SEEV Model of Visual Attention Allocation

Description

Predict the share of attention each display area receives from its salience, access effort, expectancy and value, and redesign layouts so valuable sources are salient and easy to reach.

SEEV models where people look during supervisory monitoring. Salience is a source's physical conspicuousness. Expectancy is the expected rate of relevant change at a location, its bandwidth. Value is the importance of the information or the cost of missing it. Effort is the cost of moving eyes, head or body (or fingers, for menus) to reach it, a negative factor. Combined, the probability of attending to an area is written P(A) = S - EF + EX x V. Well-trained operators scan almost entirely by expectancy times value, an optimal pattern reflecting a calibrated mental model, while fatigue and poor mental models let effort dominate. The design rule collapses the four factors to two: make valuable sources salient, and put valuable, frequently changing sources where they cost little effort. For discrete alerts, N-SEEV adds noticing factors such as onset salience and eccentricity, with steep losses beyond about 20 degrees.

Attributes

Loss type
Siloed
Effort
Days

Steps

  1. List the areas of interest the user must monitor (instruments, panels, screen regions).
  2. Rate each area's salience, expectancy (rate of relevant change), value (importance to the task) and effort (visual angle or action cost to reach it).
  3. Combine the ratings as P(A) = S - EF + EX x V to predict the relative attention share per area.
  4. Compare the prediction with the optimal expectancy-times-value distribution, or with eye-tracking dwell data.
  5. Redesign so valuable sources are salient and valuable, high-bandwidth sources are low-effort to reach.
  6. For alerts, place critical alerts within the primary field of view (about 30 degrees diameter in aviation) and use onsets.

Sources

  • Handbook of Human Factors and Ergonomics (Salvendy & Karwowski, 2021)
  • Attention and Effort (Kahneman, 1973)

Also known as

  • SEEV
  • Salience-Effort-Expectancy-Value model
  • N-SEEV
  • Noticing-SEEV

Categories

  • User Research
  • Task Analysis & Cognitive Modelling