Skill and Expertise

Skill and Expertise

Skill Acquisition

Skill Acquisition

Signature Phenomena

Problem Solving and Skill Acquisition

Overview of Various Classes of Models

Episodic Memory Based

Rule Based

Schema Based

Signature Phenomena

Power Law of Practice

Automaticity

Details of Skill Acquisition Process

Deliberate Practice

10 Years to World Class Expertise

Transfer of Training

Common Elements Description of Transfer

Use Specificity of Knowledge

Power Law

On Log-Log Scale

Stages of Skill Acquisition (Fitts, 1954, VanLehn, 1996)

Cognitive Stage (Early Phase)

Understand the domain knowledge without yet
trying to apply it.

Phase is dominated by reading and discussion.

Associative Stage (Intermediate Phase)

Attempted Use of Declarative Knowledge

Problem Solving Guided By Examples

Elimination of Errors and Misunderstandings

Ends With Initial Mastery of Skill

Autonomous Stage (Late Phase)

Continue to improve in speed
and accuracy as they practice,

End point: Automaticity

Problem Solving and Skill Acquisition

Knowledge and Skills Required for Problem Understanding

Episodic Memory

Plans or Schema

Search Control Knowledge!!!

Situation-Action Knowledge

Episodic Memory

Rules

Schema

Plans or Schema

Overview of Various Classes of Models

Models That Just Use Episodic Memory

Logan, Ross

Central role of episodic memory in skill acquisition

Models at Assume Special Representation For Procedural Knowledge (Rules)

Anderson, Newell

Need to store knowledge in form that can be
accessed very accurately and rapidly

If-Then Rules

Schema-Based Models

Norman, Reason

Highly integrated representations of complex skills

Like Schema or Scripts

Lifting the Limits (Erickson)

Acquisition of Cognitive Skills

Reading

Writing

Mathematics

Programming

Design

Limits in Human Information Processing

Attention

Memory

Problem Solving

Changes in Performance with Instruction and Practice

Automaticity

Skilled Memory

Cognitive Skills

Lifting the Limits

Development of Expertise

Acquisition of Cognitive Skills

Relationships to the Acquisition of Athletic Skills

Two Kinds of Knowledge

Declarative

Procedural

Rules and Production Systems

Relationships between problem solving and skill acquisition

Two Kinds of Representations

Declarative Knowledge

Images, Linear Orders, Propositions, Schemas

Knowledge of What

Facts, recipes, etc......

Procedural Knowledge

Productions

Knowledge of How

Riding a bicycle

Productions

IF the goal is to drive a standard transmission car

and the car is in first gear

and the car is going more than 10 miles per hour

THEN shift the car into second gear

Automaticity

Description of

IF (Goal and a specific situation)

THEN (do actions)

IF condition THEN action

Condition- Action Pair

Recognize-Act Cycle

Test Working Memory

Fire Rule Whose Condition is Satisfied

Sequences of Rules

IF the goal is to drive a standard transmission car

and the car in neutral

and the car is standing still

and the road is clear

THEN shift the car into first gear

and accelerate

IF the goal is to drive a standard transmission car

and the car is in first gear

and the car is going more than 10 miles per hour

THEN shift the car into second gear

and accelerate

IF the goal is to drive a standard transmission car

and the car is in first second

and the car is going more than 25 miles per hour

THEN shift the car into third gear

and accelerate

LEARNING

IF have rule that fires, THEN do action

OTHERWISE problem solving

Successful Problem Solving Episodes are Saved in the Form of Rules

CONDITION

Goal of Problem Solving Episode

and Current Situation

ACTION

Action that Successfully Solved Problem

The Declarative-Procedural Distinction

Declarative Knowledge:

Flexible use of knowledge

Not committed to a particular use

Easily acquired and forgotten

Procedural Knowledge:

Efficient use of knowledge

Optimized for specific use

Acquired by doing, role of practice

Production rules tend to carve up a task at its natural joints

One rule for each natural unit

Rules are the units in which the skill is acquired

Goal Structuring

Production rule conditions not only make reference to certain external situations but also specify certain goal conditions.

Experimental Evidence for Two Long-Term Memories

1) Reportability: declarative knowledge

2) Associative Priming: declarative knowledge

3) Retrieval Asymmetry: procedural knowledge

4) Acquisition:

Declarative knowledge comes from direct encoding of the environment

Procedural knowledge is compiled from declarative knowledge through practice.

5) The retention functions for the two types of memories are independent.

6) There have been a number of recent demonstrations of dissociations of declarative and procedural memory in amnesiacs and other populations.

Implicit memory

Skill acquisition in HM

Factors Effecting Practice

Spacing!

Part vs Whole Learning

Independence

Necessary Subskills

Knowledge of Results

More is not necessarily better

Human tutors vs the class room

Intelligent Computer Aided Instruction

TESTS OF ASSUMPTION THAT PROCEDURAL KNOWLEDGE
IS
REPRESENTED AS RULES

Speed Up From First to Second Correct Action

Look at A Given Step In A Problem

Measure Time to Complete Step

Assume Store Rule After First Successful Attempt

Second Attempt is Much Faster

Time to Learn A New Procedure Function of Number of Rules

Transfer a Function of Number of Shared Rules

METHOD

Polson, Muncher, and Engelbeck (1986)

TASKS

- Word Processor Utility Tasks

* change default and document format parameters

* check spelling

* duplicate diskette

- Perform Task by:

* make series of menu selections

* fill in parameter on menu for task

* exit menu to perform task

DESIGN

* 3 pairs of similar task

* vary training orders of pairs

PROCEDURE

- Learn Each Task to a Strict Criterion

- Very Constrained Training Procedure

- Anticipation Method

MODEL FITS

INDIVIDUAL DATA POINTS (n = 1079)

TRAINING Time =
Learning Time +{20 sec per New Rule}
Execution Time +{3.1 sec per Rule}
Individual Differences in Learning and Performance

Accounts for 85.2% Variance In Training Times

Page 1