, symbolic diagrammatic input
• Flexibly bridge students work needs across
  formal, informal, mobile contexts
• Minimize extraneous interface complexity load
  associated with system distractions- which
  undermine high-level planning, integrative
  thinking problem solving

7
Theoretical FrameworksCognitive Load Theory

• Extrinsic complexity of instructional interfaces
  compete for working memory resources
  with intrinsic complexity of main
  learning task
• Easier to acquire new schemas if instructional
  interfaces minimize demands on students working
  memory
• Major goal- measure minimize extrinsic
  complexity so students can devote
  mental resources fully to learning
• Human-centered design principles can minimize
  load (Oviatt, ACM Multimedia 2006)
• Example Multimodal UIs expand working memory
  size by distributing auditory visual
  processing across separate brain areas

8
Interface Design that Minimizes Students
Cognitive Load

• During learning tasks, students cognitive load
  is high
• Educational applications are excellent forcing
  function for designing low-load interfaces,
  because effective learning
  requires focus
• Areas like mathematics (e.g., geometry) are not
  well supported by existing
  interfaces
• Requires focus problem solving, rather than
  mechanical tasks
• Requires expressing translating between diverse
  representational systems (linguistic, numeric,
  spatial diagrammatic) to support
  clear thinking

9
Study on Interfaces for Math Education

• Goal Design new interfaces for mathematics that
  minimize cognitive load
  support focused problem solving
• Participants 20 high school geometry students
• Experienced with GUIs, interested in technology
• Varied math skills (low-high)
• Geometry problems completed (difficulty low-very
  high)
• Performance using 4 interfaces compared
• Existing hardcopy work practice (paper pencil),
  PP
• Digital stylus paper interface (Anoto-based),
  DP
• Pen-based PC tablet interface, PT
• Graphical PC tablet interface (free choice
  keyboard, mouse, pen, simplified equation
  editor), GT

10
Math Education Study Methods (cont.)

• Counterbalanced order (interfaces, problem sets)
• Analyzed auto timings, problem correctness,
  speech think-aloud data, self-reported
  preferences, etc.
• Assessed cognitive load as dynamic information
  processing (attention, fluency, meta-cognitive
  skills, memory, etc.)
• Sample hypotheses
• Interfaces that support existing work practice
  will minimize load improve geometry performance
• Interfaces generating higher load will increase
  achievement gap between low- high-performers
  (expanding digital divide)
• Pen-based interfaces will stimulate greater
  fluency, which aids clear thinking creative
  problem solving

11
(No Transcript)
12
Think-Aloud Examples during Problem Solution in
Different Interface Conditions

• Paper Pencil Example of good focus on math
  problem Catches an error and reworks
  problem solution
• Digital Stylus Interface Example of math
  solution Comments about chunkiness
  of pen
• Pen-Tablet Interface Examples of (1) High-level
  meta-cognitive math comments (2) Low-level
  localized math comments (3) Interface comments
  about lasso symbol
• GUI-Tablet Interface Start out on problem, but
  many interface distractions intervene before they
  return to math problem err on it

13
Task Completion Time Attentional Distractions
Increase in Tablet Interfaces, Compared with
Paper Ones
Distracted attention based on frequency of
interface comments during think-aloud
Time to complete math problems
14
As Interface Distractions Increase,
Meta-Cognitive Control Drops in Graphical Tablet
Interface,with Sharper Decline in Low-Performing
Students
Decline in high-level self-reflective math
comments
Decline in advance diagramming when planning
problem solutions
15
Ability to Solve Problems Correctly Remember
Content Deteriorates only in Low-Performing
Students when using Tablet Interfaces,
especially in Graphical Tablet UI
Low performers made more math errors in
graphical tablet UI
Low performers forgot more math content in
tablet UIs
16
Performance-Preference ParadoxMeta-Cognitive
Skills in Low and High Performers
Low performers who made more errors with tablet
interfaces nonetheless preferred them, while
high performers preferred paper interfaces
Percent problems correct
Reported preference
17
Overall Fluency (Words, Digits, Symbols
Diagrams) for High- vs. Low-Performing
Students Using Different Interfaces
18
Typical Self-Report Responses for Different
Interface Conditions

• Paper pencil
• If I had to focus and do my best, Id use paper
  and pencil.
• Im most used to it, and it was easiest, most
  efficient, and accurate.
• Digital stylus
• I stayed focused, just like paper and pencil.
• It was closer to my normal process, so I was not
  distracted.
• Id like it better if it was less chunky a
  digital pencil I could erase.
• Pen tablet
• It was cool. I loved using the pen. I could
  draw, erase, undo, and change things easily.
• It was distracting because it was so fascinating
  to play with.
• My writing looked funnier. It was hard to write
  nicely.

19
Typical Self-Report Responses for Different
Interface Conditions

• GUI tablet
• Sometimes clicking and choosing the right symbol
  was confusing and distracting. When I used it, I
  was focusing on the computer, not the problem.
• It was not that helpful at solving problems.
  Okay for input though.
• It was just faster to write.

20
Conclusions

• Performance best with interfaces more similar to
  existing work practice (DP gt PT gt GT)
• Digital paper interface (DP) best supported
  students performance (especially speed,
  correctness, attention memory)
• Pen-based interfaces (DP, PT) supported
  expressive fluency, meta-cognition correctness
  better than graphical ones (GT)
• Low-performing students experienced more
  cognitive load than high performers on same
  interfaces and tasks (drop in
  correctness, memory meta-cognition)
• Low-performers did not benefit equally from new
  digital tools
• Cognitive Load Theory provided powerful framework
  for predicting rank-ordered performance by
  interface type

21
Implications Future Directions

• Digital divide only is conquered when everyone
  can use new interfaces effectively to
  advance their performance
• Both cognitive load digital divide issues will
  require special attention in future educational
  interface design
• Further work now on
• Longitudinal tracking of interface use over time
• Other domains (e.g., science)
• Digital stylus paper interface design
• Situated classroom testing (e.g., impact on
  learning, generalization social/organizational
  issues)
• Mobile collaborative usage patterns
• Diverse student populations (disabled, First
  Nations)

22
Other Recent ProjectsCollaborative Peer
Tutoring withFlexible Digital Paper Pen Tools
23
Other Recent ProjectsI SEE- Immersive Science
Education For Elementary Kids
Children spontaneously asked 100-300 questions
during 1-hour session while conversing
directly with digital fish during
information-seeking dialogues about marine
science
24
Video of Michelle (9 yrs), who asked over 300
questions during 1-hour session