|
| |
The Game of Evolution
From: Cambridge University Press
| By:
Jack CohenIan Stewart |
EDITOR'S INTRODUCTION |
The idea that evolution follows a distinct pattern and direction does not necessarily mean that it is predetermined. But can we therefore talk about evolution as having a goal or aim? Mathematician Ian Stewart and biologist Jack Cohen investigate whether our ideas about evolution are themselves constrained by our ideas in other areas of thought. |
 volution strikes many people as being highly implausible, because today's world is so amazingly complicated and finely tuned.  They find it hard to see how such a wonderful, interlocking 'machine' could have come about through the blind workings of chance. What they often fail to bear in mind is that today's world is a snapshot of a game that has been in progress for five billion years. If it were possible to watch the entire game, the current position would make much more sense. We can't do that, but we can make reasonable inferences based upon our understanding of chemistry, biology, geology, the fossil record, and anything else that seems to be implicated in the game. It is a game that began when the first replicating molecular network accidentally assembled, and it has never stopped. It is the game of Survival, and you can consider it as having either countless trillions of players (all living creatures) or only one (the global ecosystem). Our language lacks the words, and our science lacks the concepts, to describe this game adequately, so we give it a label: Evolution. Putting a label on a concept doesn't mean that you understand it, but at least it makes you recognise that there's something to understand. |
Evolution is a very curious game, going well beyond our idealised mathematical formulation. It has trillions of players (or one, the global ecosystem), and no fixed rules. Players (or The Player) are free to make whatever moves the constraints of physics and chemistry permit. The 'success' or 'failure' of those moves is determined not by a referee with a rulebook, but by whether the player--or its progeny--get to keep playing. In a sense the real players are lineages chains of organisms, each a parent of the next. |
The goal of evolution
Biologists will tell you that evolution has no goal, and on one level of description they are right. It has no preset purpose, it contains no coded representation of its own future. But neither does water, and it still flows downhill, not up: this shows that dynamics in phase space can impose an overall directionality on processes, goals and purposes notwithstanding. And on a different level of description evolution does have a goal, one that exists only in retrospect. The goal of evolution is to stay in the game. Players who (unwittingly) achieve this goal continue playing; all the rest become losers in the most brutally literal sense--they die without breeding. |
'Goal' is of course the wrong word. The usual sense of 'goal' is a reductionist one: look inside the system and you will find a 'search image' of the future which acts as a cause of present behaviour. The word applies to a sparrow seeking an earthworm, or to a human seeking religious enlightenment, but not to evolution. 'Unforeseen destination determined by contextual constraints' is what we mean, but the only common word with that meaning seems to be 'destiny', which has all sorts of misleading mystical overtones. 'Attractor', in the dynamical sense, probably comes closest. It is a bit like firing a shotgun at a ping-pong ball: the individual pellets do not have an inbuilt drive to hit the target--indeed they are totally unaware that there is a target--but nevertheless those few pellets that happen to be travelling in the right direction do hit it. They are thus singled out for 'success' by the choice of context. The twist in evolution is that it creates its own context, and 'hitting the target' is not an arbitrary choice, but the essence of the game--the unique precondition for the game to continue. |
Evolution, then, is a five billion year-old self-modifying planetary scale game which carries around partial records of its own past. And just as it has a (contextual analogue of a) well defined goal, so it also has a (contextual analogue of a) winning strategy. Like the goal, this strategy becomes apparent only in retrospect: make winning moves. A winning move, of course, is one that lets you stay in the game. And the way you find out what moves win is again rather like firing a shotgun at a ping-pong ball: you try as many moves as you can, and sometimes one of them works. Because evolution is a learning process, and because today's creatures are the descendants of players that have consistently made winning moves, the creatures that inhabit present-day Earth have become pretty good strategists. They don't know what strategy they are playing, but they play one that wins often enough to be useful. Individuals may die, or even whole species, but the Game goes on forever and it gets wilder and more convoluted as it does so. Their strategy is encoded in their genes and passed on from their parents. |
Back to Darwin
The idea of evolution as a game also sheds light on one of the commonest objections to Darwin's theory: the fact that many of the structures found in today's plants and animals have a stunning simplicity of 'concept' and 'design'. Think of petals as an 'advertisement' to attract pollinating bees, or the eye as a light-detector, or the heart as a pump. And if we look at these structures and their simple functions at more detailed reductionist levels, they are revealed as being even more complex. Petals are actually leaves whose function has been subverted in complicity with insect evolution. A simple beating heart resolves into a mass of millions of interconnected muscle fibres, all of which have to be synchronised by electrical pacemaker waves before they can produce an effective beat. The information processing that goes on between the retina of the eye and the visual cortex in the brain is incredible: the eye is in no sense a simple camera. |
Why does nature assemble huge, intricate networks of delicate complexity in order to achieve simplicity? It would, on the face of it, make more sense if living creatures behaved in incredibly complex ways, as well as being built in incredibly complex ways. Or, given that their behaviour is (on the whole) relatively simple, that their internal structure was comparably simple. However, very few things work like that. |
An analogy with technology suggests why there should be such a tendency towards increasing complexity. Human technology is a method for achieving simple, easily stated goals (fly to the Moon, cure cancer, breed tomatoes that can be transported without sustaining damage) by highly complex means. We can see why there is such a mismatch between the simple and the complex in that case. The overall goal may be simple, but the means for achieving that goal are dictated by what is possible within the laws of physics, chemistry, and biology. We no longer get to choose, and the best we can do is to cobble together various ill-fitting bits and pieces that relate, more or less directly, to our goal. In solving one problem we create others, which must be solved in turn. In order to keep the wind from our faces in a speeding car, we add a windscreen, but then the rain sticks to it so we have to invent the windscreen wiper. If it's not raining then we can't use the windscreen wiper to get mud off the windscreen, so we add windscreen washers--artificial rain. The water in the reservoir turns to ice in winter, blocking the flow, so we put in chemical additives to prevent freezing ... |
Contextualising nature
It is tempting to explain the burgeoning complexity of life in the same way, but there is one final obstacle. Technology is driven by human goals, but all biologists learn at their tutor's knee that nature has no goals, no purposes, no intentions. Evolution, they tell us, has no inherent drive towards any particular direction--it just happens. So the analogy with technology is misleading. |
Yes, of course nature does not have goals, purposes, or intentions, for those are human attributes. But it may well have an overall dynamic, a flow, a tendency to behave in one way rather than another, even a limited kind of 'predestination', because everything in the universe is affected by the context in which it operates, and subject to rules that limit its possibilities. To see what we mean, imagine a theory of the flow of water that is founded upon the molecular nature of the liquid rather than the conventional continuum model favoured by fluid dynamicists. Suppose further that the only observations available are those of water in its natural habitat--streams and rivers, rainstorms and lakes--and that the surrounding landscape is invisible. We observe the water and try to explain its patterns as it meanders past a randomly strewn boulder or drops off an eroded ledge in a waterfall. And the story we tell is one of contingency and selection. |
At root, the flow of water is merely the random wanderings of molecules. There is no purpose, no goal, no order; just stochastic jiggling. However, as the molecules jiggle around, some of them tend to accumulate, whereas others tend not to. So the first kind do accumulate, whereas the second rapidly jiggle themselves elsewhere until eventually they accumulate too. As a consequence of this random jiggling plus selection based on the purely contingent factor of accumulation, the mass of water moves. It could, it seems, go anywhere. |
However, things look very different when you can see the landscape. The places where water molecules tend to accumulate are contingent upon their surroundings, for sure, but once you know the surroundings you can predict what will happen. Water accumulates where the potential energy is lower--in short, it runs downhill. The geography surrounding landscape imparts a definite direction. The random jigglings are important for only one reason: they render the water fluid, so that it can flow into those lower regions. Without the random jigglings, motion would be only potential, but with them it becomes actual. And while the mechanism whereby individual molecules select where to accumulate is also contingent, it inherits direction from the landscape. Individual molecules can climb uphill, but on the whole they don't. |
Evolutionary theorists have the wrong view about determinism in biology because they have the wrong view of physics. They think that physics tells us that the flow of water is completely deterministic--as if the mathematical equations used by fluid dynamicists to describe the flow of a fluid were what the fluid itself actually does. From this mistaken view they derive, by analogy, a wholly misleading contrast between Darwinian principles and physical laws, and this confuses them when they start thinking about the global constraints that affect evolution. 'Goal' and 'purpose' are not the right words, to be sure, but there are dynamic effects in evolution, resulting from constraints. Mutations make phenotypes fluid enough to change, selection implements particular changes preferentially, but the overall result is more like water flowing through a landscape. |
However, it is an invisible landscape, formed out of the nearby 'potential' phenotypes and constrained by context--a mathematical 'phase space'. It includes not only what happens, but what could have happened instead. It may sound metaphysical, but such imagery lies at the core of how physicists and mathematicians currently think about all dynamics. Energy surfaces are also invisible landscapes--you can't see them. Who cares? Dynamical phase space is just as real as ordinary space--it makes itself felt by constraining the potential dynamics so that it carries out the behaviour that we actually observe. So evolution does have a preferred 'direction', and on the whole this will be the direction (in phase space) of increasing complexity. And so organisms become ever more complicated--most of the time. |
Occasionally evolution reverses that tendency, however, introducing simplifications, such as control of body temperature in mammals. This got rid of huge amounts of genetic contingency planning for developing eggs in lakes that were warm at noon but chilly at midnight. Evolution can introduce simplifications, and it does so when the system has become too baroque. Otherwise, increasing complexity is the norm. |
If it ain't baroque, don't fix it. |
|
| |
