The investigative team focused their data collection strategies into three distinct phases: Index of Cognitive Efficiency (ICE), leadership relations, and task and team mental modeling. The ICE Index of Cognitive Efficiency (overall score) was collected using Spaceflight Cognitive Assessment Tool for Windows (WinSCAT) which is a BHP Behavioral Medicine “surveillance” tool. For leadership relations in phase 2, the crew members completed a network survey asking about their leadership relations (i.e., “Who do you rely on for leadership?”) every five days. For phase 3, each individual’s task and team mental model was collected every day except Sundays, using the elicitation method. Crew members were asked, on a scale of 1 (totally unrelated) to 7 (very strongly related), to report their perceptions of the relationships between a list of 8 task elements, which produced 28 dyadic values for each crew member for each day.

RESULTS:
Phase 1
The research team developed a computational model that successfully predicted crew shared mental models based on personal, task, social, and situational factors. This model can be used in future mission planning and mission support. Comprehensive meta-analysis of prior research established that shared mental models are a robust predictor of team performance. Though moderators were found, the relationship between shared mental models and performance was always positive. The implication being that long distance missions should monitor and promote shared mental models among the crew, among mission support, and between the crew and mission support. Leveraging research previously conducted on Skylab missions demonstrated that crew shared mental models, and shared mental models between the crew and ground can be assessed using text analysis of unstructured text; in this case mission transcripts of spoken communication were used.

Phase 2
The development of the computational model operationalizing shared mental models as network ties among crew members was finalized, followed by running “whatif” computational experiments to explicate how leadership network structures affect mental model development in teams over time. Fitting the empirical data to the computational model showed that crew relations (e.g., leadership), crew composition (e.g., cognitive ability, agreeableness diversity), and mission task design (e.g., workload, task importance) are consistent predictors that could be targeted for SMM countermeasure development. Results of computational experiments suggest that shared leadership results in the most shared mental models, followed by hierarchical leadership, and then connected leadership; factionalized and disenfranchised leadership structures result in the least shared mental models. These findings suggest informal leadership is a useful crew property to monitor, and that the formation of factions, or occurrence of disenfranchised crew members should trigger crew support countermeasures as these structures in the crew pose a threat to mental models and, by extension, performance. When evaluating how crew and ground schemas about communication networks affect information sharing, this experiment found crew members often overestimated the presence of communication links between members who are part of the same formal group and underestimated the presence of links between people who are in different groups. This underscores the importance of accurate relational schema as a precursor to effective information sharing and performance.

Phase 3
In the first portion of this phase, the research team developed a prototype dashboard for the detection of shared cognitive architecture and explored the correlation between shared mental model and team performance in HERA. Investigators found substantial positive correlations of shared mental models and both dimensions of team performance. Though shared mental models are a strong predictor of team performance across mission stages, some nuanced shifts were observed: (1) shared mental models are the most strongly predictive of performance on creative thinking tasks before and after communication delay; (2) on problem solving tasks, shared mental models increase in predictive power continuously throughout the mission; (3) of all the outcomes, fluency is the most predicted by shared mental models. The second portion of Phase 3 evaluated isolation, confinement, and network position affect network routing accuracy. Using a network routing task, network acuity among HERA crews was significantly higher than mission control members in Campaign 4 and Campaign 5, but not Campaign 3. Additionally, communication delays did not impact network acuity among HERA crews, but reduced acuity among mission control members. Further, crew members who scored lower on the personality characteristic of conscientiousness had higher network acuity in both Campaign 3 and Campaign 4. Finally, crews’ network acuity was associated with personality characteristics of openness to experience, agreeableness, and neuroticism in Campaign 4 but not Campaign 3. Overall, the findings suggest that selecting crews with high network acuity will play a key role in alleviating the risk of information-sharing failures within a multiteam system under conditions of communication delays.