Signal Detection Theory
Also known as: SDT, Detection Theory
Signal Detection Theory (SDT) is a framework for analyzing how observers detect signals embedded in noise, accounting for both sensory capacity and decision-making bias. Developed by Green and Swets in the 1960s, it provides a principled method for measuring sensitivity and response criteria separately, making it foundational in psychophysics, perception research, and diagnostic decision-making.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use Signal Detection Theory when analyzing perceptual detection, diagnostic accuracy, or any binary decision under uncertainty. It is essential when separating sensory sensitivity from decision bias, when designing experiments with signal-present and signal-absent trials, or when examining how external factors (stress, fatigue, incentives) affect detection without changing underlying perception.
Strengths & limitations
- Separates sensory sensitivity from response bias, providing clearer interpretation than hit rate alone
- Model-based framework applicable across diverse domains: vision, audition, medical diagnosis, security screening
- Provides meaningful metrics (d-prime, criterion) that remain consistent across different base rates or payoff structures
- Flexible: works with confidence ratings, multiple categories, and complex decision environments
- Assumes normally distributed noise and signal distributions; violations can distort d-prime estimates
- Requires sufficient trials for stable estimates of hit and false alarm rates; sparse data produce unreliable values
- Sensitive to extreme proportions (100% hits or 0% false alarms); corrections needed to avoid infinite z-scores
Frequently asked
What is d-prime (d'), and why does it matter more than hit rate?
d-prime is the standardized difference between the distributions of signal-present and signal-absent trials. It matters because two observers can have identical hit rates but vastly different sensitivities: one might detect signals well and have a conservative bias, while another has poor sensitivity but a liberal bias that inflates hits. d-prime separates true sensitivity from bias.
What does a criterion (c) of zero mean?
A criterion of zero means the observer is unbiased—they set their decision threshold at the point where signal-present and signal-absent likelihoods are equal. Positive c indicates conservative bias (require strong evidence to say 'yes'), negative c indicates liberal bias (say 'yes' easily).
How do I handle cases where hit rate is 100% or false alarm rate is 0%?
Use logit corrections or add a small continuity correction (e.g., (hits + 0.5) / (total signals + 1)) before calculating z-scores. This prevents infinite values and provides stable estimates when data are sparse or extreme.
Can I use SDT with confidence ratings instead of yes/no responses?
Yes. Confidence-rating data generate a Receiver Operating Characteristic (ROC) curve, which provides sensitivity estimates across multiple criteria. This yields richer information about the observer's internal decision processes.
Sources
- Green, D. M., & Swets, J. A. (1966). Signal detection theory and psychophysics. Wiley. link ↗
- Macmillan, N. A., & Creelman, C. D. (2005). Detection theory: A user's guide. Lawrence Erlbaum Associates. link ↗
- Swets, J. A., Dawes, R. M., & Monahan, J. (1996). Psychological science can improve diagnostic decisions. Psychological Science in the Public Interest, 11(1), 1-26. link ↗
How to cite this page
ScholarGate. (2026, June 3). Signal Detection Theory. ScholarGate. https://scholargate.app/en/psychology/signal-detection-theory