[ACT-R-users] PokerBot competition at ICCM2004

[Christian Schunn]

NEW ANNOUNCEMENT REGARDING ICCM2004

ICCM 2004 Competitive Symposium: PokerBot World Series

Game playing seems to satisfy a basic craving of 
human cognition by exercising its fundamental 
abilities in a competitive setting.  Therefore, 
it provides an excellent benchmark to study and 
evaluate cognitive models in tractable yet 
naturalistic settings that are simple and formal 
yet reproduce much of the complexity of real 
life.  Poker is probably the most widely played 
card game, with endless variations played by 
millions of adherents from casual players 
gambling pennies to professionals competing in 
million-dollar tournaments.  Unlike other games 
that emphasize one particular aspect of 
cognition, poker involves a broad range of 
cognitive activities, including:


 Reasoning under uncertainty (opponents' cards)

 Dealing with probabilistic outcomes (future cards)

 Decision-making with multiple options (chips used for bets)

 Individual differences (different styles of play)

 Inference of intent (from opponents' bets)

 Intentional deception (bluffing, sandbagging)

 Pattern recognition (detecting trends from flow of game)

 Social and emotional aspects (dealing with winning and losing)

 Economic behavior (factoring impact of amount of bets)

Because of the range of cognitive activities 
involved, poker provides a broader and more 
challenging test for cognitive modeling than 
other games such as chess that focus on a more 
restricted range of mechanisms (e.g. search). 
Despite the complexity of aspects involved, it 
remains a highly tractable domain, partly because 
it abstracts away from computationally demanding 
perception and interaction problems.  Poker is 
increasingly being played in online gaming 
communities where the need for challenging, 
cognitively plausible agents is increasing. 
Poker therefore provides a challenging domain at 
the intersection of fundamental research 
questions and potential mass application.

Open competitions such as Robocup 
(http://www.robocup.org/) and the DARPA grand 
challenge (http://www.darpa.mil/grandchallenge/) 
have recently provided a promising mechanism to 
make progress on long-term research issues while 
focusing on a practical, measurable goal.  To 
that end, we are organizing a competitive 
symposium at ICCM-2004 with the ultimate goal of 
producing a fully functional, human-like poker 
agent.  The competition will take place in two 
parts, emphasizing the dual aspects of cognitive 
modeling: the production of accurate models of 
human cognition that are not mere post-hoc 
descriptions of human performance but can perform 
actual tasks in a fully functional manner.  The 
first part will involve head-to-head, 
winner-take-all play between models to test their 
functionality.  The second part will involve 
evaluating the cognitive validity of the models 
in quantitative and/or qualitative manner, such 
as comparison to archived human games and Turing 
Test judgments.

The variant of poker used is Texas Hold'em, a 
popular, fast-paced game that is the centerpiece 
of the World Series of Poker 
(http://www.binions.com/worldseries.asp). 
Potential participants should contact the 
organizers by email (clebiere at maad.com) by April 
1st to register their interest and obtain 
detailed rules of participation.  Participants 
must register to the conference to be able to 
submit a player but free registration will be 
offered to the winner(s).  The best models will 
be presented at a symposium during the conference 
where progress on research issues will be 
discussed.

For more information on ICCM2004 more generally, 
see http://simon.lrdc.pitt.edu/~iccm/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/act-r-users/attachments/20040225/83c42d39/attachment.html>