Detection of visual signals consisting of multiple information sources: A signal detection analysis
Abstract
Two signal detection experiments using visual displays were performed. The displays consisted of analog gauges arranged in a horizontal line. The first experiment used three signals embedded in noise. Each of the three signals was a unique pattern of gauge values. After briefly viewing a display the observers either reported that any of the signals occurred (one-of-m signal detection) or specified which of the signals (if any) had been presented (one-of-m signal recognition-detection). At most one signal was presented on any trial. The results indicate that: (1) performance on a one-of-m detection task can be predicted from performance on a single signal detection task, and, (2) one-of-m recognition-detection performance can be predicted either from one-of-m detection performance or from single signal detection performance. The second experiment investigated how observers combine (weigh) the information present in individual sources (display elements) to form their decision. Either three, six, nine or fifteen display elements were presented at one time. The diagnosticity of the display elements was manipulated by changing the difference between the mean of the signal and of the noise distributions. It was predicted that observers would weigh a source in proportion to its diagnosticity, but this was only confirmed when three sources were used. This effect was small when compared to an optimal weighting strategy. For larger numbers of sources, observers applied equal weights regardless of the diagnosticity. Several possible explanations for this finding are considered. These experiments extend our knowledge of how humans process and use complex visual signals consisting of multiple, differentially diagnostic information sources.
Degree
Ph.D.
Advisors
Sorkin, Purdue University.
Subject Area
Psychology|Experiments
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.