Friday, 3 October 2014

Darwin Does Physics

John O. Campbell

This is a excerpt from the conclusion of the new book Darwin Does Physics

This account of the applicability of the Darwinian paradigm to the physical sciences is dependent on a context within universal Darwinism. Universal Darwinism may be most straightforwardly characterized as the fact that numerous theories across a broad range of science identify a Darwinian mechanism as responsible for the creation and evolution of their subject matters. While the adoption of this model within any one area of subject matter is remarkable, the fact that it has been widely documented throughout science makes it a coincidence demanding an explanation.

Within the social sciences Darwinian evolutionary theories are perhaps a consensus mechanism for explanations of cultural evolution. Most of the academic fields within the social sciences have a sub-school bearing the ‘evolutionary’ prefix including evolutionary psychology, evolutionary linguistics, evolutionary economics, evolutionary archaeology, etc..  

Even in the humanities Darwinian mechanisms are rife. Indeed a near consensus view within the philosophy of science considers science itself to be a Darwinian process; one which accumulates knowledge through an evolutionary process.

A powerful and compelling Darwinian/Bayesian model explains much of neuroscience. The Bayesian brain school views our mental models such as sight in terms of models that are constantly being updated with sensory information. The models which survive this Darwinian process are the ones we ‘see’.

In biology there is no scientific alternative to the Darwinian model. Of all the subjects making up the scientific enterprise only the physical sciences remain largely devoid of Darwinian explanation. Yet even here a few Darwinian theories have been proposed to explain physical phenomena at a fundamental level. 

I have described some of these Darwinian theories within the physical sciences and have suggested possible avenues by which this paradigm may contribute to the next generation of fundamental physical theories. 

Universal Darwinism is merely the observation that Darwinian explanations are common across all fields of science.  I have advocated a possible explanation for this coincidence involving information, inference and the accumulation of knowledge.

The most influential recent idea in physics may be that information is more fundamental than those entities such as matter and energy which physics has traditionally understood as most fundamental. The concept of information was first introduced to science by Claude Shannon in 1948. In a revolutionary tour de force he defined information and proposed information entropy as a measure of ignorance or uncertainty. Information entropy has subsequently been shown as equivalent to thermodynamic entropy which is at the heart of the second law and of all physics.

Significantly Shannon defined information in terms of probability assignments. Unless an entity has assigned probabilities to the outcomes of an event that entity cannot receive information about the event. Further this definition of entropy involves a probability distribution or a model of the event. 

Shannon’s great discovery was quickly followed by a re-birth of the mathematical field of Bayesian inference. Central to this field is Bayes’ theorem which describes the precise way in which a model should be updated when new information is received so that the model retains the greatest accuracy possible given the available information.

From this perspective information and entropy are not simple and may thus seem unlikely candidates to play a fundamental role in physical theories. In essence they involve one entity modelling another; one entity having some knowledge of an aspect of its environment and the ability to update this model as it receives new information. 

I propose that this complex knowledge entity, which includes information, probabilities, models and updating, be regarded as an inferential system which functions to accumulate knowledge. I have also argued that knowledge is a requirement for the existence of complexity and further that Darwinian processes are the physical implementation of inferential systems. Thus I attempt to explain the fact of universal Darwinism as due to nature’s dependence on the Darwinian process to accumulate the knowledge which is required for the existence of the complex entities studied by science.

Understanding Darwinian processes in this light moves us beyond its common ‘reproduction with selective retention’ definition and even the more sophisticated description by Dawkins and others in terms of ‘replicators’ and ‘vehicles’. In my view Darwinian processes invoke the mathematics of inference within reality. Thus they involve information, probability, models and updating. More generally they provide the mechanism by which reality is created and evolves.

The first intuition that information is at the basis of physics may be attributed to John Wheeler in about 1990. Since then it has proved remarkably productive, finding confirmation in areas such as the holographic principle, quantum information theory and the CFT/AdS duality. However within fundamental physics the consensus view concerning information is highly confused resulting in what E.T. Jaynes described as a standard of logic which would be considered a ‘psychiatric disorder’ in any other field of science.

Basic to this confusion is a failure to accept the fundamental finding that information must have a physical representation. The consensus view of quantum theory denies this and claims that the information of quantum theory is mathematical and has no physical representation outside of the human brain. This misunderstanding has resulted in endless attempts to link quantum phenomena to human consciousness. 

If an actual underlying physical representation of quantum information were discovered it would put an end to this nonsense but the scale at which this representation exists is likely far below the scale at which science is currently able to probe. 

This was also the case with biology at the beginning of the 20th century when Mendelian ‘genetics’ was considered merely a calculational device and not to have any physical representation outside of human minds. Of course since then biological information has been found to have a physical representation in the form of DNA, a physical representation many orders of magnitude smaller than the biological phenotypes which it models. 

Since the discovery of DNA knowledge within biological science has exploded resulting in the theoretical paradigm of Neo-Darwinism. Biology is now understood as an inferential system where knowledge that models strategies for reproductive success is accumulated within organisms. In turn organisms themselves are constructed according to the specifications of this model.

Unfortunately physics may be stuck in a situation similar to that of biology before its revolution. If, as I have argued, the role of information within fundamental physics must be within the context of an inferential system, physics might profitably look to better understand Darwinian processes such as biology for analogies to guide its development. I have suggested a number of biological analogies which may be useful in this regard.

The Darwinian/Bayesian view portrays, existence on all levels of organization, as extremely unlikely. The single path to existence is through knowledge and knowledge may only be gained by experience in reality. This boot-strapping processes of evolving complexity appears to have quantum roots; at the limits of our empirical abilities appear a few miraculous mechanisms, the four fundamental forces, which allow one entity to experience or relate to another. Although the ability of one entity to experience another is wondrous it is also limited; only a scant description is ever possible.

Experience, by itself, is not sufficient to produce complexity. Knowledge must be extracted from that experience, knowledge in the form of strategies to avoid the harsh constraints of the second law of thermodynamics and knowledge to out-fox competitors with evolving strategies of their own. 

Bayesian inference provides the mathematical framework to describe this process of knowledge accumulation and nature has followed those guidelines in constructing Darwinian processes to create and evolve complex structures throughout reality.Given the basic abilities to experience and to learn from experience which we find at the level of fundamental physics, the complexities of atoms, chemistry and cosmology rapidly unfold. 

The evolution of life is a landmark in the cosmic evolutionary scenario. With life and the genome of organisms a second repository of knowledge was produced that further propelled the evolutionary search for complex forms. On our planet at least this process appears to have accelerated and the breakthrough of a biological knowledge repository has been rapidly followed by the formation of neural and cultural repositories.

This view provides a homey take on our individual predicaments. We are born into an historical context. On the basis of this initial state and our experiences in the world we try some things. We are seven billion people trying things which seem plausible to each of us given our experience. Our lives are experimental, they probe the boundaries of the unknown. Each of the many challenges we may face from feeding ourselves to being a better parent are largely met in this kind of experimental manner.

We share our identity as experimental beings with all other complex entities; this attribute does not make us special. What makes us special is the scope of our experiences. We are probing the bounds of our ignorance on many fronts from experiencing the evidence of the big bang to experiencing Shakespeare and we are accumulating knowledge at a rate, accelerating so rapidly, that some believe we are approaching a kind of knowledge singularity .  

As far as we know many aspects of our neural knowledge and almost all aspect of our cultural knowledge are unique and are possessed only by one species and only on planet earth. Although we are the product of natural processes we participate within nature at a spectacular level of complexity which may be unique. We are experiments which almost certainly, at least in exact detail, have never been attempted in the universe before.

There is a grandeur in this view which surpasses even that expressed by Darwin concerning the origin and evolution of species. We now have reason to believe that the same simple mechanisms which he first discovered underlie the creation and evolution of all complex entities found in nature and that our species is a distinct landmark within this process of universal evolution.