Random Science Question Thread!

[Byrthnoth]

Force feedback holodecks and sensory immersive VR face a modified version of the same problem as #6 from the previous post, which is why AR is going to be better than VR for the near future.

Their requirements are a little less stringent because you could imagine some kind of full body suit that would emulate arbitrary physical barriers.

[BerenTebogo]

It's difficult to predict how far away brain implant prosthetics are because there are a lot of technological hurdles to jump, some of which currently look insurmountable:
1) Onboarding process - Current brain implant prosthetics correlate an action you intend to do with a movement. So there is a long training period where the researcher says "imagine moving your arm in this way" and they have you do it a dozen times and record what your brain activity looks like. Then the next movement, then the next, then maybe combinations, etc. It's an arduous process to say the least to map brain activity to intent.
2) Implant stability - ...and occasionally either your brain signature of intent or the implant itself will change and you have to remap it.
3) Movement precision - Where/what you're moving
4) Brain implant technical issues - infection prevention, implant form factor, number of channels/design, wireless?, etc.
5) Force control - How much pressure your fake muscles put out
6) Tactile feedback - What the prosthetic feels.

People are making incremental progress on 1-4 and they might eventually hit the point where brain implants are worthwhile for people who don't have tetrapelgia. Perhaps that point will be in 20 years (after FDA approval and such).

The big problems are #5 and #6. No one has good solution to and, it turns out, those two problems are pretty important to solve if you want to really use a fake hand. Without #5, you have to automate all force control in the arm and the user won't necessarily be able to control it. Without #6, everything is based on visual feedback. I just graduated from a university at the forefront of brain implant development, went to a lot of talks on this because it interested me, and I can say with confidence that we are a long, long, long, long, longlonglong way from solving either of these problems. #5 may fall out as #1-4 improve, but without tactile feedback there's no way for the user to do force control on unknown objects. I have seen the foremost lab present their research on tactile feedback, and I say to you with confidence that I don't expect to see it working in my lifetime.

Re #1

I agree there is a long rampup time and am very familiar with the process you are describing for training brain implants, but in my (completely out of field) opinion, it's just a matter of sample size. Right now we have very few (afaik) of these trials in process, but with enough of them I feel confident that, barring every brain being a unique snowflake of activity, we'll have better maps of where the activity per action is happening and training will be less arduous. I still put this as ~20 years out though.

#2 Again, larger sample sizes will help us hone in on how/why this happens and what, if anything, can be used as a workaround (redundancies?).

#3 Sample size.

#4 Valid

#5 Sample size of tests done by squeezing n objects of known tensile strength with different inputs.
#5a regarding automation and where control is:
I would imagine (key word) that throughout the average day the amount of pressure we exert on any object falls within a very narrow range of exerted pressures. Obviously if you are picking up a delicate baby bird you aren't going to exert the pressure you exert on your water bottle, but to probably a decent approximation, the pressure you exert on your water bottle, stapler, mouse and keyboard, door handle, steering wheel, fridge door, handshake, wife's tit, etc are all within a very narrow range of pressures.

It would be annoying - but possible - to train extra cues for much harder or lower pressure. Whether it be involving specific motion of other limbs in tandem with an action (an and statement somewhere), or a specific thought.

Again, it's a problem, but it's not one that should slow down implementation - it's one that should evolve over time with implementation. Just like how cars didn't have airbags for the majority of the time they were in production, tech gets better over time.


#6 I don't know.

I don't want to downplay how hard some of these problems are - because they are very hard. But that's what science is, hitting a hard problem with many, many, many hammers until you get to a confidence interval of what the right answer is. Most of these problems can be converted into problems of data sets and then once you just have large enough data sets, statistics takes over. You know this as well as anyone.

Edit: I misread the problem regarding 5. Editing in some extra stuff to go with prev. response.

[BerenTebogo]

Force feedback holodecks and sensory immersive VR face a modified version of the same problem as #6 from the previous post, which is why AR is going to be better than VR for the near future.

Their requirements are a little less stringent because you could imagine some kind of full body suit that would emulate arbitrary physical barriers.

Important point here is that immersive VR and holodeck stuff have NOTHING to do with images being tactile and reactive, and 100% to do with making your brain think they do.

Projected electromagnetic light can never, and will never, have mass or substance. (I know you know this but just clarifying for everyone else).

[Byrthnoth]

Oh yeah, you'd obviously actually need to generate real force somehow (or get a brain implant and solve #6). The most bare bones sci-fi version I can imagine would be some kind of highly manipulatable EM field inducing forces in a glove.

As far as prosthetics:

First, I'd like to clarify that prosthetic design thus far has been engineering rather than science. The labs undertaking these are typically in engineering departments and their final focus is a working product rather than an understanding of why their product works. It's minutiae, but explains why I don't think they're going to suddenly stumble upon the underlying organizing principle of interpreting motor cortex with incredibly low resolution sampling. They're doing the minimum to get a result, but that minimum might be the maximum currently technically possible.

But yeah, we seem to agree that 1-4 are just a matter of time and are likely 20 years out. I think #5 is going to require much higher resolution sampling of neurons as well as answering the question "how is force coded in motor cortex?" but that's a technical hurdle and one conceptual hurdle that will probably be overcome near-simultaneously. #6 is going to be the kicker that prevents movie-cyborg-like prosthetics.

[BerenTebogo]

Oh yeah, you'd obviously actually need to generate real force somehow (or get a brain implant and solve #6). The most bare bones sci-fi version I can imagine would be some kind of highly manipulatable EM field inducing forces in a glove.

As far as prosthetics:

First, I'd like to clarify that prosthetic design thus far has been engineering rather than science. The labs undertaking these are typically in engineering departments and their final focus is a working product rather than an understanding of why their product works. It's minutiae, but explains why I don't think they're going to suddenly stumble upon the underlying organizing principle of interpreting motor cortex with incredibly low resolution sampling. They're doing the minimum to get a result, but that minimum might be the maximum currently technically possible.

But yeah, we seem to agree that 1-4 are just a matter of time and are likely 20 years out. I think #5 is going to require much higher resolution sampling of neurons as well as answering the question "how is force coded in motor cortex?" but that's a technical hurdle and one conceptual hurdle that will probably be overcome near-simultaneously. #6 is going to be the kicker that prevents movie-cyborg-like prosthetics.

Yeah. I don't think movie-cyborg-like prosthetics are happening anytime soon. 100% agree with that. That's up there with fairy tale 100 year out minimum shit. And we probably are looking at the problems (even if agreeing) from a different angle being that you are biology and I am looking at it from the engineering side of it. Most of my responses are along the lines of "You didn't have an arm before, now you can eat, drink, and fuck. If you only do them as well as a retarded 12 year old - whatthefuckever" because, let's be honest, if you go from having no hand / arm / leg to literally anything better than one of those grabby reptar-claw things, you are gonna be a satisfied customer. Horses were better than walking everywhere, even though they shit a lot and sometimes kick you in the face and you die.

Once the initial neuron mapping problem is cracked and becomes routine, then it is just enhancement and improvement, but the problem itself will still be solved. This actually made me think of when we discovered the first Exoplanets some years ago (hard evidence) and we were excited as fuck and rolling in like a handful a year. Now we don't even blink anymore as we catalog the existence literally hundreds or thousands of new planets every year in like a 10-20 year turnaround.

[Byrthnoth]

Yeah, we'll be able to roll something out in the next two decades, probably, but everyone should still try to keep their arms.

[Byrthnoth]

Not a science question, but I talked to a faux nutritionist last week and he brought up a ton of debunked nutrition myths including one that I'd almost forgotten about. The myth was that you should avoid Sodium, "the silent killer." As confirmed by pretty much every careful study of the topic, this is almost entirely bullshit.

The current recommendation is 1.5-2g/day (1 teaspoon of fine salt), but you can probably eat a normal amount (4-5 grams) of sodium/day without any problem (approximately a tablespoon of fine salt). If you're cooking at home, you're very unlikely eat 1 tbsp of salt in a day worth of servings. However, even with processed food it's pretty unlikely as long as you don't eat like shit. There's 2.7g of sodium in an entire 1lb bag of Lays potato chips. Based on google nutritional facts, the dreaded TV dinner has ~1g at most. Unless your diet is entirely stuff like dried meat (Slim jim = 8g/lb) and Kraft singles (5g/lb), which is a rough approximation of McDonalds cheeseburgers, you probably don't have anything to worry about.

http://www.thelancet.com/journals/la...467-6/abstract
http://www.cochrane.org/CD009217/VAS...scular-disease
http://www.nejm.org/doi/10.1056/NEJMoa1311889
http://www.nejm.org/doi/10.1056/NEJMoa1311989
etc.
If you want more drawn out pop sci articles on the topic, type "salt bad? 2016" into google and good luck.


I find this myth particularly offensive because people try and cook at home as a healthy alternative and, in their zeal for health, don't add salt to their food at all or try to limit themselves to a tsp/day. Their food consequently tastes like shit and they go back to eating McDonalds or what-have-you. Salt is calorically free and we love it. Use a reasonable amount of salt in everything that it improves. As long as you aren't badly ill, your kidneys are really fucking good at buffering ions, particularly ones that are as fucking life essential and omnipresent as sodium.

[Zetanio]

Not a science question, but I talked to a faux nutritionist last week and he brought up a ton of debunked nutrition myths including one that I'd almost forgotten about. The myth was that you should avoid Sodium, "the silent killer." As confirmed by pretty much every careful study of the topic, this is almost entirely bullshit.

The current recommendation is 1.5-2g/day (1 teaspoon of fine salt), but you can probably eat a normal amount (4-5 grams) of sodium/day without any problem (approximately a tablespoon of fine salt). If you're cooking at home, you're very unlikely eat 1 tbsp of salt in a day worth of servings. However, even with processed food it's pretty unlikely as long as you don't eat like shit. There's 2.7g of sodium in an entire 1lb bag of Lays potato chips. Based on google nutritional facts, the dreaded TV dinner has ~1g at most. Unless your diet is entirely stuff like dried meat (Slim jim = 8g/lb) and Kraft singles (5g/lb), which is a rough approximation of McDonalds cheeseburgers, you probably don't have anything to worry about.

http://www.thelancet.com/journals/la...467-6/abstract
http://www.cochrane.org/CD009217/VAS...scular-disease
http://www.nejm.org/doi/10.1056/NEJMoa1311889
http://www.nejm.org/doi/10.1056/NEJMoa1311989
etc.
If you want more drawn out pop sci articles on the topic, type "salt bad? 2016" into google and good luck.


I find this myth particularly offensive because people try and cook at home as a healthy alternative and, in their zeal for health, don't add salt to their food at all or try to limit themselves to a tsp/day. Their food consequently tastes like shit and they go back to eating McDonalds or what-have-you. Salt is calorically free and we love it. Using a reasonable amount of salt in everything you make. As long as you aren't badly ill, your kidneys are really fucking good at buffering ions, particularly ones that are as fucking life essential and omnipresent as sodium.

The only time you really need to avoid sodium is if your doctor tells you to. You are hurting yourself if you eat too little in most cases. Most weight loss centers will keep their food VERY low in sodium but that's only because they are catering to some pretty unhealthy bodies and they are liable for the patient's health. You can add salt but you can't take it out.

[Byrthnoth]

Right, always listen to your doctors.

It's worth noting that going down to the currently recommended amount of sodium (below 3000mg/day) may actually increase the probability of heart attacks in hypertensive populations, per the linked Lancet meta-analysis. The Cochrane website is down right now, it seems. Above I linked a follow-up to their original metaanalysis, but here is that original study's abstract ( http://www.ncbi.nlm.nih.gov/pubmed/21735439 ), which finds that sodium restriction increases the risk of all-cause death in those with congestive heart failure and otherwise sodium levels have no discernible effect on anyone.