Results from a meta-analysis of studies using personality constructs to predict military aviation training outcomes are reported. From the 26 studies that reported effects of personality as predictors of aviation training outcome, the constructs of neuroticism (K = 7), extroversion (K = 8), and anxiety (K = 4) appeared most frequently. Meta-analysis effects were derived using both random effects and artifact distribution model. Uncorrected effects from the random effects model produced the largest mean effect for neuroticism (r meta = -.15), followed by extroversion (r meta = .13), and anxiety (r meta = -.11). Corrections for predictor reliability and range restriction produced the greatest increase in the validity coefficient for neuroticism (r corr = -.25), implying more psychometrically reliable and sensitive instruments could substantially improve the predictive validity of personality assessments in aviation selection contexts. The results confirmed the hypothesis that neuroticism and its facet anxiety would be negatively related to training success, and that extroversion would share a positive relationship with training success in military aviation.

We consider an extended notion of reinforcement learning in which the environment can simulate the agent and base its outputs on the agent’s hypothetical behavior. Since good performance usually requires paying attention to whatever things the environment’s outputs are based on, we argue that for an agent to achieve on-average good performance across many such extended environments, it is necessary for the agent to self-reflect. Thus weighted-average performance over the space of all suitably well-behaved extended environments could be considered a way of measuring how self-reflective an agent is. We give examples of extended environments and introduce a simple transformation which experimentally seems to increase some standard RL agents’ performance in a certain type of extended environment.

Linguistic Geometry (LG) is a mathematical theory that captures the decision making processes similar to those of the great chess masters. In the current demonstration LG is applied to a military combat game that allows users to test their strategies against an intelligent simulated force while receiving feedback regarding the accomplishment of the task and amount of resources spent. Additionally, a usability study was conducted on LG-based software (LG-PROTECTOR) indicating that the software was easy to learn and remember and has an interface conducive to training. The next generation of LG software (LG-EXPERT) is currently under development to provide increased capabilities that will help train users to “Think Like A Commander” as they practice rules of engagement and learn effective courses of action. Additional training implications are discussed.

We consider an extended notion of reinforcement learning in which the environment can simulate the agent and base its outputs on the agent's hypothetical behavior. Since good performance usually requires paying attention to whatever things the environment's outputs are based on, we argue that for an agent to achieve on-average good performance across many such extended environments, it is necessary for the agent to self-reflect. Thus, an agent's self-reflection ability can be numerically estimated by running the agent through a battery of extended environments. We are simultaneously releasing an open-source library of extended environments to serve as proof-of-concept of this technique. As the library is first-of-kind, we have avoided the difficult problem of optimizing it. Instead we have chosen environments with interesting properties. Some seem paradoxical, some lead to interesting thought experiments, some are even suggestive of how self-reflection might have evolved in nature. We give examples and introduce a simple transformation which experimentally seems to increase self-reflection.

No abstract