Kevin Burns designed a unique deck of cards for his research

Dealing with a deck of double-sided cards, Kevin Burns is studying how people make diagnoses and decisions in command and control.

In the Gulf of Sidra off the coast of North Africa, radar aboard a U.S. Navy destroyer spots an approaching aircraft traveling at high speed and low altitude. Originating from somewhere in Libya, the aircraft is headed directly toward the ship. To the ship's commander, the approach has all the earmarks of an aircraft with hostile intent. Is it a threat? What should he do? How the commander diagnoses the situation and decides on a course of action—what support systems might assist him—are very much the concern of MITRE's Kevin Burns.

Burns, principal investigator in a research project called Mental Models, seeks to explain how people think and act under uncertainty. Although of critical need for military command and control, the benefits of Burns' work might extend to dealing with terrorist attacks, commercial airline emergencies, nuclear power plant accidents, fire fighting, complications during surgery, or any other threat needing timely, decisive action. "Our goal," Burns explains, "is to model what goes on in the minds of decision makers so we can design support systems that can help them out."

Since Burns can't readily create war games in his lab, instead he creates card games that pose the same sorts of cognitive challenges. But his games, including TRACS (Tool for Research on Adaptive Cognitive Strategies), are not played with a standard deck of cards. Instead, TRACS uses a special deck of double-sided cards, "marked cards," if you will (see Figure 1), where the backs are black "tracks" that give clues to the suits (red or blue) on the fronts, like radar reports give clues to target IDs. Why double-sided cards? Because standard playing cards provide either all information, if they are dealt face up, or no information, if they are dealt face down, yet warfare is a game of partial information."Command and control is a serious game," says Burns, "and to win a commander must achieve some confidence in the situation and forecast the consequences of his actions." These are also the challenges of TRACS, which is why it is subtitled, "The Game of Confidence and Consequence," but could also be subtitled "The Game of Command and Control."

As a research tool, TRACS allows Burns to investigate different facets of human thinking in games with different sets of rules. So far, Burns has focused on how people assess probabilities in a solitaire game called Straight TRACS. The game is like walking through a minefield where a player chooses between two tracks (left or right) on each step by judging the odds that each track will (if turned over) match the color of a third card in the middle (Figure 1). In another game, called Spy TRACS, the player gets help from simulated teammates, but these "spies" are of limited reliability (like real spies), so the player must aggregate odds from two sources (track and spy) to make a diagnosis (red or blue?) and decision (go left or go right)?

Figure 1	Figure 2

After months of collecting experimental data, and many more months of analyzing it, Burns has developed mathematical models of the "mind sets" and "short cuts" that govern human judgments in TRACS. These models capture basic limits on how much information people can handle at once, as well as cognitive biases that prevent people from extracting as much certainty as they should from the information they have. Based on his detailed models, Burns has developed a visual display, called "Bayesian Boxes," that helps people overcome their natural biases in aggregating probabilities. Bayesian theory specifies how to extract the most certainty from uncertain data, which is what a commander must do in determining the intent of an approaching aircraft based on radar tracks and other clues. The theory derives from an unpublished essay by the 18th century Rev. Thomas Bayes, which was discovered after his death by the French mathematician Laplace. "The essence of the Bayesian approach," explains a recent article in the Economist, "is to provide a mathematical rule explaining how you should change your existing beliefs in the light of new evidence." Burns and other cognitive researchers have shown that, while people are naturally Bayesian in familiar situations, the math does not agree with the mind under novel and stressful conditions.

"Bayesian Boxes" (Figure 2) is a colored calculator that pictures probabilities in such a way that people can see both how certain they should be and why. "Showing why is the key," says Burns, "because results on a screen won't change people's minds unless they understand the reasons." In fact, adds Burns, failure to show why is the downfall of most support systems that people refuse to use.

The necessity of properly changing beliefs in light of new evidence as a situation unfolds was never more evident than in the tragic encounter of the USS Stark, and in a separate incident with the USS Vincennes, both taking place in the Persian Gulf in the late 1980s. The Stark was fired upon and hit by missiles from an Iraqi jet fighter, which resulted in the deaths of 35 Navy sailors. The Vincennes shot down Iranian commercial Flight 655, thinking it was an attacking war plane, which resulted in the loss of all 290 passengers and crew.

Hindsight is 20/20. But Mental Models research, and its eventual tools, could bring foresight and insight to the bridge of a Navy destroyer or anywhere else where decision science can enhance the natural powers of the mind. For now, it's back to the laboratory where Burns' research is moving from the minefields of Spy TRACS into the battlefield of Poker TRACS—a war fought against intelligent adversaries, but with cards and chips instead of radar and weapons.

—by Tom Green

Related Information

Websites

• Mental Models in Naturalistic Decision Making

Solutions That Make a Difference.®
Copyright © 1997-2013, The MITRE Corporation. All rights reserved.
MITRE is a registered trademark of The MITRE Corporation.
Material on this site may be copied and distributed with permission only.

Privacy Policy | Contact Us