Intel, do tell

I know I said no more meta posts, but just a quick explanation. I had a death in the family (second in a month), which happened literally the day after I posted last. So, things have been exceptionally crazy.

To tide you over, here is one of the more trumpeted pieces of BCI news.


Apparently Intel is now in the techno-voodoo-Kurzweilian prediction business, because they're saying we'll all be controlling our computers with brain signals by, and I quote, "2020". What's better is that they are basing this claim on fMRI findings. Is this reasonable, I mean, are we really that close to this amazing new world? In a word: OMGWTFNONONO. A) No one is going to wait a week for their BCI to magnetize. B) MRI requires that the whole head be subjected to a strong magnetic field, which means a gigantic, neck crushing apparatus. C) Power. Unless we are all driving out own buses filled with batteries, not even close. D) MRI. Magnets. Computers. Railroad spikes. Manhole covers. Nuff said. E) Temporal fidelity. Imagine playing a game with 400-500ms lag. That's also assuming there are massive improvements in processing and predictive algorithms, because right now it is between 3-10 SECONDS (deCharms, 2007). F) Consumer tech will push for better interaction, not more exotic. If I can control a mouse with my arm, why get a BCI that does worse? Things like multitouch at least provide something new beyond simple 2D control. G) Just, just... *facepalm*

BrainGate2 clinical trial is go

That's right! Everyone's favorite Neural Interface System, filled to the brim with Brain-Computer Interfaceyness, BrainGate, is continuing clinical trials, sans Cyberkinetics. Massachusetts General Hospital will serve as the home base, with close ties to Brown and Harvard universities and the Providence VA Hospital. Leigh Hochberg takes the helm as PI and lead investigator along with John Donoghue. The project is being called BrainGate2 in part to denote advancement in the systems, techniques and understanding used in designing and implementing an NIS. People looking for more information, including participant enrollment, can check out the below links and Braingate2.org.

Clinical trials page on BrainGate2.org website
ClinicalTrials.gov entry

Brown University release
GenEngNews article
Boston Globe article
Providence Journal article
PhysOrg article

Reddit catch
Twitter catches

BCI on House

If you love House, that great TV show starring Hugh Laurie as the cantankerous Dr. Gregory House, you should have noticed that they used a BCI in the most recent episode (season 5 episode 19). I was waiting for Hulu to have it up, but got enough emails and calls that I thought I would mention it today. Typical TV BCI, EEG-ish and mystical in operation, but the episode was pretty well done, with internal dialogue from the patient and no spelling of words as fast an able-bodied person could say them, but got a little cheesy. Still, this is an example of where an EEG system would make more sense than an implanted one. Cursor up for yes, cursor down for no.

Honda: Our in-house research can't read

Apparently, Honda has decided that they are the first to use a BCI (NIRS and EEG combo in this case) to control a robot (video here-and, oh, is it worth it). Not the first to use this technique, but the first to do it, ever. Okay, let's pretend that the Fetz 1969 experiments drove a 'device' instead of a robot (I'm not even sure how far back a device driven by surface electrode input might have been), but this is massively misleading. Sure this is just a stupid company statement, but Google "first robot controlled by a BCI" in two days and this crap will be the #1 result. *facepalm* I'm sure I can't really have an impact on such a large corporation's ridiculous press statement, so instead enjoy this video of ASIMO falling down stairs:



Thanks Katie for the heads up.

EEG consumer BCIs spanked by the Schwartzinator

Forbes has a fun little story on the state of consumer BCIs that hits on the major consumer players, but also throws real BCIs researchers John Wolpaw and Andy Schwartz into the fray. Each echos the officially sanctioned position of their respective scalp electrode versus implanted microwire/fancier probe electrode camp.

I prefer to call this the Head-Computer Interface versus Brain-Computer Interface debate. I try to be even keeled on the subject. EEG is obviously a worthwhile endeavor because there's none of that messy surgery involved, and thus appeals to a very large audience (at least in the near term). Hey, if you get a signal that can be controlled by the user reliably and with an acceptable level of precision, you have something useful. BUT, EEG is a degraded, garbled, sloppy signal prone to nearly limitless interference sources. Anything that can bork an implanted BCI can bork an EEG-based system, but EEG throws the doors open to the world because you are essentially wearing an antenna (or several).

One major problem with EEG is the fallacy that somehow destroyed information can be recovered by some form of fancy filtering. This is simply not so. Think of the electrode as a point of convergence for all electromagnetic signals of a measurable intensity. Even after narrowing frequency bands and implementing funky probabilistic decoders, any squiggle can be the convergence of several squiggles of indeterminable sources. In other words, at a specific time, a signal of amplitude +5 can be two signals of +4 and +1, -2 and +7, or 5 signals of +1-4-2+3+7. And don't even start with harmonics.

The second problem is population size. I know we like to think that motor cortex responds to only movement. It makes life easier. What life? Life in La-La-Land. MI responds to visual stimuli, movement preparation, auditory stimuli, imagined movements, movement related words (heard, internally rehearsed or spoken), reward, attention, cutaneous and proprioceptive feedback, and a bunch of other factors I'm not even mentioning. Until the impact of these influences is understood and quantified, there will not be any BCI that translates the neural activity for "move my arm to point x,y,z". EEG will never have the fidelity to isolate the differences feedback has at the single neuron level, so the nature of the recorded signal will never allow the 'direct' mapping of neural activity to output. That is, the activity that once gave rise to movement can never be harnessed with an acceptable degree of control to recreate the movement. Yes, I am using the word never. Never. There, I said it again.

Like I said, this isn't an attack on EEG, just reality. Different uses for different technologies. I can drive a car on a road. That doesn't mean I can drive a road, or that a road has no use.

BCI for the kids

Yup, it's finally happened. The Consumer Electronics Show (CES) is going on in Las Vegas, NV and I'll post a research/cool gadget wrap up later. But, I thought it was worth noting that there are not one, but TWO BCI devices being marketed towards children. Both use EEG, so I'm sure they won't offer a frustrating and imprecise experience (note the sarcasm).

The most notable is the Mattel MindFlex. The user wears a headband with what looks like electrodes over the temples (+ ground to the earlobes), and navigates a ball around an obstacle course. The folks over at Switched (part of the Engadget universe) tried it out and found it almost completely unworkable. More coverage here, here, and a video here.

The second is marketed as The Force Trainer, a Star Wars toy, and all it does is train you to raise and lower a ball in a tube. Wheee! Ball goes up! Wheee! Ball goes down! What will your child will do after he or she 'masters' this game or decides that it's boring? I don't know, but whatever it is, I'd bet you could find out within 5 minutes of giving them this. These Gizmodo reader comments pretty much sum it up:
simi510: does it work or is a gimmick? IMO both. It works as a gimmick.
hofmann: My kids would honestly play with that for 5 minutes and then beat each other with it, and they actually like Star Wars.

What IS interesting about these toys is that there will be a generation of kids that have this idea of using neural signals to control external objects. If you read this blog, you probably grasp the basic concepts behind BCI systems pretty 'natively', but ask your average Joe on the street, and chances are you would have to explain the most basic ideas repeatedly before he/she could wrap their head around it.

The second interesting bit to come out of this is that no one knows the long term impact of 'training' yourself to repeatedly activate particular rhythms. I'm not being alarmist - oh noz! Johnny is dead from over training his brain! - but if you had a kid play with one of these for days, weeks, months, what changes might occur? We know that repeatedly engaging in 'mind altering experiences' (meditation, drug use, chronic stress, etc) can alter behavior significantly. What happens when a kid eats, sleeps and breathes The Force Trainer is unknown. Would it make them more docile and calm, or unable to focus on critical thinking? I guess we'll find out!

Rockin tidbits

First, as I type, I'm watching the premier episode of Prototype This, and I gotta say I really like it so far. Check it out on Discovery Channel. As a bonus, the first episode features the Emotiv EEG BCI system. The task is to design a car that is driven by biofeedback (they started with the idea of shutting off a car in response to road rage, but it seems to be drifting towards the more fun idea of controlling the car's navigation). They also visit BioPac - which makes a number of biofeedback-related systems and sensors. Perhaps they should get one of these to add to the fun (sample video below).

The most important thing about this episode is that it highlights a concept that everyday people don't think about much, namely that any controllable signal can be translated into a computer input, and thus tied to anything a computer can control. This is a great concept to drive home for BCI enthusiasts and general hacker types. It also hits the real life issues introduced by using niche/'vertical market' devices. We know those problems all too well (and by we, I mean JDS, or as we call him "The BrainGate").



And a plane, which is the awesome-er-est:



Next, just a reminder that there are two available sites for news stories which are constantly updated .

For neuro/bci items:
http://www.google.com/reader/shared/user/12240293210094451962/state/com.google/starred

For tech items:
http://www.google.com/reader/shared/12240293210094451962

Required reading

Hitting several sites today is a new Fetz lab study where cortical recordings were used to directly drive paralyzed muscles in a monkey. More coming on the story later, but this will be one of those classic papers in the field.

Paper
Mentioned here, here, here, here, here, and here.

As a side note, we did something similar when I was working with Phil Kennedy. We hooked up a surface electrode to a patient's arm and used to the recorded signals to drive the stimulator. It was a very brief proof of concept type experiment, but worth noting in light of this study.

Consciousness obliterated

Birbaumer and Kubler have a paper in Clinical Neurophysiology on the idea that completely locked-in patients essentially slip off into unconsciousness once they become completely locked-in.

Objective: To investigate the relationship between physical impairment and brain–computer interface (BCI) performance.
Method: We present a meta-analysis of 29 patients with amyotrophic lateral sclerosis and six patients with other severe neurological diseases in different stages of physical impairment who were trained with a BCI. In most cases voluntary regulation of slow cortical potentials has been used as input signal for BCI-control. More recently sensorimotor rhythms and the P300 event-related brain potential were recorded.
Results: A strong correlation has been found between physical impairment and BCI performance, indicating that performance worsens as impairment increases. Seven patients were in the complete locked-in state (CLIS) with no communication possible. After removal of these patients from the analysis, the relationship between physical impairment and BCI performance disappeared. The lack of a relation between physical impairment and BCI performance was confirmed when adding BCI data of patients from other BCI research groups.
Conclusions: Basic communication (yes/no) was not restored in any of the CLIS patients with a BCI. Whether locked-in patients can transfer learned brain control to the CLIS remains an open empirical question.
Significance: Voluntary brain regulation for communication is possible in all stages of paralysis except the CLIS.

They examined 29 ALS patients (+ 6 others) who had been trained on a P300 based BCI system prior to the onset of completely locked-in syndrome. Once CLIS set in, they were no longer able to communicate using the system. Literature related to two other non-invasive techniques were used in addition to the P300 based system - low cortical potentials and sensory-motor rhythms. While performance across the board was "usable" (66% success is hardly usable, but I digress), once completely locked-in, patients' performance dropped to chance (except for one individual).

I have a better hypothesis. Use of an EEG based system becomes impossible once completely locked in. EEG relies on large scale, detectable changes in millions of neurons. Locked-in syndrome likely results in large scale changes to cortical topography, due to the lack of feedback - both reinforcing and resultant - resulting in slow wave potential differences unique to this, what might be called 'normalized', state. Nevertheless, an interesting and somewhat calming idea - no pain or negativity, you just drift off into nothingness.

The logical follow-ups are to take this up and down a level. As anti-fMRI as I am (not too strongly, but I think the studies waaay overstate their significance usually), throw them in the magnet and condition different modalities on different answers. Given a multiple choice question, choice A = imagine seeing fireworks, choice B = imagine hearing a train whistle, choice C = imagine throwing a baseball, etc. Just in testing the ability to before these imagined tasks should be enough to assess their ability to respond to a command. Then to provide some very rudimentary, slow communication, use the above association technique. And by down a level, I of course mean single unit recording.

A final thought. Since language is so tightly tied to the motor system, it is possible that communication becomes obfuscated by some severe, hybrid aphasia. If that is the case, the question becomes whether consciousness relies on language. And if that is the case, how does that change the way we see animals, robots and babies (or baby animal robots!). Just something to think about the next time you hippie liberal academics are smoking up at the drum circle. :P

AWOL-ish and thoughts on experimental design

I know posts have been sparse here (the Digital Trends people haven't heard from me in months, really) but like usual, I am being overly ambitious on large projects and having to justify it while secretly plotting even larger ones.

Really, it comes down to having left the research world for 4 years, I expected some basic experiments to be completed. They're obvious, elemental, and simple, but act as cornerstones to further development. They weren't. Honestly, when I returned to academia, I didn't know what kind of project to work on and after a few talks with my advisers, it became clear that next to nothing I perceived as the logical next steps was accomplished. I actually remember the exact moment when it hit me - during a meeting when I was asking for thesis project suggestions. I recall literally stopping, staring off across the room, and thinking that either a) it was really a verify this in greater depth type of idea or b) Wow.

It isn't 'bad' - the push has been towards taking state of the art techniques and philosophies built from monkey work and applying it to the human side. Makes perfect sense. The problem is that there should always be a parallel thread building on the basic/elemental side of the equation. Working in people is different from monkeys (obviously), but the two biggest differences, and it takes some serious hammering to get them into your skull, are that you have the ability to convey concepts (not just processes) through language and you have access to unobserved processes. That means:

a) No training is needed for task familiarization (not talking procedural memory or 'mastery')
a1) Replication of monkey experiments should be performed at breakneck speed.
(Erm, unfortunate pun, but I leave it in because it correctly conveys the idea. Just noting that I did catch that.)
a2) Papers based on replicating monkey experiments require less effort (tasks already coded, expected results and caveats already mapped) and report less significant findings (i.e. - they become "Worked in monkeys. Yup, works in people." comparative studies).
a3) Should require either a novel functional outcome ("We can therefore do X with a BCI.") or provide insight on a monkey-impossible phenomenon ("Think about grandma's apple pie while you do X.").

b) Almost any experience once relegated to the world of 'cognitive neuroscience' becomes admissible. I know what it means to "recall the scent of a new car", so do you. Most of us do. There might be significant variance in what that scent actually is, but in understanding 'the brain's response to imagined new car smell', we all know the subject/object of reference. Take out these accepted experiential elements because of the artificial distinction in neuroscience sub-fields, and I would argue that ascribing any neural activity to any unobserved process (preparation, reward, saliency, etc.) must be eliminated, too. Sorry, no ascribing unobserved forces to activity, since you have no ability to say with complete certainty that monkeys operate in a conceptual/cognitive manner similar to humans. You're just anthropomorphizing mental or neural processes based on statistically significant relationships in a constrained environment, and you're dissociating environmental cues, not internal representations. Does that colony of monkeys happen to be full of synaesthetes? How many times do you hear, "Yeah, the monkey just didn't want to work today"? Or, "Yeah, the monkey got bored."? So, the data is collected only when the monkey "wants" to work?

So, in short, human BCI experiments should incorporate 'cognitive neuroscience' techniques in order to enrich the functional/translatable significance of the outcomes. I love to use the example of crossing the street. A BCI user should be able to recall crossing streets, imagine crossing the street in front of them, tell a joke about a particular chicken crossing a street, point across the street, and want to get to the other side of the street without stepping into traffic. Otherwise, we're doomed to becoming glorious reflex arcs.

(And for extra credit, imagine a world in which Sartrian angst is always acted on. Scary.)

Anyhow, I think that the state-of-the-art and elemental streams need to always be examined in parallel (even better in concert, but tougher to design and control). And, having already spent too much time typing this, I'll leave to you to consider why that is (hint: there are practical, moral, and theoretical reasons). Enjoy the brain twisting. :D

Hochberg article in NEJoM

Leigh has an article in this month's New England Journal of Medicine related to The Project. Required reading, of course.

Damn you Hack A Day!

Looks like Hack A Day has a link to a complete teardown of the OCZ NIA. I'm not much of an electronics guy, but it looks like they almost have all the components pegged. Each little bit is labeled and it looks like there is a programming head (that mystery connector I noted), though no word on whether anyone has successfully written to the PIC.

Interesting to note, that the device is USB 1.1 only (well USB 2 'full speed'), meaning 12Mbps max transferred. Sampling rate is 3.9kHz, 24-bit. Like I mentioned, there is an amp that measures the difference in voltage between the outer electrodes, presumably for eye movement detection. Lots of other neat stuff, so have a look.

Another interesting NIA thing I found was a guy who has been 'training' with this and does a decent job of moving forward and shooting, though not much else. He, too appears to have grounding issues.

OCZ NIA gets nekkid

Some pics to whet your appetite (much better pics to follow)...

Meh on the construction. Look ma! No metal. Good thing they're not trying to conduct electri... oh wait...
The amp (sans warranty!)


3 Prongs for the headband piece. Probably the center channel grounded via USB and the side channels recorded differentially.

Note the mystery connector.

OCZ NIA in da house

It was here when I got home, so yay!


Initial impressions:
- Very nice packaging
- Decent build quality
- Obviously a work in progress on the hardware side (anytime I see holes in the frame that were obviously used for an external power plug, but lacking a connector behind them, I know things are in their version 1.0 form)
- Software isn't horrible, but definitely needs more options (for instance, the 'Pong' game used to train can only be played with the muscle signal)
- The glance control is abysmal (eye movement detection/EOG)
- Grounding is problematic - I could only use it is I had my hand on the box
- Amplifier overload is waaay too sensitive - if you wiggle out of the detection range, there is a lockout of several seconds
- Needs more and better manual controls
- Needs more and better feedback - depending on how you count them, there are 8 control signals, but you only get an analog trace for muscle activity

All in all, it is pretty much what I expected. Right out of the box it is nearly impossible to control, and you really need to sit with it for a good hour to get a feeling for how to modulate the signals. I will say that it has a decent amount of potential, but from what I can tell, it will not be replacing a mouse anytime soon.

For instance, once I began to get 'good' at the Pong game I started thinking of funny ways to bring up the device in lab meeting. Of course, that made me smile, which blew my control right out the window. When I realized what was happening, I tried to suppress the smile, which must have engaged antagonist muscles because it was just as bad. I had to relax and stare blankly at the screen, which kinda scared me.

Why? Because in order to use this interface, I had to be completely devoid of emotional reaction. Holy Dystopic Futures, Batman! Some nutjob will interpret this as ALL BCIs requiring this 'feature', but that isn't the case. Still, I couldn't help but think of zoned out emo teens staring blankly at their screens, and they disaffectively advance military robots on the other side of the world.

So, there needs alot more work on an adaptive filter for the NIA. That's a problem with most EEG (and less for EMG) devices, and really a requirement when compared to intracortical recording. Basically, EEG is listening in on millions of neurons, and those neurons have 5,000 - 10,000 synapses. The resulting slow wave rhythms are the aggregate signal of all that mess.

People like to use the stadium analogy - that you can tell what is happening in a sports match by listening to the cheer of the crowd and knowing who the home team is (EEG), but you get more information by listening a single person (single unit activity) - but that analogy has one major flaw. It ignores that there are signals of far more diverse frequency and amplitudes, and that inhibitory signals are part of the mix. In essence, you're listening to a stadium at 2,000 feet with an "As seen on TV" sound amplifier, with an unknown distribution of home-to-away fans, of which a random distribution have laryngitis or transplanted dolphin vocal cords, and many of which are watching other games on their portable TVs and cheering more for that game as much or more than the game you are observing. And then there are random nutjobs just running around screaming for completely unknown reasons. (Security must be very lax at this stadium!)

See why single units are better? No nutjobs. Oh wait, no, I mean more info per channel. Pull out 20 random fans and you might be able to figure out things like the score.

Anyhow, enough EEG ranting. If you can filter out all the garbage, which you can't, you could theoretically get a good signal. But, you can't. So devices like the NIA should be looked at purely from a functional standpoint. If I could do a little training, have the cursor follow my eye (for games, this would be the mouse/look control), assign two keys to muscle activity (forward and backward/W and S keys), and alpha waves to left click (shoot), I would be happy. As it stands, I can't. I can get some very basic muscle, poor neural, and essentially no EOG control. That makes me a sad panda.

I'll play with it some more over the week and post regarding any changes. Ah, operantly conditioned association... you will have to be my friend a little longer... (Hats off to the Fetz crowd!)

(I should mention that if I had to decide today between the OCZ Actuator and Phil Kennedy's system, I would go with Phil's. I never posted a review because of some concern regarding the use of lab items for review purpose, which I think is fair, but the Neural Signals device (EOG, EEG, EMG, piezo-switch, light touch switch) was really snazzy. Unfortunately, I can't seem to find a link to it anywhere on their page.)

Neural Impulse Actuator incoming!

Okay, screw OCZ's PR department. The dry electrode based Neural Impulse Actuator BCI aimed at gamers has been out a couple weeks now, so I ordered one. It should be here later today. I am not expecting much, but hey, I thought I should own the first consumer BCI. They run $150-200. Buy.com has the lowest price right now, @ $150.

Also, I coined a new word today: Neurd. :D

Welcome to Consumerism 2.0: No idea what we're doing

(Note: After writing I realized this was much more rambling than intended, but, meh, enjoy the stream of consciousness.)

Techdirt has an interesting blurb on the idea sure to make any transhuman giggle with glee and Kurzweil smile with approval. The article posits that planned obsolescence is dead. Apple was probably the last to really make a go with it using redesigned computer cases, a la iMac, but tech moves too fast and we live in a global market, making exposure and availability non-issues in product design. Also, any real improvements in the overarching technology is announced far enough ahead of time so that OEMs can queue up to the the first to offer them, and competition is very much alive and well, leading companies to show their cards as soon as they have reasonable certainty that the products or services are feasible.

Anyhow, it's an interesting point to consider. I don't think the issue is that tech development or discoveries are happening too fast for the market, I think the nature of the markets and the complexity of 'reference' products (ie - instead of traditional metrics like crop harvest, we've added PC sales, cell phone subscribers, etc to the mix) has fundamentally changed. Hell, look at HDTV. Market penetration is just starting to hit a level of real significance, and even with that, how many people that you know of have a Blu-Ray player? How many HD channels are there on your TV? Undoubtedly a large part of this slowness to adopt is due to outdated business models making their last stand (eg - RIAA and MPAA) or crazy departures from traditional models with hopes of hitting the right plan (eg - the Internet bubble burst of 2001).

Maybe it is that the nature of the Internet is more Socialist than Capitalist. Maybe information itself is like this, or human social tendencies. People try to peg down what makes modern culture such a departure from the past, and they all seem to converge at 'faster communication'. True, but you can't ignore that communication rarely stops at any one receiver - it propagates through the system/society. I guess what I mean is that there is no grape vine anymore, just a series of babbling guys and gals with megaphones.

So, why is this on a BCI blog (yes, I know I've been wandering anyways lately, but there is a point)? Because BCI design requires traditional tech, which moves by the above mentioned mysterious means, and basic science, which has not fully embraced this new paradigm. Academia as a whole hasn't, and the approach to a product's design versus understanding the underlying nature of it are in complete discord. What's more, initiatives like the NIH's push for translational research over basic science don't help. At best they take scientists uncomfortable with the idea of making use of findings and slowly nudge them to thing of practical applications, but what it really means is a nice rewrite of their grant's intro and discussion.

To make matters even more difficult, the academic market in science is nearly uncustomizable - what works for molecular biology doesn't for engineering. Who in their right mind would develop a product to be used by 15 labs? Sure you could charge through the nose, but eventually, you'll lose out to budget concerns or lack of manpower to implement the one new feature that 4 of those 15 labs are requesting. Try to get things rolling with a big company and wait for cries of Big Corporation sullying the good name of academia. Try to get a more science oriented company to take on a problem and you get EndNote. Sure you CAN use EndNote for keeping track of publications, but that has got to be the most obnoxiously buggy, overpriced, and generally annoying program I've ever been forced to use. I fricken HATE EndNote. But it isn't all Thompson's fault - why the Hell do we still care about publications? Are we still at the point where we need magazines to tell us what is hot is science for the Fall Season? Oh, and how many other publications are there out there where YOU pay the PUBLISHER for accepting your content, once reviewed by others? "We're not even sure this is very good, so we'll send it to some other people in your field. We won't pay them, bu they will tell us if it is worthy, and then maybe you can pay us a couple thousand dollars, and we'll add it to the magazine we make a profit off of." None, and publishing in every profession serves the same purpose - career advancement/recognition. (I'll give J of Neuroscience a pass, since it funds a society that supports meetings, but it is still a ridiculous model.)

I'm starting to get off topic here, but the idea in general is that basic science needs a revolution in the way it is harnessed, supported, and integrated into companies or collaborations between researchers and companies. Basic science negates the possibility of planned obsolescence by its very nature, and the technologies that support it do as well. The result has been many smaller companies with highly specialized products at high prices. That's why I predict that BCI development will remain the territory of small, highly specialized companies. Regardless of what the numbers say about the number of people that can use one, the pace of development is set by basic science, not mass market technology, which incurs high costs for low material gains. It is a Lose-Lose model right now because no one knows how to push research into the next major phase that I think all researchers know they are on the cusp of, nor do we know how to operate a traditional company in the modern/connected world (at least, not responsibly). Most people assume that you develop a device, some magic occurs, and you are off and running a business, but that is becoming less and less the case.

Don't get me wrong - there are plenty of chances to spin science into a viable company that makes a few people rich and a bunch happily employed. But, the chances that it will change the world are slim to none because the means simply aren't there. As the death of planned obsolescence has pulled some of the old tricks of product design out from under the businesses that relied on it, science based business has suffered from stagnation, which may be a boon. Rather than scramble reactively, we're thinking of ways to push past overspecialization. And I know just how we'll do it... but you'll have to subscribe to DirectNeuralInterface v2 for the low low monthly rate of just $2000 to find out ;)

Monkey self feeding BCI control

Alright, I get it. Several people have sent along links to Andy Schwartz's Nature publication, so here's the info. Thanks to Remy Wahnoun and Natalia "Get Moose and Squirrel" Bilenko for doing my work for me and compiling the links, and being the first to send info along.



http://www.nature.com/nature/journal/vaop/ncurrent/full/nature06996.html
http://news.bbc.co.uk/2/hi/science/nature/7423184.stm
http://technology.newscientist.com/channel/tech/dn14000-robomonkeys-use-brain-power-to-grab-a-bite.html?feedId=online-news_rss20

The work reported involves real time decoding to drive a robot arm, which fed the monkey. I haven't had time to run through the whole thing, so here's the "shoot from the hip" version of my thoughts. I'll update these

Interesting points:

- The monkey was trained to use the arm with a joystick, and followed by, the part I thought was very interesting, various levels of pre-programmed control assistance while using brain signals. I will expand on this later, but it is an interesting task design for several reasons.

- Having seen some of the raw video at the Neurobotics 2007 workshop, there is much more obviously correlated arm motion. They have an entire, substantial paragraph devoted to this, since they probably knew it would be a criticism.

The three points they make are:
1) Movements were with the arm ipsilateral to the electrodes.
2) The movements were delayed by up to a second.
3) Moving is required, since BCI control has been demonstrated before without it.
They cite some supplementary info and write, "... monkey’s hand movement was only loosely coupled to prosthetic control." The video I saw before tells a slightly different story. There was a pulling movement 100% of the time (every trial), and when the arm didn't reach the monkey's mouth it would 'paw' or 'dig' (repeated movements similar to pulling an invisible object toward the body) repeatedly until the arm reached the monkey's mouth. You can see this pretty clearly in the video linked above (watch the hand in the plexiglass tube). Yes, there is a delay, but the whole situation would be interesting to examine. It may not require the 'pulling' at onset, but monkeys very often use both hands to eat. Those of you with monkeys, go give them a treat. They will grab it with they right hand (usually), and then pull it towards their body and engage their left hand as they do so. Feeding it almost always a bimanual task for macaques, explaining the ipsilateral activation and the delay. Reason #3 above is kind of a brush off reason - hey, movement is needed to use a BCI, so no big deal. The three cited papers are all in humans, all were tetraplegic, all lacked some degree of propriocentive feedback, two didn't use PVA or multielectrode arrays, none used an arm, none had confirmed the lack of covert movements with EMG, and the list goes on.

- The trial was very organic and continuous, which is great. From my brief skimming, it looks like simultaneous control of the hand aperture needed to be maintained throughout the task, hinted at when they observe the the hand slowly opening along the path to the food.

In general this is a nice, brief summary with some very interesting points (perfect for a Nature Letter). Obviously they are preparing some studies, and this was released to wet out appetites. The 'organic' nature of the task will open it up to many criticisms, but more studies like this are needed in order to understand the dynamics in play when dealing with a experiences that lack built-in boundaries and environmental awareness. Definitely good thought fodder.

BCI news items

First, pointed out by the lab, Nicolelis is in the news again. Monkey brains in North Carolina controlling walking robots in Japan. Nice. Again, dissociation of neural activity and decoder performed (monkey stopped walking, but could keep the robot walking while remaining still).
No mention of a paper with some objective measures, but a pretty decent write-up available on the NYTimes site. There's even a nice schematic and video.

Interview by PhysOrg with EEG mogul Fatourechi, related to a paper published in JNE. Take a gander here.

Hot off the press!

And just when I went back to dig up more stories, this hits the front page:

Ted Berger, rocker of many socks, has released a report on the state of BCI. It is long, it is free, and I have only skimmed it, but good stuff. Here's the breakdown:

The report contains three overall findings on Brain-Computer Interface (BCI) work worldwide:

-- BCI research is extensive and rapidly growing, as is growth in the interfaces between multiple key scientific areas, including biomedical engineering, neuroscience, computer science, electrical and computer engineering, materials science and nanotechnology, and neurology and neurosurgery.

-- BCI research is rapidly approaching first-generation medical practice—clinical trials of invasive BCI technologies and significant home use of noninvasive, electroencephalography (EEG-based) BCIs. The panel predicts that BCIs soon will markedly influence the medical device industry, and additionally BCI research will rapidly accelerate in non-medical arenas of commerce as well, particularly in the gaming, automotive, and robotics industries.

-- The focus of BCI research throughout the world was decidedly uneven, with invasive BCIs almost exclusively centered in North America, noninvasive BCI systems evolving primarily from European and Asian efforts. BCI research in Asia, and particularly China, is accelerating, with advanced algorithm development for EEG-based systems currently a hallmark of China's BCI program. Future BCI research in China is clearly developing toward invasive BCI systems, so BCI researchers in the US will soon have a strong competitor.

It is over 200 pages, and at first glance a really nice report of the state of the art. (Several other very prominent authors, as well.) Found the report news release here, and the actual reports (free to everyone!) is here (6MB pdf). For anyone asking, "Could you recommend a good general overview of the field?" this would be it.

The Neurosky is falling!

Into fun-filled children's toys, that is! Everyone's favorite EMG, I mean no, wait TOTALLY EEG from one dry electrode because they have either signed a pact with Satan or they aren't using brain signals, has been in talks to integrate Neurosky into several toys designed by Sega Toys.

Now, I love me some BCI in the consumer marketplace, but this is starting to get a little out of hand. It was funny when Emotiv showed off the Cap of Infinite Dorkiness(tm), but Neurosky better post some concurrent EMG recordings that show they are actually using EEG before I believe it (I'm talking wet electrode, established 'real' techniques). Until then, I'll say it again, you are only getting covert muscle movements, not EEG. There's a reason EEG studies use 128 electrodes with skin abrasion and conductive gel. Oh wait, no, it is kinda fun to torture the volunteering undergrads, but no, no, it does have a real purpose.

Bah! Anyways, here's the media blitz. Here here here.