In the future, lab mice will live in computer chips, not cages?(undark.org)
undark.org
In the future, lab mice will live in computer chips, not cages?
https://undark.org/2020/07/02/future-lab-mice-computer-chips/
78 comments
This is why I think the quote "All models are wrong, but some are useful" is so meaningful. Lots of bright eyed comp sci majors come into statstical genomics and bioinfomatics PhD programs every year, and all learn the hard lesson that wet lab still exists for a reason.
Not understanding your underlying conceptual model (be it biology here, or maybe healthcare, or natural resource exploration) will lead a lot of poor assumptions, limiting the impact of all the hard work.
Not understanding your underlying conceptual model (be it biology here, or maybe healthcare, or natural resource exploration) will lead a lot of poor assumptions, limiting the impact of all the hard work.
I would argue we can't even model those things. Non-linear, dynamic systems are not predictable beyond a certain point even if you have a perfect model. Uncertainty with models of dynamic systems like biological ones, increases exponentially as a function of time. And they are highly sensitive to initial conditions, so if your input is off from reality by even the tiniest amount, reality will veer off dramatically from your expected results, even if you have the perfect model. Chaos theory is a thing. We can use models to learn about systems and how they might work, or how we might think they work, but we can't use them to model reality effectively enough to determine treatments.
While individual trajectories of non-linear systems are not predictable, they all end up on an attractor and usually all we are interested in, is the attractor.
That is when performing a test on a mouse we are not measuring what a particular mouse would do, but what the mice do on average. So as long as model behaves like a mouse it can be useful, even though it would not behave like any particular mouse.
Also it's important to remember that the question is not about all or nothing, when we get a model most likely instead of 100 experiments on real mice we will be doing millions on virtual mice, and then 10 on real mice to verify the results.
That is when performing a test on a mouse we are not measuring what a particular mouse would do, but what the mice do on average. So as long as model behaves like a mouse it can be useful, even though it would not behave like any particular mouse.
Also it's important to remember that the question is not about all or nothing, when we get a model most likely instead of 100 experiments on real mice we will be doing millions on virtual mice, and then 10 on real mice to verify the results.
> Programs and algorithms can only encode models (which are themselves imperfect) of what we already know.
The Game Of Life seems to contradict this claim. Also worth considering, computational physics and weather/climate simulations.
Scientists of all sorts already do learn new insights by creating simulated models of systems. The simulated systems will continue to get more accurate.
The Game Of Life seems to contradict this claim. Also worth considering, computational physics and weather/climate simulations.
Scientists of all sorts already do learn new insights by creating simulated models of systems. The simulated systems will continue to get more accurate.
> As bad as animal (and human) studies and trials can be, the beauty of them is we don't really have to know how or why things work, we can just do the thing and if the results are good, we have a new treatment.
If you didn't have at least a broad idea of why something might work, you would most likely never have "done the thing". You need to narrow the search space. Simulations can help here.
If you didn't have at least a broad idea of why something might work, you would most likely never have "done the thing". You need to narrow the search space. Simulations can help here.
ML models sometimes cannot be explained by humans and contains information we do not know. But they can be contained in a program or algo.
This is, quite frankly, ridiculous. Yes, computational models do have a place in science and drug discovery, but I don't know anyone in the field who seriously thinks they will reduce the number of animals used in medical research.
If anything, you may see something like the Jevons paradox, where when you increase the efficiency of something (in this case, animals used in medical research), usage actually goes up.
If anything, you may see something like the Jevons paradox, where when you increase the efficiency of something (in this case, animals used in medical research), usage actually goes up.
> If anything, you may see something like the Jevons paradox,
Thank you for that. I think it might help explain why no matter how large my hard drives get I seem to have no problems filling them as the sizes and number of files seemed to increase to keep me constantly upgrading.
As for the lab mice, if we ever do reach the point where models are enough to replace the need for live animals it does open the door for even the most casual experimentation that might be considered too cruel and inhumane to be performed on living creatures. I'm not sure what kind of useful results could come from that kind of research, but at least then it could be explored without moral considerations or causing trauma to the researchers who would otherwise have to become torturers.
Thank you for that. I think it might help explain why no matter how large my hard drives get I seem to have no problems filling them as the sizes and number of files seemed to increase to keep me constantly upgrading.
As for the lab mice, if we ever do reach the point where models are enough to replace the need for live animals it does open the door for even the most casual experimentation that might be considered too cruel and inhumane to be performed on living creatures. I'm not sure what kind of useful results could come from that kind of research, but at least then it could be explored without moral considerations or causing trauma to the researchers who would otherwise have to become torturers.
That's probably what the beings who are simulating us were thinking.
I don't think it'll happen tomorrow, but computer technology hasn't finished advancing...nor has our medical knowledge. Perhaps within the next century bits will turn out cheaper, faster, and more effective, perhaps even enabled by quantum computing. Biology seems massively parallel, and we've only just gotten started with parallelism in computing.
That literally makes no sense. How exactly would a quantum computer assist us in simulating problems like drug transport and binding? (quite a few problems are purely empirically solved right now) How exactly is biology massively parallel? I suppose you could argue a cell is just a weird form of SIMD but that is a vast oversimplification. We’ve also been doing parallelism pretty much since computers were invented (and even before that, assembly lines were a thing) it’s just that now we really care cause our one trick of shoving transistors in doesn’t work so hot anymore.
A cell is a three dimensional environment where many weirdly shaped things interact with each other. It's not entirely unlike calculating a 3D video game. A quantum computer would assist us similarly as it helps in breaking encryption - it helps you search the problem space for a solution (e.g. protein shape) faster. Genomics is basically statistics.
| Genomics is basically statistics.
As someone in a bioinfomatics Ph.D., yes, there is statistics in genomic, but if you don't understand the underlying biology, you're going to be chasing wild statstical gooses. And even then, we sorta all treat our discoveries as temporary, needing wet lab work to help validate our claims. There's just something you cannot simulate.
As someone in a bioinfomatics Ph.D., yes, there is statistics in genomic, but if you don't understand the underlying biology, you're going to be chasing wild statstical gooses. And even then, we sorta all treat our discoveries as temporary, needing wet lab work to help validate our claims. There's just something you cannot simulate.
I've never met a single person in my life that did or even thought anything worth a damn that wrote off complicated shit in the sort of blasse way people like you do. X is basically Y, why so complicated? It's easy to say everything is easy and simple when you're aware that you won't be contributing shit and more likely will be an annoyance to someone who will.
I didn't mean it's simple at all! but that most work is done using statistics.
> That literally makes no sense. How exactly would a quantum computer assist us in simulating problems like drug transport and binding?
You can use quantum computers to simulate quantum mechanical systems, such as ones studied in chemistry. https://dl.acm.org/doi/10.1147/JRD.2018.2888987
Indeed, simulating quantum systems was the reason quantum computers were proposed by Feynman in the first place!
You can use quantum computers to simulate quantum mechanical systems, such as ones studied in chemistry. https://dl.acm.org/doi/10.1147/JRD.2018.2888987
Indeed, simulating quantum systems was the reason quantum computers were proposed by Feynman in the first place!
I don't know about the state of current tests, but it doesn't seem to me like the author is necessarily proposing to switch left and right to use computer models instead of a real mice.
Instead, the way I'm reading it, they are recommending to use computer models whenever possible, and only resort to live animal testing when the limits of the modelling are reached.
Instead, the way I'm reading it, they are recommending to use computer models whenever possible, and only resort to live animal testing when the limits of the modelling are reached.
It's a waste of time. There isn't a model that can sufficiently simulate the chaotic behavior of biological cellular systems and the complex interactions with potential chemical treatments. Models are Mickey mouse representations of reality.
It's not completely a waste of time. Animal research is extremely expensive in terms of time, man hours and money[1]. Whatever shortcut you can take to detect ineffective treatments before reaching this stage is arguably going to cost you more time and money than raising a colony of rats for testing your hypothesis out.
[1]: I own 4 (once 5) pet rats. They are only a small colony and, while I love them, I don't keep them at lab standards (for example, I don't a vet sleeping on the couch 24/7 in case of emergencies), yet, in a bad month, I could end up with spending 1000+$ for their health. Now, imagine that for the health standard of lab rats/mice and the numbers required for research.
[1]: I own 4 (once 5) pet rats. They are only a small colony and, while I love them, I don't keep them at lab standards (for example, I don't a vet sleeping on the couch 24/7 in case of emergencies), yet, in a bad month, I could end up with spending 1000+$ for their health. Now, imagine that for the health standard of lab rats/mice and the numbers required for research.
There are companies who make money by wasting time like that. I'm no expert in this field but from what I gather sometimes Mickey Mouse Models can help too. Economics is built on such models.
The GDP of an economy can be described in 5 variables. The non-linear dynamic processes like what occurs at the cellular level and at the organism level in a biological system, cannot be modeled with any degree of certainty. The complex chemical interactions of medicines with complex organisms like mice cannot be modeled with any degree of certainty, neither in the short term nor the long term. Uncertainty is an exponential function of time in such systems.
https://en.m.wikipedia.org/wiki/Chaos_theory
https://en.m.wikipedia.org/wiki/Chaos_theory
Who here has watched as much Black Mirror/Westworld as I have and know how this ends? My question is, if a robot is engineered well enough to replicate the "real" thing, then when you test on it, aren't you just shifting the torture from "living" to another thing yet to be granted the rights of the living? If you answer "no", then you have to admit that the model is incomplete and it does not in fact feel pain and is thus in an inadequate replica. Also agree with other comments that this is and will remain computationally infeasible at least until production quantum computers arrive.
You can ask the reverse question - if the Robot version of a mouse is sufficiently similar to a mouse, maybe the mouse also doesn't actually feel pain, and is really a kind of biological robot.
Maybe so are you.
Dualism holds that living entities are special, and can never be replicated artificially. So your question can actually be turned into dualism vs monism (which is not a question with a settled answer).
Maybe so are you.
Dualism holds that living entities are special, and can never be replicated artificially. So your question can actually be turned into dualism vs monism (which is not a question with a settled answer).
This would run into the "here is a hand", or rather "here is a fist" objection.[1]
Pain is the ground reality that is to be explained. If you manage to explain away pain rather than explain it, you haven't actually answered the question at all. If your answer at the end is "pain is not real", this doesn't prove that pain is not real, it proves that your premises were wrong.
[1] https://en.wikipedia.org/wiki/Here_is_one_hand
Pain is the ground reality that is to be explained. If you manage to explain away pain rather than explain it, you haven't actually answered the question at all. If your answer at the end is "pain is not real", this doesn't prove that pain is not real, it proves that your premises were wrong.
[1] https://en.wikipedia.org/wiki/Here_is_one_hand
If a computer prints on the screen "I'm in pain", is it actually in pain?
If you tell me "I'm in pain", are you actually in pain?
If you are sedated with a memory blocker, and are then hurt, but you don't remember it afterward, were you actually in pain? (This isn't an academic question - many medical procedures simply paralyze the patient and add a memory blocker, but don't provide much analgesia. The patient can not express pain, and does not remember it. So does the pain they felt matter?)
You want an explanation of the wrong "ground reality", you are discussing the wrong question, it's not "is pain" real.
It's all about the subjective experience of pain, and can a robot share in that? Can a person in a coma share in that? Can an animal?
If you tell me "I'm in pain", are you actually in pain?
If you are sedated with a memory blocker, and are then hurt, but you don't remember it afterward, were you actually in pain? (This isn't an academic question - many medical procedures simply paralyze the patient and add a memory blocker, but don't provide much analgesia. The patient can not express pain, and does not remember it. So does the pain they felt matter?)
You want an explanation of the wrong "ground reality", you are discussing the wrong question, it's not "is pain" real.
It's all about the subjective experience of pain, and can a robot share in that? Can a person in a coma share in that? Can an animal?
> If you tell me "I'm in pain", are you actually in pain?
> If a computer prints on the screen "I'm in pain", is it actually in pain?
Maybe. If you SMS me that you are in pain, are you in pain? "I'm in pain" is evidence of someone being in pain. But being in pain, the referent, is just being in pain. There's no additional indirection.
> If you are sedated with a memory blocker, and are then hurt, but you don't remember it afterward, were you actually in pain?
Yes.
> You want an explanation of the wrong "ground reality", you are discussing the wrong question, it's not "is pain" real.
No, this is an issue that was actually brought up.
> > maybe [...] you [...] [don't] actually feel pain
All reason is grounded in experience. If I know anything, it's the things that I experience directly. Pain is probably the most direct experience there is.
edit:
> It's all about the subjective experience of pain, and can a robot share in that? Can a person in a coma share in that? Can an animal?
Yes. No. Yes.
These are not difficult questions to answer... You're phrasing it like there's some kind of philosophical difficulty here, but I don't think there is. If you define pain as an aversive sensory experience, ie. a sensory experience that makes you wish it would stop (this definition covers disgust and discomfort as well, and I think that's valid), then it seems clear that anything that can form goals (and yes, a memory blocked sedated person can form goals, they just can't act on them) can experience pain. I think the objection to computers is just that we rely too much on our empathy to determine when something "is in pain". I think an NPC in a game that moves away from a fire is experiencing pain, even though its mind is extremely simple, just like an amoeba would when it avoids a harmful chemical. It doesn't have such a thing as a qualia of pain, maybe (I'm not convinced!) - but that's hardly necessary.
I feel we must overcome our distrust of functional descriptions. Pain is a functional pattern, and not a very complex one.
> If a computer prints on the screen "I'm in pain", is it actually in pain?
Maybe. If you SMS me that you are in pain, are you in pain? "I'm in pain" is evidence of someone being in pain. But being in pain, the referent, is just being in pain. There's no additional indirection.
> If you are sedated with a memory blocker, and are then hurt, but you don't remember it afterward, were you actually in pain?
Yes.
> You want an explanation of the wrong "ground reality", you are discussing the wrong question, it's not "is pain" real.
No, this is an issue that was actually brought up.
> > maybe [...] you [...] [don't] actually feel pain
All reason is grounded in experience. If I know anything, it's the things that I experience directly. Pain is probably the most direct experience there is.
edit:
> It's all about the subjective experience of pain, and can a robot share in that? Can a person in a coma share in that? Can an animal?
Yes. No. Yes.
These are not difficult questions to answer... You're phrasing it like there's some kind of philosophical difficulty here, but I don't think there is. If you define pain as an aversive sensory experience, ie. a sensory experience that makes you wish it would stop (this definition covers disgust and discomfort as well, and I think that's valid), then it seems clear that anything that can form goals (and yes, a memory blocked sedated person can form goals, they just can't act on them) can experience pain. I think the objection to computers is just that we rely too much on our empathy to determine when something "is in pain". I think an NPC in a game that moves away from a fire is experiencing pain, even though its mind is extremely simple, just like an amoeba would when it avoids a harmful chemical. It doesn't have such a thing as a qualia of pain, maybe (I'm not convinced!) - but that's hardly necessary.
I feel we must overcome our distrust of functional descriptions. Pain is a functional pattern, and not a very complex one.
This comment doesn't make any sense to me. Why would the existence of a robot copy imply a lack of pain? How can you argue that whether humans feel pain is an unsolved question? Shall we experiment on you instead?
Many philosophers are confused. Free will is a solved problem (we have free will regardless of the metaphysical premise, unless you're using a weird definition of free will, and even I can prove it for any given metaphysical premise), yet they're still working on it.
If somebody links "free will" to "determinism", they're very confused about what free will is.
There are a lot of philosophers doing genuinely useful work, but so much of the field is pre-occupied with trivial (and non-trivial) variations of the "if a tree falls in a forest, does it make a sound?" koan.
If somebody links "free will" to "determinism", they're very confused about what free will is.
There are a lot of philosophers doing genuinely useful work, but so much of the field is pre-occupied with trivial (and non-trivial) variations of the "if a tree falls in a forest, does it make a sound?" koan.
I must be quite confused then, but how is free will not linked to determinism?
There's a proper proof, which is too long to fit in this margin (i.e. I'm too lazy to find a link), but I can do a much shorter one:
1. Non-determinism is equivalent to determinism plus extra, random input.
2. The addition of random noise does not bestow free will upon something that didn't have it without the random noise.
3. 1 and 2 imply no non-determinstic systems can possess free will if deterministic systems cannot possess free will.
4. 3 implies free will is independent of determinism.
1. Non-determinism is equivalent to determinism plus extra, random input.
2. The addition of random noise does not bestow free will upon something that didn't have it without the random noise.
3. 1 and 2 imply no non-determinstic systems can possess free will if deterministic systems cannot possess free will.
4. 3 implies free will is independent of determinism.
See my reply here: https://news.ycombinator.com/item?id=23727782
Easiest way I can come up to go with it is to admit we sometimes exploit life for the greater good, and also swear to yourself that you do your best to keep it as natural, efficient and minimal as possible. Otherwise you don’t get to even grow and eat single grain of wheat do you...
I think you are missing something obvious here. Pain does not exist in any objective sense, it is a construct created by the brain in response to certain stimuli as a survival mechanism. If you were creating a simulation of a human body to use in medical experiments then surely you would not give it a brain that functioned in any meaningful sense.
I would not say that a body incapable of feeling pain is an "inadequate replica". A human or animal under anaesthetic does not feel pain - are they inadequate. You can certainly perform surgery on an unconscious person, and you can administer medications to them.
[edit] I should have added that I am totally in agreement that we are not even close to being able to simulate a human body with any degree of accuracy.
I would not say that a body incapable of feeling pain is an "inadequate replica". A human or animal under anaesthetic does not feel pain - are they inadequate. You can certainly perform surgery on an unconscious person, and you can administer medications to them.
[edit] I should have added that I am totally in agreement that we are not even close to being able to simulate a human body with any degree of accuracy.
Maybe pain is a part of the treatment as any body/antibody response. It is a push notification mechanism designed in a million of years of evolution for a reason.
But I agree that it might be deprecated soon.
I'm not sure these questions can be answered philosophically. After thousands of years, philosophy can't even agree on the the morality of things that already do exist.
Science fiction and moral philosophy depict "model organisms" as physical replicas of organic ones for a reason. Our moral ponderings are driven by moral feelings, and they need to trigger feelings for the reasoning to be interesting. The neocortex court only debates when limbic prosecutors press charges.
Ultimately, the task here is software that answer questions, not software that is a mouse. If software those two converge, we have a moral dilemma. Seems unlikely though.
Science fiction and moral philosophy depict "model organisms" as physical replicas of organic ones for a reason. Our moral ponderings are driven by moral feelings, and they need to trigger feelings for the reasoning to be interesting. The neocortex court only debates when limbic prosecutors press charges.
Ultimately, the task here is software that answer questions, not software that is a mouse. If software those two converge, we have a moral dilemma. Seems unlikely though.
Realistically this unlikely to be a problem in this century, however you are raising some valid points nonetheless. A lot of,if not most,capabilities animals, including humans, have are related to one thing: survival. Ability to move rocks, climb trees,or build sophisticated structures boils down to body's need to get food and preserve itself from destruction. These aspects are often overlooked when modelling AI and the whole focus is on pure logic. I sometimes think that one day we'll end up with universal AI and the first thing it will ask will be sometime like: "why should I do it?".
If you don't do it our funding will be cut and we'll have to power you down.
And that's where it gets complicated. You'd either need to create it in a way that you could inflict these negative reactions, like pain,thirst,fear,and so on.,as otherwise it would respond with " OK, go on then,not that I care anyway".
If you torture someone then use a time machine to restore him to the initial state, did you do anything bad?
You can't do this to anyone real but you can easily save and then restore the state of the simulation at any point.
Torturing someone in a simulation is not any different than imagining someone being tortured, or performing lots of arithmetic operations.
So the only logical thing is to admit that torturing/killing/hunting a species to extinction is bad not because of pain, but because of stealing computational resources belonging to someone else/some other sophisticated computational process.
You can't do this to anyone real but you can easily save and then restore the state of the simulation at any point.
Torturing someone in a simulation is not any different than imagining someone being tortured, or performing lots of arithmetic operations.
So the only logical thing is to admit that torturing/killing/hunting a species to extinction is bad not because of pain, but because of stealing computational resources belonging to someone else/some other sophisticated computational process.
> Torturing someone in a simulation is not any different than imagining someone being tortured,
Isn't it? Our imaginations are predictive, not simulative.
Isn't it? Our imaginations are predictive, not simulative.
Sure, our imagination does not have enough computational power to be simulative now, but in principle there is nothing preventing existance of a living creature with vastly more powerful brain.
If you're assuming imagination ~= simulation for an argument, you need to state the assumption up-front. That assumption doesn't hold for humans, nor for anything reasonably imaginable for people in our culture.
> Torturing someone in a simulation is not any different than […] performing lots of arithmetic operations.
Near as we can tell, the universe itself just performs lots of arithmetic operations.
I'm going to give an example similar to:
> If you torture someone then use a time machine to restore him to the initial state, did you do anything bad?
to make a point. If you create a person, torture them, then incinerate the person to restore them to the original state, did you do anything bad? If you're privileging "the present state of the universe, excluding historical events, even when evidence of those historical events is apparent in the present state of the universe", you're taking a very niche ethical viewpoint: one where anything can be undone by killing everyone involved.
> So the only logical thing is to admit that torturing/killing/hunting a species to extinction is bad not because of pain, but because of stealing computational resources belonging to someone else/some other sophisticated computational process.
I'd call this the principle of explosion, but I don't actually see how this can even follow from what you said earlier.
> Torturing someone in a simulation is not any different than […] performing lots of arithmetic operations.
Near as we can tell, the universe itself just performs lots of arithmetic operations.
I'm going to give an example similar to:
> If you torture someone then use a time machine to restore him to the initial state, did you do anything bad?
to make a point. If you create a person, torture them, then incinerate the person to restore them to the original state, did you do anything bad? If you're privileging "the present state of the universe, excluding historical events, even when evidence of those historical events is apparent in the present state of the universe", you're taking a very niche ethical viewpoint: one where anything can be undone by killing everyone involved.
> So the only logical thing is to admit that torturing/killing/hunting a species to extinction is bad not because of pain, but because of stealing computational resources belonging to someone else/some other sophisticated computational process.
I'd call this the principle of explosion, but I don't actually see how this can even follow from what you said earlier.
The example with creating a person and then destroying it is different, because after creation it can continue to live independently from yourself. But if you create a person in your computer, that person can exist only as long as you provide the computational power for it to exist. In a way it is similar to abortion, where destroying a fetus is ok because it cannot exist on it's own, and the mother cannot be forced to provide her resources to it.
Ian M Banks explored this in the culture series. When doing speculative simulation of how another civilization will react to situation, what do you do with all those little simulated individuals when you've got the data you need. If you simulated them with enough detail, isn't turning the simulation the same as genocide?
This is less of a "we will replace natural meat with lab-grown meat" claim and more of a "we will eventually be able to travel to the moon using only a megajoule of energy" claim. It would take a vast change in our basic understanding of physics to effect such an innovation. The crux of the matter is that mouse bodies don't simulate mouse bodies, they are mouse bodies. The idea you could have something the size of a mouse body that could fully simulate a mouse body is ridiculous. Biology is computationally efficient.
The University of Nevada used a 360 TFLOP supercomputer to simulate a mouse brain. They managed to simulate the neurons firing once every ten seconds. In reality neurons fire an average of 200 times per second. Even if their model was fully accurate, that means you need 720PFLOPS to simulate just the brain of a single mouse.
Barring computers based on completely different principles or a revolution in the way we understand physics, a installation at a single facility will never be able to fully simulate hundreds of experimental mice.
The University of Nevada used a 360 TFLOP supercomputer to simulate a mouse brain. They managed to simulate the neurons firing once every ten seconds. In reality neurons fire an average of 200 times per second. Even if their model was fully accurate, that means you need 720PFLOPS to simulate just the brain of a single mouse.
Barring computers based on completely different principles or a revolution in the way we understand physics, a installation at a single facility will never be able to fully simulate hundreds of experimental mice.
It's absolutely amazing to me how efficient and powerful brains are. Just the simple action of say, moving from one side of a yard to another involves realtime sensor information, motor control and feedback, threat assessment and decision making, previous experience recall and more. In mere miliseconds even in mice.
It gets insanely more complicated for humans. If you decide to drill a hole in a piece of wood, your brain will process a ridiculous amount of information just for that single purpose, plus it's always running primitive functions like any other animal. Realtime.
I don't see any computer doing that anytime soon.
It gets insanely more complicated for humans. If you decide to drill a hole in a piece of wood, your brain will process a ridiculous amount of information just for that single purpose, plus it's always running primitive functions like any other animal. Realtime.
I don't see any computer doing that anytime soon.
Sounds like map-vs-territory confusion to me. A model of a mouse in a computer isn't a mouse, the point of doing the experiment is to uncover whether what you thought the model predicted was in fact what happened, because it's possible you didn't incorporate some aspect of the real mouse.
Simulations are useful for making calculations. Say I'm convinced that buildings work a certain way and I have a finite element model of one. I don't know what the predictions are for a new building of a certain shape, so I shove it in a computer to tell me. The computer tells me what I should predict assuming that my finite elements ideas were correct. What it won't tell me is whether there's some aspect of that model that doesn't correspond to what the building is actually gonna be made of.
Simulations are useful for making calculations. Say I'm convinced that buildings work a certain way and I have a finite element model of one. I don't know what the predictions are for a new building of a certain shape, so I shove it in a computer to tell me. The computer tells me what I should predict assuming that my finite elements ideas were correct. What it won't tell me is whether there's some aspect of that model that doesn't correspond to what the building is actually gonna be made of.
Doing drug development without animal testing is like trying to modify a huge, messy, legacy code base without integration tests.
Our computer models are like unit tests. They are relatively self-contained and quick to run. However, they cannot test all the interactions and stuff we do t know about, so we have animal testing as a kind of integration test to make sure stuff works like expected.
And a phase 3 clinical trial is like pre-prod. When you think everything works, you expose test it in the real world with very careful monitoring and plan to keep it from going out into prod if problems are found.
Our computer models are like unit tests. They are relatively self-contained and quick to run. However, they cannot test all the interactions and stuff we do t know about, so we have animal testing as a kind of integration test to make sure stuff works like expected.
And a phase 3 clinical trial is like pre-prod. When you think everything works, you expose test it in the real world with very careful monitoring and plan to keep it from going out into prod if problems are found.
Why do they want to model mice, when they the whole medical industry don't know shit about the mice they use for experiments?
They use mice to experiment with, BUT through breeding the current lab mice are more resilient then humans. Which can imply that certain drugs getting passed for human testing because they did well with mice. Today's lab mice are more resilient, but the same drug can damage humans.
Bret Weinstein talks about this on JRE: https://www.youtube.com/watch?v=pRCzZp1J0v0&t=9597s
They use mice to experiment with, BUT through breeding the current lab mice are more resilient then humans. Which can imply that certain drugs getting passed for human testing because they did well with mice. Today's lab mice are more resilient, but the same drug can damage humans.
Bret Weinstein talks about this on JRE: https://www.youtube.com/watch?v=pRCzZp1J0v0&t=9597s
Given the current state of our nematode simulation tech (http://openworm.org/), "the future" might take a while to arrive.
As a fellow animal researcher: this article is pretty much devoid of informed or carefully considered thought about the issue.
Many commenters have noted that biological systems are mind-numbingly complex, and we're not even close modeling these systems in a way that's useful for testing novel hypotheses of biology researchers.
The field in which I work, neuroscience, frequently does employ computational modeling, but this modeling is primarily used to provide a mechanistic link between inputs and outputs observed in an in vivo experiment. Present a sensory signal to an animal, observe something about neural dynamics in a brain region, THEN use simplified biologically-inspired elements to provide a plausible mechanism between the two. Even the state-of-the-art in silico model is always child's play compared to the real thing. We definitely don’t know enough about that real thing to build a model that would circumvent its need. This author also works in a neuroscience lab – they should know better.
Many commenters have noted that biological systems are mind-numbingly complex, and we're not even close modeling these systems in a way that's useful for testing novel hypotheses of biology researchers.
The field in which I work, neuroscience, frequently does employ computational modeling, but this modeling is primarily used to provide a mechanistic link between inputs and outputs observed in an in vivo experiment. Present a sensory signal to an animal, observe something about neural dynamics in a brain region, THEN use simplified biologically-inspired elements to provide a plausible mechanism between the two. Even the state-of-the-art in silico model is always child's play compared to the real thing. We definitely don’t know enough about that real thing to build a model that would circumvent its need. This author also works in a neuroscience lab – they should know better.
This completely bypasses the revolution that will take place long before this: Organs on a chip. With the development of organoids [0] we made the first step. Things like the lung-on-a-chip [1] are steps towards a sort of mini body that mimics many functions and simulates the body not only in the molecular biological way but in the very important physical way (which, it turns out can not be decoupled from the molecular biology).
Why would we put mice in computer chips if we can have pseudo human bodies in the real world?
[0] https://www.youtube.com/watch?v=2SG5ivm6jkw
[1] https://wyss.harvard.edu/media-post/lung-on-a-chip/
Why would we put mice in computer chips if we can have pseudo human bodies in the real world?
[0] https://www.youtube.com/watch?v=2SG5ivm6jkw
[1] https://wyss.harvard.edu/media-post/lung-on-a-chip/
> Why would we put mice in computer chips if we can have pseudo human bodies in the real world?
Because a lot of the existing data was gathered on mice.
Because a lot of the existing data was gathered on mice.
One can look at it like how us developers look at testing.
You have multiple steps on the way from code to delivery. There are unit test, more realistic test frameworks, test environments and then finally tests at a customer / usage by a customer.
In the same vain I could see (and even hope) some mice tests could be decreased or removed as better models and simulations gets created. That would be quicker and more cost efficient for everyone. But in the end you want a real world test before giving it to a human. And maybe we could find better things than mice for that as well, but right now I think that is the best we have got. I am not a medical researcher though so I don't know if there are alternatives out there.
In the same vain I could see (and even hope) some mice tests could be decreased or removed as better models and simulations gets created. That would be quicker and more cost efficient for everyone. But in the end you want a real world test before giving it to a human. And maybe we could find better things than mice for that as well, but right now I think that is the best we have got. I am not a medical researcher though so I don't know if there are alternatives out there.
A friend of mine works on simulating mouse organs. We had some pretty great talks over beer about modeling and cross model validation. Eventually he'd rant about all models are wrong but some models are useful.
If I remember correctly, they had some success in rejecting animal studies. Things like, we don't know what this will do overall, but we're pretty sure the kidneys or liver will fail like X, so don't bother with the wet lab. Or possibly try the wet lab at small scale so they can validate the model.
We're far far away from full simulation. But we know a whole lot about mice. I internalized those conversations as, my friend writes type checkers for drugs against mouse hardware.
If I remember correctly, they had some success in rejecting animal studies. Things like, we don't know what this will do overall, but we're pretty sure the kidneys or liver will fail like X, so don't bother with the wet lab. Or possibly try the wet lab at small scale so they can validate the model.
We're far far away from full simulation. But we know a whole lot about mice. I internalized those conversations as, my friend writes type checkers for drugs against mouse hardware.
This is apparently different from the lab-on-a-chip research which seems to seek to miniaturize organs into a microfluidic environment. I'm not actually sure if they are using organ cells on a scaffold or if they can discard the scaffold for microfluidics and cells alone even after reading a few papers on the subject.
Personally I think living 2D animals in a petri dish would be metal as fuck.
Personally I think living 2D animals in a petri dish would be metal as fuck.
I worked with a guys who said 'one of the reasons' he quit medical research was because of the amount of mice he had to kill on a daily basis. He said, "the quick way is to snap their necks with the back of your pencil." as he demonstrated on the desk the technique of how to hold the pencil a certain way.
This isn't the right point to make.
There are a lot of unnecessary animal experiments because the way in which experiments are planned is very unprincipled. in silico approaches have great potential to improve the efficiency of animal experiments.
I guess now would be a good time to recommend 'Permutation City' by Greg Egan... Great book!
The brain is not an algorithm and can’t be described as such.
Its pointless to use animals in research, it makes no logical sense. Even if the research is successful after 10 years of animal cruelty ... you still have to test it on humans.
What about all the chemicals that never get to human testing due to harmful effects coming up in animal testing? Is that not a net positive from a human perspective?
Sure - though plenty of chemicals hurt animals but don't hurt us. One small example: theobromine and caffeine (in chocolate) kill cats and other animals that lack the ability to metabolize these alkaloids rapidly enough.
Oddly enough, our own cat loves chocolate and (following a near miss & visit to the vet) now appears to resent having had that avenue of pleasure permanently blocked.
Oddly enough, our own cat loves chocolate and (following a near miss & visit to the vet) now appears to resent having had that avenue of pleasure permanently blocked.
Are there absolutely no alternatives to test that ?
Not as far as our brightest scientists in the relevant fields have been able to figure out. What do you propose?
Test with synthetic proteins ?
https://www.peta.org/issues/animals-used-for-experimentation...
https://www.peta.org/issues/animals-used-for-experimentation...
Yes, those techniques are what we are moving towards and we are slowly getting there, but for the time being they can't be universally applied.
You do understand you are just defending big pharma and cosmetics. I really doubt any new ground breaking discovery has come from animal testing.
I've never seen a more text book example of changing the subject when you don't have a real answer.
sk0g and hobofan point out a key benefit of animal testing, explain why there is no alternative at present, and rather than addressing that or acknowleding the point, you respond with non-sequiturs.
Try to learn to separate what you want to be true "animal testing is pointless" vs what is actually true "animal testing has benefits that cannot be sensibly found any other way". Then we can productively move towards eliminating animal testing. But if you won't even listen to others points, how do you hope to convince them of anything?
sk0g and hobofan point out a key benefit of animal testing, explain why there is no alternative at present, and rather than addressing that or acknowleding the point, you respond with non-sequiturs.
Try to learn to separate what you want to be true "animal testing is pointless" vs what is actually true "animal testing has benefits that cannot be sensibly found any other way". Then we can productively move towards eliminating animal testing. But if you won't even listen to others points, how do you hope to convince them of anything?
Animal testing is all over biochemistry and biology, is not just used for pharmaceutical and cosmetics testing. Many ground braking discoveries have been (and are still being made) via animal testing. There are whole subdisciplines that are based around animal testing like developmental or behavioural biology.
Though you can avoid it if you really want to, if you study biochemistry, you will learn how to do animal testing and are taught about all the ethics involved, how to do the testing in the most humane way, and how to avoid (or minimize) doing it in the first place if possible.
I've been going through the list of recent Nobel Prizes and the research for these involved animal testing: Nobel Prize for medicine 2018 (mice), 2017 (flys), 2014 (rats), 2012 (frogs, mice). Older research made before we had good tools to work with cell cultures, etc. will probably skew even more towards animal testing.
Though you can avoid it if you really want to, if you study biochemistry, you will learn how to do animal testing and are taught about all the ethics involved, how to do the testing in the most humane way, and how to avoid (or minimize) doing it in the first place if possible.
I've been going through the list of recent Nobel Prizes and the research for these involved animal testing: Nobel Prize for medicine 2018 (mice), 2017 (flys), 2014 (rats), 2012 (frogs, mice). Older research made before we had good tools to work with cell cultures, etc. will probably skew even more towards animal testing.
Citation needed. What specific things were discovered ? Are they useful to humans ? Could they be done without testing ?
All that I read are pointless science articles about aging in mice as if that applied to humans.
All that I read are pointless science articles about aging in mice as if that applied to humans.
> Are they useful to humans?
No, probably not. I'm sure they are just giving Nobel Prizes for Medicine out for funny discoveries rather than for findings that form the basis of most of our current and future medicine.
> Could they be done without testing?
Sure they could! But scientists just love to make animals suffer!
No, probably not. I'm sure they are just giving Nobel Prizes for Medicine out for funny discoveries rather than for findings that form the basis of most of our current and future medicine.
> Could they be done without testing?
Sure they could! But scientists just love to make animals suffer!
> Nobel Prizes for Medicine out for funny discoveries
Yes, just like they give oscars to pointless movies. Even roman numerals had its fervent defenders.
Yes, just like they give oscars to pointless movies. Even roman numerals had its fervent defenders.
Can you name 1 important medical discovery that has NOT come from animal testing?
None that we know of yet.
Programs and algorithms can only encode models (which are themselves imperfect) of what we already know. There's countless things we know that we don't know, and yet more that we don't know that we don't know.
As bad as animal (and human) studies and trials can be, the beauty of them is we don't really have to know how or why things work, we can just do the thing and if the results are good, we have a new treatment. Lots of super common drugs, even paracetamol, are like this.
Of course, knowing and accurately modelling the action is great, but it's not necessary.