Deciding on Mars: The effects of isolation on autonomous team decision-making

Long distance missions, like Mars, hinge on the ability of autonomous crews comprised of diverse experts to make high quality decisions throughout the mission. How well do analog crews perform on decision tasks involving distributed expertise? Are there mission phases where performance is particularly problematic? Does dissent within the crew improve information sharing? Five parallel space-relevant decision making tasks requiring crews to leverage distinct information to make a team decision were developed. Tasks were designed using the hidden profile paradigm. Each task presented the crew with a problem and 3 decision options. Each crewmember received some unique information and some information known by all crewmembers. In total, the crew received 29 or 30 pieces of information about each decision option. In keeping with hidden profile tasks, information was distributed to crewmembers so that a majority, if not all, of the individuals prefer the worst option. Only if the crew combined unique information can they reach the optimal decision. The preference structure of the task was validated on a crowdsourcing website participant pool (N = 3,184). The set of tasks was administered during NASA's Human Exploration Research Analog (HERA) in Campaigns 4 and 5. Six 4-person crews lived and worked in an 80-m³ habitat for 45-day missions. Unique information sharing and decision quality were assessed on mission day (MD) -4, 6, 14, 20, and 34. Findings show crew decision making suffered in isolation and confinement. The best performing crew correctly solved 60% (3 out of 5) of the tasks, whereas the worst performing crew correctly solved only 20% (1 out of 5) of the task. The decision-making performance of the crews peaked in the second quarter of the mission with a 71% success rate and had a low of 17% on MD 34. Crew information sharing also peaked on MD 14. A manipulation to create dissent within the crew improved the amount of unique negative information shared by the crew (ruling out inferior options), but did not affect the amount of unique positive information they shared (needed to rule in superior options). These findings suggest space crews will benefit from team decision training and protocols for making team decisions that mitigate these performance decrements. The tasks developed here provide a useful way for future analog studies to evaluate the efficacy of training and protocols.