| Home | About Edge| Features | Edge Editions | Press | Reality Club | Third Culture | Digerati | Edge:Feed | Edge Search |

Implications of Natural Selection and the Laws of Physics

Lee Smolin, Richard Dawkins, Nicholas Humphrey, Brian Goodwin, Jaron Lanier, George Johnson, Marcelo Gleiser, and Lee Smolin (2) on Implications of Natural Selection and the Laws of Physics.

From: Lee Smolin
To: Richard Dawkins

I have two kinds of questions for Richard Dawkins about the further implications of evolutionary theory. These are about his responses to the ways that some of us physicists have been thinking about natural selection, both to try to apply our methods to biology as well as to use it to try to resolve the puzzles of our own science.

The first is about the possible applicability of the mechanism of natural selection outside of biology. Dawkins himself has invented the notion of the meme, an idea which propagates from mind to mind and survives as long as it reproduces itself. But there are a few of us who have wondered whether it might apply on a much larger scale, perhaps even to the laws of physics themselves. This seems a possibility that might be looked into as the current status of our search for a unified theory of all the interactions is that the best candidate so far-string theory, seems to have a large number of different solutions, each of which describes a different world. These worlds may differ in dimensionality as well as in the kinds of elementary particles and forces that are observed. They are something like different phases of the fundamental theory, rather analogous to the different phases of water, except there are many more. We cannot draw definite conclusions about this because we only have an approximate form of the theory under our control, but it seems possible that a principle is needed to pick out which combinations of particles and forces we find in our world that is not in the fundamental theory itself.

In this case it seems natural to wonder if some kind of historical explanation might account for how the actual particles and forces we see were picked out of a large set of possibilities. Such an explanation would have to account for the fact that it seems that the present set allows a much more complicated world-in terms of the existence of a large variety of different kinds of atoms and molecules-than the average set. This makes possible galaxies, stars and, of course, life. Some people have tried to account for this with the anthropic principle, but a few of us have felt that it might be possible to do better and explain it through a genuine mechanism of natural selection whereby regions of the universe reproduce themselves, with some small variations in the properties of the elementary particles and their forces. Given plausible assumptions about physics at very small scales, you can actually make such a theory work. I will not go into this here, but one can actually make a testable theory along these lines.

This idea was anticipated by the American pragmatist Charles Sanders Pierce, who wrote about a hundred years ago:

"To suppose universal laws of natural capable of being apprehended by the mind and yet having no reason for their special forms, but standing inexplicable and rational, is hardly a justifiable position. Uniformities are precisely the sort of facts that need to be accounted for. Law is par excellence the thing that wants a reason. Now the only way of accounting for the laws of nature, and uniformity in general, is to suppose them results of evolution."

I would thus be very curious what Richard Dawkins thinks of the possibility that the logic of natural selection might apply to the laws of nature and the history of the universe itself.

My second question is what Dawkins thinks of the possibility that collective effects can occur in systems in which many species are evolving together via natural selection. For example, Per Bak, Maya Paczuski, their collaborators and others have simple models of many species evolving together in which the system as a whole reaches a self-organized critical state in which one sees collective effects including punctuated equilibrium and a power law distribution of extinctions (both of which are observed in nature.) To physicists such as myself these do not seem in conflict with the point of view of Dawkins, but it seems also necessary to take these kinds of effects into account as they can occur in a system of many selfish genes interacting in such a way that their "extended phenotypes" overlap. I have the impression from some conversations and remarks that some evolutionary theorists dismiss this mistakenly as a kind of mystification. I would be very curious what Dawkins thinks of these issues.

From: Richard Dawkins
To: Lee Smolin

I have long been intrigued by Dr Smolin's idea of a natural selection of universes. It is the only correct way known to me of applying the word 'evolution' to cosmology. Usually, when people speak of the 'evolution' of the universe, they mean 'development' (an individual animal develops, changes in its own structure, it does not evolve. A lineage evolves, and it is an individual sequence of successive developments. Authors speak of stars 'evolving'. Stars do not evolve, they develop. For things to evolve, they have to give birth to a changing lineage of daughter things.

The reason Smolin's idea is interesting is that it may answer the challenge, "The universe is too good to be true. It looks like a put-up job." But, note that the Smolin hypothesis cannot be used to account in particular for the BIOLOGICAL part of that "too good to be true". Smolinian selection may account for the fact that our universe has the necessary constants, dimensionality and laws to last for last for a long time (not fizzle out or crunch immediately its initiating bang), long enough to spawn daughter universes (and INCIDENTALLY long enough to breed life). But Smolinian selection cannot account for the fact that our universe is specifically congenial to life, or to intelligent life, or to us. My negative conclusion would break down only if life itself is in the habit of engineering the spawning of daughter universes. As far as I am aware, this hasn't been suggested, but it is, I suppose, a theoretical possibility that daughter universes are generated as a consequence of the fooling around of highly evolved physicists.

Note that any Darwinian theory depends upon the prior existence of the strong phenomenon of heredity. There have to be self replicating entities (in a population of such entities) that spawn daughter entities more like themselves than the general population. This appears to be true of the Smolin model of Darwinian universes. It is certainly true of the normal Darwinian selection of genes I suspect that it is not true of the "collective effects" mentioned at the end of Smolin's note. I suspect that these collective effects are best handled in an alternative way. Replicators (such as genes) flourish in their environment, but what is often forgotten is that an important part of the environment is the other genes. Therefore you get mutually compatible partnerships evolving, NOT because the partnerships themselves are units of selection but because the lower level units (genes or whatever the replicators are) are selected for their mutual compatibility. I suspect that this may be what Smolin means when he says 'in a system of many selfish genes interacting in such a way that their "extended phenotypes" overlap.'

But this may not be very coherent since I am suffering from flu (like most other people in England at the moment).

Richard Dawkins

From: Nicholas Humphrey

Smolin refers to C.S.Pierce as having anticipated the idea that the universe evolved by some kind of selection. But there was a much earlier statement of a similar idea by Denis Diderot in his "Letter on the Blind, for the Use of Those who See" published in 1749. In this "Letter" he imagines the blind Cambridge mathematician and atheist, Nicholas Saunderson, arguing furiously with a clergyman about whether or not the existence of order and beauty in nature implies the existence of a divine creator. In an extraordinarily prescient passage, he first discusses how living animals might have evolved by competitive elimination of "unfit" forms, and then goes on to argue that what was true of animals might have been true of the universe as a whole. The following is Jonathan Kemp's 1937 translation:

"You may imagine, if you wish, that that order which impressed you has always existed. But leave me free to think it has done no such thing, and that if we went back to the birth of things and of time, and perceived matter in motion and chaos becoming unravelled, we should encounter a multitude of shapeless beings instead of a few highly organized beings. . . I can maintain to you that . . monsters annihilated one another in succession; that all the defective combinations of matter have disappeared, and that there have only survived those in which the organization did not involve any important contradiction, and which could subsist by themselves and perpetuate themselves. . .

"But why should I not believe about worlds what I believe about animals? How many worlds, mutilated and imperfect, were perhaps dispersed, reformed and are perhaps dispersing again at every moment in distant space, which I cannot touch and you cannot see, but where motion continues, and will continue, to combine masses of matter until they shall have attained some arrangement in which they can persist. O philosophers, transport yourselves with me to the confines of the universe; move over that new ocean, and seek among its irregular movements some trace of the intelligent Being whose wisdom so astounds you here!"

From: Brian Goodwin

I think that Lee Smolin's ideas about the differential abundance of universes that have different degrees of self-reproductive potential are extremely interesting, not to mention other similaritites he has identified between biological and physical processes such as excitability in galactic dynamics and the properties of autonomous agents. However, there is a fundamental point that needs to be clarified in order to pursue these analogies (possible homologies). In physics, the objects of investigation (e.g., the elements, superconductors, galaxies) are all regarded as natural kinds - that is, structures generated by dynamic processes that have distinctive intrinsic natures described by the causal factors at work in their production and maintenance. In the current biological view of evolution (i.e., Darwinism, or NeoDarwinism), the fundamental objects of study (species) are not natural kinds. They are historical individuals, accidental conglomerations of parts (characters originally, now genes) that have passed the survival test. The philosopher of biology, David Hull, has given definitive expression to these concepts. So looking for analogies between evolutionary and physical processes may get bedevilled by this basic difference of explanatory mode between the two subjects: biology as a historical science (no natures, only contingencies), physics as a rational science that has both contingent factors (e.g., initial and boundary conditions) and causal dynamics that explain intrinsic natures. My own view is that Darwinism is a scientific aberration that needs to be embedded in a more comprehensive dynamic theory that will bring it into line with the rational tradition in science. Then life and its manifestations can be understood as expressions of intrinsic organisational principles in particular states of matter (those we call living). Species, the attractors of this dynamic order, are then natural kinds, though of course their manifestation requires particular contingencies, just as does carbon or water.

In relation to Lee's particular enquiry, natural selection operating on universes that vary with respect to reproductive potential would be a perfectly natural expectation. What this shows is that natural selection, seen from the perspective of dynamical systems, is a statement about dynamic stability: those entities that make more of themselves (other things being equal) will tend to predominate. But natural selection in biology does not explain anything about the entities to which it gives rise, except that they survive (tautology), and this is as far as current evolution theory goes. Some of us are exloring explanations of life and its expression in species and other taxa at a somewhat deeper level than this, akin to the causal explanations used in physics. This of course is within the tradition that goes back to Goethe, Cuvier, Geoffroy St Hilaire and includes W. Bateson, D'Arcy Thompson, Needham, and Waddington. A recent book dealing with the conceptual issues is: Form and Transformation: Generative and Relational Principles in Biology, by Gerry Webster and Brian Goodwin. Cambridge UP, 1996. Excuse the advertising. I hope that Lee's query will initiate a dialogue on all these issues.

Brian Goodwin

From: Jaron Lanier

The most intriguing passage in Lee Smolin's note is the assertion that inheritance, selection, and evolution among universes is a testable idea. I would love to know more about this.

It's possible to derive mathematics from the most minimal initial ideas, and that makes it all the more annoying that the physical universe seems to arise from quirky and seemingly irreducible features. The program of finding an evolutionary reduction of the arbitrariness of cosmology is vulnerable, however, to falling into an infinite regress. There would have to have been some proto-cosmology analogous to the primordial soup that launched life, and then the question would be whether THAT had evolved. (This is not a problem for biology, since biology doesn't have to explain its origin from a void, only from chemicals.) Is it possible to pose ultimately simple, non arbitrary, initial conditions that could give rise to an evolution of physical universes?

It might be the case that there's only a layer of evolution that has risen from non-trivial initial conditions that were themselves not evolved, but that would make our universe out to be even more capricious than we had initially feared.



From: George Johnson

The exchange between Dr. Smolin and Dr. Dawkins reminded me of a theme that obsessed me while I was writing Fire in the Mind. If you'll allow me to quote from the book:

"Traditionally, biology has been seen as a historical science, while physics is regarded as a search for absolutes. Physicists seek that which is constant throughout the universe. Biologists are supposed to be content to pick their way through the accretion of mechanisms and mechanisms built on top of mechanisms that evolution happened to lay down on earth, to describe natural artifices -- organisms -- that, with a different roll of the Darwinian dice, would be unrecognizable to us. But this division between physics and biology seems to be breaking down. Biologists like Stuart Kauffman are looking for timeless truths, principles of complexity -- laws of the organism that might be reflected in all creatures, domestic or extraterrestrial, and even in metaorganisms like societies and economies. Conversely, physicists are seeking signs of contingency in the way the universe happened to crystallize from the big bang. Perhaps the particles and forces we observe and the laws they obey are 'frozen accidents,' just like biological structures. If so, it would be no more required that we have neutrinos than that we have hemoglobin, no more necessary that we have four fundamental forces than twelve ribs and thirty-three vertebrae."

From: Marcelo Gleiser

On chapter 6 of The Dancing Universe I start a discussion of how premature it is to attribute "intelligent design" to the Universe, when we don't even understand our own intelligence. I identify the discovery of "complex patterns" in Nature as a consequence of a brain that is the product of natural selection; as identifying complex patterns is one of our brains foremost powers, "finding" them is easy too. In a sense, this means that we are trapped within our one modes of functioning, so that our reconstruction of Nature is inherently "human".

From: Lee Smolin

The idea of cosmological natural selection is one that was concocted out of desparation, out of the apparent failure of string theory to lead to unique predictions for the spectrum of elementary particles. It took me a long time to take it seriously, I have recently been pleased to see that some biologists and astronomers do seem to think its plausible, if, of course, unproven.

As for Richard Dawkin's point that cosmological natural selection could not account for life in the universe, I think that there is a possibility that it might, or at least could go a long way in this direction. The reason is that carbon chemistry seems to play an important role in the processes that govern star formation. The clouds out of which stars form cool themselves to temperatures of as low as 10 degrees above absolute zero through processes in which carbon and oxygen play the key role. The main mechanism of cooling is believed to be rotational transitions of CO, the clouds are also filled with dust grains that are made largely of carbon, these serve both to shield the interiors of the clouds from star light and as sites for molecular binding. This is not nearly all of the role that carbon chemistry plays in astrophysics, but it gives a hint: it seems that the existence of carbon chemistry might be explained by cosmological natural selection because without it not so many massive stars that become black holes would be formed. Other conditions necessary for life, such as the fact that the universe has stars might also be explained by cosmological natural selection.

For me this is enough for the present, but two people I know have suggested that intellegent life might make a universe more fit, because they would make many black holes. Louis Crane, a mathematician who has contributed a lot to topics related to quantum gravity, has suggested that far in the future, after all the stars have burnt out, intellegent beings would make small black holes so as to provide energy and keep warm from the Hawking radiation. A great many small black holes would be necessary for this! Edward Harrison, a distinguished astrophysicst, has suggested that intellegent being, having figured out the game, might make black holes just to increase the probability of life like ourselves in future universes.

I myself prefer to let people far in the future worry about such things, and concentrate on the question of testing predictions from the original idea, on which some progress it seems can be made.

About Richard Dawkins response to my query about collective effects in natural selection: I am trying to suggest that his view is completely compatible with the behavior of the models of Bak and Kauffman. It is true that heridity strictly speaking is not necessary to get the kinds of collective effects one sees in some of the models of Per Bak and collaborators, where the only attribute things have is a number corresponding to fitness, and the only rule is that things with a lower fitness number are most likely to become extinct and be replaced by an entity with a random fitness. This is enough to get effects like power law distributions of extinctions (which is apparently what is observed and punctuated equilibrium.) But this is only a very simple model designed to explore such effects, the idea is that when the fitness is defined by the actual properties of the organism and its relation to its environment, it gets its meaning from heridity. That is, it is possible to model systems with heridity by using only an abstract attribute of fitness. But the model would not be relevant to biology were there not heridity. So I agree with the point that the genes are the units of selection, and that this is sufficient to understand collective effects, such as "partnerships". But I do also suspect that one needs to concepts that come from the study of self-organized critical systems to understand the general systematics of the history of life, such as rates of extinction.

What Brian Goodwin says is very provocative, because of his use of the concept of "natural kinds". The idea is that as he puts it, there are "structures that are generated by dynamical processes that have inrinsic natures described by the causal processes at work in their production and maintanence." Certainly there are things like this, and it is important to study them from this point of view. But I think there are some important things we don't know about them: Just how complex can such a thing be, in the absence of heridity, or control under some entity utilyzing coded information (i.e. DNA, RNA)? There certainly are such systems, such as the disks of spiral galaxies, that do appear to have spontaneously organized themselves. As there are enormous numbers of examples, it really seems that galaxies might be considered natural kinds. But they are much less complex than living things. Second, there perhaps are intrinsic organizational principles in non-equilibrium statistical physics, and I am among those who think the search for them is worth undertaking. But the burden is on us to find these principles and understand their scope. Third, I am not sure how powerful a concept of natural kinds can be in the absence of a general framework of natural selection. Perhaps I am reading in, but I suspect this concept carries with it a kind of Platonism, in which one imagines that one could enumerate and classify before hand all the possible "natural kinds". I think it is worth wondering whether this is even possible in principle. If not, then a science of natural kinds might be itself necessarily at least partly historical, and thus tied to the framework of evolution by natural selection.

Finally, to Jaron Lanier's point, yes of course, but there is nothing to prevent us from going step by step. If cosmological natural selection, or an idea like it, is going to be useful, it is going to be because 1) it leads to real predictions about what would happen if the parameters that come into fundamental physics were chosen differently, that are not contradicted by what we can deduce from our best astrophysical theory and observation and 2) our understanding of the fundamental principles of physics, (one candidate is string theory, or better the unknown exact theory behind string theory) leaves us in a situation in which there are many possible versions of elementary particle physics consistent with it. For the present, both of these conditions appear to be satisfied.

| Home | About Edge| Features | Edge Editions | Press | Reality Club | Third Culture | Digerati | Edge:Feed | Edge Search |

| Top |