What does current research show about shared mental models in virtual teams, and how are they measured?

Why shared mental models matter more when teammates are apart

A shared mental model is the common picture teammates carry of what the task requires and how the team works. When people share that picture, they can coordinate without spelling everything out. In virtual teams, where members are geographically separated and often lack visual cues, that implicit coordination becomes harder—and more valuable. The 2025 systematic review of studies from 2004 to 2023 concludes that diverse types of shared mental models impact team functioning in virtual settings [1].

Real-world evidence from emergency medicine shows what happens when that shared picture weakens. In ambulance missions, the dispatcher is physically separated from the crew but coordinates, organizes, and conveys critical information. Analyzing voice recordings from actual missions, researchers found that high-complexity missions had lower frequencies of team competencies and coordinating mechanisms than low-complexity ones [2]. The authors concluded that the lack of visual input from a team member during interaction could lead to process loss, with the team breaking into sub-units [2]. That is a proposed explanation, not a proven cause, but it fits the pattern: distance removes cues that normally keep everyone on the same page.

How researchers actually measure a shared mental model

There is no single standard instrument. The 2025 review coded studies by the communication technology used, team characteristics, type of shared mental model, and the input-process-output framework, and found that measurement methods lack standardization [1]. In practice, researchers pick a method that fits their setting, which makes comparing results across studies difficult.

One approach listens to what teams say. In the ambulance study, researchers inferred shared mental models from the frequency of coordinating mechanisms and team competencies in voice recordings of real missions [2]. This captures behavior as it happens, but it measures communication patterns rather than the underlying knowledge directly.

Another approach asks teammates to make the same judgment. A 2025 feasibility study in team sports developed a virtual-reality measurement in which players watched video clips and indicated both their own decision and their partner's decision; when teammates' decisions matched on a clip, that counted as a shared mental model for that situation, and the average across clips formed the team's score [5]. The pilot showed that shared mental models were measurable this way—teammates agreed more often than chance [5]. This is a situational measure, designed to get closer to what people actually do than a questionnaire would.

Questionnaires remain in use as well. In the same sports study, questionnaire-based shared mental model scores correlated with team trust and with subjective team performance, even though the virtual-reality measure did not confirm the assumed influences in the main structural model [5]. That divergence is a useful warning: different measurement methods can tell different stories about the same team.

What changes the picture: task complexity and the limits of each method

Task demands appear to shift how shared mental models play out. In the ambulance missions, the profile of communication differed between high- and low-complexity missions, with more frequent use of coordinating mechanisms and team competencies in low-complexity missions [2]. The study also found that shared mental models and closed-loop communication were the only coordinating mechanisms observed, and leadership and team orientation the only competencies exercised [2]. That narrow range may reflect the specific emergency setting rather than virtual teams in general.

Measurement choices carry their own scope limits. The virtual-reality method was tested with handball teams, not workplace virtual teams, so its promise for organizational settings is suggested rather than established [5]. The ambulance study used voice recordings from real missions, which gives ecological validity but captures only what is spoken [2]. And the 2025 review's central finding—that methods lack standardization—means that a result from one measurement approach may not transfer cleanly to another [1].

Where measurement is heading, and what is still missing

The push is toward situational, behavior-based measures and toward consistency. The 2025 review calls for innovative approaches and standardized measurement to move the field forward [1]. The virtual-reality study positions itself as a starting point for more situational group research, arguing that situational measures have the highest ecological transfer to behavior [5].

A related line of work on human-AI teams reaches a similar conclusion from a different direction: a theoretical review found that a lack of consistency in terms and of effective means for measuring shared mental models in human-AI teams currently impedes realization of the concept [4]. That review also notes it remains unclear how the shared mental model construct may differ when some teammates are AI agents, how and under what conditions such models form, and how they should be quantified [4]. For virtual teams of humans, the practical takeaway is that shared mental models are clearly useful to understand, but the field has not settled on one trustworthy way to count them.

Sources used in this answer

Related research questions