ABSTRACT
Several studies have suggested recently that a more dynamic conflict resolution mechanism in the ACT-R cognitive architecture (Anderson & Lebiere, 1998) could improve the decision-making behaviour of cognitive models. This part of ACT-R theory is revisited and a new solution is proposed. The new algorithm (Optimist) has been implemented as an overlay to the ACT-R architecture, and can be used as an alternative mechanism. The operation of the new algorithm is tested in a model of the classical Yerkes and Dodson experiment on animals’ learning. When Optimist is used, the resulting model fits the data better than the previous model (e.g. R 2 increases from .85 to .93 in one example).
