Linking Cognitive Science to Education: Generation and Interleaving Effects

1
Linking Cognitive Science to Education
Generation and Interleaving Effects
Lindsey E. RichlandRobert A. BjorkJason R.
Finley University of California, Los
Angeles Marcia C. Linn University of California,
Berkeley

• XXVII Annual Conference of the Cognitive Science
  Society
• Stresa, Italy
• July 23, 2005

2
Introducing Desirable
Difficulties for Educational
Applications in
Science
www.psych.
3
Desirable Difficulties (Bjork, 1994)

• Design principles that have been found, in
  laboratory research, to impair performance during
  training but enhance performance at a delay

4
Desirable Difficulties (Bjork, 1994)

• Spacing rather than massing study
• Interleaving rather than blocking practice on
  separate topics or tasks
• Varying contextual cues
• Reducing feedback to the learner
• Testing rather than re-presenting
• Laboratory studies using simple material

5
Learning versus performance

• What we can observe is performance, what we must
  infer is learning
• and the former is an unreliable index of the
  latter
• Instructors are, therefore, susceptible to
  choosing less-effective conditions of learning
  over more-effective conditions
• And as learners, we, too, are susceptible to
  confusing performance with learning

6
Goals of the IDDEAS project

• Do such findings extend to realistic educational
  materials and contexts?
• And, more broadly, what design principles are
  fundamental in optimizing educational materials
  and practices?

7
WISE (Web-based Inquiry Science Environment)
http//wise.berkeley.edu

• A software system for science instruction
• Advantages as a tool for teachers
• Supports authoring and customization
• Contains a library of tested projects
• Transportable
• Gathers embedded assessments of student progress
• Advantages as an IDDEAS research tool

8
(No Transcript)
9
(No Transcript)
10
On-going studies

• WISE Platform
• Laboratory studies, UCLA
• Classroom studies, UCB
• Design Principles
• Laboratory studies, UCLA

11
Interleaving Effect
(e.g. Shea Morgan 1978)
12
Interleaving

• Motor tasks patterns, force production, bank
  machine transactions (Lee Magill, 1983, Simon
  Bjork, 1990 Charles et. al, 1990, Jamieson
  Robers, 2000)
• Sports badminton, volleyball, baseball (Bortoli
  et al, 1992, Goode Magill, 1986, Hall et al,
  1994)
• Abstract learning tasks mazes, tracking
  (Carleson et.al, 1989, Jelsma Van Merrienboer,
  1989, Jelsma Pieters, 1989)
• Logic rules, boolean operators (Schneider et al,
  1995, Carleson Yaure, 1990)

13
Generation effects Example with simple
Laboratory materials (Hirshman Bjork, 1988)

• Read condition
• Presented Bread Butter
• Participant responds Bread Butter
• Generation condition
• Presented Bread B_tt__
• Participant responds Bread Butter

14
Recall of Butter

• Read condition at 30
  min. delay
• Presented Bread Butter
• Participant responds Bread Butter13
• Generation condition
• Presented Bread B_tt__
• Participant responds Bread Butter41

15
Undergraduate Laboratory Experiment 1

• Explore interleaving and generation effects with
  realistic educational material
• In a controlled laboratory environment

16
Undergraduate Laboratory Experiment 1

• Participants 83 UCLA undergraduates
• Material
• Adapted an existing WISE module about
  habitability of other planets
• 2 Sets of Information to be learned in WISE
  module (divided into single slides)
• Mass of a Planet
• Distance of a Planet from its Sun
• Procedure
• 1 hour spent on WISE module
• Post-test at 2-day delay

17
Undergraduate Laboratory Experiment 1

• 2 x 2 Between-subjects
• IV1 Presentation Order Blocked vs. Interleaved
• MMMMMDDDDD vs.
• MDDMDMMDMD
• IV2 Embedded Study Events Read vs. Generate
• Mercurys mass is less than Earths mass. vs.
• Mercurys mass is ___ than Earths mass.

18
(No Transcript)
19
Exp. 1 Post-test, 2 Day Delay

• Simple sentence-completion Information that was
  presented and re-studied via either generation or
  reading
• E. g. The amount of heat and light emitted by
  the sun in our solar system has increased by
  ____ since the beginning of earths history.
• Integration questions within a topic (mass or
  distance)
• E. g., Would an object weigh more on the planets
  in our solar system made mostly of gas or made
  mostly of rock? Why?
• Integration questions across both topics (mass
  and distance)
• E. g., Imagine a planet that is smaller than
  Earth and that was located 1.5 AUs from its sun,
  which is the same strength as the Earths sun.
  How would this planets potential for life
  compare to Earths?

20
Exp. 1 Read vs. generate performance on
sentence completions
21
Exp. 1 Blocked vs. Interleaved by type of
Posttest Integration Question
22
Undergraduate Laboratory Experiment 2

• Participants 44 UCLA undergraduates
• Same material
• All interleaved order
• Sentence level generation questions
• E.g., Describe in a sentence how the size of one
  planet's mass can affect another planet.

23
Undergraduate Laboratory Experiment 2

• Between-subjects
• IV Knowledge required for successful generation
• Within-topic integration e.g. Mass only vs.
• Between-topic integration Mass Distance
• DV Post-test at 2 day delay

24
Exp. 2 Single-topic vs. Topic-integration
generation
25
Undergraduate laboratory studies Reflections

• Findings promising but not straightforward
• Possible benefits of interleaving
• Definite benefits of generation for a specific
  item
• Possible broader benefits of generation
• Many questions remain
• Repetitions
• Relatedness of topics
• Feasibility
• Extending the chain of evidence
• Parallel Studies laboratory classroom

26
Undergraduate laboratory studies Reflections

• Advantages of Collaboration (complex systems
  approach)
• What is realistic?
• What is the goal of our research?
• Technology as bridge Catalyst

27
For More Information

• IDDEAS http//iddeas.psych.ucla.edu
• WISE http//wise.berkeley.edu
• Jason Finley jfinley_at_ucla.edu

28
Exp. 2 Illustrative responses