The Meaning of "Life"

By Sean Carroll | September 8, 2007 11:21 am

John Wilkins at Evolving Thoughts has a great post about the development of the modern definition of “Life” (which, one strongly suspects, is by no means fully developed). Once we break free of the most parochial definitions involving carbon-based chemistry, we’re left with the general ideas that life is something complex, something that processes information, something that can evolve, something that takes advantage of local entropy gradients to make records and build structures. (Probably quantum computation does not play a crucial role, but who knows?) One of the first people to think in these physical terms was none other than Erwin Schrödinger, who was mostly famous for other things, but did write an influential little book called What Is Life? that explored the connections between life and thermodynamics.

Searching for a definition of “Life” is a great reminder of the crucial lesson that we do not find definitions lying out there in the world; we find stuff out there in the world, and it’s our job to choose definitions that help us make sense of it, carving up the world into useful categories. When it comes to life, it’s not so easy to find a definition that includes everything that we would like to think of as living, but excludes the things we don’t.

Milky Way

For example: is the Milky Way galaxy alive? Probably not, so find a good definition that unambiguously excludes it. Keep in mind that the Milky Way, like any good galaxy, metabolizes raw materials (turning hydrogen and helium into heavier elements) and creates complexity out of simplicity, and does so by taking advantage of a dramatic departure from thermal equilibrium (of which CV readers are well aware) to build organization via an entropy gradient.

Update: Unbeknownst to me, Carl Zimmer had just written about this exact topic in Seed. Hat tip to 3QD.

  • lylebot

    Does the Milky Way reproduce?

  • Sean

    No, but neither do a lot of people.

  • island

    Not to brag, (too much), but I’ve been saying this for a very long time before Lenny every latched onto it, and Wilikins has known about this for just about as long:

    Course… I was naive back then as my terminology betrays, and thought that an entropic anthropic principle would be welcomed by science… lol@me.

    Jame Kay, Eric Schneider, Dorion Sagan, and Scott Sampson have also made the local connection to this:

    This is Sagan and Kay’s book:

  • island

    Does the Milky Way reproduce?

    Does the universe evolve?


  • lylebot

    No, but neither do a lot of people.

    Sure, but most have the capability, and they’re part of a group (Homo sapiens) in which the prototypical member is able and willing. I’m just saying that ability to reproduce is a pretty common part of the definition of “life”, but you ignored it in this post.

  • Lee Kottner

    I’m having trouble with the “make records and build structures” part, which precludes anything but intelligent life (unless you include fossil records and stuctures like external shells or coral colonies). Surely the presence of something like bacteria found elsewhere in the solar system would be regarded as “life” under most definitions. And I have to agree with lylebot about the reproduction part. No life without some kind of reproduction. I’ll even give straight-up replication the nod. Ya gotta kick against entropy somehow.

  • Low Math, Meekly Interacting

    I’d say the ability to generate a reasonable fascimile of oneself and evolve over generations of replications are probably THE only universal qualities of life anyone will be able to agree upon. Those attributes generally require the ability to extract energy and raw materials from the evironment self-sufficiently, but entities like viruses make even those criteria controversial to this day.

  • island

    The trick here is “self-regulation”:

    The Gaia hypothesis is an ecological hypothesis that proposes that living and nonliving parts of the earth are viewed as a complex interacting system that can be thought of as a single organism. Named after the Greek earth goddess, this hypothesis postulates that all living things have a regulatory effect on the Earth’s environment that promotes life overall.

    Ecosystems are systems where energy flows and material cycles are maintained in anapparently stable, but non-equilibrium state through a process of self-regulation. Such adefinition does just apply to biological systems, it can also apply to systems that involveentirely physical processes. We discuss how four systems, each operating on very differentspatial and temporal scales, each exhibit these features of an ecosystem. These are, in orderof increasing magnitude, the cell, the forest, the Earth and the Galaxy. In particular, wediscuss how the process of star formation across the spiral arms of galaxies works as anecosystem. The carbon abundance plays a crucial role in both the self-regulation and theevolution of the system. We suggest that spiral galaxies may be the first ecosystems toform in the Universe after the Big Bang.

    Ecosystems are systems operating far from thermal equilibrium, driven by flows of energyfrom one part of the system to another, whose raw materials are continually re-cycled asthey pass through the components of the system. Ecosystems operate autonomously, by aprocess of self-regulation. Their flows of energy mean they cannot remain static with time.Although autocatalytic cycles may exist at times (when the system returns to its startingstate), as the Universe evolves so must the system if it is to be maintained. That evolutionoccurs by a process of natural selection, though the use of that term should be widened toinclude purely physical mechanisms as well as those involving biological organisms.These common traits are exhibited, though in vastly different ways, by the four systems wehave discussed; the cell, the forest, the Earth and the Galaxy. These systems each operateon very different spatial and temporal scales from each other. Each system also showsstructure over a range of scales, they are ‘critical systems’. A system showing structure onall scales cannot be in thermodynamic equilibrium, but must continually change with time,as energy flows through it, from one scale to another. This provides a link between theecosystems we have described, as they cannot act completely independently of each other.The Universe is displaying the characteristics of self-organisation through the ecosystemsoperating at different levels within it (see Smolin 1997 for a fuller discussion of thisbehavioural aspect). It has produced the Galactic ecosystem, which has produced the Earth,which has produced life.

    The self-evident prediction here is that the structure of the “eco”-balanced universe is also “self-regulating”.

  • island

    “i” said:
    The self-evident prediction here is that the structure of the “eco”-balanced universe is also “self-regulating”.

    Rats I forgot to include that Eddington also thought that the cosmological constant version of the general-relativistic field equation expressed the property that the universe was “self-gauging”.

    That’s just a “minor little detail” thought, and everybody thinks that Eddington was nuts anyway, so no big deal… 😉

  • archgoon


    What about non-breeding worker ants?

  • Sean

    “Reproduction” is not at all a crucial aspect of what we think of as “life.” If a self-aware golem spontaneously generated out of the ooze, walked around scratching itself, played checkers, wept at La Boheme, and gently passed away without ever reproducing, it would be a useless definition indeed that refused to admit that such a being was alive.

  • Jason Dick

    I’ve always thought that a good definition would be as follows:

    The entity in question can be classified as a member of a group of entities whose evolution through time can be accurately described through the theory of evolution. This requires that at least some members of the group in question are capable of reproduction, and all members of the group were produced via reproduction (that is, produced by other members of the same group). It also requires that there be some sort of selection process that makes it so that certain members of the population are more likely to have successful offspring than others, based upon variation within the population (this means, for example, that your typical computer virus is not alive, because though it reproduces, there is no variation introduced by the process of replication for selection to act upon).

    This definition handily removes stars, planets, and all other astronomical objects from being alive. It would include viruses (unlike some definitions of life), and would also include any computer program that has the requisite properties (frequently used in evolutionary computation).

  • Dirk Vertigan

    Self-Replicators play a much more fundamental role in physics than most people realize.

  • marty

    three words: replicating entropy pumps.
    can anyone do it in two words?

  • Reginald Selkirk

    Larry Moran brings up an interesting historical tidbit:
    Waiting for the paradox

    In the mid-20th century a number of physicists became involved in biology, hoping to discover the new physical laws that made “life” work. Eventually, to their great surprise, it became clear that “life” does not require new physics or new chemistry.

  • island

    I think that “life as we know it” can be distiguished from other layers of similar energy forms by the fact that life as we know it metabolizes raw materials (turning hydrogen and helium into heavier elements) and creates complexity out of simplicity, and does so by taking advantage of a dramatic departure from thermal equilibrium… more efficiently than any other level of comparison.

    As this applies to intelligent life… among other things, we make matter/antimatter pairs much more efficeintly than black holes and supernovae do as we more “naturally” maximize work.

  • Count Iblis

    I agree with Island #8 and Sean #11. What we mean by a living organism is, in general, some sort of machine that takes care of itself (tends to maintain homeostasis).

    A star can perhaps also be considered as a very simple machine (it regulates its energy production). So, I guess its just a matter of degree. There exists a spectrum of machines ranging from very simple systems (galaxies, stars) to very complex systems (humans).
    Of course, a star when described exactly is not at all simple. But considered as a machine it is simple.

  • Doug

    A version of ‘What Is Life?’ by Erwin Shrödinger is available on the web.

    There may be a subtle difference between:
    What is life?
    What is existence?

    Both, probably, may be treated as mathematical objects through mathematical game theory?

    Who needs von Neumann Algebras when there is a von Neumann Minimax Theorem?

  • Low Math, Meekly Interacting

    Golems are the product of intelligent design and/or miracles, neither of which seem to occur in nature. The only way known to get from single cells to opera fans is through evolution, which requires slightly-less-than-perfect reproduction. The only hypothesis I’m aware of that doesn’t resort to the miraculous to explain how we got from molecules to cells in the first place was the spontaneous generation of the first pre-biotic replicators. Perhaps those were single molecules that could catalyze their own reproduction. No one really knows. It’s hypothesized that the pre-biotic replicator or replicators increased in complexity through reproduction, mutation, and natural selection. Molecular evolution, if you will. Were those putative molecular self-replicators alive? Depends on who you ask at this point. They might satisfy the minimum criteria, according to some.

    I think the idea of synthetic organisms does throw a spanner in the game of defining life, though. We’ve been watching the evolution of simulated organisms for decades now, and studying them is an important part of research into evolutionary theory. Are those little cellular automata and their ilk alive? Does answering “no” render the “evolving replicator” definition useless? Does it matter that it’s a snippet of computer code instead of a “natural phenomenon”? Tough call to make.

    I suppose it’s conceivable that some day we will make a machine that is completely self-sufficient, and perhaps even self-aware. Would it be alive? It’s hard for me to have an opinion at this point since nothing of the sort is known to exist, and we don’t know if it ever will.

    Right now, though, given what we know of how things actually happened in the “real world”, the way life arises at all seems to be though the spontaneous creation of a self-replicating entity that evolved greater complexity, speciated, etc. Theoretically, reproduction and evolution are at the very foundation, and without them you don’t get past square one.

  • jeff

    Once life has been adequately defined, someone will inevitably attempt to explain it’s meaning. Oddly enough, the question of meaning goes back even further than the question of definition – perhaps all the way back to the first creature in the universe with enough sentience to climb to top of a small mound, survey it’s surroundings, and think to itself, “what’s all this, then?”

  • Jolly Bloger

    I’ve always thought that a good _general_ definition of life is anything capable of undergoing Darwinian evolution. Of course I have to say general, because that definition excludes things like worker ants, mules, and childless humans. I quite like the extended version Jason wrote above.

    Computer viruses, or programs specifically intended to imitate life, are tricky because of the environment in which they exist. They don’t so much exist inside a computer or on a hard drive as within a theoretical mathematical matrix and set of algorithms, of which the computer is a model. A particular magnetic domain cannot be said to be part of a piece of code the way a cell or even a carbon atom is a part of a conventional life form, because the code itself can freely move across bits. The magnetic landscape is more like an environment than parts of an organism. So what is a computer virus actually MADE of? It’s just information, isn’t it? Do we also have to add corporeality to the definition of life? An idea can evolve in a Darwinian fashion too, but we call that a meme rather than another form of life.

  • Jason Dick

    Bear in mind that information is all that any life is: we are nothing more than a transient configuration of atoms. The specific atoms that make up our own bodies are entirely replaced every few years. What we are is not the matter that makes us up, but rather the information that tells that matter which structures it is to take.

    That being the case, I have no problem calling the right kind of computer program alive. It’s just a different medium than our own carbon-based life.

    Bear in mind that your typical, run-of-the-mill computer virus is not alive, because all it does is replicate itself. There is no variation introduced in this replication upon which natural selection can act. But people have been working with simulated life for some time, and it turns out that evolutionary algorithms can be very useful in finding good solutions to problems. Personally I wouldn’t hesitate to call such a program alive, as it would be difficult to come up with a functional definition that would exclude such things.

  • The Celestial Toymaker

    In theory, a computer could be programmed to reproduce itself and the program running on it.
    But the idea that life could evolve as a program, or the universe is a “computer simulation” would seem to violate the “principle of least action”, as stated by Maupertuis:
    “Le Grand principe que la nature, dans la production de ses effets, agit toujours par les voies les plus simples.”
    I suspect that anything that bypassed nucleosynthesis and the evolution of replicant organic molecules would have led to a much more rapid increase in the entropy of the universe.
    So ulitmately, the localised low entropy that made the “computer life” possible would break down faster than its ability to replicate itself.
    In order to be adaptive and self-sustaining, it would need to evolve towards more intelligence, but that evolution would go towards infinity in any computer based on floating point arithmetic (continuum number)
    So there’s almost certainly a cut-off point in the evolution of intelligence.

  • island

    You have to think about the entropy of a closed expanding system that has an increasing negative pressure component that is inceasing at an accelerating rate, but is most apparently being counterbalanced/regulated by in equally effective increase in gravity.

    You don’t like it, and you want to project a one-sided feature into this coin, but that’s not what the geometry most apparently indicates without assumptions that aren’t necessarily justified, and since there is physics that does justify the process for what is observed… there is a most peferred theory, until proven otherwise by a complete theory of quantum gravity that doesn’t require some modification to the negative energy soloutions and the vacuum state.


  • Quasar9

    As Sean has pointed out Mules are alive but cannot reproduce
    one of the drawbacks or benefits of ‘hybrids’ – depending on your preference

    Are sterile women (and men) not ‘alive’
    Are women on contraception not ‘alive’
    Are men who’ve had the snip not ‘alive’
    and what did monkeys evolve from?

  • The Celestial Toymaker

    Quasar: “As Sean has pointed out Mules are alive but cannot reproduce”

    Aren’t mules made of DNA?

  • bob

    Sean’s golem may be fantastical but it demonstrates that the quality of awareness is sufficient to define life.

    And now that it appears that self-organizing, helical, information-processing structure may be a feature of inter-stellar space, the golem or an SAS or Boltzmann Brain is not so fantastical after all.

  • The Celestial Toymaker

    re #27 Awareness doesn’t help much either. Is a bacteria ‘aware’?

    I think the B.B argument (whether applied to organic or computer based “life”) is a negative demonstration of big bang nucleosynthesis and the theory of evolution.

    It shows that the extraordinarily improbable is always overtaken by the near inevitable – biological life forms based on carbon will evolve into life under the right conditions

    Long before a BB could emerge out of a chance quantum fluctuation, the universe would have moved to a state where molecular structure was impossible, unless there was a new big bang.

    An infinite growth in organisation and information just can’t happen.

  • Count Iblis

    Also, we can think of whether a group of organisms that lives together form a “superorganism”. E.g. we humans consist of cells and are clearly organism in our own right. Is an ants nest also a superorganism? Or groups of people living together?

  • Pedro J.

    What about a sufficient condition for a system to be considered “living”. Barrow and Tipler proposed a system at least composed by subsystems capable of self-reproduction in some enviroment and containing information which is preserved by natural selection.
    The Anthropic Cosmological Principle. §8.2

  • Lee Smolin

    Dear Sean,

    This is a great post, can I offer a definition of life from my first book, Life of the Cosmos, p 156:

    “A living system is

    A a self-organized, non-equilibrium system (see below), such that

    B its processes are governed by a program which is stored symbolically, and

    C it can reproduce itself, including the program.”

    On p 155 I defined a self-organized, non-equilibrium system to be “a distinguishable collection of matter, with recognized boundaries, which has a flow of energy, and possibly matter, passing through it, while maintaining, for time scales long compared to the dynamical time scales of its internal processes, a stable configuration far from thermodynamic equilibrium. This configuration is maintained by the action of cycles involving the transport of matter and energy within the system and between the system and its exterior. further the system is stabilized against small perturbations by the existence of feedback loops which regulate the rates of flow of the cycles.”

    The point is that the biosphere as a whole, and perhaps a stable disk galaxy may satisfy the definition of a self-organized non-equilibrium system, and, while that is interesting and worth noting, living organisms made of cells, satisfy the further conditions B and C. On the other hand, computers, computer programs, and robots do not satisfy A, even if they can be built to reproduce themselves.

    One could also emphasize the existence of a semipermiable membrane, the flow of matter through which is under the control of the cell-as it is possible that the self-construction of such membranes played a key role in the origin of life, as discussed in adap-org/9709004.



  • bob

    re #28: I don’t know whether bacteria are “aware” although they could be and a discussion of what constitutes “awareness” would be an interesting next question. I personally find the term preferable to “consciousness” which carries a lot of baggage.

    But my point was that awareness is a sufficient, not necessary, defining characteristic.

  • Count Iblis

    Lee, it is not so clear to me why robots (once they are fully capable of maintaining themselves and taking care of all the facilities that support them) would not satisfy A. It seems to me that if A is not satisfied then they would not be able to last very long…

  • jeff

    it is not so clear to me why robots (once they are fully capable of maintaining themselves and taking care of all the facilities that support them) would not satisfy A

    Not only material robots, but also software simulations of arbitrary complexity, with subsystems that are not at “equilibrium”. Does the containing system have to be our universe? Or is there something inadequate about Turing equivalence (perhaps properties that the universe has which Turing machines do not?), making “life” (as per Lee’s definition) impossible in a software environment?

  • Lee Smolin

    Count, by robots I mean robots as they exist now, such as those that do repetitive tasks in factories. I do not claim it is impossible that we could construct artifical living things.

  • The Celestial Toymaker

    I read Lee’s book a few years ago and quite enjoyed its arguments, while not agreeing with them all. But I’d rather not thumb through it again. So I’ll just talk off the top of my head instead!

    I don’t really see how galaxies can be described in any sense as “living”, any more than the sun is. They are just self organised collections of matter that are able to reduce entropy over very long timescales.
    Only by a very circuitous argument could they be said to reproduce themselves, although there is a form of evolutionary cycle at work.

    So let’s say that they form a precondition for life e.g carbon, oxygen & nitrogen nucleosynthesis and the supply of radiant energy against the background of cold space. Life is a thin film of exotic chemistry based on that.

    I’d see any “awareness” that life has developed as an evolutionary mechanism, but not a definition. The sense perception mechanism would be better.
    It enhances survivability, but is something an inanimate self organised system can’t have.

    Once you can store sense perceptions and have the ability to process them, you have the rudiments of awareness – i.e. a mechanism where the organism is able to evaluate the effects of different courses of action.

    As to the “robotic argument”, I agree it’s a theoretical possibility that we could construct artificial living things, but I don’t think the reverse is true, for the reasons outlined up the thread.

    Hence it’s not an argument that the evolution of life, or even a spontaneous machine organism could result from such a mechanism. I suppose there’s a rather unsettling argument that in the interests of survival, humans might increasingly try to incorporate themselves into longer-living machine intelligences. That way lies the Borg……

  • Sean

    I still don’t see why the ability to reproduce has anything to do with whether we would classify something as “alive.” I’m pretty sure I can imagine coming across a novel organism and judging that it was unambiguously alive without having any idea whether it could reproduce. Why is that property so important?

    Otherwise, Lee’s definition does a pretty good job at formalizing the information-processing part of the definition. The “running according to a program” aspect is an interesting idea that I don’t think I’ve heard before. I would have to think about whether there could be counterexamples where there is no such program even though we would all agree that something was alive.

  • Amara

    I’ve written about a dusty plasma physicists’ neat idea for life on Clifford’s blog here , where we discussed scalings for life: here and there was a related thread on CV here.

  • Freiddie

    Life definitely includes things that can ask this question (“the meaning of life”), right?

  • macho

    The argurments against reproduction make sense for individuals, but how do you make sense of evolution (another one of your criteria) without some kind of reproduction? Your Golem doesn’t evolve — neither do I. Evolution usually refers to the change of an entire population/species, not an individual. And it always assumes the reproduction of new individuals within this group.

  • Low Math, Meekly Interacting

    It’s very difficult for me to imagine how to produce something with the other qualities of life “naturally” if that something lacks the ability to reproduce (which it must have inherited), because the alternative appears to necessitate some kind of miraculous saltation from the primordial schmutz. From the get-go, if we all agree that “life” has a kind of…er…negentropic complexity that is self-sustaining, how could it possibly come to be without a miracle (or an inventor)? Evolution is all my poor little brain can come up with that fits what we know about the natural world. And if something doesn’t reproduce, how can it evolve? So, again, if reproduction isn’t part of the formula, I have a hard time imagining how there could be life at all. Sure you get a celibate or sterile individual here or there, but even if they don’t reproduce, they owe their very existence to reproduction if they are “natural”.

    Always fun to ponder this question! I agree that before humanity can get serious about exobiology, we needs us an answer.

  • Russ Abbott

    You may be interested in my paper “Emergence Explained” (Complexity, Sept/Oct 2006) and my 2 page abstract submitted to the upcoming Workshop on Philosophy and Engineering “Bits don’t have error bars.”

  • Sean

    I can certainly imagine an individual evolving. Besides which, I’m by no means certain that evolution (in some Darwinian sense) is a necessary part of the best definition of life (which I don’t claim to have).

  • Quasar9
  • The Celestial Toymaker

    #37 Actually, I think the reproduction definition is essential to defining life, as we know it (Jim), whereas running a program is a highly questionable definition.

    Programs are artefacts of created by pre-existent living creatures that have evolved and reproduce. Running a fetch-execute cycle of stored instructions doesn’t seem to be the way living things operate and nor is it proven that the universe (as a whole) is algorithmic.

    Max Tegmark’s article/paper called the “Mathematical Universe” has some useful discussion on this topic and of course, it relates to Alan Turing’s original preoccupations with thinking machines.

    Unless he operates under a pseudonym here, it’s a shame that Tegmark doesn’t participate in these discussions – I’d like to see a Tegmark, Carroll, Smolin smack down.

  • Count Iblis

    I think it’s wrong to include things like reproduction and evolution in the definition of life. These are essential mechanisms making it possible for organisms to have appeared naturally (without interference of intelligent beings), but not essential for a thing to be alive.

    Why not just keep Smolin’s point B (see post # 37):

    “its processes are governed by a program which is stored symbolically”

    The “which is stored symbolically” is not necessary i.m.o.

    If an organism is identified with a certain program then from the perspective of that organism, or from our perspective, it doesn’t matter much if you modify it so that it doesn’t satisfy the other criteria anymore.

    I would still be the same person if I were replaced by a robot with a computer in its head that implements my neural network.

  • Count Iblis

    I made a typo, Smolin’s post is #31 not #37

  • Elliot

    Here is another stab at a definition from Jerome Rothstein

    “In the broadest sense, he describes generalized intelligent life forms as self-replicating, computer-controlled heat engines that are able to play survival games.”

    Note this definition is of “intelligent” life not just life.


  • macho

    I’m going to go back to the reproduction argument, since I think this is central.

    1) In defining an living organism, we need to think a bit more broadly than individuals.
    The behaviour or characteristics of the entire “biosystem” is key.

    2) Considering the larger organism, all life as we know it shares one key trait — its processes are ultimately focused on continuing the existence of the life form.

    Keeping humans out of the equation so that we can avoid politically and emotionally charged areas, it’s still important to consider the larger organism. There are plenty of examples in the plant and animal world where individuals
    cannot or choose not to reproduce themselves (wolves, ants, bees) yet the
    overall group is designed to keep the species going.

    There are also many examples of self-sacrifice and altruistic behavior in animal groups, but these behaviors are also tied to the overall survival of the species.

    In multicelled creatures the role of individual skin cells or liver cells is not necessarily reproduction, but the overall organism coordinates the behavior of the subunits to keep the larger creature going.

  • Count Iblis

    Another related problem is how to decide if some piece of matter contains a certain specified organism.

    This seems to be a rather trivial exercise once the organism is defined e.g. as a machine that executes some program. However, let’s now consider the fact that information is conserved according to the laws of physics. The present state of the universe is related to the state the early universe was in via a unitary transformation. So, in principle, one can say that we exist also in the early universe (but subjectively experiencing the present state of the universe, of course ).

  • B

    I am wondering whether one can come up with a sensible definition without employing any scales at all? Imagine a living system that ‘operates’ (acts?) on timescales such that we don’t take notice of it? That might de facto be impossible, but the reason for this would be a scale dependence of certain features, like formation of structures, stability, equilibration times etc.

    Besides this, the problem with the notion of ‘life’ is that in absence of a definition we might not all understand the same under it. E.g. read the comments above and the issue of ‘reproduction’. Sean (#11) thinks reproduction isn’t a crucial requirement as long as the ‘living thing’ can scratch itself, whereas Lee (#33) includes it. Whether or not it’s a requirement to be included in what Sean calls the ‘best’ definition is not a question one can really answer with yes or no, since a definition is exactly what one is looking for. One or the other definition might be more or less in agreement with ones intuition, or ‘better’ depending on what we think it should cover and be useful for. At least for me, my understanding if ‘life’ is influence by the 7 points I learned at school (see Wikipedia: Life, that includes reproduction).

    @ # 27,28 Calling on ‘awareness’ doesn’t help either, since it’s just another word that lacks an appropriate definition.

    I’d put my hopes on complexity that could turn out to be a useful characterization of essential features, and one that could be properly definable.

    Occasionally I can’t avoid having the impression that life is characterized by the urge to do completely useless things, such like writing comments on other people’s blog or books about the meaning of liff. But more seriously, how would evolution work without trial and error? Or is that just an illusion and part of the ‘program’ we’re running on?

  • Paul Valletta

    There are different levels of structures that form part of “life” systems, as Sean state the milky way may not be alive?..but it certainly contains living organisms. Those living organisms contained within a standard milky way, also contain living organism systems ( bacteria inside humans, or head-lice outside humans ).

    The chemical/molecular structures are evidently more complex than we can label them, can we catagorize the source of “life” as being:

  • jeff

    Running a fetch-execute cycle of stored instructions doesn’t seem to be the way living things operate and nor is it proven that the universe (as a whole) is algorithmic.

    But that’s not the point. Lots of things aren’t what they seem to be, and Turing machines are equivalent to many things that don’t look like them at first glance.

    The real question is: Is there any property above and beyond a theoretical Turing machine that is required for “life”, whether or not the universe has this property or not? If the answer is no, then software life is possible. If the answer is yes, then you need to identify that special property.

  • Jason Dick

    Well, I guess I’d like to provide an argument as to why reproduction should be an essential component of the definition of life (not the reproduction of the individual, naturally, but of at least some members of the group at large).

    First, let’s consider the purpose for generalizing the definition of life. Our foundational purpose, I claim, would be to see if we should behave towards these potential organisms as we behave towards living things here on Earth. This is the foundational purpose behind forming categories, after all: so that we can simplify our interactions with the world around us, to better understand it and better interact with it.

    If you accept this proposition, then it follows that the best general definition would necessarily be a functional definition. As such, I claim, we should be concerned not with what life is, but rather with what life does. And the fundamental behavior that all life on Earth follows is properly described by the theory of evolution. One might also add a few other things, such as the “entropy pump” definition, but I suspect that they would be redundant. One addition that may not be redundant is that life is capable of affecting its own environment (this would likely eliminate the evolutionary programs I mentioned previously).

    If we make use of another definition than a functional definition, we might end up including something in the category of life which behaves in a fundamentally different manner from the life we recognize here on Earth, which, in turn, makes the inclusion in the category rather unhelpful.

  • Chemicalscum

    Sean #11

    You mean like Hoyle’s Black Cloud

    Of course his Black Could and Sean’s Golem have the same improbability as Hoyle and Wickramasinghe’s “tornado in a junkyard” created 747. They meant this as an argument against abiogenesis not evolution. However this is naive, in that it ignores that abiogenesis is itself a stepwise process dependant on selective processes. Firstly chemical evolution which is not dependant on a replicator but is instead a secular evolution dependant on competing kinetic systems resulting in complex polymers. This provides a basis for the evolution of small self-replicating molecules which start the evolution of replicators.

    My point is really, that for any form of life to evolve anywhere with any chemistry (OK here I refer to chemistry, but life forms not based on chemistry using other physical processes instead, where the energy is to high for the electromagnetic force produce molecules may be possible) there must be a replicator. So the formation life is dependant on a replicator to come into being.

    From a modal realist point of view it is possible to argue that Hoyle’s Black Cloud and Sean’s Golem exist somewhere in the multiverse, however they are so highly improbable that the probability that they or anything like them exists in our universe is vanishingly remote. Therefore any life form we are likely to encounter must be the result of evolution and therefore a replicator.

  • Neil B.

    I think that capacity for awareness is something found only in “life”, but not necessarily the converse (not all life is aware.) I know, “awareness” is hard to define, but so oddly enough is “stuff” as I have pointed out here before (and check my link if you’ve forgotten.)

  • Low Math, Meekly Interacting

    “As such, I claim, we should be concerned not with what life is, but rather with what life does.”

    I think this is superbly put. Because what life “is” is a collection of matter. So is a rock. It’s properly distinguishing the two that is the confounding problem, and how that collection of matter comes to be distinguishable at all is of paramount importance. Without resorting to vitalism, what have we got? To me talking about something living without reproduction and evolution seems a bit like talking about something living without there being “life”.

  • Elliot

    I belive the ability to replicate a similar version should be part of the operational definition of life. I believe there should be a distinction between a complex information processing object, and an object that extracts energy from the environment and within its boundary the second law of thermodynamic is locally violated. I think there is an important but subtle distinction.


  • Kepler2

    How can such little creatures as we know if the
    galaxies are alive or not? Do the bacteria in
    our intestines know they are part of a much larger
    living system?

    Galaxies interact in ways that appear to be alive.

    Carl Sagan once said that intelligent beings were
    a way for the galaxy to know itself. I wonder if
    little collections of tiny creatures on scattered
    balls of rock are enough to do the trick?

    Maybe in those energy fields surrounding the really
    big black holes in galactic centers are more “worthy”
    of such massive entities as galaxies. See Gregory
    Benford’s SF novel Eater for a related concept.

    We are too buried in the “trees” called stars to
    know enough about the “forest” called the Milky
    Way galaxy. Living galaxies – would not surprise
    me one bit.

  • Andre
  • Arun

    If a self-aware golem spontaneously generated out of the ooze, walked around scratching itself, played checkers, wept at La Boheme, and gently passed away without ever reproducing, it would be a useless definition indeed that refused to admit that such a being was alive.

    You just previously admonished us that our definitions are supposed to make sense out of the stuff that is there, not the stuff that we imagine but doesn’t exist.

    If such a golem could exist, or angels or God we’d have a problem too.

    So it seems to me that you’re not doing what you’re saying.

    Wasn’t one of the crucial advances in biology the experimental refutation of spontaneous generation? So it became clear that biology did not have to deal with these kinds of golems. Now you’re saying biology is useless unless it can deal with these kind of impossibilities.

  • Jolly Bloger

    Living galaxies is a really interesting idea, and not theoretically impossible either. If galaxies as a whole are able to undergo Darwinian evolution, then I would say that yes, they are certainly alive. The problem is with the evolution.

    First, are galaxies able to produce copies of themselves with high copying fidelity, but some slight random changes to their properties (the changes themselves able to be passed down to the next ‘generation’)? I don’t think they are. There is a theory about black holes being able to evolve (I think I read it in Hawking… not sure) but I can’t think of a mechanism for galaxies to pass on their individual properties. Doesn’t mean it can’t be done though.

    Second, there would have to be some means of competition. The above paragraph is about the heritable random mutations, the other side of evolution is natural selection which requires competition for scarce resources. Again, I can’t think of how this could play out among galaxies, and even if it could, they are simply too remote to interact. Galaxies so rarely come close enough to one another to be able to have any effect, competition and thus evolution and thus life (ignoring the golem or black cloud as philosophical considerations, all natural life surely requires evolution to form).

    Finally, and related to the remoteness in space of galaxies, there has not been enough time. Life on earth has been ticking along for billions of years to get to where it is. For much (most?) of that time, each generation (from birth to reproduction) was on the scale of minutes or hours, and even now on the scale of a few dozen years for the longest generations. If galaxies were able to evolve, surely each generation would take them millions if not billions of years, and they’ve only had about 14 billion in total. Not enough for any real ‘life’ to form from galactic ‘nonlife’.

    If there is a way for galaxies to satisfy any definition of life without having evolved I’d be extremely interested to hear it.

  • Elliot

    What about a “boundary” or membrane? Does there not need to be a clearly defined surface separating something that is alive from the rest of the cosmos? I believe there does. But does it need to be a solid boundary or can a gravitationally bound glass cloud potentially be living.


  • Elliot

    another related definition by Stuart Kaufmann who has deeply investigated this area.

    “An autonomous agent is something that can both reproduce itself and do at least one thermodynamic work cycle.”

    are all autonomous agents alive? open question
    is everything alive an autonomous agent? probably


  • Arun


    Or else we could say that since physics doesn’t cover many events in Harry Potter stories, it is quite a useless science.

  • Quasar9

    If the measure of life is the ability to produce life
    then the (whole) Universe or multiverse is alive.

    Even if biological life became extinct in the Universe
    that would not automatically make the Universe dead
    The Universe exists before & beyond biological life.

  • Elliot

    correction to #63 above “gas” cloud not “glass” cloud.


  • Neil B.

    Maybe it’s time to remind everyone: words and the boundaries of definitions are a snare: there are all sorts of things with all of the properties we ascribe to “life”, some that have some properties in various portions but not others (like the sterile living beings), etc. That what things are, and we shouldn’t fall under the territory of names.

  • Elliot


    Agreed but there should be some precise “test” determining if a particular thing alive or not alive. Deconstructing the components should in principle help in identifying what that test is. It should be a bit more scientific than for example the story about legal definition of pornography where someone said to paraphrase “I can’t give you a definition but I know it when I see it”.


  • Jolly Bloger

    I can see fuzzy boundaries and definitions within the category of “life”. For instance, the boundary of ‘species’ is definitely not rigid. You could very easily have a broad spectrum of creatures between chimps and humans – in fact such a spectrum has existed, just not all at once. It’s an accident of nature (not exactly an accident, but by no means logically necessary) that species are usually so separate from one another.

    But the boundary between life and non-life surely must be clear, mustn’t it? To say that “life is that from which life comes” and then to declare the entire universe alive is skirting the issue a bit. It’s skewing definitions and bordering on metaphysics. Surely we can come to a satisfying definition that includes all life and excludes all non life without getting too mystical.

  • Jolly Bloger

    “If the measure of life is the ability to produce life”

    One more comment on this: it’s circular reasoning. You can’t include the concept of life in its own definition, you’ll end up with gobbledygook like “the universe is alive”.

  • Quasar9

    “You can’t include the concept of life in its own definition”

    lol – no wonder you call yourself the jolly blogger.
    If there is no life in it – we tend to call it DEAD and/or inanimate.

    “If the measure of life is the ability to produce life …”

  • B

    @71/72 The criterion X can reproduce X isn’t an empty statement. The non-circularity lies in ‘reproduce’ (F ‘reproduces’ F under derivation is a perfectly useful definition for the exponential function). It would be a circular reasoning if you’d say something like ‘life is what is aware’ and vice versa define ‘awareness is what characterizes life’.

  • Alan

    Whats wrong with “replicates”

    Everything that does “replicate” is life and everything that doesn’t “replicate” is not.

    Making “man” = life and “a virus” = life and “the universe” = not life

  • Alan

    If you can think of any reason why that very simple definition is not ok then mail me

  • Jolly Bloger

    Quasar, I’m the Jolly Bloger, not Blogger 😉

    B, I agree it isn’t meaningless to say that a quality of life is the ability to replicate itself, but I still say you can’t include a word in its own definition. That’s just basic lexicography. To say that the definition of life is “something that produces life” is certainly meaningless because you still haven’t said what life is!

    Alan, your definition has some of the same problems as more in-depth ones above as it excludes non-reproducing organisms like childless humans or worker ants. In fact, humans don’t replicate at all. Replication implies a clone, asexual reproduction. Also, your definition would include any kind of Turing machine (a robot programed to build robots just like itself). Some would claim those do constitute life, but I think most people have a gut feeling that there’s something more involved.

  • B

    blog threads replicate themselves all the time.

    See also: Meme.

  • Jolly Bloger


    Again, one further comment. It’s one thing to say that life is that which reproduces ITSELF and quite another to say “life is that which produces life.” To use your exponential function analogy, its valid to define it as “F reproduces F under derivation” but you can’t define an exponential function as “anything that makes an exponential function.”

    If you built a machine that would assemble molecules to make lemmings, so you have a big device with perfect fully formed lemmings popping out, the lemmings are life but the machine is not.

  • B

    Hi Jolly: I was about to say the same as you wrote in #78. You can include ‘itself’ in a definition. I.e. take 0 = x* 0 for all x as a definition for 0. Whether using a word in its own definition or not is good praxis is arguable. but you can’t define an exponential function as “anything that makes an exponential function.” No, but the problem is not the second use of the word ‘exponential function’ but the verb ‘make’. You could very well say an exponential function is defined through it’s derivative also being an exponential function, i.e. f’ = f. Zero is a number that yields again zero under multiplication with an arbitrary other number etc.

  • Elliot

    Is it possible that some of the difficulty in this exercise is that we are attempting to give a “physical” definition to life and that life as we know it is biological (2 levels of complexity away from pure physics via chemistry) is generating some of the confusion. Just a thought.


  • Jolly Bloger

    Nono, the sentence “an exponential function is defined through it’s derivative also being an exponential function” is not the same as f’=f. With the latter, you’re already taking as a given the definition of a function and the definition of a derivative, so there’s a very strict relationship between f and f’.

    You still couldn’t say “an exponential function is something that produces an exponential function under derivation” because who knows what an exponential function is? Armed with only that definition you might think that f=x^2 is exponential. You cannot exclude it using only your definition. As long as you have a differentiable function, your definition will be logically valid, so its just as useful on f=x as it is on a real exponential function defined the usual way as f’=f.

    It’s like defining ‘green’ as “the colour of green things.”

  • Jolly Bloger

    We’re getting away from the actual discussion with all this philosophical talk about what is a definition. I’d like to move back to the issue.

    Elliot, thats just the thing – we’re trying to come up with a definition that goes beyond biology. Biology itself is defined as the science of life, so that’s only as good as we’re able to identify what life is. Life “as we know it” (i.e. anything described by biology) is a small subset of what we can conceive of as life. All life that we know is carbon based, for example, and most is written in the exact same language of DNA. A comprehensive definition that includes all possible forms of life must go beyond this.

  • Elliot


    My point is that life may be an emergent phenomenon not subject to an all or nothing physical definition. In other words it pops up here and there not necessarily carbon based. If life is to be given a definition or a “physical life test” developed (this thing is/is not alive) doesn’t that imply that life “may” reflect some as yet undiscovered physical law that predisposes it’s development?

  • B

    Jolly, I have tried to explain you why an auto-reference of the object-to-be-defined can indeed make sense in a definition (what you doubted in #71). Whether or not an object with a given definition actually exists, or whether it is unique is another question (and one that needs to be shown in both examples I’ve given above). As I’ve said earlier, if you have a problem with my use of words, just do whatever you like, there’s a reason why I like maths better than philosophy. I agree that my above wording isn’t satisfactory, as the word ‘an’ should have more precisely referred to the same function (otherwise you could define a class of function whose derivative is also in that class). Sorry about that. Since all of this is irrelevant for the discussion anyhow, and it seems you’ve gotten the point with the self-referral, I’ll drop the issue here.

  • Heresiarch

    Do galaxies reproduce? We’ve not been able to observe one through its life cycle, so we can’t know.

    Emergence, complexity, self-organization, and so forth are transparent revivals of the discredited doctrine of Vitalism. Not that there’s anything wrong with that . . . .

    And so on. For endless amusements on the topic of life’s origin and scope, be sure to meander through

  • Self-Replicating Post

    Self-Replicators play a much more fundamental role in physics than most people realize.

  • Jolly Bloger

    For the record, emergence, complexity, and self-organization are not at all tied to vitalism, they are all involved in perfectly scientific formulations of life through evolution without invoking some hocus pocus 17th century life energy. Those concepts are also important in economics and computer science.

    B, I agree it looks like we sort of agreed all along, we just misunderstood each other. I still maintain that it is strictly improper, according to the laws of language, to use a word in its own definition. You never see this in a dictionary :) (unless the word is referring to an above definition of the same word).

  • Torbjörn Larsson, OM

    Old but not dead thread in a fascinating subject. Most of my 2c has been said, but I can amplify:

    I agree with Jason Dick that a functional description is better. My own motivation is that it fits established theory better. Also, in biology distinctions are fuzzy. For example, we are forced to use several species concepts. So we can and probably must use several life concepts.

    [For living sexual populations we can use the “biological species” concept. (Different populations who can’t or won’t mate belong to different species.) But that doesn’t describe asexual populations or fossils.]

    Therefore I propose the following modern concept of life (in analogy with “concept of species”):

    An organism is the unit element of a continuous lineage with an individual evolutionary history.

    The model for the definition is an organism as the current slice in a continous process. Thus combining the idea of life as individual and life as process with evolution and an implicit assumption of am inheritance mechanism, and a robust definition of organism. Quite a few birds knocked down with one stone.

    This definition excludes organelles and such replicators as prions because they have entered dependent niches, as they are subsumed into an organism. But viruses are individual organisms under the definition, as they coevolve instead.

    Similarly such things as memes, stars and galaxies are excluded, as they have no inheritance. While software/hardware such as implementations of genetic algorithms are included. (As noted, such populations may or may not inhabit other environments than we do.)

    Finally, to amplify:

    Viruses and cells from different extant and extinct domains could have crossed the Darwinian threshold several times from a progenotic state. Also, early total extinctions are consistent with the observation of early and so likely easy abiogenesis.

    These crossings are initiated by diverse selfish elements, a mechanism that is often observed in genes. Evolution, especially past a progenotic state, is a competitive [sic!] and robust process which we should expect to be common elsewhere.

    Maybe we will meet singular existences, biological or mechanical, that aren’t described thusly. But I believe the way to bet is that they are rare and bound on an extinction path. We could probably approximate “life” with “populations that obey evolution” as above. Other characteristics are coincidental.

  • Torbjörn Larsson, OM


    Emergence, complexity, self-organization, and so forth are transparent revivals of the discredited doctrine of Vitalism.

    Acute characterization – I haven’t seen it expressed so sharply before.

    Evolution isn’t a “ladder of progress” or other misplaced Victorian concepts, but a symmetric process that can go both ways. The initial conditions was asymmetric, but after 3-4 Ga the majority of biomass is still simple Virus and Bacteria. Simple is better, and I believe biologists think that it is energy requirements that constrain size.

    There is a coincidental tail diffusing towards complexity in lieu of the constraint pushing back. Simpler and faster reproducing Bacteria have outcompeted Archaea to inhabit extreme environments and Eukaryota to exhibit complex phenomes. As these things go, we are the scum of Earth. 😛

  • Low Math, Meekly Interacting

    I’d say even the least complex prokaryotes and viruses are not at all simple compared to your average glob of abiotic matter.

    And even protists share many, many genes with multicellular organisms as large and complex as ourselves. Indeed, biology reveals a remarkable level of recapitulation throughout its putative history. What one sees with multicellular organisms is a new layer of complexity over what can be achieved with a genome surrounded by a membrane, and this arrangement is almost totally constrained by size, most notably the ratio of surface-area to volume.

    I don’t think there is a “better” in evolutionary terms. When there is selective pressure (and even in its absence, due to genetic drift), life tends to fill up all the niches, but there’s no reason to expect any kind of symmetry in terms of number of species or mass of a particular species in those niches, and no value judgement we can readily place on such metrics so long as the species in question is viable from generation to generation.

  • Low Math, Meekly Interacting

    I would hasten to add that if somebody starts in with the vitalistic hocus-pocus of life’s “fundamental theory” or the woolier gestalt concepts, run away screaming, of course. But I think it’s too much to dismiss all talk of complexity and emergence as mysticism. I admit it’s not so well-defined, but IMHO “more is different” is one of the pithiest little scientific observations one is likely to come by, and derserves consideration.

  • Low Math, Meekly Interacting

    Hmmm. Just read the Seed article. I have a feeling I would have been one of the folks yelling at Cleland if I were there. I find the idea that we should give up on defining life until we find extraterrestrials to be rather vacuous, really. That sounds like opening our minds to the point our brains fall out to me. It’s also not at all clear to me why we need any more theory than natural selection to make a reasonable stab at being comprehensive. What the hell else is supposed to give rise to this thing we refuse to even try to define until we find it under a rock on Titan scratching itself? What are we even looking for, then? I also find the argument against Darwinian evolution’s reliance on genes to be specious, quite frankly. If there is not some information in this magical bag of enzymes (how one gets something as complex as an enzyme as a mere constituent is completely ignored, apparently, which itself is outrageous), of what significance is “change”? If it gets dry the water in the puddle evaporates, and the mud turns to dirt. That’s environmentally-driven change. So fricking what?

  • Mark R.

    It’s a little strange to me, defining “life”. To do so, you have to abstract yourself, or separate yourself from it. We might be running up against that whole “nothingness” problem again, only in a variation.

    Part of it seems to depend upon where we choose to look, and on what scale. Cells of the body, and each one alive, composed of un-alive atoms? And we thinkers, an unlikely result of the collective efforts of these cells?

    I suppose if a cell in our body might contemplate its existence, it would have to observe the things around it, or better yet, travel outside the body, finding, perhaps, a gargantuan being, beyond comprehension, scratching itself.

    Or, if planets could talk to one another, might the Earth brag, oooo, look what I’ve done: look at my hairdo, as all we little critters now go springing out into the space surrounding it.

    Everything seems so interdependent upon everything else. It’s all connected, if even just from the Big Bang. It makes me wonder if asking the question, is something alive, is really all that important. What does seem important, though, is a general respect shown toward all our constituency, and whatever constituency we might find ourselves a part, on whatever scale.

    We have suns whose component processes manufacture carbon and metals. We have plants whose component processes manufacture oxygen. We have ourselves whose component processes manufacture thinking about stuff like this.

    And we’ve likely got big clusters of all this stuff going on everywhere. I wonder, is a delineation between life and non-life anything? Or are we just looking for something we recognize to be something like ourselves — something we can, in some more intimate way, relate to? And if that’s the case, maybe it’s just as good to ask the question, what are we, to wonder such things, or to wonder at all?

    Maybe we don’t really need to abstract ourselves from life to define what it is and is not, creating boundaries where none seem, on various scales, to exist.

    I just love to death, though, that we really don’t want to be all alone in it all, though.

  • Doug

    There is an interesting paper on a computer model for one of the primary nucleosides.

    Rainer Glaser, Brian Hodgen, Dean Farrelly, and Elliot McKee
    ‘Adenine Synthesis in Interstellar Space: Mechanisms of Prebiotic Pyrimidine Ring-Formation in Monocyclic HCN-Pentamers’
    Article, Astrobiology 2007, 7, 455-470. PDF. (Publication in Print: June 2007)
    Online Visualization: Chime Displays and Reaction Animations

  • Heresiarch

    Back to that Vitalism thing: The olde conception at least honored the notion of causality. The vitalist doctrine DID propose (various versions of) a “life energy,” but the contemporary versions (complexity, emergence, self-organization, etc.) don’t EVEN do that. They propose that nature violates entropy with NO cause whatsoever. It just happens.This is rational thinking? Czech it out:

  • Pingback: Are Cities Just Very Large Organisms? | Cosmic Variance | Discover Magazine()


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Cosmic Variance

Random samplings from a universe of ideas.

About Sean Carroll

Sean Carroll is a Senior Research Associate in the Department of Physics at the California Institute of Technology. His research interests include theoretical aspects of cosmology, field theory, and gravitation. His most recent book is The Particle at the End of the Universe, about the Large Hadron Collider and the search for the Higgs boson. Here are some of his favorite blog posts, home page, and email: carroll [at] .


See More

Collapse bottom bar