A TEST OF ECONOMIC-MODELS OF STOPPING POLICY IN VISUAL-SEARCH

Optimization models of human visual search performance can be constructed that assume that the human operator maximizes an objective function, subject to given constraints. Two experiments were performed to validate one such optimization model, using ten students as the subjects. They were instructed to inspect printed circuit boards for one or two faults under different defective rates and cost structures. It was discovered that human inspectors adopted a systematic search strategy. They stopped searching after either one or two complete scans of the circuit board, an optimal strategy in terms of the search economic model. Their stopping time was affected only by the fault type in a one-fault search. In two-fault search, two types of search strategies were apparent in subjects' performance. Four subjects searched for the easy fault first, then searched for the difficult fault. Six subjects searched for both faults simultaneously. The Embedded Figures Test (EFT) was found to categorize these two types of subjects.