Training Requirements Analysis for Tactical Domains

Abstract : The goal of this research program was to develop and test a model of proficient performance. Three aspects of proficient performance were studied: recognitional capacities, perceptual learning, and the use of analogies. We studied the use of analogical reasoning for generating predictions in technological environments. Three models of analogical reasoning were considered and rejected. The first is the standard a:b::c:d model employed by test-makers. The second stems from the use of analogical reasoning to generate new scientific hypotheses. Seven Air Force engineers were interviewed; all had used comparison cases as analogues for the task of predicting reliability of subsystems for the B-1. Neither model was able to account for the performance of the engineers. A third model claims that analogical reasoning is based on similarity matches, and is probabilistic. This model was rejected on conceptual grounds -- the processes it relies on are inadequate for the task. To replace these models, we developed a new theory of analogical reasoning, showing its basis in standard forms of deductive logic. We also were able to define the conditions under which analogical reasoning will generate formally valid conclusions. This work is relevant for any area in which comparisons play an important role, such as the domain of technological improvement involving design changes in automobiles, aircraft, etc. The research performed has implications for a number of applied areas, such as the development of methods for generating predictions under conditions of uncertainty, the design of programs for training personnel to reach high levels of proficiency, and the development of automated decision aids to support experienced C2 personnel.