Posture and VR in Passive BCIs

Passive BCIs have captured the imaginations of researchers and innovators for 15 years now, envisioning a world where humans can implicitly, and thus seamlessly, communicate with machines. While extensive scientific literature exists on passive BCIs, practical commercial implementations are still in their infancy. That's why we are thrilled to present the findings from a recent study that sheds light on the real-world potential of passive BCIs by looking at the impact on their capabilities when combined with an upright posture and VR (virtual reality) head-mounted displays.

In this study, led by Diana Gherman, we aimed to understand the impact of posture and task presentation modality on the EEG signal and its implications for passive BCI performance.

‍

In a carefully designed experiment, 22 participants underwent tasks while their EEG data was recorded. Each experienced four conditions: computer screen modality in sitting posture (CS-sit), computer screen modality in standing posture (CS-stand), virtual reality modality in sitting posture (VR-sit), and virtual reality modality in standing posture (VR-stand). Our well-established paradigm involving using mental calculations as a means to generate mental workload, as well as periods of relaxation, was presented either on standard computer screens or through virtual screens in a VR environment. Muscle activity, signal-to-noise ratio, and frequency band changes were analyzed to examine the impact of posture and modality on the EEG signal.

The results unveiled intriguing insights into the interplay between posture, modality, and the EEG signal. The influence of posture on parieto-occipital alpha and frontal theta signals, vital for workload assessment, was marginal. While we did find muscle activity contributing to signals of interest, our spatial filters were effective in eliminating these artifactual contributions. This suggests that despite variations in muscle activity, the fundamental signals utilized for workload detection remain consistent.

Another significant aspect investigated was the impact of modality, specifically the use of VR headsets, on the EEG signal. The study revealed distinct patterns of noise associated with VR modality compared to traditional computer screen modality. This reinforces the importance of understanding and managing the additional disturbances brought about by VR technology to ensure reliable and accurate BCI performance.

Remarkably, despite the observed signal variations caused by posture and modality, the study's findings showcased the robustness of simple calibration algorithms used in passive BCI classifiers. The results demonstrated consistent classification abilities between modalities and postures. This implies that the core features utilized for workload detection are less influenced by posture and modality variations. However, further research is needed to explore the impact of different types of features and domains on passive BCI performance in dynamic contexts. That is something that our colleague, Dr. Marius Klug, is highly interested in and actively exploring.

The implications of this study extend beyond the laboratory environment and into the realm of practical BCI applications. By understanding the contextual feasibility of BCIs and the underlying signal variations, we can unlock the true potential of non-invasive BCIs in real-world scenarios. This knowledge serves as a foundation for future advancements, enabling us to refine and optimize BCI technologies for diverse user populations and dynamic environments.

By continually expanding our knowledge and pushing the boundaries of innovation, we can harness the power of passive BCIs to revolutionize human-machine interactions, which is one of the main goals Zander Labs is dedicated to reach in the upcoming years.

We invite you to join us on this adventure! Let's collaborate and shape a future where the mind and technology converge seamlessly, empowering individuals from all walks of life.

‍