What is life?
(This post is part of the attempt to do a write up after every Cambridge_UK meetup, of something raised at the meetup.)
What is life?
Eliezer, in “The First World Takeover”, talks about the search for objects with increasing numbers of bits of functional complexity. He framed the difference between life, and the previous non-living universe, as being the introduction of self-replicators who could search for self-replicators with an improved ability to search.
In terms of organisms that share genes among a population via sexual reproduction, we have to consider this search mechanism (evolution by natural selection) to be a property of a population rather than of a single individual.
And, indeed, when we take into account the speed limit and complexity bound for evolution, what we’re looking for is a population who search faster than the natural decay rate imposed by copying fidelity, genome size and how many harmful mutations get dropped each generation by the fraction of the population that don’t reproduce.
And that is relative to the environment the population happens to find themselves situated in. Radiation affects mutation rate. Harshness of the environment (including competitors) determines whether the population is viable (replacement rate at least equals the death rate) or whether they will spiral down into extinction.
So maybe, rather than asking whether an individual or system of individuals counts as a life form, perhaps a more well defined question would be to ask whether they count as a viable searcher relative to a specific environment.
Why does that distinction matter?
Are prions living? How about viruses? Crystals?
We know that a turing complete machine can be implemented in Conway’s Game of Life and that, given a very specific environment, crystals can not only self-replicate, but also pass on information to their ‘off-spring’. So it would make sense to say that a system of crystals might be constructed that would be ‘alive’ relative to a specific environment that permitted the system to evolve—to ‘search’ not just for a variant replicator that might be improved in some limited way (such as replicating faster), but search unlimited bits of search space, including for an improved searcher.
Whereas a virus or prion would not count, unless in a particular environment the information they pass from generation to generation is able to alter the ‘search’ mechanism (the machinery that replicates them) in a way that can improve how it ‘searches’.
The topic was raised at the meet-up, by the way, over the question of under what circumstances it would make sense to count as being alive, self-replicators made of synthetic RNA. A definitive answer wasn’t reached. Any opinions?
(Please note, I’m not an expert in these areas, just writing up the report, so any mistakes are mine, not those of the meet-up participants who generated the specific ideas.)
Surely what this shows is that the word ‘life’ doesn’t correspond to a category in reality except in the fuzziest of ways. So taboo it. What aspects of ‘life’ are you actually interested in?
I’m personally interested in several aspects or questions...
How far off is humanity from being able to synthesise a living being purely from matter that did not come from another living being?
Many people hold that living beings should be granted a right to not be wantonly deprived of life, other things being equal. But what are the attributes a being requires to qualify for such moral ‘protection’ ?
If an AI can be said to be alive, is it still alive when the execution of the code is temporarily suspended? If it is a scale, is one whose code has been slowed to one clock cycle per year less alive?
Morality isn’t based on alive-ness, it’s based on sentience, IMO. Beings have moral weight when they have preferences about how the universe should be.
Saying that moral weight is based on sentience is IMO largely a tautology. Sentience is mostly the word we use for “whatever poorly defined features of a mind give it moral weight”.
Short version of my other response: Sentience and life are probably both nonsense words, but if we’re picking a nonsense word to define rigorously and care about, it should be sentience.
Even granting that, it at least expresses that moral weight is a function of a mind, which is not entirely tautological.
Hence the word “largely”.
Yes, but saying that everything alive deserves moral consideration is a different position.
Is “preference” a word we have any idea how to define rigorously?
I have the increasingly strong conviction that we ascribe emotions and values to things we can anthropomorphize, and there’s no real possibility of underlying philosophical coherence.
Short answer: Rigorously? I don’t know.
But I know that the quality that causes me to care about something, morally, is not whether it is capable of reproducing, or whether it is made of carbon. I care about things that are conscious in some way that is at least similar to the way I am conscious.
No, I don’t know what causes consciousness, no, I don’t know how to test for it. But basically, I only care about things that care about things. (And by extension, I care about non-caring things that are cared about).
I’m willing to extend this beyond human motivation. I’d give (some) moral standing to a hypothetical paperclip maximizer that experienced satisfaction when it created paperclips and experienced suffering when it failed. I wouldn’t give moral standing to an identical “zombie” paperclip maximizer. I give moral standing to animals (guessing as best I can which are likely to have evolved systems that produce suffering and satisfaction)
I give higher priority to human-like motivations (so in a sense, I’m totally fine with giving higher moral standing to things I can anthropomorphize). I’d sacrifice sentient clippies and chickens for humans, but in the abstract I’d rather the universe contain clippies and chickens than nothing sentient at all. (I think I’d prefer chickens to clippies because they are more likely to eventually produce something closer to human motivation).
Don’t worry—I am not under the impression my moral philosophy is all that coherent. But unless there’s a moral philosophy that at least loosely approximates my vague intuitions, I probably don’t care about it.
The main point, though, is that if we’re picking a hazy, nonsense word to define rigorously, it should be ‘sentience,’ not ‘life.’
(edit: might be meaning to use the word “sapient,” I can never get those straight”)
The fact is that the meanings different people use for sentient vary much more than for sapient.
Interesting.
I read you as arguing for a narrower class that didn’t include the chicken. I’d sacrifice Clippy in a second for something valuable to humans, but I don’t really care whether the universe has non-self-aware animals.
I believe chickens are self-aware (albeit pretty dumb). I could be wrong, and don’t have a good way to test it. (Though I have read some things suggesting they ARE near the borderline of the what kind of sentience is worth worrying about)
A common test for that (which I’m under the impression some people treat more like an ‘operative definition’ of self-awareness) is the mirror test. Great apes, dolphins, elephants and magpies pass it. Dunno about chickens—I guess not.
That would test a level of intelligence, but not the ability to percieve pain/pleasure/related-things, which is what I’m caring about.
Then self-aware is quite a bad word for it. I suspect that fish and newborn babies can feel pain and pleasure, but that they’re not ‘self-aware’ the way I’d use that word.
Nociception has been demonstrated in insects. Small insects.
Edit: Not to mention C. elegans, which has somewhere around three hundred neurons total.
So the Buddhists were right all along!
(FWIW, I assign a very small but non-zero ethical value to insects.)
Anthropomorphizing animals is justified based on the degree of similarity between their brains and ours. For example, we know that the parts of our brain we have found are responsible for strong emotions are also present in reptiles, so we might assume that reptiles also have strong emotions. Mammals are more similar to us, so we feel more moral obligation to them.
Are cryonically suspended people alive? Are people in a coma alive?
Precisely. We could just taboo the word “life” and ask direct questions such as “what legal rights does it make sense for society to afford cryonically suspended people”, but that misses the point that the rest of society needs to be interfaced with, and they are not going to taboo “life”. If you’re trying to make the case in the public area that a cryonically suspended person deserves legal right RRR, which was previously available only to ‘living’ people, then you are going to get asked questions such as “well, what should ‘life’ mean?”, and it will be helpful to have a pre-prepared answer ready to hand.
Just a random thought: if we did that, we’d probably be using information from another living being. Does it matter so much where the actual atoms come from?
Various religions predict that humans can’t create life, that being a power reserved solely for deities (according to their theology).
We don’t have to understand what they mean by “life” or “create life”, nor does what they mean have to be well defined, in order for us to understand that it is (to them) an important issue and that when scientists do this it will have real social consequences, which are worth thinking about and planning for in advance.
Well, obviously not; all atoms of the same type are the same. I assumed he meant molecules. (As in, able to synthesize all biologically relevant molecules. All molecules of the same type are the same too, of course.)
-- K. Eric Drexler, Engines of Creation
Reminds me of Arguing By Definition; you can argue about the definitions of life and find one that includes (or excludes) prions or viruses. Does it really matter? When you call something ‘life’, all kinds of connotations come in; but of course the phenomena does not change.
Baby d-
[is killed]
-3 karma for a meeting report? Yeah, that’s just the sort of incentive LW needs.
Yeah, that seems pretty wrong.
I think part of the problem is the title doesn’t say “Meeting Report”, so people just clicked, skimmed, and then judged the content as if it were a random rambling rather than a report.
From the link: “If one star burned longer or brighter, that didn’t affect the probability distribution of the next star to form.” This is false, brighter stars die sooner and emit more stuff in space, including heavier elements, thus affecting both the probability of formation and the makeup of the future stars. You can even take this further, arguing that this amounts to reproduction (with synthesized elements instead of gametes), and that stars are a form of life (they self-replicate). This has been explored in science fiction, of course. In this way, stars also transform the interstellar space into more (or less) conducive to further star life. They also resist entropy, as long as there is some material to make more stars out of.
In fact, I dare you to argue that stars are not alive by your definition.
I’m thinking now of Lee Smolin’s hypothesis that universes reproduce via black holes, which is why our universe seems fine-tuned for their production.
Huh. Lacking any kind of background to evaluate that critically, but it’s a neat concept.
It’s in Wikipedia. His original detailed theory was disproved, but many still find the idea of interest.
There is a nice public polemic between Smolin and Susskind, with some accessible parts in it.
‘Life’ is a term used by biologists to describe that subset of chemical phenomena which is their purview.
Don’t they use different definitions, depending upon whether they are asking if a human is alive, if a human’s arm is alive, or if a single cell within a human’s arm is alive?
(Okay, so, it’s obvious that if you have to explain a joke, the joke was never that funny. Still, here’s my explanation. Your post seemed like something someone would write if they were trying to find necessary and sufficient conditions for a term which probably doesn’t even refer to a natural kind. So, I made a silly summary of a possible meaning as use definition.)
However, based on your response, I don’t understand why you would ask “What is life?”. You seem to already understand the problematic nature of that particular concept.
(Also, you definitely should have titled your post “What is the meaning of ‘life?’”)
Our intuitive concept of ‘life’ conflates two concepts: the ability to reproduce and complexity. Many would consider a machine made by an advanced civilization (but incapable of reproduction) as ‘alive’; meanwhile, one might hesitate to call prions ‘alive’, despite their capacity to replicate, merely due to their simplicity.
It’s always relative to a certain environment. Human beings and most animals can’t survive outside of the current biosphere. In that respect we’re no less independent from certain peculiar conditions than viruses are. We both depend on other living organisms in order to survive.
Maybe redefine life against a continuum of how unlikely, complex the necessary environmental conditions are that sustain it?
Some autotrophic cells might rank at one currently known bound while higher animals would be on the other end.
Yes, sorry, that’s another thing I missed off the write-up. Rather than “life” being a binary “you’re either a living organism or you are not”, it might be better looked at as a scale. One possible measure being the range of environments in which you are a viable searcher; another being the percentage you control of the parts of the environment that are relevant to your replication.
The alternative viewpoint suggested was rather than asking how alive a single organism is, ask what is the required organism+(other organisms or part of the environment) that should be considered to have the necessary attributes to count (collectively) as alive.
Is this correct? Imagine a least-convenient-genome where every mutation causes instant death (but the mutation rate is low enough that some offspring survive). Then the decay rate is nonzero but the search rate appears to be 0.
My point is that “search” in this context seems to imply seeking out new innovations, not merely maintaining a stable population. The search rate only appears relevant to survival when the environment is changing.
“Rate of search divided by rate of change of environment” might be an interesting metric (where the “environment” obviously includes other populations of replicators who have their own search rates). Possible division-by-zero error if neither the phenotype nor the environment seem to be changing very much.
It seems to me that both a genotype and phenotype is needed to qualify for labelling something as alive. It’s difficult to see how any form of natural selection could operate on inheritance and variation without a genotype. So that would rule out crystals and so forth.
Are other approaches to life any more productive at categorising whether an individual being is alive, such as Dawkins’ “non-random survival of randomly varying self-replicators” or Lehninger’s “living organisms preserve their internal order by taking Gibbs free energy from their surroundings” ?
The synthetic RNA in question was the family of ligase-polymerase ribozymes described here: http://www.sciencemag.org/content/332/6026/209.short