Description
Measures the smallest perceivable change in a stimulus (size, colour, delay, price, sound) and the Weber fraction that predicts how it scales.
Weber observed that the just noticeable difference is a constant proportion of the stimulus magnitude (delta I / I = k). Psychophysics offers rigorous procedures to measure difference thresholds: the method of constant stimuli, method of limits and adaptive staircases such as the 2-down-1-up rule that converges on about 70.7% correct. Forced-choice designs (2AFC) control for response bias. UX makes constant decisions about perceptible differences, such as latency improvements, animation durations, colour states, haptic intensities and price changes, without measuring thresholds. JND testing answers questions like 'will users notice this 80 ms speedup?' with defensible data.
Attributes
- Discipline
- Cognitive Psychology & Perception
- Loss type
- Unmigrated
- Stage
- Design
- Data
- Quantitative
- Effort
- Days
Steps
- Define the stimulus dimension, the reference level and the task (e.g. 'which of two versions felt faster?').
- Choose a procedure: two-alternative forced choice with an adaptive staircase (e.g. 2-down-1-up) for efficiency.
- Randomise presentation order and interval, and include catch trials.
- Run each participant until the staircase has 8-12 reversals; estimate the threshold from the mean of the last reversals.
- Fit a psychometric function across participants and compute the Weber fraction at several reference levels.
- Translate thresholds into design specs (e.g. improvements below the JND are not worth shipping as perceptible features).
Sources
- Gescheider, Psychophysics: The Fundamentals (3rd ed., 1997)
- Levitt, Transformed up-down methods in psychoacoustics (1971), JASA
- Fechner, Elemente der Psychophysik (1860)
Also known as
- JND
- Weber's law
- Difference threshold
- Psychophysical staircase
Categories
- User Research
- Cognitive Psychology & Perception