Signal detection theory

Signal detection theory models how observers distinguish signals from noise under uncertainty. It separates sensitivity to evidence from the decision threshold used to classify it.

Sensitivity index d′: A measure of separation between signal-present and signal-absent evidence distributions, commonly expressed in standard-deviation units. It quantifies discriminability independently of the observer’s response criterion.
Yes–no detection task: An experiment in which an observer reports whether a signal was present on each trial. This task directly yields hits, misses, false alarms, and correct rejections.
Statistical decision theory: The study of choosing actions under uncertainty using evidence, probabilities, and consequences. Signal detection theory applies decision-theoretic reasoning to perceptual classification.
Threshold theory: A class of detection models in which stimuli below a fixed sensory threshold produce no detectable response. Unlike signal detection theory, it treats detection as crossing a sensory boundary rather than deciding amid overlapping evidence.
Radar: A system that detects and locates objects by transmitting radio waves and analyzing returning signals. Wartime radar operators faced the practical problem of separating weak echoes from noise.
Receiver operating characteristic: A curve plotting true-positive rate against false-positive rate across decision thresholds. Varying the criterion traces the trade-off predicted by signal detection theory.
Two-alternative forced-choice task: A task requiring selection of which of two intervals or locations contained the stronger or relevant stimulus. It estimates discriminability while reducing the influence of a yes–no response criterion.
Statistical hypothesis testing: A procedure for evaluating data against competing statistical hypotheses using a decision rule. Detection tasks resemble tests of signal absence versus signal presence.
Classical psychophysics: The study of relationships between physical stimuli and sensory experience, including measurement of detection and discrimination. Traditional threshold methods often summarize performance without separating sensitivity from response bias.
David M. Green: An American psychologist known for developing and applying signal detection methods in psychophysics. His work helped establish modern experimental and analytical practice in the field.

Linked from 25 pages