Measuring Situation Awareness in a Collaborative Game

It seems that ThoughtLink is working on a very similar project of ours. Together with the Center for Naval Analyses (CNA), they conducted an experiment to measured the effect of different modes of communication and visualization on a distributed team's shared situational awareness. This project was done for the Defense Advanced Research Projects Agency (DARPA) in support of the Wargaming the Asymmetric Environment (WAE) program. They investigated collaborative situation with the following factores: team members had to share information to do well, their decisions could be directly and easily recorded, the measure of their decisions would describe the degree of shared situational awareness of the team. Situation Awareness is according to Endsley (1995):

the perception of the elements in the environment within a volume of space and time, the comprehension of their meaning, the projection of their status into the near future, and the prediction of how various actions will affect the fulfillment of one's goals.

The game they used is ScudHunt:

SCUDHunt is a simple, short, abstract game of command, control, communications, intelligence, surveillance and reconnaissance (C3ISR) played by a team of 4 to 7 players. The game requires group decision-making and allocation of scarce resources under conditions of time pressure and uncertainty.

The goal of the game is for team members to correctly determine (within a specified timeframe) where three Scud launchers are located, on a 5 by 5 grid. Each team member controls a different sensor, and team members must share their sensor results in order to identify the launcher locations. In each turn, team members decide, typically in a collaborative process, where to locate each sensor. The results for a sensor are returned to the team member controlling it. Results include: X - launcher found; O - no launcher in the square; ? - not sure. Some sensors can also be killed or temporarily disabled on a turn.

At the end of each turn, based on the search results to date, each team member nominates at least three grid squares in which the launchers might be located. The overlap among the nominations reflects the team's shared situational awareness. If there is no overlap, every team member will vote for a different set of grid squares. If there is complete shared situational awareness, each member will vote for the same set of grid squares.

We measured SSA as: the total number of nominated squares divided by the number of unique grid squares nominated. Team SSA scores vary from 1 to the number of team members. As an example, given 4 team members each of whom nominates three squares, if there is minimal SSA, each member will nominate different sets of squares. Then the total number of nominated squares is 12 (4 members x 3 squares/member) and the number of unique grid squares is also 12 (there is no overlap), and the team's SSA score is 1. If the same team has complete SSA and each member votes for the same set of grid squares, then their team score will be 4: 12 squares nominated overall divided by 3 unique nominations.

There is a full report about this: Gaming and Shared Situation Awareness. Why do I blog this? in the European project about 'mutual modeling' we are working on, there is also this idea of using a collaborative game as a paltform to extract information concerning collaborative behavior. My interest towards this game is twofold:

  • to know how researchers benefit from using a game as a platform to extract data, eliciting collaborative behavior
  • to see which results they obtained about the use of communication tools and their impacts on situational awareness