trajectories (i.e., actions), thereby generalizing the lessons learned from the previous two examples. Gibson and Crooks' key insight was that the constraints on behavior can be identified by developing a functional description of the work domain. Rather than adopting a context-free language to describe objects in the work domain, Gibson and Crooks used an action-relevant (i.e., functional) language to describe those objects. For automobile driving, terms such as obstacle, collision, path, and destination play a key role.
When these functional possibilities are integrated, the result is a field of safe travel. Several examples are given in Figs. 7.3 to 7.5. This field description of the work domain represents the possible paths that the car may safely follow. Note that the field of safe travel is a description of the work domain, not of the task. The object of description is the environment, not the actions taken on that environment. The field merely describes what is possible from a functional point of view. It is still up to the driver to choose a particular trajectory on any one occasion. In short, the field describes constraints on action, not action itself. As a result, field descriptions inherit the advantages of the constraint-based approach described in chapters 3 and 5.
Table 7.1 provides a summary of the relationship between the examples discussed in this section. We hope we have convinced you of the important and unique value of work domain analysis. Of the several examples that we reviewed, J. J. Gibson and Crooks' ( 1938) field of safe travel was the closest to our needs, but it was developed for the particular case of automobile driving. What we require is a modeling tool that has the same general characteristics (and thus, advantages), but that can be applied to