Open AccessThe subject of the work is simulation of human cognitive processes. Authors propose a computational model of artificial grammar learning – the task allowing researchers to explore processing of environmental statistical regularities in humans. A number of the experiments are presented in the literature demonstrating that several cognitive systems take part in human information processing. The aim of the present work is to describe the model, consistent with the accumulated empirical data.
The proposed model implements the dual-system approach in the form of formalized algorithms. It contains two independent blocks that process upcoming information independently. Authors describe the principles of the interaction of these blocks allowing us to simulate human behavior in different task conditions. The simulation results are compared with human experimental data represented in the literature and obtained by the authors.
The model appeared to be in good agreement with the experimental data. Successful and unsuccessful aspects of the model are described and their possible reasons are discussed. In contrast to most of the current dual-system models, the presented model does not contain neural networks. The advantages of dual-system approaches are described, and symbolic and connectionist approaches to cognitive modeling are discussed. Dual-system models allow us to describe the dissociations between implicit and explicit components in human experience. Connectionist models are criticized for their complexity. Authors suggest that at the same explanatory power, symbolic models should be preferred rather than connectionist ones. They allow better understanding what information processing mechanisms that take place in the human mind.
The results of the presented work can be applied in construction and testing of theoretical models in psychology, and in the development of cognitive architectures based on the human information processing mechanisms.