Comment by rolleiflex | Hacker News Reader

Comment by rolleiflex | Hacker News Reader

I’ve read the study just now - it does appear that:

a) The chemical put into the retina is surprisingly biocompatible, in that it does not appear to have caused anything out of the ordinary to happen. It works by effectively converting IR light to visible light - so it does not involve ‘sending more data’ to the brain, or ‘opening another colour band’ or anything like that. It is seen as just plain old green light, because the crystal structure of the chemical basically converts 980nm~ IR to 500~nm visible green light. So no uber-neurohacking going on here.

b) The procedure they are using to put the chemical in place is fairly run-of-the-mill medical procedure for humans for treatment of eye diseases, albeit a little icky.

So, I’m surprised, to say the least, but this could actually work with humans as a fairly harmless enhancement. I’m still waiting for the other shoe to drop, but haven’t seen one in the study itself for the time being. (Disclosure: not an expert, but I follow related fields)

Wow this is awesome. I bet they could design a different particle to fix some kinds of colorblindness, for example by blocking/converting the wavelengths of light that tend to be confused between red and green.

> does not involve ... ‘opening another colour band’

That's not so obvious to me. If this chemical were uniformly spread across the retina, then I'd agree. But if it's sufficiently clumpy, or ideally attached to every second green-sensitive cone, then it absolutely would give (N+1)-colour vision (N=3 in humans). On first glance I can't figure out what the paper has to say about this distribution... and even if it were uniform in their mice, perhaps the next step would be to make it non-uniform somehow.

The earlier way to do something similar was by genetically altering some cells, [1]. If you inject a virus (carrying the modification) at a low concentration, then you do get just half the cells. Then the brain learns (in a few weeks) which cells are now giving what information, and you get more colours. [1] did this in monkeys, IIRC they are N=2 and the virus added the gene for one of the human receptors. I don't think it was safe enough for humans.

[1] https://www.nih.gov/news-events/nih-research-matters/gene-th...

I don't know enough about the subject matter to judge if this comment is correct or not, but it is polite, on-topic, and therefore shouldn't be downvoted.

still it would not be possible to discriminate if the green that i see is normal green or IR green. it would be rather obvious in the night, when color cones are not used, but during the day it would be confusing. I 'm thinking "Aww that s a beautiful tree ... oh , it's a trash fire"

No, that's exactly what the monkeys learned to discriminate. After a few weeks they pass monkey-specific colour-blindness tests, which wild-type monkeys cannot.

I don't think you are born with different wiring to the three different kinds of cones, your brain just learns to lump all the red ones together, so that a red object stays the same colour in different parts of your visual field. The claim is that the monkeys learned which of the previously-identical cones now had a new colour.

yes it is possible to create objects that can discriminate between colorblind and wild type. but from the point of view of the colorblind person (the augmented person) it will still be a source of confusion.

Just to be clear, the point isn't that we can distinguish which monkeys are treated, it's that the monkeys can distinguish many colours which they could not before. The treated ones behave like trichromats.

Do they lose any sleep philosophising about how these new colours are possible, and what to call them? Who knows. But I believe they can still do all the usual monkey things.

> it's that the monkeys can distinguish many colours which they could not before

That's the point they can not distinguish more colors. Their retinal cones still have the same spectral response. They cannot distinguish between a normal 500nm green leaf and a 900nm infrared source.

Ah, you mean the mice? In the near-IR study it's not clear (to me, yet) whether all the cones are altered, or just some of them. If all of them are altered, then for sure, same N=2 as normal mice.

The monkeys are in a different study, the one I linked a few posts above.

sorry i was referring to the nanoparticles. Indeed rhodopsins are used so routinely in neuroscience that one wonders why it has not been used to treat colorblindness in humans yet.

OK! I was told was that inserting genes with a virus like this had a tendency to cause cancer, which is why it wasn't done on humans (least of all for something not life-threatening). But I don't know if that is still accurate.

The nanoparticles sound to me like they could be safer? Would be interesting to know whether they can be non-uniformly spread, so that only half the green cones can see one.

Came to say exactly this, I read the paper several times and could not find any surface level issues. It might make things appear greener as they would get extra equivalent light by having both a visible and NIR component.

Given that, and the tactical advantage of seeing NIR in low light situations, someone is going to try this on people. And the results of that study could have a lot of impact.

Small but important correction: the chemical is injected into the retina, not the cornea.

I was thinking this was cool because if it works in the cornea, it could possibly be injected into a contact lens and achieve the same effect. But it looks like that won't work because the effect depends on the chemical being in (very) close proximity to photoreceptors in the retina.

That makes sense, it needs to be where the image is focused in order to produce more than just a haze or blur. If it was on the lens or cornea it'd probably just produce a washed out brighter background that made night-vision worse.

Fixed. Thank you!

Look at a bright enough 980 and you will see blue, I assume as it up-converts to 490. Not sure if it was the beam exapnder or my eyes...

Look at a bright enough 980 and you won't see anything else!

Very true. I find myself gently explaining laser safety way too often for comfort, even in lab environments. We don't have an intuitive understanding of the dangers. I my case, I knew the divergence and had calculated my expected wattage before taking a peek back downrange... but ya.

I knew someone that briefly turned on a laser engraver with their arm in it. Not sure of the wavelength, but she claimed to feel it go through and could smell burned flesh. It was fairly new tech at the time and the doctor couldn't tell her much. She had some mild tingling afterwards, but nothing major.

She didn't tell her boss because she had disabled the safety switch, for some reason that seemed good at the time.

Prob a CO2 laser, out at 10000nm. When discussing laser safety, the primary concern is eye damage. Burning flesh etc is bad, but that heals, and it takes way more energy. Eye damage is nearly instant, it does not heal, and lasers you can dig out of consumer electronics or buy >class 3b are hecka-capable of seriously injuring someone. Even the standard 3b lasers are no joke, and often output light outside the visible range in addition to the visible beam.

This is caused by a tech called Second harmonic generation, or frequency doubling.

"A process in which two photons with the same frequency interact with a nonlinear material, are "combined", and generate a new photon with twice the energy of the initial photons. In some cases, almost 100% of the light energy can be converted to the second harmonic frequency. These cases typically involving intense pulsed laser beams passing through large crystals, and careful alignment to obtain phase matching. In other cases, like second harmonic imaging microscopy, only a tiny fraction of the light energy is converted to the second harmonic."

https://en.m.wikipedia.org/wiki/Second-harmonic_generation

Reply on news.ycombinator.com