A STOCHASTIC MODEL OF SIGNAL DETECTION UNDER CONDITIONS OF TEMPORAL UNCERTAINTY.

Abstract

The report describes a reformulation of the problem of analyzing detection data obtained when signal presentation times follow an arbitrary intersignal interval distribution. The approach is based on revised definitions of hits, misses, correct negatives, and false positives which are natural in this situation. It permits the construction of functions analogous to the isosensitivity and isobias curves of classical detection theory directly from observed time series of signals and responses. A model of the processes underlying the generation of signal-response time series is described, and the probability densities of interval durations associated with hits, misses, correct negatives, and false positives are presented. The areas under these distributions, and two related densities, are interrelated. An expression for the expected number of responses per signal is given along with suitable estimation procedures. The proposed model incorporates arbitrary intersignal interval and response latency distributions. These were replaced respectively by exponential and triangular distributions in a test of the model using data obtained from a simple auditory detection experiment in which signal rate and intensity were varied. (Author)

Document Details

Document Type
Technical Report
Publication Date
Jun 24, 1968
Accession Number
AD0693868

Entities

People

  • James N. Cronholm

Organizations

  • United States Army Medical Research Laboratory

Tags

DTIC Thesaurus Topics

  • Construction
  • Demographic Cohorts
  • Detection
  • Intensity
  • Intervals
  • Probability
  • Signal Detection
  • Uncertainty

Readers

  • Sensor Fusion and Tracking Systems.
  • Statistical inference.
  • Theoretical Analysis.