How Smart Virtual Domains can Accelerate Learning

1
How Smart Virtual Domains can Accelerate Learning

• Wayne Zachary, Ph.D.
• CHI Systems, Inc.
• 1035 Virginia Drive, Suite 300
• Ft. Washington, PA 19034 USA
• wzachary_at_chisystems.com

2
Research

• To steal one idea from one person is plagiarism.
  To steal many ideas from many people is
  research.
• -attributed to Frank Westheimer
• Who am I stealing from?
• in no particular order and definitely an
  incomplete list
• Lev Vygostky
• Saul Amarel
• B.C. Bloom
• Paul Feltovich
• Robert Hoffman
• Kurt van Lehn
• Allen Collins
• Gary Klein
• Gregory Bateson
• Susan Chipman
• Susan Frankel

3
Introduction

• Accelerating What? Learning What?
• Not facts, relationships, tasks, heuristics
• Rather, work/role expertise within a domain
• Defining the space
• What is work/role expertise?
• Are there commonalities across domains?
• How is it acquired ontogenetically?
• How is that acquisition traditionally
  facilitated?
• Accelerating the process

4
Expertise

• Expert has the ability to consistently generate
  high quality judgments in the full range of
  naturalistic domain situations, from common to
  rare
• Expertise is
• domain- dependent
• knowledge-based
• takes people years to develop expertise

5
Expertise Across Domains
All happy families resemble one another, each
unhappy family is unhappy in its own way
Tolstoy (Anna Karenina) Is there a common
meta-structure to expertise, across domains?
6
Expertise Is Developed Internally in Across
Domains

• Cognitive structure
• Development of interrelated
• Problem/domain representations (increasingly
  abstract), and
• Reasoning strategies that rely on those
  representations
• Not All domains afford the opportunity for
  expertise formation

7
Expertise is Developed In Social Context

• Expertise is
• cognitive but its development is not (just)
• arguably H.S.s dominant adaptation
• involves holistic integration of range of human
  adaptive capabilities which are inexorably
  socially-based
• Expertise development
• socially situated
• acquired through interaction with others
• multi-dimensionally experiential
• involves demonstration, trial, practice, play,
  action
• locally accretive
• new knowledge is internalized accretively to
  existing knowledge without deconstructing it
• builds systematically through space of
  representations and reasoning strategies
• internalization is itself a process not an
  event
• is a series of steps not a smooth, continuous
  curve

8
Facilitating Expertise Acquisition

• What works...
• One-on-one tutoring
• Discussion and discourse
• Exploration through direct interaction
• Practice and rehearsal
• Scaffolding and support
• These are all socially situated activities, but
  they only set the stage for the process. What
  facilitates the process?
• recognizing and seizing the teachable moment
• What makes one moment teachable and not another?
• the learners Zone of Proximal Development (ZPD)
  -- the distance between the Learners actual
  development and the potential development
• teachers, in the broad sense, have to wait for
  the teachable moment to occur and often miss it.

9
How to Accelerate Expertise Acquisition?

• Create teachable moments and capitalize on them
  with smart virtual domains
• shorten each step by focusing the expertise
  acquisition process
• How? By realizing that Builds systematically is
  not just another education-ese metaphor
• Amarels work showed that the ontogeny of
  expertise acquisition can be mapped out in
  computational terms
• Such a Knowledge Ontogeny Map can
• drive the recognition of ZPDs and teachable
  moments through observation of learner behavior
• provide a roadmap for the virtual environment, so
  that it can steer the learner toward teachable
  moments
• provide context and content for the synthetic
  characters of various types and the social basis
  for learning that the provide
• The smart virtual domain grows with and adapts to
  the learner, providing accelerated learning
  longitudinally

10
Accelerated Learning Vision

• A persistent virtual environment focused on
  helping an individual learn to become expert in
  specific work/role
• The virtual domain adapts to the individuals
  evolving expertise through us of an ongoing
  knowledge map and smart algorithms which track
  the learners local ZPD through the many paths by
  which expertise can evolve in that domain
• A broad cast of synthetic characters create and
  exploit learnable/teachable moments in concert
  with the virtual domain itself to accelerate the
  learning pace
• These moments are exploited using a range of
  proven teaching approaches implemented within the
  virtual domain and applied opportunistically to
  fit the local ZPD.

11
Apply Multiple Strategies to Compress Each Step
competent for evolution to next level
able to play/practice/rehearse
skill
able to begin interaction
time to competence
12
Can You Say That in a Picture?
Smart Virtual Environment
ZPD recognition and instantiation
learning associate
teammates
Coach/critic
supervisors
range of synthetic characters
range of social contexts
The virtual domain and synthetic characters
within it are driven by the knowledge ontology
map and the ZPD recognition process to offer
different social contexts for learning and to
utilize each one to capitalize on teachable
moments in the leaners local ZPD.
13
Technology Maturation Makes This Vision Possible

• Synthetic characters
• rational purposive behavior
• emotions and personality
• speech and pragmatic language use
• variable levels of expertise
• Virtual environments and domains
• adaptive objects and interaction rules
• instructional/control agents
• flexible/reusable art, physics, etc.
• Knowledge Representation
• meta-level markup languages
• pattern-based reasoning
• Computation
• fast, distributed, graphical, multimodal

14
So What? Slide

• Theory
• integrates socially-situated and
  knowledge-centric approaches
• Suggests ways to address learning at all phases
  of expertise acquisition
• Accelerates learning by controlling the
  environment in which learning occurs
• Research recommendations
• Develop and demonstrate knowledge ontogeny
  language,ZPD recognition algorithm
• Design, conduct experiments to validate
  accelerated learning with ZPD identification
• Proof of concept implementation in flexible
  virtual domain

15
Thank you!

• Questions?