BCI lets completely “locked-in” man communicate
arstechnica.com
arstechnica.com
My grandmother would cry (tears would flow) when some family members came by and spoke about her condition in front of her. My mother would jerk her hand away from my father's when he was too fidgety with her. It would appear that they were both experiencing some form of consciousness, just trapped and likely very scared. So I took care of the swatting away of others and spoke softly to them to let them know they weren't alone.
I wish they could have expressed their needs in that moment.
I dunno, the request for Tool, very loud gets me a little misty eyed.
Unfortunately it seems like this kind of thing pops up every 5-10 years so I’ve become quite skeptical of it.
One might worry that a good enough heuristic (i.e. word completion) will generate sensible sentences even from noise input. However, in this case, he is typing his son's name.
The difference here may be that I think in many of the cases I’m thinking about it might not have been locked-in so much as people thought to be in comas or with other serious damage he where the truth is they probably weren’t capable at all.
That doesn’t seem to be the case here. He seems to lack the physical ability to communicate but I’m not sure there is a reason he would lack the mental ability.
Having said that, I've just skimmed the Nature article. But what stands out is that they had 86 days of unsuccessful communication before they changed strategies and then all of a sudden started reporting success. I really hope this works but we should remain skeptical for the possibility of abuse here. Other commenters in this thread have reported that the researchers have committed misconduct in their brain research.
The only thing I fear more than being in the discussed patient's situation is being in their situation and having my thoughts incorrectly interpreted at best or outright manipulated at worst.
For example:
Make one person roll a dice and tell the number to the patient ensuring that nobody else can hear it. Kill the person that roll the dice to ensure the result is secret. Make another person read the BCI and write down the result. Repeat this 18 times and then look at the video and compare the dice and the BCI. It's a success if they guessed correctly at least 9 times. The expected number of correct answers is 3, and the probability to guess 9 randomly is only 0.1%.
If they do this only once, live streaming the experiment it is enough for me, but perhaps it's better to have a lower probability for a random guess. Also, I like that the "success" threshold is one half of the experiments. It's easy to explain, not too much in case it's not perfect, and less than a half is not impressive.
This reads like an excuse to kill 18 people...
The dude who was tortured with 12 years of Barney would like a word with you
As of today we barely have a good grasp on treating viral illnesses (and even bacteria is coming to bite us back). I don't foresee any effective treatment for this type of illness in the next few decades.
Touch screens are also low fidelity input devices. The lessons learned on streamlining BCI inputs could have wider applications. Like building a web app over dialup.
In particular, the second video shows sessions where he is spelling out words with the audio-interactive system
https://www.sciencedaily.com/releases/2010/09/100907071249.h...
2014: https://www.technologyreview.com/2014/07/09/172124/a-speech-...
There were immensely innovative studies in brain research all over the past 50-60 years, but this is still breathtaking.
But through a lot of hard work, we gave someone living in a black mirror-style horror the start of being able to communicate again.
Give it 20 years and I'm sure we'll have a standard box to stick on someone's head to let them communicate reasonably well (or at least enough to order a cold beer).
IMO it is less about which purpose this specific device is used for, and more about what this breakthrough has the potential to enable us to do in the future, using this tech as the basis to build on top of (assuming it is proven a success and all the kinks are worked out).
If there was a standardized "smart hat" that I could put on my head that would allow me to control my AR glasses in an efficient way, that would be crazy. Because right now, that's one specific concern regarding AR usability, since current solutions with buttons and touch surfaces and voice controls and making hand gestures are awkwardly inconvenient and feel clunky.
UI/UX paradigm for certain consumer device categories completely shifts. A lot of accessibility problems get resolved. Hardware itself on a lot of consumer devices can become much simpler and more robust, since there is no need for much hand-driven interaction.
And that's just a singular the most obvious example I could've thought of. There are plenty of potential amazing things that could be enabled by this tech, even if the specific use case for it that was presented in the article might not "provide better quality of life".
Modern resuscitation is like nothing I have ever seen before. Over the past 100 years, we have gained an almost complete understanding of cardiac electrophysiology, acid-base balance, hematology. In the worst-case scenario that you get hit by a car in a major US city, you will likely have TXA forced into your bloodstream through an intravenous catheter within 13 minutes to counteract the body's incorrect hyperfibrinolytic reaction. Your heart rhythm will be constantly analyzed and treated according to ACLS which has essentially been studied and modified to death by the AHA in a gradient descent pattern - we're stuck in a local minimum where the antidysrhythmic of choice alternates between amiodarone and lidocaine. When you arrive at the hospital, a FAST ultrasound scan will be performed where the culmination of materials science and informatics comes together to quickly identify blood in the retroperitoneal space without incurring the logistics of getting a trauma patient into a CT scanner. All this while a neurologist checks 12 of your cranial nerves and someone checks the tone of your rectal sphincter - because evidence shows us this is an important sign. In the case that you have an intracranial bleed, you'll be followed with MRI (which I'm convinced is the closest we've gotten to Star Trek scanner technology) to identify the penumbra of a stroke based on the deoxygenation of hemoglobin. We can now scan a variety of elements with magnetic resonance spectroscopy besides just precessing hydrogen atoms, which is helpful in identifying neoplastic lesions without the daunting task of a brain biopsy.
Not everything is gloom-and-doom at the financing of some big company.
"isotonic crystalloids" - Solutions such as normal saline used to restore volume proportionally to the bloodstream and intracellular spaces (as opposed to hypotonic or hypertonic, isotonic solutions just match the normal blood's osmolality)
"sedation agents" - The patient is out, you keep 'em out, ideally without tanking their blood pressure. Different than paralysis.
"electrophysiology" - Electrical conduction through the heart.
"acid-base balance" - Various chemical processes and enzymes in the body only work well within a certain pH range, and long-term, only a narrow pH range is compatible with life.
"hematology" - Blood and its components, including the cascade of enzymes and clotting factors involved in coagulation.
"TXA" - Tranexamic acid - An anti-anti-clotting agent, which stops the body from breaking down blood clots. In the context of trauma, it allows your body to form the needed clots to avoid imminent exsanguination.
"ACLS" - Advanced Cardiac Life Support - The AHA algorithm for identifying and treating life-threatening heart rhythms that most paramedics, RTs, nurses, etc. and above are trained on. In practice, this means that an unresponsive patient or code situation can be run to a certain standard with the staff who are immediately summoned in any clinical area of any hospital within about 30 seconds.
"antidysrhythmic" - In the ACLS algorithms, it's a drug (usually a potassium channel blocker or sodium channel blocker or both) that acts on the cardiac action potential to do something that tends to make lethal dysrhythmias stop and return to a normal sinus rhythm. How exactly, you'd need to ask a cardiologist or specialized pharmacist.
"FAST [ultrasound] scan" - Focused assessment with sonography for trauma - it's just an ultrasound of the flank area to detect blood around the kidneys ("retroperitoneal space") which has a high mortality rate and usually requires emergency surgery.
"cranial nerves" - Your brain can communicate with the rest of your body either through your spinal cord, or through the cranial nerves which are direct interfaces for things like smell, sight, tongue movement, etc. Checking these is an important part of determining of a brain injury or spinal cord injury is present.
"rectal sphincter tone" - If it's present, then it's unlikely that spinal cord injury above a certain level has occurred.
"intracranial bleed" - It's bleeding, in your brain.
"magnetic resonance spectroscopy" - The original MRI technique was based on spinning all the hydrogen atoms in your body such that they aligned head to toe, then detecting electromagnetic induction in a coil as they were jarred back and forth at a megahertz frequency. Nowadays, we can do this with other atoms like phosphorus in the same way.
"precessing" - The pattern of relaxation of an atom similar to a spinning top where it goes from the aforementioned align orientation (forced by a strong magnet to align head to toe), and gradually relaxes to its random orientation while spinning at the Larmor frequency (megahertz frequency of the corresponding type of atom).
"penumbra" - The border between healthy living brain tissue and dead brain tissue at the outside territory of a brain injury post-stroke.
"neoplastic" - Uncontrolled cell growth, similar to cancer except not all neoplasms are cancer, some are benign tumors which will never be able to metastasize/spread to other parts of the body.
The decreasing life expectancy in the US is probably, ahem, largely attributed to a medical system that is not particularly well-optimized for boosting mass life expectancy (despite being world-class in absolute quality of care, I assume).
If you look closely, past decades have brought immense breakthroughs such as
- the first completely synthetic new antibiotic
- dramatic improvements in spatial and temporal resolution of imaging systems (MRI, PET etc.)
- actually useful robotic assistance for surgery
- easy, simple, and widespread techniques that are nevertheless life-saving, including a broad awareness for important events such as cardiac arrest and stroke
- much better, faster and more accurate laboratory work, down to the DNA level
And that's only a fraction of it which a layman such as myself can come up with from the top of my head!
And then you mention life expectancy, which is quantity of life.
For example, the longest Czech word: Nejnezdevětadevadesáteroznásobovávatelnějšími = [do something] by those most not multipliable by 99.
Capital is short for capital city.
Vacuum is also short for vacuum cleaner.
I'll grant you Aussehen - that's a single word in english. But longer than the German one.
I'd be pretty shocked if a German speaker thought a tortoise had a Schild or really was a Kroete, in the same way I'd be shocked if an English speaker said that a bush baby was an actual baby. Non-compositional noun compounds occur in both languages.
One thing that provides a lot of contextual information in German are the inflections and suffixes, i.e.
"Köchin" - female cook
"Koch" - male cook
"Köchchen" / "Köchle" ... - diminuitive cook
"Kochs" - possessive / "the cook's ..."
etc.
As an aside, German also relies heavily on modal verbs (nearly as much as English does), far more than even related languages such as Spanish. There are many distant or unrelated languages with more informative verb conjugations/inflections.
Real languages exist somewhere on a spectrum. Languages in the Indo-European family show an empirical historical tendency to become less synthetic and more analytic over time and via contact with other languages. German is more synthetic than English, but less synthetic than Russian, and much less synthetic than very old IE languages such as Vedic Sanskrit. English is more analytic than German, but less analytic and more synthetic than Afrikaans. Some non-IE languages such as Finnish are extremely synthetic.
[1] https://en.wikipedia.org/wiki/Synthetic_language , https://en.wikipedia.org/wiki/Analytic_language
https://en.wikipedia.org/wiki/Analytic_language
>>For example, Mandarin Chinese has many compound words, giving it a moderately high ratio of morphemes per word, but since it has almost no inflectional affixes at all to convey grammatical relationships, it is a very analytic language.
Within the Indo-European family that includes Latin and Sanskrit, the empirical historical process seems to have been that the Proto-Indo-European language was highly synthetic, and as time went on, less synthetic daughter languages such as Latin and Sanskrit developed from it by degrees. These in turn gradually developed by degrees into even less synthetic / more analytic daughter languages (Italian, Hindi, etc.)
Why this happened is an open theoretical question. Pidginization (which would roughly correspond to 'lack of isolation'), and/or phonetic changes that impacted discernment of synthetic features are leading candidates.
Based on the description I would classify Afrikaans as synthetic, not analytic.
Afrikaans has an origin as a pidgin Dutch spoken by slaves and other lower classes. Afrikaans speakers can communicate effortlessly with Dutch and Flemish speakers for example. Our grammar is decidedly germanic and we make use of morphemes in stead of separate words.
Held: hero Heldjie: diminutive hero Helde: plural
Heldin: female hero Heldinnetjie: diminutive female hero Heldinne: plural
Although it should be noted that explicit female versions of nouns is falling out of favour to some degree.
English and Afrikaans still have remnants of grammatical gender and case, but it's not really enough to be considered a prominent feature of the language.
She said that, as spoken, information density is fairly similar from language to language. Languages that express the same information using more phonemes tend to be spoken using more phonemes per unit of time, and it what (in some circumstances though not all) can contribute to the perception that people speaking in one language speak much faster.
I suppose that might not help much when the issue is writing though, or any circumstance where phonemic production is at a fixed rate regardless of language.
Languages with denser information are spoken slower. Vice versa.
The position of the verb in that language is fixed so it needs to *always* stand at the end of the sentence.
I remember translating news articles from BBC Persian and having to read five lines of text before reaching the verb. I love suspense.
Either way, you could take shortcuts like leaving out vowels (works pretty well in most languages and is even the default in some, e.g. in Arabic).
I would postulate that the loop starts by selecting the yellow block and waiting for him to select or reject it. If he selects it, he is able to cycle through the letters from left to right, ending at the symbol all the way to the right on every layer to exit that block. Then it keeps cycling down until he selects a new colored block. You can see this in practice if you rewatch the bit where he asks his son if he wants to watch a movie with him.
The article also notes that the man gave feedback to improve the system, so I would guess that it's a mix between a limitation of the way this system is set up, how fast the system can accurately distinguish a yes from a no, and how fast he is able to keep up, and for how long.
It would be interesting to see this system be implemented in a better way, I'm sure you could speed this up by dynamically changing the letter layout or even suggesting words on the fly, in the same way spelling suggestions work on other devices.
"...‘(son’s name) moechtest du mit mir disneys die hexe und der zauberer anschauen auf amazon’ – ‘would you like to watch Disney’s witch and wizard with me on amazon’ on day 461"
Edit: seems his requests could have been conceded through his feeding tube:
"The first message was a simple thanks to Birbaumer and the rest of the team. Other messages related to the man's care preferences: asking for a head massage or for more gel on his eye (which was prone to dryness) and requesting a higher head position when visitors were present. He even made suggestions for improving the performance of the spelling system.
Eventually, he could make specific dietary requests for his feeding tube: soup with meat, sweet pea soup, and curry with potato. And he did request a beer and for his caretakers to play his favorite band, Tool, very loud."
Unfortunately he's locked-in because of progressive muscle atrophy, so recovery isn't likely.
You can definitely still experience alcohol. There was a GQ article about Richard Norris, recipient of a face transplant, who went to the store and got himself some Wild Turkey which he ingested via a tube to his stomach.
> He reaches into the brown bag, pulls out a bottle of Wild Turkey. [...] "For my throat," he says. He can’t take over-the-counter medications, he says. Too risky with his meds. There’s a backpack at his feet. He opens it, pulls out some tubing and a wide syringe, about a half-inch across, the kind you use to give medicine to horses. [...] He hooks the syringe to the tubing, lifts up his shirt. [...] He opens the Wild Turkey and starts pouring.