Maze-Solving as a Performance Measurement Tool for Human Operations under Time-Stress

This report documents a series of experiments to investigate the usefulness of maze-solving as a performance measurement tool for humans under time-stressed conditions. Three experiments were carried out: the first to establish an appropriate difficulty level for the mazes, the second to test the effect of repeated exposures to the same set of mazes on performance, and the third to measure the effect of time-stress on maze-solving performance. Subjects were asked to solve a maze by guiding a small dot from the beginning of the maze to the goal. Subjects had control over the dot direction but not over its speed. Stress was induced by increasing the dot speed. Results from the first experiment indicated that the largest and most complex maze configuration considered showed the most reliable difference between dot speeds, and this configuration was used in both of the subsequent experiments. The second experiment demonstrated that the same set of mazes could be presented to subjects four or five times with no appreciable learning effects. The third and last experiment quantified the differences in performance due to dot speed. There were significant and reliable differences in score among dot speeds. Errors made in solving mazes were examined qualitatively as well, and the errors found were symptomatic of a shortened planning horizon: incorrect turns at early critical decision points, and attempts to return to the centerline of the maze prematurely. (Author)