Developing an EEG-based Instantaneous Neural Feedback System for Neuroadaptive Research

Professor Ouyang, Guang   (Principal Investigator (PI))

Dr Dien Joseph   (Co-Investigator)

Dr Lorenz Romy   (Co-Investigator)

42

2021-10-01

774060

Developing an EEG-based Instantaneous Neural Feedback System for Neuroadaptive Research

Bayesian Optimization, EEG, ERP, Neuroadaptive Research, Online neural feedback

PsychologyNeuroscience

Humanities & Social Sciences (H)

17609321

General Research Fund (GRF)

2021

On-going

1 To develop an instantaneous neural feedback (INF) system based on brain responses. One way to assess cognitive states is to observe the patterns of neural activation in response to external stimuli. The brain’s response pattern has been shown to reflect the cognitive process more precisely than the spontaneous activity, which contains more irrelevant activity and heterogeneous components. However, obtaining a reliable brain response pattern to a specific stimulus requires a highly complex routine comprising design, set-up, and sophisticated offline processing, typically costing hours of labor. Conventional cognitive research segregates raw data collection and data processing into two organizationally independent endeavors. The aim of this project is to develop and validate a system that can reliably obtain various aspects of brain response patterns in an instantaneous manner. The instantaneously obtained brain response patterns will then serve as indicators of cognitive states, providing feedback to support innovative research or practices, e.g., allowing the development of neuroadaptive research paradigms. The features of the instantaneously obtained brain response patterns will include average event-related potentials (ERPs), latency-jitter-corrected ERPs, trial-to-trial brain response variability, and event-related spectral perturbation. These multifaceted brain response features will be integrated into the system and visualized in a user-friendly way. 2 To validate the utility and effectiveness of the INF system in neurocognitive researches that incorporate neuroadaptive paradigms. In the majority of conventional neurocognitive research paradigms, neither the participants nor experimenters have access to the pattern of brain responses during data collection, as preprocessing requires a high degree of sophistication. There are several disadvantages to the segregation of online data collection and offline processing and analysis, one of which is that it does not allow neuroadaptive experiments to be conducted. For example, it would be interesting to know how feedback of a participant’s own brain response modulates top-down attentional control, which may affect subsequent cognitive processes, or how it might enhance cognitive and learning processes. Therefore, a system capable of providing both instantaneous and reliable neural feedback would have profound significance, although it would need to undergo a strict and thorough process of validation before it could be used in scientific research. To demonstrate the validity, feasibility, and benefits of brain response pattern-based neural feedback in neurocognitive research, we will design strictly controlled experiments. The first experiment will use cognitive task paradigms to investigate the effects of neural feedback on cognitive processing, and the second will examine how reliable the brain response pattern indicates cognitive load in real time. 3 To apply a Bayesian optimization framework for searching individually optimal experimental parameters to maximize the stimulus-evoked neural response characterized by event-related potentials. The relationships between the features of a stimulus input and the brain’s response are extremely intricate and high-dimensional. In sharp contrast, the scope of a conventional, strictly controlled cognitive experiment is extremely narrow, meaning that such experiments have limited efficacy and efficiency in revealing complex brain-stimulus relationships and their underlying mechanisms. Bayesian optimization offers an efficient way of making inferences about stimulus-brain response relationships based on active data sampling in an online mode, which requires online data acquisition of brain response patterns. Therefore, one of the advantages of building the INF system is to enable the application of a Bayesian optimization framework to leverage neuroadaptive research, which is one of the major objectives of the proposed project. Another obstacle in neuroimaging-based cognitive research is the individual differences in brain-stimulus relationships, which leads to difficulties in analyzing the results using simple methods that assume a high level of cross-subject homogeneity. In this regard, the Bayesian framework will also be applied to efficiently search the individually optimal parameters to generate the strongest neural response activation, which may help to tackle the issue of individual differences in cognitive research that examines between-group effects. 4 To provide training for researchers to help them to acquire the knowledge and skills needed to conduct neuroadaptive research. At a late stage of the project, the intellectual materials collected will be assembled and organized for educational purposes, i.e., to train researchers in the theoretical and technical aspects of conducting research studies based on neuroadaptive paradigms. The intellectual materials will include the theoretical fundamentals of neuroadaptive research, the principles of the methodologies concerned, the structure of the system and data-processing pipelines, and the training data. Theme-focused training workshops will be organized to conduct the training activities.