Erik Hollnagel

Ph.D., Professor, Professor Emeritus

Goals-Means Task Analysis (GMTA)

The Goals-Means Task Analysis was a proposal for a systematic approach to analyse activities (functions) based on the goals-means principle. A detailed description of the GMTA was presented in the book Human Reliability Analysis, published in 1993. The method had been developed from the late 1980s, partly in conjunction with some international research projects sponsored by the European Space Association (ESA).

Background

The GMTA was based on cognitive systems engineering (Hollnagel & Woods, 1983) and cognitive task analysis (Woods & Hollnagel, 1987). It also referred to the underlying principles of multilevel flow modelling (MFM, cf. Lind, 1990). The idea of goal-means decomposition is very old, and can be found in Aristotle's Nichomachean Ethics. In psychology, the obvious reference was the General Problem Solver developed by H. A. Simon, J. C. Shaw and A. Newell in 1959 (see Newell & Simon, 1972), and the Structured Analysis and Design Technique (SADT) developed in the early 1970s (Marca & McGowan, 1987). The GMTA has influenced the development of both CREAM and the FRAM.

Principles

The Goals-Means Task Analysis Method (GMTA) uses the following basic concepts: goal, task or just task step (corresponding to the means), precondition, execution condition, postconditions, and timing conditions.

A goal describes the state that is to be achieved by the acting agent. Goals are either defined as independent or derived goals. A derived goal is the result of the decomposition of another (parent) goal. The distinction between independent and derived goals is not absolute but pragmatic, and depends on the chosen level of description. Just as some task steps are considered elementary in a practical sense, so some goals will be considered as independent goals or top goals. Unless this is done there will be infinitely many levels of recursion.
A goal is achieved by a set of task steps. This set need have only a single member. A task step can be composed of other task steps in which case it serves as a label or reference to a set of task steps (a routine, a group, a procedure). Each set of task steps may be concatenated as a larger task step; conversely, each larger task step can be decomposed into the constituent task steps. Task steps describe activities that can be carried out if the pre-conditions are true. In this way a task step which has a pre-condition will create a sub-goal which expresses the pre-condition.
A precondition describes the conditions that must be satisfied for a task step to be carried out. When a precondition is found, it is considered as a sub-goal (a derived goal). This feature makes the GMTA recursive. A precondition can be followed by one or more task steps which, when carried out, will achieve the precondition.
An execution condition describes the conditions that must be satisfied while a task step is carried out. An execution conditions is thus always attached to a task step. In cases where the execution condition is implied by the description of the task step, it need not be defined explicitly. Most execution conditions will therefore refer to temporal conditions and relations to other task steps.
A postcondition describes the side-effects which are deemed to be important because they are known or assumed to have consequences for e.g. other preconditions - either in a positive (synergistic) or negative (antagonistic) way.
The timing or sequence conditions of a task step describe constraints on the sequencing (starting and stopping) of a task step.

References

Hollnagel, E. & Woods, D. D. (1983). Cognitive systems engineering: New wine in new bottles. International Journal of Man-Machine Studies, 18, 583-600.

Lind, M. (1990). Representing goals and functions of complex systems. An introduction to multilevel flow modelling. Technical University of Denmark, Institute of Automatic Control Systems.

Marca, D. & McGowan, C. (1987). Structured Analysis and Design Technique. McGraw-Hill.

Newell, A. & Simon, H. A. (1972). Human problem solving Englewood Cliffs, NJ: Prentice-Hall.

Woods, D. D. & Hollnagel, E. (1987). Mapping cognitive demands in complex problem-solving worlds. International Journal of Man-Machine Studies, 26, 257-275.