Human-System Modeling:
Some Principles and a Pragmatic Approach
Michael A. Freed & Michael G. Shafto
Cognition Research Group
Human-Automation Integration Research Branch
NASA-Ames Research Center
mfreed@mail.arc.nasa.gov
mshafto@mail.arc.nasa.gov
Abstract
There are a number of formalisms and architectures for modeling human performance, but
there is little guidance on how to go about building useful human models. This is a serious
problem since human modeling is difficult and full of pitfalls. The intended application of the
model should play a strong guiding role in model development. We are building an
engineering model intended to help interface designers predict usability problems -- in
particular, to alert designers to features of an interface that may increase the risk of certain
kinds of human error. In reflecting on our experience in building this model, we have
developed several inter-related principles that have been helpful in directing our investment of
time and effort. Taken together, these principles suggest a methodology for the development
of human performance models for complex human-machine systems.
www.siencdirect.com