Cognitive science research can improve online learning in several ways:
1) By focusing on learning as integrating information rather than just adding it, and using techniques like problem-based learning, explanations, and assessments as instructional tools to promote deeper learning.
2) By using online platforms to conduct randomized controlled trials to test techniques grounded in cognitive and social psychology for increasing motivation, teaching learning strategies, and fostering a growth mindset.
3) By leveraging the control and scalability of online environments to iteratively improve educational interventions through precise measurements and targeted feedback based on principles of cognitive science.
ISYU TUNGKOL SA SEKSWLADIDA (ISSUE ABOUT SEXUALITY
How can Cognitive Science improve Online Learning & Education?
1. How can Cognitive Science
improve Online Learning?
Joseph Jay Williams
Joseph_Williams@berkeley.edu
www.JosephJayWilliams.com/education
CognitiveScience.Co/learn
www.LearningResearch.net
1
6. • Consider:
– Scenario 1: General invading a fortress can’t use full force, part of force insufficient
– Scenario 2: Doctors destroying an internal tumor can’t use strong rays, but weak
rays aren’t enough
• Transfer extremely low (Gick & Holyoak, 1984)
• Motivated MBA students appeared to learn extremely well, but
failed to transfer to face-to-face in real life (Gentner, Loewenstein, &
Thompson, 2003)
• Transfer is so rare that it’s not a plausible goal (Detterman, 1993)
6
Transfer is rare
7. • Problem Based Learning (Hmelo-Silver, 2006; Needham & Begg,
1998; Schwartz, 1998)
7
How do you…?
Is it possible to…?
Before: Start with Questions & Problems
9. Statistics problem
• Learn a university’s ranking system from examples
(Schwartz & Martin, 2004, Belenky & Nokes, 2011)
9
Sarah was ranked higher.
Rule for ranking John Ranked? Tom
Higher personal score 85%
> 69%
Points above average 6%
> 4%
Points below maximum –5%
> –18%
Number of deviations
above average (Z-score)
0.75
< 1.3
85% in History
Min of 67%, Max 90%
Class Average 79%
Standard Deviation 8%
69% in Physics
Min of 42%, Max 87%
Class Average 65%
Standard Deviation 3%
>
11. Key Findings
• Explaining increased:
• Discovery of principles (Williams & Lombrozo, 2010)
• Use of existing knowledge (Williams & Lombrozo, 2013)
• Explanation’s effect was selective:
• Same or even worse memory
• Impaired learning if patterns were unreliable (Williams, Lombrozo,
Rehder, in press)
• Similar effects in 5 year olds (Walker, Williams, Lombrozo, & Gopnik,
under revision)
12. Online Mathematics Exercises
• Khan Academy:
• Explain why that solution is correct.
• Here is another student/teacher’s explanation.
• Grade both.
• Rate how similar they are.
13. Benefits of explanation
• Instructor guided & learner generated
• Learning without feedback
• Abstract principles
14. • The “Testing Effect” (Roediger & Karpicke, 2006)
14
Immediate test: Study+Study ~= Study+Test
After hours, days, weeks: Study+Study < Study+Test
Learners claim: Study+Study > Study+Test
After: Use Assessments as Instructional Tools
15. • Mixing Effect (Rohrer, 2009)
• Ten Benefits of Testing (Roediger et al, 2011)
15
After: Efficient Assessments use Mixing Effect
16. • Support Randomized Experiments or A/B Testing
• Precise delivery & control
• Quantitative measures of learning
• Ecological validity
• Evidence-based decisions
• Fidelity
• Scalability
• Iterative improvement
16
Real-World Laboratories
18. • Beliefs about intelligence (Dweck, 2006)
• Do you agree that…
– Your intelligence is something very basic about you that
you can’t change very much. (Fixed Theory).
– No matter how much intelligence you have, you can always
change it quite a bit. (Malleable Theory).
• Teach a malleable theory?
18
Increasing motivation
19. 19
Randomized Controlled Trial: Middle, High School, Community College.
(Paunesku, Romero et al, 2011; 2012)
Increasing motivation
20. • Before: Preparatory questions
• During: Explanations
• After: Applying concept
Currently applying these principles to learning a “Growth Mindset”
Next directions:
• Create video versions for MOOCs, Khan Academy, students
• Change feedback in Khan Academy exercises
20
Cognitive + Social Psychology
22. • Sophisticated uses
• “Theory”: Fishing vs.
Toolkit for Problem-solving
22
Content Exercise
Online search
23. • Rate the plausibility of each answer.
• Predict Accuracy.
• Rate similarity.
• Grade explanations.
23
1 2 3 4 5 6 7
“Best” vs. “Big” Data
24. Resources
• Selection of Cognitive Science research applicable to (Online) Education
• www.josephjaywilliams.com/education
• Wiki with resources on Online Education: sites.cognitivescience.co/learn
• Contact between researchers & practitioners: www.learningresearch.net
• Comprehensive Wiki & Newsletter on K-12 Ed-Tech
www.edsurge.com
• Policy Prescriptions for schools
National Center for Education and the Economy www.ncee.org
Surpassing Shanghai, by Mark Tucker
• Institute of Education Sciences “evidence-based education”
www.whatworks.ed.gov
• E-learning in Industry & Workforce, Online Corporate Training & Development
www.elearningguild.com
www.astd.org
From Trinidad.
Websites:
Participation QUESTION: Where does you knowledge about learning & education come from? Student’s strategies? Teacher’s approach? Instructional designer. Policy maker?
DO AN EXPERIMENT! GENERAL….
Bucket model of the mind
“Instructionism”
Integrating into webpage.
EXAMPLES: Khan Academy. MOOC on power search.
Learning to program, about product specifications, sales strategies, accounting procedures, management skills.
Normally, we absorb content.
But actively processing should help, you have probably had the experience of understanding something better after explaining it to someone else, why a solution is correct – trying to teach is the best way to learn.
Extensive evidence for this.
Why? Why does explaining why help learning?
How do you know when to ask learners for explanations?
A question I’ve explored.
Two views. General boost –pay attention & spend more time, more motivated.
Or Selective. Drive people to discover principles.
Many studies in real-world, hard to understand, so we did the first lab study with artificial materials.
Image of robots.
Explaining behavior.
Learning about statistics concepts like variability.
In lab or online, adult participants learn material.
Asked a “why?” question. E.g. Explain category membership. Explain why someone behaved in a certain way. Explain the right answer.
Control: Matched it for time. Choose any strategy, describe, write out or say aloud their thoughts.
Measure learning.
Why is that an X?
Why does Y do…?
Why is Z the right solution?
TL: Add figure showing data for consistent items so that you have two matched figures, one with anomaly (showing this effect) and one for consistent, which I presume will be flat
Findings: Discovery of principles.
But depends on their knowledge.
Can *hurt* learning.
Choose cases carefully.
HAVE EFFECT IN WORKPLACE TOO.
HEALTH BEHAVIORS.
HAVE EFFECT IN WORKPLACE TOO.
HEALTH BEHAVIORS.
All agree it’s extremely important to teach learning strategies, ability to learn. But how successful? Transfer is very rare.
Assessments are hard.
Excellent precedent to go for ambitious measure like grades – it’s cutting edge, game-changing research, on a whole new level.
Practical and policy context – teachers will adapt educational technologies that help them meet their goals.
Workplace environments, have to look for similar measures – e.g. Do people put in more hours, more ready to ask for advice, to report errors?
Improve learning strategies – does sales performance increase, less training required, fewer hours debugging?
*Go for MOOCs, because grades & all the data already measured. Can we insert short training videos?
They can get practice using it on specific content.
Train people in educational habit of explaining. Extending the work I mentioned to incorporate insights from reciprocal teaching. Explain *to another person*.
Use learning principles to guide design of platforms, features & assessments