Distributional assumptions and observed conservatism in the theory of signal detectability

Laurence T. Maloney, Ewart A C Thomas

Research output: Contribution to journalArticle

Abstract

The theory of signal detectability typically fits data from Yes-No detection experiments by assuming a particular form for the noise and signal plus noise distributions of the Observer. Previous work suggests that estimates of the Observer's sensitivity are little affected by small discrepancies between the assumed distributions (usually Gaussian) and the Observer's true underlying distributions. Possibly for this reason, estimates of the Observer's choice of criterion or likelihood ratio suggesting suboptimal performance have also been taken at face value. It is, for example, commonly accepted that human Observers are conservative: They are said to choose criteria corresponding to likelihood ratios that are closer to 1 than the ratios produced by optimal criteria. We demonstrate that estimates of likelihood ratio can be markedly biased when the distributions assumed in estimation are not the Observer's true distributions. We derive necessary and sufficient conditions for an optimal Observer to appear conservative when fitted by distributions different from those governing his choices. These results raise a fundamental question: What information about the Observer's underlying noise and signal plus noise distributions does the Observer's performance in a Yes-No detection task provide? We demonstrate that a small number of isosensitivity (ROC) curves completely determines the Observer's underlying noise and signal plus noise distributions for many familiar forms of the theory of signal detectability. These results open up the possibility of a semiparametric theory of signal detectability.

Original languageEnglish (US)
Pages (from-to)443-470
Number of pages28
JournalJournal of Mathematical Psychology
Volume35
Issue number4
DOIs
StatePublished - 1991

Fingerprint

Detectability
Gaussian distribution
Politics
Noise
Observer
Experiments
Likelihood Ratio
Normal Distribution
ROC Curve
Estimate
Receiver Operating Characteristic Curve
Demonstrate
Biased
Discrepancy
Choose
Necessary Conditions

ASJC Scopus subject areas

  • Applied Mathematics
  • Experimental and Cognitive Psychology

Cite this

Distributional assumptions and observed conservatism in the theory of signal detectability. / Maloney, Laurence T.; Thomas, Ewart A C.

In: Journal of Mathematical Psychology, Vol. 35, No. 4, 1991, p. 443-470.

Research output: Contribution to journalArticle

@article{cb545e03b4ca4ef48e68c9f77ad06e9b,
title = "Distributional assumptions and observed conservatism in the theory of signal detectability",
abstract = "The theory of signal detectability typically fits data from Yes-No detection experiments by assuming a particular form for the noise and signal plus noise distributions of the Observer. Previous work suggests that estimates of the Observer's sensitivity are little affected by small discrepancies between the assumed distributions (usually Gaussian) and the Observer's true underlying distributions. Possibly for this reason, estimates of the Observer's choice of criterion or likelihood ratio suggesting suboptimal performance have also been taken at face value. It is, for example, commonly accepted that human Observers are conservative: They are said to choose criteria corresponding to likelihood ratios that are closer to 1 than the ratios produced by optimal criteria. We demonstrate that estimates of likelihood ratio can be markedly biased when the distributions assumed in estimation are not the Observer's true distributions. We derive necessary and sufficient conditions for an optimal Observer to appear conservative when fitted by distributions different from those governing his choices. These results raise a fundamental question: What information about the Observer's underlying noise and signal plus noise distributions does the Observer's performance in a Yes-No detection task provide? We demonstrate that a small number of isosensitivity (ROC) curves completely determines the Observer's underlying noise and signal plus noise distributions for many familiar forms of the theory of signal detectability. These results open up the possibility of a semiparametric theory of signal detectability.",
author = "Maloney, {Laurence T.} and Thomas, {Ewart A C}",
year = "1991",
doi = "10.1016/0022-2496(91)90043-S",
language = "English (US)",
volume = "35",
pages = "443--470",
journal = "Journal of Mathematical Psychology",
issn = "0022-2496",
publisher = "Academic Press Inc.",
number = "4",

}

TY - JOUR

T1 - Distributional assumptions and observed conservatism in the theory of signal detectability

AU - Maloney, Laurence T.

AU - Thomas, Ewart A C

PY - 1991

Y1 - 1991

N2 - The theory of signal detectability typically fits data from Yes-No detection experiments by assuming a particular form for the noise and signal plus noise distributions of the Observer. Previous work suggests that estimates of the Observer's sensitivity are little affected by small discrepancies between the assumed distributions (usually Gaussian) and the Observer's true underlying distributions. Possibly for this reason, estimates of the Observer's choice of criterion or likelihood ratio suggesting suboptimal performance have also been taken at face value. It is, for example, commonly accepted that human Observers are conservative: They are said to choose criteria corresponding to likelihood ratios that are closer to 1 than the ratios produced by optimal criteria. We demonstrate that estimates of likelihood ratio can be markedly biased when the distributions assumed in estimation are not the Observer's true distributions. We derive necessary and sufficient conditions for an optimal Observer to appear conservative when fitted by distributions different from those governing his choices. These results raise a fundamental question: What information about the Observer's underlying noise and signal plus noise distributions does the Observer's performance in a Yes-No detection task provide? We demonstrate that a small number of isosensitivity (ROC) curves completely determines the Observer's underlying noise and signal plus noise distributions for many familiar forms of the theory of signal detectability. These results open up the possibility of a semiparametric theory of signal detectability.

AB - The theory of signal detectability typically fits data from Yes-No detection experiments by assuming a particular form for the noise and signal plus noise distributions of the Observer. Previous work suggests that estimates of the Observer's sensitivity are little affected by small discrepancies between the assumed distributions (usually Gaussian) and the Observer's true underlying distributions. Possibly for this reason, estimates of the Observer's choice of criterion or likelihood ratio suggesting suboptimal performance have also been taken at face value. It is, for example, commonly accepted that human Observers are conservative: They are said to choose criteria corresponding to likelihood ratios that are closer to 1 than the ratios produced by optimal criteria. We demonstrate that estimates of likelihood ratio can be markedly biased when the distributions assumed in estimation are not the Observer's true distributions. We derive necessary and sufficient conditions for an optimal Observer to appear conservative when fitted by distributions different from those governing his choices. These results raise a fundamental question: What information about the Observer's underlying noise and signal plus noise distributions does the Observer's performance in a Yes-No detection task provide? We demonstrate that a small number of isosensitivity (ROC) curves completely determines the Observer's underlying noise and signal plus noise distributions for many familiar forms of the theory of signal detectability. These results open up the possibility of a semiparametric theory of signal detectability.

UR - http://www.scopus.com/inward/record.url?scp=38149144377&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=38149144377&partnerID=8YFLogxK

U2 - 10.1016/0022-2496(91)90043-S

DO - 10.1016/0022-2496(91)90043-S

M3 - Article

VL - 35

SP - 443

EP - 470

JO - Journal of Mathematical Psychology

JF - Journal of Mathematical Psychology

SN - 0022-2496

IS - 4

ER -