THE MENTAL STATE FORMALISM OF GMU-BICA
Alexei V. Samsonovich, Kenneth Alan De Jong, Anastasia Kitsantas · International Journal of Machine Consciousness · 2009
GMU-BICA, the biologically-inspired self-aware cognitive architecture developed at George Mason University, continues to be a useful prototype for various intelligent artifacts, including intelligent tutoring systems, yet the underlying formalism of mental states used in its design was never described in detail. The present theoretical work aims at filling this gap, focusing on the top cognitive level (mental states) and leaving detailed description of the lower level (schemas), as well as non-declarative components, for future publications. Among the distinguishing features of the GMU-BICA mental state formalism are: (i) a subject-centered view of the world, (ii) the multiplicity of mental perspectives simultaneously represented in working memory, each playing its unique functional role, and (iii) the limited span of awareness. This model is consistent with human psychology and gives testable predictions. The work explains, through analysis of examples, how the framework can be used to build computational models of self-regulated learning, and why in this case it is expected to unleash a new for artifacts power of human-like cognition and learning.