Improving military change detection skills in a virtual environment the effects of time, threat level, and tutorials

Abstract

The first one hundred days of combat are the most dangerous time for a soldier due to uncertainty and unfamiliarity with his area of operations. Developing a desktop virtual trainer for change detection could reduce the threat to soldiers by improving their ability to detect signals among the cluttered, noisy neighborhoods of the current operational environment. Building upon previous change detection and signal detection work, this thesis explores the use of Army Virtual Battlespace 2 as a prototype-training tool for change detection. Leveraging an Army-owned distributable trainer would potentially benefit soldiers prior to deployment. This research team conducted an experiment that tested fifteen participants over four weeks. Each participant explored the virtual environment twelve times. Researchers analyzed correct detections, false alarms, user confidence, threat levels, and tutorial group assignment. This first attempt at developing a military-oriented virtual trainer resulted in statistically significant improvement in detection percentages, user confidence, and decay of false-alarm rates over time with p-values less than 0.01. The results showed no significance in the use of an in-simulation tutorial or target threat level. Future work should expand on this foundational research to determine whether the skills developed using this trainer transfer to real-world change detection.