The final resting place for all this research… Ron Laughery, Ph. D

The final resting place for all this research…

• Ron Laughery, Ph.D.

• University of Colorado

Items to be covered…

• What is the problem this research is trying to solve from an operational perspective?

• What is the basic human performance modeling and simulation approach that this research will feed?

• What is the specific tool and architecture that we are working to advance?

• What are the issues in moving this research into practice?

What is the problem this research is trying to solve from an operational perspective?

• The £5,000,000,000 question…

• In about 1995, Robin Miller, an operational analyst with the MoD asked us this question and made this statement at a meeting:
• “A question I get all the time can be summed up as this – should we invest £5B in new kit, or should we instead invest that £5B in training? If your models can’t help me answer that question, you’re not doing your job.”

• We are trying to ensure that we are doing our job in Mr. Miller’s eyes

What is the basic human performance modeling and simulation approach that this research will feed?

• In military and civilian systems, decisions are increasingly being made on the basis of model based analyses

• System effectiveness depends upon…

Two basic approaches to modeling human/system performance

• Reductionist

• Breaking human activity and interaction with the system into discrete activities

Advantages/disadvantages of reductionist modeling approach

• Advantages

• Intuitive
• Level of detail determined by need
• Basic data are usually available or easily obtained
• Consistent with many military systems and operational analysis models

• Disadvantages

• Often requires extensive subject-matter expert input

Second approach to modeling human/system performance

• First principled/cognitive models

• Based on theories of the underlying mechanisms that facilitate human behavior

Advantages/disadvantages of the first principle approach

• Advantages

• Requires less data input from either experiments or subject matter experts
• More first-principle based and, if component models are valid, easier to defend

• Disadvantages

• Model construction can be quite cumbersome for simple tasks
• We don’t have enough real first-principle models of human performance

A strategy that has worked- a hybrid approach

• The flexibility of reductionist models combined with the power of first principles of human behavior is the formula for success

Reductionist modeling with Task Network Modeling

• Largely involves the extension of a task analysis into a network defining sequencing

Going from a task network to a running computer model

• Add timing information and task/system interdependencies

Add human decision making strategies

• Any defined branch point represents a need for a decision

• Logic and rule sets, goal seeking, naturalistic

Then, develop a scenario, equipment model and/or links to other simulations

Run the model to collect human/system performance data

Combining First Principles of human behavior with Task Network Models

• For the past 16 years, we have been embedding and linking first principle models of human performance into our tools including

• Cognitive workload and human response
• Micro models of human time and accuracy
• Human error and system response to error
• Performance shaping factor effects
• Linkage to anthropometric, biomechanical models
• Goal driven task scheduling
• Naturalistic Decision Making
• Situation awareness modeling
• Integration of cognitive engineering models such as ACT/R

• Predicting training effects is still the weakest link!

IMPRINT Architecture - Operations Modeling

Who Has IMPRINT?

• Army

• Navy

• Air Force

• Other Government

• Contractors

• University

Current IMPRINT Implementation: Stressors by Task Type

Approach to modeling human response to stressors

Use task network models to study aggregate effects of PSFs

What view of training is in IMPRINT now?...

What we really need for a reasonably accurate representation of training…

• We need these functional relationships…

• For different task types (the taxonomy)
• For different “types” of training

Big questions…

• Purpose of models

• Design of optimal training systems
• Design of systems considering training

• Taxonomies

• Training environment
• Task type

• Scope/complexity of tasks studied

• Do small tasks scale to large tasks?

• How do we treat Retention