Instruction in Dynamic Tasks Based on a High-Fidelity Cognitive Architecture

摘要

Initial research was performed on the Brute synthetic task for unmanned flight (based on the Predator system). This research indicated that spatial orientation was a major difficulty that people had in performing this task. Many participants made many errors in their directional judgments. A series of experiments were performed systematically investigating the difficulties people had in integrating a map view of a terrain with a camera view available from a plane. Two strategies were identified for bringing the map view and the camera view into alignment. One strategy involved mentally rotating the camera view until it was in alignment with the map view. The second strategy involved calculating the offset of various targets in the camera view and adding that offset to the direction of orientation in the map. Cognitive models in ACT- R were developed that implemented both of these strategies and fit to eye movements of participants. Part of this effort involved developing an imagery module for ACT-R that can be used more generally to model navigation. The final report describes in more detail the nature of the experimentation, the results, and the ACT-R model.