Developing Learning by Teaching Environments that support Self-Regulated Learning Gautam Biswas, Krittaya Leelawong, Kadira Belynne, Karun Viswanath, Daniel.

Developing Learning by Teaching Environments that

support Self-Regulated Learning

Gautam Biswas, Krittaya Leelawong, Kadira Belynne, Karun Viswanath,

Daniel Schwartz, and Joan Davis

Dept. of EECS & ISIS, Box 1824 Sta B, Vanderbilt UniversitySchool of Education, Stanford University

International Conference on Intelligent Tutoring Systems 20042004

OutlineOutline

► IntroductionIntroduction► Learning by Teaching: previous workLearning by Teaching: previous work► A new approach: Betty’s BrainA new approach: Betty’s Brain► A computational architecture for Betty’s A computational architecture for Betty’s

BrainBrain► ExperimentsExperiments► ConclusionConclusion

IntroductionIntroduction

► Intelligent Tutoring SystemIntelligent Tutoring System Curriculum sequencingCurriculum sequencing Intelligent analysis of student’s solutionsIntelligent analysis of student’s solutions Interactive problemInteractive problem Do not set learning goals and to apply strategies

► The learning process must help studentsThe learning process must help students Constructivist learning Exploratory learning metacognitive strategies

IntroductionIntroduction

► Learning-by-teaching is an open-ended and self-directed activity Exploratory and constructivist learning Good learners bring structure to a domain by

asking the right questions Good teachers build on the learners’ knowledge

to organize information

► Learning by teaching paradigmLearning by teaching paradigm Students teach computer agentsStudents teach computer agents

Learning by Teaching: previous Learning by Teaching: previous workwork

► Limitation Limitation knowledge structures is difficult to uncover, analyze,

and learn outcome feedback is less effective than cognitive fee

dback►On the positive

increased motivation► suggestion

use metacognitive strategies to promote learning

A new approach: Betty’s BrainA new approach: Betty’s Brain

► Structure of the learner-as-teacherStructure of the learner-as-teacher► Embody 4 principles of designEmbody 4 principles of design

Teach through visual representations that organize the reasoning structures

well-known teaching interactions to organize student activity

The agents have independent performances that provide feedback on how well they have been taught

Low start-up costs of teaching the agent

A new approach: Betty’s BrainA new approach: Betty’s Brain

A new approach: Betty’s BrainA new approach: Betty’s Brain

► Four agentsFour agents the teachable agent, Betty the mentor agent, Mr. Davis the student agent the environment agent

► All agents interact through the Environment Agent

► Each agent only contacts to the Environment Agent

A new approach: Betty’s BrainA new approach: Betty’s Brain

► ScaffoldScaffold Well-organized online resourcesWell-organized online resources Structured quiz questionsStructured quiz questions

► FeedbackFeedback Provide hints on how to know and how to teachProvide hints on how to know and how to teach

A new approach: Betty’s BrainA new approach: Betty’s Brain

►When a student begins teachingWhen a student begins teaching Set goal on what to teachSet goal on what to teach Gain the relevant knowledgeGain the relevant knowledge

►MentorMentor Seek further informationSeek further information

► BettyBetty QuizzesQuizzes ReportReport

A computational architecture for A computational architecture for Betty’sBetty’s

►Each agent has a Monitor, Decision Maker, Memory, and an Executive

ExperimentsExperiments

► 3 equal groups of 15 students3 equal groups of 15 students Pretest with 12 questionPretest with 12 question Six 45-minutes sessions over a period of 3 weeksSix 45-minutes sessions over a period of 3 weeks Posttest with 12 question was the same as Posttest with 12 question was the same as

pretestpretest

► seven weeks laterseven weeks later A memory testA memory test A preparation for future learning transfer testA preparation for future learning transfer test

ExperimentsExperiments

ExperimentsExperiments

ExperimentsExperiments

ExperimentsExperiments

ConclusionConclusion

► SRL in understanding and transfer perform SRL in understanding and transfer perform goodgood