Signal Detection Theory

Description

Separates a person's ability to discriminate signal from noise (sensitivity, d') from their tendency to respond 'yes' (criterion), clarifying whether failures are perceptual or strategic.

Signal detection theory models decisions under uncertainty where a person must distinguish signals from noise, such as a real alert among false alarms, a phishing email among legitimate ones or an anomaly in a chart. Every response falls into one of four outcomes: hit, miss, false alarm or correct rejection. From hit and false alarm rates, SDT estimates sensitivity (d'), the ability to discriminate, independently of the response criterion, which reflects bias toward saying yes or no. This shows whether a design improves perception or merely shifts caution, and supports ROC analysis. In UX, it suits alert design, security warnings, content moderation and AI-assisted decision tasks. Rigour comes from designed trial sets with known ground truth, enough trials per condition (often 50 or more), and reporting both d' and criterion.

Attributes

Loss type
Latent
Effort
Weeks

Steps

  1. Define the detection task, the signal and the noise, with known ground truth for each trial.
  2. Build a trial set with a realistic or controlled base rate of signals, with 50 or more trials per condition.
  3. Run participants through trials, recording yes or no responses and optionally confidence ratings.
  4. Compute hit and false alarm rates per participant and condition, applying corrections for extreme rates.
  5. Calculate d' and criterion (c or beta), and ROC curves if confidence ratings were collected.
  6. Compare designs on sensitivity and bias, and consider the costs of misses versus false alarms.
  7. Translate results into alert thresholds, interface design or training.

Sources

  • Sensation and Perception
  • Lost Methods in UX
  • Van Drongelen (Signal Processing for Neuroscientists)
  • MacKay
  • Forward-Thinking Playbook (Gopinath)
  • Detection Theory: A User's Guide (Macmillan & Creelman)
  • The Inquiry Series, Ch 2 — Signal, Noise & the Detection of Intent
  • Lost Methods in UX — A Complete Treatise (Gopinath)
  • Green DM & Swets JA (1966) Signal Detection Theory and Psychophysics
  • Macmillan NA & Creelman CD (2005) Detection Theory: A User's Guide. 2nd ed
  • Wickens CD et al. (2015) Engineering Psychology and Human Performance

Also known as

  • SDT
  • d-prime analysis
  • Detection theory
  • Signal Detection Theory (SDT)

Categories

  • User Research
  • Cognitive Psychology & Perception