Model, taxonomy and methodology for research employing electroencephalography-based brain-computer interface games

Abstract

The rapid expansion of Brain-Computer Interface (BCI) technology, aligned with the advancements on the fields of Physiological Computing (PC), Human- Computer Interaction (HCI) and Machine Learning (ML), allowed for the recent development of applications outside of clinical environments, such as education, arts and games. Games controlled by electroencephalography (EEG), a specific case of BCI technology, benefit from both the fields of BCI and games, since they can be played by virtually any person regardless of physical condition, can be applied in numerous contexts, and are ludic by nature. Despite these recent advancements, there is still no solid theoretical foundation to aggregate the terminology and methods of these fields, since current models and classification schemes can represent characteristics of either BCI systems or games, but not both. In this sense, this work presents a general model for representing EEG-based BCI games; a taxonomy for classifying primary studies of the field; and a methodology for conducting scientific studies using those games. The proposed model is intended to help researchers describe, compare and develop new EEG-controlled games by instantiating its components using concepts from the fields of BCI and games. The CoDIS taxonomy was constructed based on an expanded version of this model, which considers four aspects of EEG-controlled games: concept, design, implementation and study, each with different dimensions to represent a variety of characteristics of such systems. Based on both the model and the taxonomy, and guided by the principles of empirical research, a new methodology called PIERSE was developed for the planning, implementation, execution and reporting of scientific experiments that employ EEG-based BCI games.