|
Scientific Work Useful research on human judgment and decision making requires a blend of both rigor and relevance. On the one hand, the rigor of controlled experiments and formal mathematics is needed to build models that are credible. On the other hand, the experimental tasks must have relevance to practical problems if the models are to be applicable. Our approach seeks to balance rigor and relevance by performing experiments in a "synthetic task environment", which is a laboratory context that simulates the cognitive challenges of "naturalistic" decision making (Zsambok & Klein, 1997). |
|
|
Our synthetic task environment is a new family of card games called TRACS: Tool for Research on Adaptive Cognitive Strategies (Burns, 2001; 2004). TRACS is a played with a special deck of two-sided sides cards, where the backs provide clues to the fronts. This unique feature of TRACS (did you ever see a game of two-sided cards?) offers researchers the dual advantages of rigor and relevance. |
|
With respect to rigor, the back of each card constrains the identity of the front (unlike standard playing cards with non-informative backs). This constraint makes the games more tractable to mathematical analyses of optimal solutions, which are needed to benchmark cognitive performance. With respect to relevance, the TRACS cards and rules are designed to simulate the probabilistic and dynamic challenges of naturalistic decision making (Zsambok & Klein, 1997) in military, medical and other practical domains. For example, in one version of the game, the backs of the cards (called tracks) are analogous to the radar images that a military commander must diagnose (e.g., Friend or Foe?) to make decisions and take actions. The TRACS games can be played online in the TRACS Arcade and TRACS Casino. More background on the art and science of the game, and details on the cards and rules, are available at www.tracsgame.com and Burns, 2004. All of the games use the same deck of two-sided cards (see below), but the deck size and makeup can be adjusted for various levels of difficulty and different numbers of players. The fronts of the cards show colored sets (Red or Blue) called "treads". The backs of the cards show black shapes (triangle, circle or square) called "tracks". There are only six types of track/tread cards in the deck, but there are multiple copies of each card type (and a non-uniform distribution of card types) in the deck. This structure defines the track/tread odds in a "full deck" (at the start of the game), but the odds change in play as tracks are turned to reveal their treads (so you are no longer playing with a full deck).
Our initial experiments (Burns, 2002a; 2002b) were performed with the simplest game (played solitaire) called Straight TRACS. The task is matching game, kind of like walking through a mine field where you have to turn over one of two tracks (left or right) at each step, trying to match a given tread (Red or Blue) for that step. The object is to minimize the risk of "strikes" (mismatches) on a trip through the deck. The challenge is to count cards and update odds to make the best choice on each turn. Subsequent experiments have (see A Measured Result), and future experiments will (see Where We Are Going), use other versions of the game that require more complex judgments and choices, especially the TRACS Casino games. References Burns, K. (2001). TRACS: A Tool for Research on Adaptive Cognitive Strategies. At www.tracsgame.com. Burns, K. (2002a). On Straight TRACS: A Baseline Bias from Mental Models. Proceedings of the 24th Annual Conference of the Cognitive Science Society, 154-159 (Erlbaum). Burns, K. (2002b). Dealing with TRACS: The Game of Confidence and Consequence. Proceedings of the American Association for Artificial Intelligence, Symposium on Chance Discovery. Burns, K. (2004). Making TRACS: The Diagrammatic Design of a Double-Sided Deck. Proceedings of the 3rd International Conference on the Theory and Application of Diagrams. Zsambok, C. E., & Klein, G. (1997). Naturalistic Decision Making (Erlbaum). |
