Description
Model a system's commands and observations, generate the minimal mental model that lets an operator control it without surprises, check real mental models against it and derive safe training units.
The system and the user's mental model are both described as labelled transition systems with commands (user-initiated) and observations (system-initiated). A mental model controls the system if, after any shared sequence, the commands it expects are available and every possible observation is expected; full control adds knowledge of all available commands. A minimisation algorithm generates the smallest full-control model, which defines what training and documentation must teach; if none exists, the design cannot be controlled without surprise. Checking a user's or manual's model against the system exposes automation surprises and mode confusion. The full-control model can also be decomposed into basic units, each safe on its own, giving a learning sequence and a learning-complexity metric.
Attributes
- Discipline
- Human Factors & Error
- Loss type
- Siloed
- Stage
- Design, Evaluation
- Data
- Analytical / expert
- Effort
- Weeks
Steps
- Model the system interface as a transition system distinguishing commands from observations.
- Generate a minimal full-control mental model by minimisation; if none exists, flag the design as uncontrollable without surprise.
- Model existing mental models, for example from manuals or training material, and check them for control and full control; failures are potential automation surprises.
- Decompose the full-control model into basic sub-models such as sequences and loops.
- Complete each unit with all observations the system can produce, so each unit is safe to use alone.
- Order units for incremental training and report learning complexity as the minimal number of units.
Sources
- The Handbook of Formal Methods in HCI (Weyers et al., 2017)
Also known as
- Full-control mental models
- Automation surprise detection with HMI-LTS
- Mode-preserving user model generation
- Learning complexity
- Learning-unit decomposition
Categories
- Accessibility
- Human Factors & Error