A PSYCHOPHYSICAL STUDY OF INTERPRETATION OF PHYSIOLOGICAL RECORDINGS.
Final technical rept. 1 Apr 64-28 Feb 66,
MICHIGAN UNIV ANN ARBOR SENSORY INTELLIGENCE LAB
Pagination or Media Count:
Studies were conducted on the ability of observers to read records of physiological character. The records were constructed of signals added to noise, with both the signal and noise designed to be similar to conditions frequently encountered in physiological recordings. The observer was presented with a trace on an oscilloscope which was generated either by signal plus noise, or by noise alone. He was asked to state whether the trace contained the signal or contained only noise. During the studies he was not permitted to see the signal without the noise. After each decision, the observer was told if he was right or wrong. In the first of two experiments, the signal had no uncertain parameters, while in the second, the signal had a relatively small starting time uncertainty. ROC curves were obtained by changing the bonus payoff schedule for the observer. Under fixed signal conditions, d, the index of detectability, was independent of drastic changes in the observers criterion. In the second experiment, the slight uncertainty in starting time was reflected in a shift of the slope of the ROC curve. The change in slope was indicative of increased variance in the S N distribution of events. Author