Signal Detection Theory

...and now for another Human Factors Minute!

Signal Detection Theory is a model for understanding decision accuracy in ambiguous circumstances developed in the 1950s as a response to WWII radar research and is used in domains ranging from psychology, medical diagnostics, and statistical decisions.

Detection theory explains an observer's ability to determine the presence or absence of a stimulus, based on how well they can differentiate the target from random input, referred to as noise. What qualifies as a signal and noise is context dependent.

Noise is anything that interferes with detection. For humans, noise is everywhere. Personal experience, expectations, physiological state among other factors affect the decision maker’s threshold to filter through internal and external sources of noise. In a life or death scenario someone may be more likely to detect faint stimuli than they would in a moment of safety, but they might also be more likely to over identify harmless stimuli as a threat.

A well adapted system demonstrates a strong ability to discern such information. A subject’s identification of the stimuli’s status falls into one of four categories:

-True positive

-False negative

-False alarm

-Correct rejection

In a real world scenario, a radar operator would be responsible for deciding if the radar screen is displaying electronic signals that represent a plane or random inputs.

Detection theory can be observed in memory recognition experiments or eyewitness identification. Witnesses tend to base their decisions on a perceived level of familiarity with the suspect. In this example, the signal is the familiar feeling generated by the known input, and the noise corresponds to a false feeling of familiarity from the new stimulus.

This has been another Human Factors Minute!