Copy the page URI to the clipboard
Steffert, Tony
(2018).
DOI: https://doi.org/10.21954/ou.ro.0000e26d
Abstract
Electroencephalography (EEG) is the measurement via the scalp of the electrical activity of the brain. The established therapeutic intervention of neurofeedback involves presenting people with their own EEG in real-time to enable them to modify their EEG for purposes of improving performance or health.
The aim of this research is to develop and validate real-time sonifications of EEG for use in neurofeedback and methods for assessing such sonifications. Neurofeedback generally uses a visual display. Where auditory feedback is used, it is mostly limited to pre-recorded sounds triggered by the EEG activity crossing a threshold. However, EEG generates time-series data with meaningful detail at fine temporal resolution and with complex temporal dynamics. Human hearing has a much higher temporal resolution than human vision, and auditory displays do not require people to focus on a screen with their eyes open for extended periods of time – e.g. if they are engaged in some other task. Sonification of EEG could allow more rapid, contingent, salient and temporally detailed feedback. This could improve the efficiency of neurofeedback training and reduce the number and duration of sessions for successful neurofeedback.
The same two deliberately simple sonification techniques were used in all three experiments of this research: Amplitude Modulation (AM) sonification, which maps the fluctuations in the power of the EEG to the volume of a pure tone; and Frequency Modulation (FM) sonification, which uses the changes in the EEG power to modify the frequency. Measures included, a listening task, NASA task load index; a measure of how much work it was to do the task, Pre & post measures of mood, and EEG.
The first experiment used pre-recorded single channel EEG and participants were asked to listen to the sound of the sonified EEG and try and track the activity that they could hear by moving a slider on a computer screen using a computer mouse. This provided a quantitative assessment of how well people could perceive the sonified fluctuations in EEG level. The tracking accuracy scores were higher for the FM sonification but self-assessments of task load rated the AM sonification as easier to track.
The second experiment used the same two sonifications, in a real neurofeedback task using participants own live EEG. Unbeknownst to the participants the neurofeedback task was designed to improve mood. A Pre-Post questionnaire showed that participants changed their self-rated mood in the intended direction with the EEG training, but there was no statistically significant change in EEG. Again the FM sonification showed a better performance but AM was rated as less effortful. The performance of sonifications in the tracking task in experiment 1 was found to predict their relative efficacy at blind self-rated mood modification in experiment 2.
The third experiment used both the tracking as in experiment 1 and neurofeedback tasks as in experiment 2, but with modified versions of the AM and FM sonifications to allow two-channel EEG sonifications. This experiment introduced a physical slider as opposed to a mouse for the tracking task. Tracking accuracy increased, but this time no significant difference was found between the two sonification techniques on the tracking task. In the training task, once more the blind self-rated mood did improve in the intended direction with the EEG training, but as again there was no significant change in EEG, this cannot necessarily be attributed to the neurofeedback. There was only a slight difference between the two sonification techniques in the effort measure.
In this way, a prototype method has been devised and validated for the quantitative assessment of real-time EEG sonifications. Conventional evaluations of neurofeedback techniques are expensive and time consuming. By contrast, this method potentially provides a rapid, objective and efficient method for evaluating the suitability of candidate sonifications for EEG neurofeedback.