Outline of Lecture I Intro to Signal Detection

Outline of Lecture I. Intro to Signal Detection Theory (words) II. Intro to Signal Detection Theory (pictures) III. Applications of Signal Detection Theory

Part 1 Introduction to Signal Detection Theory S. D. T. In Words

Signal Detection Theory S. D. T. is a procedure for measuring sensitivity to stimulation, independent of the subject’s response bias.

Detection Experiment • We want to measure a subject’s ability to detect very weak stimuli. • Signal Detection Theory requires a “Type A” experiment. • How do we know when the subject is objectively incorrect?

“Catch Trials” The subject is asked to make a response when no stimulus has been presented (also called “noise only” trials).

Not All Errors Are Equal 1. Reporting stimulus is present when it’s absent (“false alarm”). Versus 2. Reporting stimulus is absent when it’s present (“miss”).

Correct Responses Differ, Too 1. Reporting stimulus is present when it’s present (“hit”). Versus 2. Reporting stimulus is absent when it’s absent (“correct rejection”).

Absent Present Stimulus-Response Matrix Correct Rejection False Alarm Miss Hit “No” “Yes” Response

Absent Correct Rejection False Alarm Type I error Present Stimulus-Response Matrix Miss Hit Type II error “No” “Yes” Response

Signal Detection Theory S. D. T. reduces the stimulus-response matrix to two meaningful quantities. 1. Detectability (d’) - a subject’s sensitivity to stimulation. 2. Criterion (b) - a subject’s inclination to favor a particular response; bias.

Part 2 Introduction to Signal Detection Theory S. D. T. In Pictures

Distributions of Sensory Responses

Distributions of Sensory Responses Spontaneous Activity is Constant

Distributions of Sensory Responses Spontaneous Activity is Normally Distributed The “Noise” Distribution

Distributions of Sensory Responses A Mild Stimulus is Presented (d’=1) The “Noise” Distribution The “Signal + Noise” Distribution

Distributions of Sensory Responses A Moderate Stimulus is Presented (d’=2) The “Noise” Distribution The “Signal + Noise” Distribution

Distributions of Sensory Responses An Intense Stimulus is Presented (d’=3) The “Noise” Distribution The “Signal + Noise” Distribution

Distributions of Sensory Responses Sub-Threshold Stimulus is Presented (d’=0) The “Noise” Distribution The “Signal + Noise” Distribution

About d’ So, d’ is a statistic for measuring perceptual sensitivity.

About d’ So, d’ is a statistic for measuring perceptual sensitivity. Also, d’ often refers to “detectability”, and “discriminability” in perceptual experiments.

About d’ So, d’ is a statistic for measuring perceptual sensitivity. Also, d’ often refers to “detectability”, and “discriminability” in perceptual experiments. A high d’ value -----> good performance: A low d’ value -----> poor performance.

About Bias Now let’s consider THAT OTHER aspect of behavior… bias.

About Bias: The inclination to favor a particular response. Example: The inclination to favor the “yes, I see it” response over the “no, I don’t see it” response.

About Bias Signal Detection Theory assumes that Bias can be measured according to a criterion. Criterion: A rule for converting sensory activity into an overt response.

Criterion The “Noise” Distribution The “Signal + Noise” Distribution

Neutral Criterion The “Noise” Distribution The “Signal + Noise” Distribution

Absent Present Stimulus-Response Matrix Correct Rejection False Alarm Miss Hit “No” “Yes” Response

Neutral Criterion The “Noise” Distribution The “Signal + Noise” Distribution

Liberal (low) Criterion The “Noise” Distribution The “Signal + Noise” Distribution

Conservative (high) Criterion The “Noise” Distribution The “Signal + Noise” Distribution

About Bias Just as d’ is the statistic for sensitivity, Beta ( ) is the statistic for bias. When… = 1, the criterion is neutral (no bias) the criterion is low (liberal bias) the criterion is high (conservative bias)

Part 3 Applications of Signal Detection Theory

S. D. T. Applications S. D. T. can be used in perceptual discrimination experiments.

S. D. T. And Discrimination The “slow” distribution The “fast” distribution

S. D. T. Applications S. D. T. can be used in non-perceptual research, including memory experiments.

S. D. T. And Memory The “new items” distribution The “old items” distribution

Learning Check I. Draw two bell-shaped curves (Gaussian distributions) with the same mean, but different standard deviations. II. Draw two bell-shaped curves (Gaussian distributions) with the same standard deviations, but different means. III. Draw one signal-detection-theory plot for a subject who has POOR discrimination, and another signal-detection-theory plot for a a different subject is has GOOD discrimination. IV. Finally, on the SDT plots that you just completed, draw a liberal criterion for one subject, and a conservative criterion for the other. Label each of these clearly.