trends in bcis for usdesa/brendanslides.pdf · and 2011 “trends in bci research”. video...
TRANSCRIPT
Trends in BCIs for us
Brendan Allison, PhD
University of California, San Diego
COGS 189
January 8, 2019
Brain-Computer-Interface (BCI) components
BCI components (Allison, 2011)
Applications: RISER model
From Wolpaw and Wolpaw (2012); since widely adopted
This broadens the applications of BCIs from the
definitions in the 2010 “Gentle Introduction” chapter
and 2011 “Trends in BCI Research”.
Video courtesy of G. Schalk.
Work is from Schalk et al., 2017.
Common view
BCI/AN is for users with disabilities
Sci-fi view
BCI and AN is from, and for, evil
Star Trek (1966)
This is still among the mostrealistic and altruistic
portrayals of BCIs.
Widespread BCI adoption requires replacing
conventional interfaces for conventional
users in conventional settings.
Common view
TU Graz,
2010
Replacing or supplementing conventional
interfaces for conventional users in
specific settings.
→ Any user may be like a disabled user due
to situational disability and laziness.
People might enjoy a BCI in some settings
just for fun.
Emerging view
When a healthy person cannot use some natural means of communication and control in a certain situation.
Leeb et al (2007), Faller et al (2010, 2017), Aloise et al (2011) – virtual navigation
Middendorf et al (2000), Trejo et al (2006), Menon al (2011) – pilots or astronauts
Pineda et al (2003), Nijholt et al (2008, 2011,2016), Navarro et al (2011) – gamers
Allison (2008, 2010, 2011), Millan et al (2010), Zander and Kothe (2011) – other examples
Also cellphone users, mechanics, surgeons, soldiers, drivers aka Tesla seatwarmers
And (sadly) disability by laziness: remote control or cellphone users
Trejo et al (2006) Scherer et al (2008)
Sit. Disability
Widespread BCI adoption requires
dramatically new capabilities, such that gel-
based wired “ugly” systems are appealing.
Common view
Disruptive TechnologiesPractical electrodesWireless systems
Emerging view
Cosmesis
PERCEIVED Cosmesis
Common view
Schalk (2008)
Moderately disabled
Eager healthy
Mainstream healthy
Speed = critical for broader adoption
Emerging view
Ease, design, utility = key for broader adoption
• Improved hardware (esp. wireless, dry electrodes)
• Easy, available software (BCI2000 and others)
• Easy, available systems and classes (Hack-a-thons, majors)
• Improved math and signal processing
• Improved knowledge of EEGs and cognitive correlates
• Wearable computing ubiquity
• Perceived cosmesis
• Positive media coverage of AN and related systems
Catalysts
Catalysts of Broader Adoption
Why not use something else?• Faster than an unavailable interface
– That requires impractical hardware
– That cannot be easily used
– That would take longer to provide the same information (if at all)
• Easier to use than other interfaces– More portable, accessible, or convenient in real world settings
– Induced disability is major, even from laziness (TV remote control)
– More natural and intuitive
– Less training?
• The only interface capable of total privacy?
• May seem novel or fun
How much useful additional information would remain in the output of the brain-, other body-organ- , or behavior-based workload gauges after regressing out variation directly measurable [from photodiodes, a software agent counting mouse clicks, and a microphone]?
Commentary on DARPA AugCog report by Gevins and Smith, 2003.
**Note: They referred to workload monitoring here.
Why not use something else?
• “Enhance” = User experience in games
• “Supplement” = Google Glass (this was 2015)
• “Improve” = Upper Limb therapy after stroke
• “Research tool” = Cognitive neuroscience research
Case Scenarios
Image Triaging• Images shown to a user 10x per second
– EEG can identify a minority of images that are of interest
– Could reduce time for photo analysis work
From Paul Sajda’s group
Neurofeedback• Has a “bad rep” due largely to unethical claims
– Done properly, NF is good for (at least) relaxation and attention
– Many, many companies offer NF.
– How can people identify the best ones?
From company websites
“Lie” detection• P300 and related signals can detect image familiarity
– Does the accused have “guilty knowledge”?
From Farwell and colleagues
NeuromarketingCan detect how focus groups react to sounds/images
Error detection• ERN = Error-related negativity
– Real-time error correction requires good single-trial performance. Most systems are integrated with BCIs – not other interfaces
– Non-real-time HCI, usability testing
– Still a popular topic at the BCI Meeting 2018
From Schalk et al. (2000)
Alertness monitoring• Based on only 2 EEG channels, 20 years ago
– Red = missed targets; green = hit targets
– Blue = actual error rate; red = predicted error rate
From Scott Makeig and T-P Jung
Task adaptation
From Alan Pope’s group
EEG-controlled World of Warcraft (g.tec)
4 controls:
Turn left, right, move forward, perform actions like grasping
objects, attacking other objects
60 Hz LCD display with 15, 12, 10 and 8.75 Hz.
BCI overlay based on OpenGL –
can be used with any graphics application
Game adaptation
From Ewing et al. (2016)
Necomimi
Brainball
Star Wars Force Trainer
Cockroach racing
Towards “Big BCI”
BCI Meeting
2018
Sponsors
Emerging “Big BCI”
Galvani
NeuralinkFacebook
Emerging “Big BCI”• Facebook
– Announced a noninvasive 100 word-per-minute BCI.
– They announced at our 2018 BCI Meeting that it will be in 10 years.
– What’s new: BCI? Optical imaging? Natural language processing?
• Neuralink– New effort popularized by Elon Musk
– Focuses on invasive approaches, including patients
Good progress on BCI training and infrastructure, but we need more:
• Classes, majors, degrees focused on BCI, neurotech
• Many classes on BCIs; NCAN summer schools
• Publicly available training videos, websites, exercises, lectures
• Quite a lot available online
• Open Access software for BCIs
• BCI2000, OpenVibe, OpenEEG, BF++, others
• Official conferences and societies
• BCI Conferences, BCI Meetings, BCI Society!!!
• Official journals and special issues
• BCI Journal, Special Issues in Frontiers
• Make BCIs fun and available
• Museums, hackathons
Public activities for students and enthusiasts to make their own BCI applications!
Wolpaw et al., 2002
Thanks to Prof. De Sa, and to all of you for coming!
For questions, including image credts, please email me: [email protected]