1) The document discusses using augmented and virtual reality technologies to measure and synchronize brain activity between individuals.
2) Several studies have found that people's brain activity patterns can synchronize when they perform tasks together or interact socially.
3) The author proposes using EEG sensors integrated into VR headsets to measure brain activity during collaborative VR tasks and explore how VR environments and cues could enhance inter-brain synchronization.
4) Simulating brain synchronization between a human and virtual agent using EEG and computational models is also discussed as a direction for future research.
Hyperautomation and AI/ML: A Strategy for Digital Transformation Success.pdf
AR AND VR FOR BRAIN SYNCHRONIZATION
1. USING AR AND VR FOR
BRAIN SYNCHRONIZATION
Mark Billinghurst
May 30th 2019
AWE 2019
2. HMDs with Integrated EEG
● A number of HMDs available with integrated EEG
● Neurable – 6 electrodes in visual cortex
● Looxid - 6 electrodes frontal lobes
Neurable Looxid
3. AR/VR and Brain Activity
• Brain Computer Interfaces (BCI)
• Using brain activity to control AR/VR experiences
• Emotion/Cognitive Monitoring
• Measuring emotion, stress, cognitive load
• Enhancing Collaboration
• Brain synchronization
4. AR/VR and BCI
• Using brain activity to interact in AR/VR
• Control by thought, user trained response
• Hands free interaction
8. Adaptive Training
• Training content dynamically changes
depending on the learning state of the user
• Task performance
• Test performance
• Subjective measures
• None of these measure users mental activity
or cognitive load during the task
9. Background
● VR training tools are effective and skills transferable to real world
● Earlier use of EEG in VR studies
○ Interaction, Measurement, VR environment did not adapt
Zhang et al. [2017]
- EEG for cognitive load
- VR driving simulator
- Autistic children
Friedman et al. [2007]
- EEG for locomotion
- CAVE system
10. Motivation
Making VR training systems adaptive in real-time to the
trainee’s cognitive load to induce the best level of
performance gain
We are first to explore real-time adaptive VR training
systems using workload calculated from EEG
15. User Study
● Participants
● 14 subjects (6 women)
● 20 – 41 years old, 28 years average
● No experience with VR
● Measures
○ Response time
○ Brain activity (alpha power)
• 5 minutes fixed trial time
17. Results – Response Time
Increasing levels
Response Time (sec.)
No difference between
easiest and hardest levels
18. Results – Time Frequency Representation
• Task Load
• Significant alpha synchronisation in the hardest difficulty levels
of the task when compared to the easiest difficulty levels
Easiest Hardest Difference
19. Key Finding
Similar reaction time but increased brain activity
showing increased cognitive effort at higher
levels to sustain performance
20. Conclusions
• Adaptive VR training can increase the user’s cognitive load
without affecting task performance
• First demo of the use of real-time EEG signals to adapt the
complexity of the training stimuli in a target acquisition context
• Future Work
• Significantly increase task complexity
• Can predict user performance based on the cognitive capacity
• Using AR display
• See real world and more distractors
23. Hyperscanning
• The simultaneous acquisition or
recording of neural activity from two
or more individuals
• Generally concerns the study of how
two or more individuals interact in a
co-operative or competitive scenario
25. Brain Synchronization
Phase extracted from the signals
Time course of normalized EEG signal filtered in the alpha-mu
frequency band for the channels P8 of both subjects *
* Dumas, G., Nadel, J., Soussignan, R., Martinerie, J., & Garnero, L. (2010). Inter-brain synchronization during
social interaction. PloS one, 5(8), e12166.
Behaviour
27. Previous Research
• Active research since 2002
• > 140 research papers
• Excellent review
• Wang, M. Y., Luan, P., Zhang, J., Xiang, Y. T., Niu, H., &
Yuan, Z. (2018). Concurrent mapping of brain activation from
multiple subjects during social interaction by hyperscanning:
a mini-review. Quantitative imaging in medicine and
surgery, 8(8), 819.
• No hyperscanning research in AR/VR
28. Previous Studies
TASKS EXAMPLES
Imitation task Finger tracing, meaningless hand movements
Coordination/joint task
Rhythmic finger movements, unconsciously
synchronised footsteps,
Eye contact/gaze tasks Social interaction
Economic games/exchanges Prisoner’s dilemma, Trust Game
Cooperation and competition
tasks
Visual search task, Cooperation between Lovers vs
Strangers, children and parents vs strangers, same or
different gender
Interactions in natural scenarios Daily life conversations (face to face: higher sync)
29. Example Results
• Evidence suggests Hyperscanning is able to measure inter-brain synchrony
Tang, H., X. Mai, S. Wang, C. Zhu, F. Krueger and C. Liu (2016). "Interpersonal Brain Synchronization In The Right Temporo-Parietal
Junction During Face-To-Face Economic Exchange." Social Cognitive and Affective Neuroscience.
30. Benefits of Brain Synchronization
• Several potential benefits
• Improved engagement in learning
• Improved feeling of “Flow”
• Better collaboration performance
• Increased trust
• Great Social Presence
• Better group social dynamics
31. Study 1 – Finger Tracking
• Repeating classic study
• Users track opposite fingers
* Kyongsik Yun, Katsumi Watanabe, and Shinsuke Shimojo. “Interpersonal body and neural synchronization
as a marker of implicit social interaction”. In: Sci Rep 2 (2012), pp. 959– 959.
36. Hypothesis
• VR can reproduce Face to Face Brain Synchronization
• AR/VR cues could be used to enhance Synchronization
• Viewpoint sharing
• Using shared virtual cues
39. Brain Synchronization Simulation
• Using a computational brain-inspired Spiking Neural Network Architecture - NeuCube
• Modelling the brain synchronization from finger tracking
EEG data is recorded over time
(c) Mapping, Learning, Pattern Visualisation and
Classification
(a) Spatio-Temporal Input Data Stream (b) Data Encoding
EEG
recording
Computational Modelling of Data in
a 3D Brain-Inspired SNN System
…
…
…
Class A
Class B
Class C
Class N
Output Class
Classification and Knowledge Extraction
Converting EEG signals
into sequence of spikes
40. Before Tracking
Left Finger
Participant 1
Before Tracking
Right Finger
Participant
3D visualisation 2D visualisation Input Interactions
3D visualisation 2D visualisation Input Interactions
(a) (b) (c)
(a) (b) (c)
41. After Tracking
Left Finger
Participant 1
After Tracking
Right Finger
Participant 2
3D visualisation 2D visualisation Input Interactions
3D visualisation 2D visualisation Input Interactions
(a) (b) (c)
(a) (b) (c)
42. Experiment A Accuracy Per Class
%
Total Accuracy
%
EEG Data Classes Class 1 ( AL-P1) Class 2 (AR-P2)
80
Class 1 ( AL-P1) 5 0 100
Class 2 (AR-P2) 2 3 60
Experiment C Accuracy Per Class Total Accuracy
EEG Data Classes Class 1 ( CL-P1) Class 2 (CR-P2)
60Class 1 ( CL-P1) 3 2 60
Class 2 (CR-P2) 2 3 60
Classification accuracy of 10 EEG samples (5 samples per class) using leave one out cross validation method.
The classification accuracy correspond to the similarity between the two models – the higher the classification
accuracy – the more different the models are.
44. Conclusions
• Opportunities for EEG use in AR/VR
• Brain Computer Interaction
• Cognitive Monitoring
• Adaptive Training
• Enhancing Collaboration
• Brain Synchronization
• Many directions for future research
45. Simulated Synchronization
• Simulated brain synchronisation
with a virtual character / avatar
• Humans interaction with virtual
agent e.g. BabyX
• Measure human EEG and
simulate virtual character EEG
46. Technology Trends
• Advanced displays
• Real time space capture
• Natural gesture interaction
• Robust eye-tracking
• Emotion sensing/sharing
Empathic
Tele-Existence
47. Empathic Tele-Existence
• Move from Observer to Participant
• Explicit to Implicit communication
• Experiential collaboration – doing together