Investigating motor preparatory processes and conscious volition using machine learning

BackgroundConscious volition is a broad term and is difficult to reduce to a single empirical paradigm. It encompasses many areas of cognition, including decision-making and empirical studies can be done on these components. This work follows on the seminal work of Libet et al. (1983) which focused on brain activity preceding motor activity and conscious awareness of the intention to move. Previous results have subsequently faced criticism, particularly methods used to average out EEG data over all the trials and the readiness potential not being present on an individual trial basis. This following study aims to address these criticisms. ObjectivesTo use machine learning to investigate brain activity preceding left/right hand movements with relation to conscious intent and motor action. MethodologyThe data collection involved the recreation of the Libet experiment, with electroencephalography (EEG) data being collected. An addition made in this study was the choice between "left" and "right" while observing the Libet clock to subjectively mark the moment of conscious awareness. Twenty-one participants were included (four females, all right-handed). A deep (machine) learning model known as a convolutional neural network (CNN) was used for the EEG data analysis. ResultsSubjectively reported conscious intent preceded the action by 108 ms. The CNN model was able to predict the decision "left" or "right" as early as 4.45 seconds before the action with a test accuracy of 98%. ConclusionThis study has shown motor preparatory processes start up to 4.45 seconds before conscious awareness of a decision to move.