Main Subcommittee Category
EducationHow would you label your submission?
IndustryAbstract
Many scientists have been struggled to walk the line between designing an experiment with rigorous control and conducting research whose findings widely generalize. As government and industry invests funding into developing training systems and competency measurement approaches, the importance of transferability of control study findings will be key to prevent the valley of death. This debate often shows up when practitioners are challenged to translate experimental tasks from basic research to a more applied context. The n-back (Kirchner, 1958), a respected measure of working memory ability, is commonly poked fun at for disconnect from any realistic task. Recent work (e.g., Vine et al., 2020 as cited in Vine et al., 2021) has furthered these concerns. The present study expands on an adaptive training system approach presented in 2022 (Rebensky, et al., 2022), which explores the n-backs transferability via a domain relevant task, dubbed Mission Relevant Audio Cue Task (MRACT), which has been designed to mimic communication call-and-response procedures of military personnel that would challenge working memory similar the n-back. Participants were recruited and completed both tasks separately in the Driving Adaptive Research Testbed (DART) developed at an Air Force Research Laboratory. The data from these tasks were compared along relative difficulty of task, perceived difficulty (NASA-TLX), and their demands on physiological measures related to mental workload (fNIR and heart rate variability). The findings of this study reveal similarities between the two tasks but also points of substantial divergence, which have implications for basic research laboratories that train algorithms within adaptive training settings. These conclusions lead to suggestions for future practitioners looking to walk the same line when developing domain relevant tasks based on basic research. The paper and presentation will provide guidance for research and development labs on the design of tasks to test novel measurement and algorithm approaches.