Friday, 1 February 2013

What Would Happen If We Rewound Evolution?

The famous palaeontologist Stephen Jay Gould once pondered what the results would be if we wound back the clock of evolution half a billion years and observed the trajectory.  In his book Wonderful Life Gould argues that if we were given the chance to "rewind the tape of evolution and play it again" we would find ourselves living in a world populated by a different kind of animal kingdom, maybe one without primates (and as a corollary, humans).  This hypothesis (called the 'exaptation' hypothesis) is based on the observation that fitness for conditions in any current climate does not ensure long-term survival, because conditions change rapidly, and the things that had good evolvability in time x might not have good evolvability in time y. 

Arguing to the contrary, Simon Conway Morris in his Crucible of Creation propounds the view that humans are the unique yet intended goal of evolution.  While Conway-Morris does not want to imply that evolution is driven by mysterious goal-directed forces like supernatural intelligence, he appeals to the argument of 'convergence' to argue that at least in its broad outlines, the outcome of evolution is predetermined.  Convergence occurs when two lines of evolution independently develop the same or very similar structures, such as when reptiles and mammals independently evolved a fish-like body plan (as per examples like the ichthyosaur and the whale respectively).

To the greatest degree the path of evolution is one of contingency; mutations are random and cannot be predicted with foresight, and sexual reproduction is a random combination of genes.  By ‘mutations are random’, we mean that specific mutations do not arise in response to the selection needs of an organism or population. These random mutations, along with gene flow, genetic drift, and natural selection are the mechanisms that facilitate evolutionary change.  The mechanism for the fittest genes surviving is what is meant by natural selection, where natural selection is the main driving force that underpins evolution*.

To ask whether Stephen Jay Gould is right in his view that if we rewound all those mutations, migrations, genetic drifts, and natural selection, and played the tape anew, nothing close to human beings would very likely evolve again, we first need to consider the tape we are rewinding.  Stephen Gay Gould need not think of the ‘rewinding’ of evolution purely in terms of the hypothetical, because once we think of evolution as being made up of constituent parts, then in terms of replaying the tape of evolution we can observe that this has happened many times, because different species have evolved independently in similar ecological conditions.  Eyes and wings, for example, have evolved separately many times over.  So clearly we can make a reasonable estimate that a rerun evolution would bring about eyes and wings because they are such useful traits to acquire.  So we have evidence that once evolution reaches a certain level of production throughout those vast ecological niches, some traits seem highly likely to evolve again. 

In the sense being described, it is hard to justify thinking of eyes and wings as mere products of contingency; we ought to consider that the system of evolution has something more powerful underpinning it.  If understood in the right way (and there are many wrong way to understand this) the powerful underpinning It has is 'progress' through a ratchet mechanism.  People often hate the term 'progress', but if it is only taken to mean the inevitable direction that organisms tend to take, and the increased complexity that will occur with ratchet mechanisms, then (as long as we are mindful of that caveat) by progress we can mean something like “successfully surviving and reproducing”. What I think is true about evolution is that the progress is obviously not some accidental act of serendipity that keeps popping up randomly.  Why did wings evolve independently in birds, bats and insects?  Why has echolocation sprung up in bats, birds and dolphins?  Why has antifreeze evolved in independently in both the Arctic and Antarctic?  Why are eyes virtually everywhere you look (pun intended)?  It’s not just random chance – it is because progress (where progress = successfully surviving and reproducing) is built into the system.  In one sense the whole system is about survival, so through an implicitly human lens we can look at the evolutionary story and see animals getting good at surviving and reproducing.  

But here’s the most impressive thing about evolution; the constituent parts of biological evolution tend towards ratchet type progression, but not by having any direction towards better outcomes – this mechanism occurs purely through self-organising of the individual units (think of an ant colony as a great example).  Let's look at self-organisation in more detail..

Self Organisation
In self-organising systems each constituent obeys some basic rules determining its wider interaction with other constituents of the system, resulting in highly ordered behaviour in biological evolution.  The best human analogue I know is the economy, which looks after itself in a similar way, but in this case the self-organising systems are human beings obeying some basic rules of supply and demand, determining the self-organising wider economic system.  This is what Adam Smith’s term ‘the invisible hand’ means in economics.  For Adam Smith, the invisible hand acted as a social mechanism that channelled collective objectives toward meeting the needs of the people that made up a society, by ensuring competition between buyers and suppliers which channels the profit motive of individuals into providing products that society desires at prices which are rarely above cost.  What was soon observed, making the argument for laissez-faire economic philosophy strong, is that markets automatically channel self-interest toward socially desirable ends.  I don’t mean to suggest that the results of an open market economy are all due to skill – there is a lot of luck too.  But what the invisible hand does is enable people to be skilful and lucky. 

Have you ever seen the wonderful synchronicity in those birds that fly in a large flock, with each perfectly attuned to the flying patterns of the rest (see picture below)?  Like the ant colony, that is a beautiful example of self-organising structures where each individual is contributing to the whole in looking after its own interests.  I have a friend who is a computer programmer; he can use complex systems theory to simulate this flying using basic rules determining how each individual of the flock responds to its neighbours.  The most notable thing is that one doesn’t need any governing central planning principle to underpin the system – all that’s needed is the procedures for each bird to look after itself, and those beautiful flight patterns emerge in collective form. 

This is the kind of mechanism that drives evolution.  Organisms in biological evolution lock into the beneficial changes, whereby the ratchet mechanisms of natural selection facilitate successful dynamic structures that have good reproductive qualities.  These structures (like eyes, wings, and echolocation) are selected because they have the adaptable qualities needed for self-maintenance up against the tumult of a world in restless change.  This is what is meant by organisms being adapted to their ecological niche – obviously evolving antifreeze in the bloodstream is far more beneficial in the Arctic than it is in Africa.  In generalised terms, what makes this model representative in terms of complex systems theory is that it is a procedure observed in everything from physics, to information theory, right through to biological evolution and the economy.  We have mathematical systems that generate algorithms that encode successful models of the world, thereby allowing constituent parts to anticipate aspects of the wider organisation of the system but without any central planning from on high.

Prediction and Patterns
Clearly when we talk about predictability in evolution, and whether there is any inevitability in the system, the answer is; that all depends on what we’re trying to map.  Mathematical systems that are either random or chaotic on the whole can give rise to pockets of order – it just depends on the resolution of the parts on which we zoom in.  For example, here is a very evident pattern we can observe in human history.  When social and technical advance feeds population growth and population growth feeds social and technical advance, we find a kind of predictable pattern that gathers mass, size and momentum.  To see why recent progression stumbled upon us consistent with population increase, think about how the world has gone for the past 200,000 years.  For the past 199,800 years we’ve had low global populations, and humans lived in meagre conditions, with lots of primitivism, low life expectancy and frequent infant mortality.  Until recently in our 200,000 year history we have lived in pretty poor circumstances, just above the subsistence level.  Then a couple of hundred years ago something changed.  People started to become more scientific, more empirically minded, richer, and populations began to increase more rapidly (it’s still going on)**. 

But how reliable is that pattern?  Overall, not that reliable – particularly if we want indication of the supposed outcomes of replaying the tape of evolution and asking if humans will come about again.  Not only does evolution of life run concomitantly with external factors such as meteorites hitting the earth, volcanic explosions, severe droughts, and ice ages, where the fittest do not always survive – human behaviour is chaotic enough to not be predictable at a wider level.  By chaotic I don’t mean erratic (although it is erratic too) – I mean simply that (like the weather) the initial conditions of any one time do not make long term predictions conducive, because there’s no telling what technology humans will discover in the future, what global conditions will fall upon us, or which of the many possible man-made catastrophes we might engender. 

Just as people in the 17th century couldn’t have foreseen what would emerge as the result of the industrial revolution, we cannot foresee what will happen in the next few hundred years.  Given what we’ve seen in the last century, it is perfectly reasonable to predict that humans have found the method to lock into progressive techniques, and that (save for any unforeseen catastrophes) that progression will continue to increase. But we cannot know for certain, because the broader picture is beyond the immediacy of our sphere of local and present forecasting (predicting societal change is a bit like predicting the weather. – we can only do short-term forecasts). With short term forecasts we can use economic equations to predict some economic parameters and show a tend towards short term progression (short terms can exceed our lifetime), but as is the case with the weather, the chaos makes long term predictions unlikely.

Here evolution and society follow a similar heuristic – and that can be expressed with the model of computational irreducibility devised by Stephen Wolfram.  To see what X will do, one has to "run" through the full computational nexus of the system - there isn’t a more succinct analytical method of finding out what biological evolution or a society will do other than running the full computation simulation of that society – and that will take as long as the whole thing took (in the case of evolution, over 4 billion years).

You may have heard of non-linear feedback systems.  What happens in the case of societies is this; social and technological advance and a rising population feed back off each other, whereby once the ratchet mechanisms lock into place, social and technical advance feeds population growth and population growth feeds social and technical advance.  Of course, the ratchet effects of a locking into place don’t have to come in quick fait accompli bursts – they can be gradual over time; for example, over the course of about 8000 years we had creative peaks in the form of previously unseen social and technical antecedents like the agricultural revolution, the emergence of writing, philosophy, science, the printing press, mechanical engines, electricity, all of which help bring about increasing technological accomplishments, increase in prosperity and scientific paradigm shifts. 

So what we have is a system of order and progression, but also a system of chaotic perturbation that makes the overall system pretty intractable, and not amenable to succinct equations that consistently map societal change and evolution to predictable mathematical analysis. Whether it be the Cambrian explosion, the industrial revolution, or the FOXP2 gene mutation that brought about human ability of language, one can never know when the favourable junction of conditions will came into place and bring forth new horizons and vistas of potential change – which is why I agree with Wolfram that we are dealing with computationally irreducible systems here.

So in the most important sense, Stephen Gay Gould was right that if we rewound the tape of evolution we couldn’t expect to see the same outcomes, much less believe that humans are inevitable.  But Simon Conway Morris is also right that despite the overall chaotic perturbations that underpin the system, there are framework structures that are simply the best for certain adaptive purposes.

Any animal needing to adapt by surviving a freezing climate, and any vertebrate needing to swim in the water, and any creature needing to fly, is going to evolve in the same general direction only by being successful in passing on genes that favour survival.  The reason being - the genetic constraints imposed on those developmental pathways and the demands of the environment mean that there are restrictions on the evolutionary processes.  So if we reran the tape, the details would be different, but the framework probably would be more or less the same.

 * Note about randomness and chance; there is a certain chance element in the process – as I said, mutation is a process of random chance - but it is only random with respect to improvement.  Mutations enrich the gene pool whose diversity natural selection acts upon.  Natural selection is the non-random survival of randomly varying genetic codes - the survival of which depends upon their phenotypic effects as regards the process of embryogenesis on phenotypes, which cause survival or non-survival.  Survival and reproduction depends upon the passing on of the genetic code of instructions that built these organisms, equipped them and made them conducive to survival and reproduction.

** This progression can be explained by a simple rule of thumb – people innovate, improve and provide answers to problems - and the more people, the more innovation, improvements and problems solved.  The more ideas and the more people to share those ideas with, the more humans prosper.  It’s no coincidence that each half century has been progressively better than the last, and that the most recent times have been the most globally prosperous than any time in history.  That’s largely because we have 7 billion people on the planet – more ideas, more innovation, better technology, improved economic freedom, peak human liberation, and more global communication and potential to help the neediest.  When the world has 8 billion people it will be even more prosperous; when it has 9 billion, yet even more prosperity.  It’s no coincidence – the recent burst in population in the past 200 years has been the primary cause of our burst in prosperity (200 years is only 0.1% of 200,000 years).  That we have a maximal population and progression dialect in 0.1% of the entire human history suggests that there is a clear pattern at work.