On TRACS: Dealing with a Deck of Double-Sided Cards

TRACS (Tool for Research on Adaptive Cognitive Strategies) is a new suite of card games played with a special deck, where the back of each card is a clue to the front of the card. This design simulates the clue/truth structure of real world domains like medicine and warfare, where truths (fronts of cards) must be diagnosed from clues (backs of cards) in order to make decisions (cards to choose, chips to bet, etc.). Here I present the cards and rules of TRACS. I also discuss how the games have been used for computational investigations of memory limits and Bayesian inference. The methods for these studies include human experiments and agent simulations, both of which are facilitated by the unique features of TRACS. The products of TRACS research include a computational model of memory limits and a decision support system for Bayesian inference.

• Type: Conference
• Year: 2005
• Where: CIG
• Authors: Kevin Burns
• Keywords: Bayesian Inference, Cards, CIG 2005, Memory Limits