Cognitive Models of Interaction
Cognitive models describe and predict how people perceive, think, and act when using interactive systems, allowing analysts to estimate performance such as task time before any user is tested.
Definition
Cognitive models of interaction are formal or quantitative descriptions of the human perceptual, cognitive, and motor processes involved in using a system, used to explain interaction and to predict measures such as task completion time without empirical testing.
Scope
This topic covers analytical models of human performance in interaction: the model human processor, the GOMS family and the keystroke-level model for predicting expert task times, and laws of motor and perceptual performance such as Fitts's law and the Hick-Hyman law. It addresses how these models predict efficiency and inform design choices. It does not cover empirical metrics gathered from users, treated under usability metrics and measurement, nor the device-level details of pointing, treated under input and interaction techniques.
Core questions
- How can task completion time be predicted before building or testing an interface?
- What does Fitts's law say about the time to acquire a target?
- How does the GOMS family decompose tasks into goals, operators, methods, and selection rules?
- What are the strengths and limits of predictive cognitive models?
Key concepts
- model human processor
- GOMS
- keystroke-level model
- Fitts's law
- Hick-Hyman law
- predictive performance modeling
- expert task time
- operators and methods
Key theories
- GOMS and the keystroke-level model
- GOMS analyzes skilled task performance as goals, operators, methods, and selection rules; its simplest member, the keystroke-level model, predicts expert execution time by summing estimated durations of elementary actions such as keystrokes, pointing, and mental preparation.
- Fitts's law
- The time to move to and acquire a target increases with distance and decreases with target size in a logarithmic relationship, giving designers a quantitative basis for sizing and placing interactive controls.
- The model human processor
- Card, Moran, and Newell modeled the human as interacting perceptual, cognitive, and motor processors with characteristic capacities and cycle times, providing a foundation for engineering approximations of interaction performance.
Clinical relevance
Cognitive models let designers compare interface alternatives analytically, for example estimating which menu layout or control placement is faster, without running studies for every option; Fitts's law in particular guides the sizing and placement of targets in graphical and touch interfaces.
History
Card, Moran, and Newell's 1983 book brought cognitive psychology to bear on interface design, building on the keystroke-level model published in 1980 and on Fitts's 1954 law of aimed movement. The GOMS family was later compared and extended by John and Kieras, and predictive modeling remains a tool for analytic evaluation alongside empirical methods.
Key figures
- Stuart K. Card
- Thomas P. Moran
- Allen Newell
- Paul M. Fitts
- Bonnie E. John
- David E. Kieras
Related topics
Seminal works
- card1983
- card1980
- fitts1954
Frequently asked questions
- What is Fitts's law used for in interface design?
- Fitts's law predicts how long it takes to point at a target based on the target's distance and size. Designers use it to make frequently used controls larger or place them at screen edges and corners, which are effectively easy to hit, thereby reducing pointing time and effort.
- When are predictive cognitive models more useful than user testing?
- Predictive models like the keystroke-level model are most useful early on or when comparing many design variants, because they estimate expert performance times cheaply and quickly without recruiting users. They complement testing, which is still needed to capture learning, errors, and real-world behaviour that the models simplify away.