Signal detection with criterion noise: applications to recognition memory.

Area of Science:

• Cognitive Psychology
• Decision Science
• Psychophysics

Background:

• Signal detection theory (SDT) traditionally assumes criterion placement is a noise-free process.
• This assumption is fundamental to understanding decision-making in discrimination tasks.
• Existing models do not fully account for variability in decision criteria.

Purpose of the Study:

• To challenge the noise-free assumption in criterion placement within signal detection theory.
• To introduce and formalize the noisy decision theory of signal detection (ND-TSD).
• To empirically investigate the impact of criterion noise on recognition memory.

Main Methods:

• Derivation of generalized equations for the isosensitivity function and discrimination measures under criterion variability.
• Examination of ND-TSD's relationship with existing decision-making models.
• An experiment using recognition memory for word ensembles to assess criterion noise magnitude.

Main Results:

• Criterion noise is substantial and significantly contributes to the variance in the isosensitivity function's slope.
• Empirical evidence supports the notion that criterion placement is not a noise-free process.
• The magnitude of criterion noise was found to be non-trivial in recognition memory tasks.

Conclusions:

• ND-TSD offers a novel framework for understanding decision-making, incorporating criterion variability.
• The theory can explain inconsistencies in recognition memory research, such as learning effects and slope invariance.
• ND-TSD provides a mechanism to reconcile deterministic and probabilistic models of decision making.