Chapter 33

A Cognitive Systems Engineering Approach To The Design of Decision Support Systems

Philip J. Smith
The Ohio State University

Norman D. Geddes
Applied Systems Intelligence, Incorporated

 

Outline

Human Performance on Decision-Making Tasks

Errors and Cognitive Biases

Slips

Mistakes

Cognitive Biases

Designer Error

Systems Approaches to Error

Errors and Cognitive Biases — Implications for Design

Human Expertise

Descriptive Models

Normative Optimal Models

Human Expertise — Implications for Design

Additional Cognitive Engineering Considerations

The Human Operator as Monitor

Complacency

Excessive Mental Workload

Lack of Awareness or Understanding

Lack of Trust and User Acceptance

Active Biasing of the User’s Cognitive Processes

Distributed Work and Alternative Roles

Organizational Failures

Case Studies

Case Study A. Interactive Critiquing as a Form of Decision Support

The Application Area

The Design Solution

Evaluation of the Implemented Solution

Case Study A — Conclusions

Case Study Study B. Software Associates: Decisions in Real-Time, High-Stakes Settings

The Application Area

The Need for Decision Aiding

The Design Solution

Results

Case Study C. Decisions and Problem Representations

The Application Area

The Design Solution

Results

Case Study D. Reducing the Burden of Training

The Application Area

The Design Solution

Results

Conclusions

References

 

Figures

Figure 33.1: (a) Full test panel with intermediate results marked using color-coded markers provided by AIDA (shown here in black and white). For the panel shown above, at this point in the analysis it looks like anti-Fyb is present if the reaction is being caused by a single antibody (Fyb is on all the reacting cells—the 2+ and 3+ cells—and only on those cells). However, it looks like anti-E and anti-K are present if the reaction is being caused by 2 antibodies. AIDA = Antibody Identification Assistant. (b) Sample ABO and Rh panel with feedback provided to the user regarding a possible slip or mistake.