ACT-R (Anderson & Lebiere, 1998) is a general purpose hybrid cognitive architecture for developing cognitive models that can vary from simple reaction tasks to simulations of pilots navigating airplanes and operators of airtraffic control systems. ACT-R follows the approach of unified theories of cognition (Newell, 1990), in which several theories about different aspects of cognition are used in a single simulation system. Today, ACT-R has emerged as the architecture of choice for many cognitive modelling problems.
The symbolic level of ACT-R is organised as a goal-directed production system with declarative and procedural types of knowledge encoded in the form of chunks and production rules respectively. The subsymbolic level accounts for fuzzy or probabilistic properties of cognition, and it uses activation values, utilities, associations and other parameters to control the way symbolic knowledge is used. The dynamics of these parameters is described by a set of equations, that arise from neuroscience and cognitive psychology theories.
Conflict resolution is an important part of the subsymbolic level of ACT-R, and it represents a model of the decision--making mechanism in the brain. Several studies have suggested recently that a more dynamic conflict resolution mechanism in the architecture could significantly improve the decision--making behaviour of cognitive models. The current mechanism of ACT-R has been revised and a new algorithm has been proposed (see Belavkin, 2003; Belavkin & Ritter, 2004). The new algorithm is called OPTIMIST (stands for `Optimism' plus `Optimisation') and it has been implemented as an overlay to the ACT-R architecture. Thus, it can easily be used as an alternative mechanism.
This paper provides the documentation on how to load and use the OPTIMIST algorithm.
Below you can download the OPTIMIST overlay Version 2 for ACT-R Versions 5 and 4.