A Synthesis of our Ideas


Here we try to pull together the various themes into a coherent and self-consistent theory of life: what it is, what makes it unique, how it works and how it relates to the wider universe, both in origin and function.  This is of course a work in progress, so this page will change as new understanding is integrated into our developing understanding. What you see here is the current position in summary.

The information perspective has encouraged us to think of life as continuous with the wider universe and to concentrate on the essential structure and processes that make it special. Empirically, the existence of life coincides with a substantial increase in the complexity of that part of the universe where it arose. This is a part of the general increase in algorithmic complexity of the universe which has followed its growing entropy (in substantial part generated by spatial expansion). Life then has a cosmological underpinning, recognised by many cosmologists such as Paul Davies and George Ellis. Chemistry is the result of `strong' (`hard') emergence: a universal increase in order caused by an injection of information identified as the Pauli exclusion principle. This caused, by chemistry, a major unfolding of algorithmic complexity (which we identify with `autonomous information') and  life is a development of chemistry that is especially rich in opportunities for functional combinations (those which both have and give context to others in their sphere of influence). Kornberg's adage that “life is chemistry” remains true at the material level, but under this, we see the informational basis of the chemistry of life. We see how this information unfolds to create increasing algorithmic complexity from functional relationships and how these become self-sustaining and replicating, adaptive and appear to grow a sense of purpose.

In our interpretation, life is not fundamentally a product of chemistry, rather it is something that uses chemistry as a medium in which to exist. We see life as the further unfolding of algorithmic complexity beyond chemistry. The basis of this complexity is the repertoire of structures that may arise form autocatalytic systems of chemical reactions. The history of life is one of increasing complexity, caused by repeated combination of formative patterns such that they give context and thence function to one-another. But there remain many complex systems, some of which are autocatalytic, that do not constitute life. From Maturana and Varela’s pioneering work on autopoiesis, we know that a physical boundary (the tegument) that allows only selected molecules through and at a rate that can be controlled, is necessary for a system to live. This idea has been developed by Pier Luisi and colleagues to show how this boundary can be built, renewed and maintained entirely by internal processes of the contained ‘cytoplasm’.

Life is a phenomenon of the ‘perpetual organisation of matter’: it is dynamic, but held together in precise and complex organisation. This organisation is special because it orders matter into a dynamic system that is self-contained, making an identifiable whole from a collectively autocatalytic set, capable of maintaining and making copies of itself, capable of independently ordering energy into work to achieve this, and capable of making and running programs which give it autonomy. Possibly this autonomy is the most characteristic observable feature of life, distinguishing it from all non-life. It is obvious and familiar already, but so difficult to express in objective and scientific terms, that it has largely been ignored up to now. Autonomy occurs at the tegument separating the living from the non-living, where cause and effect are transformed into signal and response.

The information embodied in the universe has formed into a 'Russian doll' multi-layered structure in which each layer spontaneously creates the one above it as 'emergent properties' arise from complexity. This is by far the most elaborated in and by life, but why? Our understanding of the phenomenon was advanced by the concept of functional equivalence sets and the context downward causation put forward by, especially, George Ellis and colleagues. From this we understand elemental processes of life: homeostasis, anabolism, replication and work generation as information processes existing at a level of organisation greater than the chemical reactions constituting them. The business of information processing is commonly termed computing, but the computations of life are particular in that they are cybernetic. This should be no surprise since the pioneers of cybernetics (Wiener, von Neumann, von Foerster, Bateson and others) originally set out to understand the relationship between life, organisation, control and information. Our perspective emphasises the idea that all organisational scales of life (from molecule to ecosystem) unite in self-maintenance and reproduction. The subject of this cybernetic computation is living: life is a `program' running on itself, whose function is to output itself. This sort of recursion is familiar as autocatalysis (now it is the autocatalysis of information - a realisation first stated by Stuart Kauffman). It highlights the fact that for life, there is no distinction between the `machine’ and the `program’ - both are functional information. All information must be instantiated and embodied by differences in physical matter and/or energy. The program and the embodiment of life are both the same information: life is the program and the biochemical structure is its embodiment.

Stuart Kauffman realised and has since emphasised that to be living a system must be able to “complete at least one thermodynamic cycle”, his deeper insight was that life was made from a self-maintaining organised open non-equilibrium system and that the maintenance of this organisation required work to be done and that in turn requires the system to be an engine. Self maintenance means that it has to be formed of an autocatalytic set, so, combining the two, it must be an autocatalytic engine. The great insight was that in life, organisation catalyses organisation. From classical thermodynamics, we know that the flow of energy is made to do work only when it is constrained in degrees of freedom by organisation of matter. The work obtained can be deployed in creating and maintaining this organisation of matter. Therefore a general truth about living systems is that they are engines, whose work is used to make and maintain the organisation of the engine. This thermodynamic understanding shows one of the ways in which life is more than ‘mere’ autopoiesis: true autonomy means supplying all the work needed for self making and self maintenance from within.  

Cells compute using networks of molecules (mostly enzyme switches), but what are they computing? The answer is homeostasis and reproduction and the filtering needed to select what is needed from the environment for growth and maintenance. This is the machinery of autopoiesis. In a multi-cellular organism, networks of cells with different specialisations are computing by communicating information, filtered or amplified, in cascades reminiscent of the activity of artificial neural network computations. They too are running the algorithms needed for homeostasis and reproduction which constitute the ‘perception’ to select what is needed from the environment for growth and maintenance. If the organism has a nervous system, this will add extra computing power, via additional and complimentary ‘wetware’: neural networks do the same job as cell-signalling networks, but more effectively so as a result of the specificity of connections that can be achieved.
Now, what of a network of organisms: an ecological community? This too is a computer constantly solving the ever changing problem of homeostasis and reproduction: selecting  what is needed from the environment for growth and maintenance. The organisms specialise in different aspects of the process (functional roles) and link together in networks that are essentially (in information terms) similar to those of the molecules in a cell.
At each level, we see computation over a network of specialised components.

The theory of autopoiesis (self making and self maintenance) was originally conceived for a single cell by Maturana and Varela (published in 1980). The basic requirements for life, of homeostasis and growth, are true also for the community of cells in an organism and for a community of organisms in an ecosystem. Since all life shares this common requirement, at all levels of biological organisation, it can be used as a unifying principle. Indeed, all life can be seen as one elaborate multi-cellular complex. Human society has developed through increasing specialisation of task, so that the efficiency with which each person does their job increases - this leading to a surplus of production that can be reinvested as population growth. So it is with cells: first specialising into a community of symbiotic partners, then forming distinct organisms, the division amongst which is just an elaboration of the division of labour among cells. An ecological community is an elaborate network of cells, many taking part in organisms, but all working as a homeostatic network, manipulating and taking what it needs from the abiotic environment to perpetuate autopoiesis through information processing.

We understand life to be a process, not a thing; it is the manifestation of action; life is living. Living is the autonomous performance of at least self-maintenance and renewal. This seems to require at least a complete and selectively permeable boundary and the ability to work as an engine, to use the ordered energy generated for the selection of required material and its ordered integration into the boundary and the machinery within, which is responsible for the self-maintenance and renewal. The operation of collectively autocatalytic sets implies that to do this, a system does not require template replication (for which all known life uses the  information bio-polymers RNA and DNA). Because the environment in which a boundary-contained autocatalytic machine and engine operates is unpredictably variable in time and space, this ‘draft’ organism must also be able to regulate its actions (homeostasis) and respond to opportunities and threats: it must be capable of detecting change and changing itself to improve its environmental circumstances. The closure of the collectively autocatalytic set and of the boundary already ensure that this system is a Kantian whole (so named by Kauffman). Its ability to change itself and its environment (by moving within the environment, or altering its relationship with it through changes in the selective permeability of its boundary) give it the appearance of goal-orientated behaviour, of wilful action. The foundation of such behaviour is the homeostatic set-point, which provides the goal and the direction for adaptive behaviour. This has been described by George Ellis as downward causation through information control and appears to be unique to living things (and the devises made by humans).

There seems to be a lower bound to the complexity of a system that can achieve all this (though so-far unquantified). Given such a lower bound, at least as a conjecture, it follows that life would at least benefit from (and it may turn out to require) an information store that is separate from the working engine of the organism (for which the punched-tape of a Jacquard loom is the exemplar). Once established, it is a relatively easy step from this to a  programmable information store, by which we mean that the information (sequence) it embodies is close to independent of the energy needed to embody it. When this condition is  met, the entropy of the memory system derived from its stored information is independent of that derived from its general physical form (e.g. A-G-T-T-G-A has the same physical entropy as A-A-G-G-T-T, but has more information entropy - see here).  This is the feature that information biopolymers share with computer memory, be it in silicon, ferro-magnetic or any other medium used. Indeed it is even what gives the essential ‘computer’ nature to Babbage’s Analytical Engine. Several researchers have conjectured that this feature of computation is necessary for evolvability and that the latter is an essential part of the definition of life (especially Sara Walker). But it is not yet certain that a separate programmable information store is required for life, given the ‘composome’ model especially promoted by Doron Lancet  and colleagues, in which closed autocatalytic sets of chemical reactions can store information in an almost programmable way and can show (at least a kind of) evolution by natural selection. 
The idea of collectively autocatalytic wholes may be applied beyond the biochemistry of the cell, but this has rarely been considered to-date. What reason would cells have for forming colonies, why would colonies of specialists (somehow) become multi-cellular organisms and might communities of organisms not also be, in some respect at least, collectively autocatalytic? If there is an answer, it is probably to be found by quantifying the extent to which these higher-level organisations of life are collectively autocatalytic Kantian wholes. Two points become clear immediately. First, at levels of organisation higher than the cell, there is no external computable information store: in multi-cellular organisms a separate copy is kept with most (exceptions include mammalian erythrocytes) cells of the body and ecosystems are clearly spontaneously emergent. Second, the autocatalysis comes in the form of ecological mutualism, which in turn is connected with increased specialism and sharing of information among the parts. In other words, the collective autocatalysis, if it exists at these higher levels, is one of distributed computation, coordinated by downward control of ‘abstract’ information: it is embodied by the interactions among the parts, not within the parts themselves. We also understand that difference is the foundation of information and difference is diversity, so biodiversity is literally information embodied by biological systems. The more diverse a biological community, the more information it can embody and perhaps then, the more functional it may become, if function is quantitatively defined as the rate of self-generating process. If these processes depend (at least to some extent) on autocatalysis (in the form of mutualism), then we may explain the relationship between ecological functioning and biodiversity in general terms, that smoothly interface with information-based ideas at the sub-cellular level of description.
At the grandest scale of life on earth, the participation of organisms (especially plants and microbes) in the chemical processes of the planet has the hallmarks of homeostasis, as James Lovelock and Lynn Margulis elaborated into their ‘Gaia Hypothesis’. This and the manipulation of the environment at any lower level of organisation should never be thought of in teleological terms, but neither does it require a strict application of evolution by natural selection of random variation to explain it. Such apparent perfection and intent of purpose is nothing more than the operation of homeostasis that arises from the embodiment of an appropriate set-point in the information embodied by the interactions among components of a collectively autocatalytic system. Just how the set-point became established is one of the core mysteries we have yet to solve. The answer may be very important for these studies, since the set point is a rare case of ‘injecting information into the system’ and just as Pauli’s exclusion principle is, the set-point may be the source of strong emergence. If so it would be only the second ever to have been discovered. 




The Theme is led by
Keith Farnsworth