Reproduced from: http://www.academia.edu/11623102/THE_RISE_AND_FALL_OF_THE_MACHINE_METAPHOR_ORGANIZATIONAL_SIMILARITIES_AND_DIFFERENCES_BETWEEN_MACHINES_AND_LIVING_BEINGS

THE RISE AND FALL OF THE MACHINE METAPHOR: ORGANIZATIONAL SIMILARITIES AND DIFFERENCES BETWEEN MACHINES AND LIVING BEINGS

by Victor Marques & Carlos Brito

Verifiche XLIII (1-3), 2014, pp. 77-111.

Abstract.

Our goal in the paper is to offer both an eulogy and a critique of the machine metaphor as a theoretical resource for understanding organic systems. We begin by presenting an abbreviated history of the machine metaphor, pointing out how it was instrumental in the development of modern biology, as it provided a conceptual basis for an analytical program in the sciences of life. Then we deal with what exactly makes the machine metaphor such a successful resource, pointing to what organisms and machines in fact share in common – based on the relational approaches advanced by Varela and Rosen, we suggest that both are ʻconstrained systemsʼ. In the third part, we present an alternative way of conceptualizing living systems, bringing now the disanalogies with machines to the foreground. Reviewing the independent work of different authors, we show that there is distinct organicist theoretical camp, where the organism is generally understood as an autonomous system. Finally, we observe that many authors from that camp are now reclaiming Kant’s treatment of organisms in the Critique of Judgment, in particular the concept of «natural purpose» – but those authors do that with a markedly anti-Kantian goal: to naturalize teleology. Our conclusion is that the view of organism as an autonomous system gives us the key to a naturalistic understanding that can finally overcome the mechanical view of nature so characteristic of modern thought. The machine metaphor, despite all its undeniable contributions to the advancement of biological research, shows itself ultimately insufficient for a complex view of the phenomena of life – and discarding it doesn’t need to mean any concession to vitalism: on the contrary, it may be exactly what we need to invigorate a robustly materialist project.

1. The Machine Metaphor

It is possible to trace the origin of the machine metaphor back to the Middle Ages, particularly to the time when technological advances and the production of artifacts of increasingly complexity exerted profound cultural influence and provided new images to interpret the natural world. Mechanical parlance served well the pioneers of the systematic study of anatomy and physiology to understand the new data in terms of already familiar experiences. Technology therefore offered an intuitive picture of nature upon which science could work. With the emergence of the mechanical philosophies of the seventeenth century, and the ambition to give an account of the whole of nature in terms of inert matter interactions alone, it was only natural to think of life as nothing more than a specific type of machine, the difference between organisms and mere artificial automata reduced to a quantitative one, residing solely in the degree of complication. One should thus place the effort of developing a mechanical paradigm in the context of the historical emergence of modern science, with the successful appearance, especially after Galileo, of a new physics, opposed to classical Aristotelian physics and in an ongoing struggle with the animistic worldview.

One of the most conspicuous aspects of the conception of nature presented by Aristotle is the centrality the notion of final cause plays within it. Its teleological character resides precisely in the claim that natural objects have a function, expressing thus purpose. For Aristotle, that purpose is immanent in reality itself was clearly demonstrated, for example, in the natural tendency of living beings to develop, regenerate and respond adaptively to environmental influences. Organic entities were considered therefore both material and teleological systems, at the very same time. Aristotle was not only one of the greatest thinkers of antiquity but also one of the first, if not the very first, biologist, and his vision of Cosmos was based after the model of the organism¹.

In fact, some type of ʻpan-vitalismʼ seems, as pointed by Hans Jonas², to be the primitive hypothesis of the human interpretation of Being, pre-modern way of thinking being marked thus by a kind of ontological dominance of life. Properly modern thought, on the other hand, is distinguished precisely by the very opposite movement:

This kind of outlook changed dramatically during the metaphysical shift that took place over the course of the seventeenth century. Nature became stripped of its capacity to self-organize as an end unto itself³.

The physics of Galileo differs from the Aristotelian physics most notably by the complete absence of final causality. The conceptual shift can be viewed as follow: what Aristotle intended to explain through a natural teleology may now be better understood in terms of efficient causality and nothing more. Final causes becomes unnecessary, and reliance on them, unscientific in principle. In losing its generic status, and demoted, from something nearly universal, to a position of a particular case (a very special one), life must then be explained in terms of non-living phenomena, that is to say, embedded in the theoretical space now monopolized by the hypothesis of comprehensive pan-mechanicism.

The scientific program of mechanization of living phenomena, as a constitutive part of a naturalistic mechanical project, actually starts to gain traction from the seventeenth century onwards, and especially with the publication of Harvey’s treaty on circulation. Here we already see the analogy with machines paying off theoretical and practical outcomes, helping to produce promising research hypotheses and lines of inquiry. The actual conceptualization of a philosophically grounded and generalized machine metaphor is commonly attributed to Rene Descartes. Robert Rosen, a theoretical biologist and key figure in the contemporary discussions on the machine metaphor, locates its roots in the analogy with artificial toy automata common in the royal gardens of 17th century:

The machine metaphor was first proposed by Descartes in the early seventeenth century. It is reported that, as a young man, Descartes was much impressed by some lifelike hydraulic automata. With characteristic audacity, he later concluded from these simulacra that life itself was machinelike⁴.

The power and intellectual appeal of the metaphor stem from the fact that complicated machinery in general serves as a very practical demonstration that purely mechanical systems, when properly organized, can perform all kind of complex behaviors. Even more: we know exactly how the machine works – after all, we built it. We know precisely what it is because we constructed it; it is epistemically transparent. We can then be confident that there is nothing magical or mysterious in the operation of a machine, no matter how amazing or counterintuitive its competences are – nothing at all that doesn’t follow strictly the very universal and basic laws of physics that apply to inanimate matter, being thus fully reducible to mere efficient causality. Therefore, it is not necessary to postulate any vital principle or inner hidden telos to explain how a machine does what it does. And if by means of a machine we can mimic some characteristics and behaviors of living beings, nothing prevents us to eventually build such a machine capable to emulate all of them – the difference between one and the other would be only one of degree, not of quality. Just as there is nothing miraculous in the machine, there is also no need for something miraculous in the organism either.

It is easy to see, however, how the machine metaphor lends itself naturally to theological speculations. What is now miraculous is not that the organism functions, but that the organism exists. As the argument used to go: from a machine it can be deduced the existence of a machine constructor, a designer. The internal teleology inherent in the natural entity is discarded, but just to be replaced by the external teleology derived from a rational mind, which organizes the material from the outside so that it acquires functionality: «Final cause, the for-the-sake-of-which a creature possessed the form that it comes to have, was not lost but rather relocated. Seventeenth century metaphysicians moved final cause from within nature to the mind of God»⁵.

Incidentally, this compatibility with Christian theology has to be counted as one reason for the success of the metaphor. At the same time that it freed scientists from direct theological considerations when investigating physical nature (since material systems themselves, other than human beings, are no longer inherently sacred), it also reserves an untouchable place for God and spirit – outside nature, and thus also outside the scope of scientific investigation. Nature becomes despiritualized, disenchanted, but by the very same movement, spirit becomes dematerialized, disembodied – two absolutely distinct kinds of stuff. Kampis captures precisely this ambiguity in the mechanical worldview:

Mechanistic materialism, a philosophical pedestal for mechanistic thinking, is deistic. […] Deists assume God’s existence as creator but do not let him intervene in the world he has created. This was reflected in nineteenth century thinking, admittedly mechanistic but also religious, according to the spirit of the time⁶.

The mechanistic monism in the realm of the natural world, while entering into confrontation with a pre-modern conception of a spiritual, lively, spontaneously productive nature, reinforced another religious view, that of a transcendent creator, who by his own will externally imposes form on a exteriorized, indifferent matter, in itself essentially inert and formless. Since the new physics was based on a non-teleological conception of matter, all finality had to be externalized, expelled from nature, thereby replacing the immanent teleology by a transcendent teleology. The final, fatal blow to teleology would have to wait for Darwin to be finally delivered. The final form of the “Enlightned” theory of organism, dominant during most of the 20th century, was a combination of the machine metaphor plus a Darwinian view of evolution, with natural selection acting as a kind of “blind watchmaker” designing “moist robots”.

The machine metaphor brought us undeniable intellectual benefits. It not only brings biology closer to engineering (and to physics, as a consequence), but it also offers a thoroughly naturalistic model, a general framework for explaining the activity of living beings without recourse to any principle beyond the scope of science research and potential empirical investigation. To a large extent, the metaphor owes its vitality and popularity to the privileged role in the battle against vitalism. For the concrete practice of the scientist, however, it has had an even more important methodological consequence, as it grounds, and serves as the basis for, an analytical research program in biology. The machine metaphor accomplishes that by providing the theoretical justification for assuming that the same procedure used successfully to study machines in general is also applicable to study any organic system: dismember it into its constituent parts and characterize them individually as independent sub-systems⁷. To the extent that the organisms are similar to machines, the analytical strategy works: biological research can thus accumulate knowledge and progress forward by treating living beings as a complicated ensemble of discrete mechanisms, describing particular components and identifying particular functions.

There is no denying that this program has been immensely successful. The analytical strategy based on the machine metaphor remained at the center of the spectacular advances in, for example, molecular biology during the last century. There would be no way to explain such astounding success if there were not in fact real similarities between machines and organisms. If practice is the proof of theory, it’s mandatory to accept that, at least in some sense, organisms can be viewed as machines. But in what sense? Or, what exactly makes the machine metaphor such a good metaphor?

2. Organisms and machines as constrained material systems

We now know very well that living beings are composed of the same kinds of atoms that are found in other parts of nature, and no other special or unique substance enter in their making – organisms are made out of the exact same stuff as the rest of the entire, life-less, universe. But unlike most of the physical or chemical systems which we encounter in the inanimate world, living beings are very complex, consisting of great number of parts, diverse in form and shape, arranged in a specific and intricate fashion: to sum, organisms are material systems both heterogeneous and ordered. Additionally, each part seems to express a definite purpose, as if it was a product of design. Organisms, of course, are not the only material systems to express all those characteristics: also in the case of machines each one of them is present.

In fact, as physical entities, both organisms and machines are systems marked by certain internal heterogeneity, amidst which is possible to distinguish differentiated parts. Also, both machines and organisms are organized into functional component, thereby allowing a description of the overall behavior making use of notions such as functioning, regulation, operation and coordination.

On those grounds, Chilean biologists Humberto Maturana and Francisco Varela, following the anti-vitalist tradition, identified organisms as machines⁸. By emphasizing the mechanic nature of living beings they wanted above all to assert two basic commitments. First, their firm belonging to modern scientific tradition, explicitly rejecting any residue of animism, denying that there is anything magical or incomprehensible in the passage from non-life to life, no forces or principles whatsoever which are not found in the material universe – living beings in no way escape from the ordinary physical laws that are valid to inanimate systems. Equally important, however, is that, as Varela specially is keen to stress, just as machines, organisms are defined by their organization⁹. In this sense, what is really characteristic of machines is that they are coordinated systems of components capable of satisfying certain relations.

By adopting such abstract and most general account of machine, Varela is making a functionalistic point. He wants to draw attention to the relative independence of an abstract organization with respect to the substrate, in sum, to a principle of multiple realizability: a machine can be materialized through various physical structures, the precise, detailed material nature of components being largely irrelevant given that they are able to play specific functional roles – in short, it doesn’t matter the stuff the component is made of, only what it is capable of doing (under certain conditions). That is to say: what is really relevant about a component are the relational dispositions in interacting with other components. The organization of a machine, defined by its relations, is logically independent of the properties of the components, which are, to a certain extent, arbitrary: «a given machine can be realized in many different ways by many different kinds of components»¹⁰. Therein lies the distinction drawn by Varela between organization and structure – organization being a purely relational notion, with no connection to materiality, and structure the concrete physical realization.

In fact, that very same distinction had already been articulated and put into play by Robert Rosen in his own relational treatment of life¹¹. Rosen contrasts models of a «structural character» to «higher level» models, which deal with the «functional organization», and then calls attention to the same principle of multiple realizability, stressing that «essential features of cellular organization can be manifested by profusion of systems of quite different structure». In order to complement a merely «structural study of a biological system», which abstracts away the organizational properties, he calls for the necessity of the opposite strategy, abstracting away the structure and leaving behind only the organization, to be studied abstractly in purely theoretical terms, divorced from any particular realization. That will become his later duality between a reductionist approach – throw away the organization and keep the underlying matter – and a relational approach – throw away the matter and keep the underlying organization¹².

It’s now easy to see that both Rosen and Varela were working in the context of the developing movement of theoretical biology (inspired by ideas from cybernetics and systems theory) and the attempts to formulate a general theory of organized systems, understood as «the study of organization per se, divorced from material embodiment»¹³. A relational approach allows thus one to treat organization as «a thing in the abstract», making visible the arbitrary relation between an organization and the specific materiality of its components: «there is nothing in the components that mandates that particular organization, nor anything in the organization that mandates those particular components»¹⁴. Both Rosen and Varela were after what is special about life not in terms of substance – of what it is made of – but in terms of pure form – of how it is organized. Which is precisely why Varela in his Principles of biological autonomy uses ‘machines’ and ‘systems’ interchangeably: «machines and systems point to the characterization of a class of unities in terms of their organization»¹⁵. That also explains why both Rosen and Varela, at least in their earlier works, claim to be advancing mechanistic theories of life¹⁶.

It is not in the least surprising thus that Rosen points to the very same similarity between machines and living beings:

What distinguishes a material system as a machine, as distinct from a stone or a crystal, must somehow reflect its intrinsic organization. […] But once we talk about organization, we are in a relational context. We are basically defining machine as a material system that admits (i. e., that realizes) a relational description¹⁷.

For Rosen, a system may be called “organized” to the extent that it is possible and useful to analyze it as consisting of components, which in turn are defined in functional terms from their role in the relation to the other components of the system. The component is the organizational unit: a part with a function. The relationships between the components in determining the ordered activity of the system as a whole is the object of what Rosen calls «relational theory of systems». What machines and living beings share in common is that they are both material systems that admit relational descriptions.

What we are dealing with here is nothing like the “organizing force” of vitalism, a “something more” added to mere materiality, but, as Ashby highlights, with organization as a coordinated system of constraints:

in the past, biologists have tended to think of organization as something extra, something added to the elementary variables, the modern theory, based on the logic of communication, regards organization as a restriction or constraint¹⁸.

Organisms, however, just as man-made machines, are not pure organization, but actual material systems realized through physical structures. So how then to pass from an abstract characterization to concrete existence, or, how to think the physical realization of constraints? In his classic paper from 1968, Life’s irreducible structure, Polanyi seems to point to the right direction¹⁹. Polanyi observes that the constructor of a machine «restricts nature in order to harness its working». But how does this restriction takes place, and becomes physically effective? By imposing boundary conditions on the laws of physics and chemistry. And the same is also valid for living beings, whose components perform functions in the same way as parts of a machine do:

in this light the organism is shown to be, like a machine, a system which works according to two different principles: its structure serves as a boundary condition harnessing the physical-chemical processes by which its organs perform their functions.

It is useful to note the distinct character of these two principles: while the laws of nature are, in principle, universal, inexorable, and incorporeal, the constraints are, by definition, local and arbitrary and require a specific physical realization, they are also a product physical processes and exist in a determinate context – they are not only constructed by a previous process but precarious, vulnerable to destruction and natural decay. Constraints are physically realized by way of boundary conditions harnessing the laws of inanimate nature in order to do something (work, e.g.), and both machine and organisms are better characterized as material systems of coordinated constraints. They are constituted by components that canalize physico-chemical processes in order to realize functions. As Rosen puts it, machines and organisms are «constrained systems»²⁰.

What unites both organisms and machines is that in both cases we are dealing with materials systems in which the organizational aspect is crucial and essential – when we want to understand the machine as machine, the relevant thing to offer is a relational description (a purely material description, albeit still possible, will inevitably overlook what is actually relevant about the phenomena in question). And the same goes in the case of organisms. Without that central underlying communality, the machine metaphor would never work. As far as it is theoretical fruitful, it’s because we are dealing fundamentally with organized entities.

That is why the machine metaphor has proved so powerful and long lasting, to the point that it has managed to survive even the crumbling of two other notions that used to accompany it, and forming together a coherent theoretical system: the idea of special creation and preformationism. The recognition of the historical character of living beings and the understanding of development as a process of epigenetic nature seem to weaken the metaphor, forcing important disanalogies, since, as a rule, the appearance of a machine is connected to the activity of a conscious mind and does not involve the spontaneous emergence of order as in development, the process of differentiation that does take place during biological ontogeny. In fact, if organic bodies are machines, it is clear that they are a very peculiar kind of machines, which human beings are not yet able to build.

Howard Pattee (1971) was one of the first to conceptualize organisms as coordinated sets of constraints, and used this notion to give precise expression to the idea of interactions among different hierarchical levels of organization²¹. More recently, authors such as Kauffman (2000)²², Mossio and Moreno (2010)²³, and Deacon (2012)²⁴ are reclaiming, in a way or another, the concept of constraint in order to make explicit the specificity of living beings. It is only at this level of abstraction that the essential difference between organisms and machines shows itself. Meanwhile the components and organization of a machine are defined from the outside, with its constraints externally imposed upon the material structure, in the organism the opposite is true, since its constraints are permanently posited, generated, repaired and modified as a result of the system’s own functioning. Organisms and machines are constrained systems, but living beings are, on the top of that, also autonomous systems that produce their own constraints; that is, the constrained dynamics of the lower level elements results in the maintenance and replacement of the very constraints that control their behavior, in a closed loop.

Terrence Deacon, also recognizing that both machines and organisms are functionally organized systems, calls attention to important differences between them. While machine parts are produced separately and only put together later during the assembling process, nothing similar happens in the organic case: the organism develops spontaneously, its parts differentiating from relatively undifferentiated starting point, with no exterior assembling needed, the components grow out of a prior unity, coming to be already integrated with the ever functioning whole. Organic components are from the beginning interdependent and involved in multiple relations with other components, both contributing to the function of the rest of the system and being reproduced by the networks of organic processes. On the other hand, with conventional machines, the manufacturing process is distinct from the operation process: the machine is constructed out of pre-existing components, with purpose of performing a function that is generally not related to its own production²⁵.

In the organism we are faced with a material system in which the components depend for their very existence on the overall context of organic activity, within which they are continuously being reproduced and repaired. Each functional component is materially dependent on the context afforded by the joint and coordinated activity of the other components, in a meshwork of tangled mutual dependence and reciprocal relations. In fact, we can say that it is the organism itself that defines its own limits, that is to say, the constraints that give it a definite shape are posited in and by the actual process of functioning of the organism as a articulated whole, the result of its activity being thus its continued existence.

In addition, the organism constituents are liable, plastic, incessantly changing, constantly being modulated by the metabolism to ensure adaptive responses. The regularity and stability of the organism does not result from materials that resist deformation, its permanence is not rooted on material inertness. Unlike a machine, that maintains its form because the physical constitution of its components is such as to render the interactions between them relatively insensitive to thermodynamical fluctuations, organisms exploit non-equilibrium chemical processes generating order by self-organization.

Just as the machine metaphor has several advantages that helped foster scientific progress, it also has some important limits, that, if not properly acknowledged and explicitly criticized, can severely impoverish the conceptual space in which we make sense of complex biological phenomena. As Lewontin & Levins expressed it,

the machine metaphor creates a general program for biological investigation that is circumscribed by just those properties that organisms have in common with machines, objects that have articulated parts whose motions are designed to carry out particular functions. So the program of mechanistic biology has been to describe the bits and pieces of the machine, to show how the pieces fit together and move to make the machine as a whole work, and to discern the tasks for which the machine is designed²⁶.

3. Life as autonomy

In fact, if organisms are machines in this general sense, it is clear that we are then dealing with a very unusual kind of machine. The common characterization of living beings as “reproducing machines” is ambiguous, as it can give rise to two, quite distinct, forms of understanding the phenomena of life. On the one hand, “reproduction” could be understood as replication, that is, production of another entity with similar characteristics: the organism reproduces in the sense of multiplying itself, increasing in numbers in a given population (that is the sense that first comes to mind, for example, when we think about “selfish genes” replicating – that is to say, making copies of themselves). The most important fact about biological systems would therefore be that they are the result of a reproductive event and therefore have a genealogical history, which can also be extended into the future, as a tree of successive reproductive events expanding indefinitely. A point of key importance here is that from the notions of replication, inheritance, variation, and differential viability it is possible to deduce the principle of evolution by natural selection. According to this view, what essentially characterizes life is its connection with the evolutionary process: organisms are that which, at the population level, can undergo a process of evolution by natural selection and, at the individual level, are the result of a prior evolutionary process. Evolution by natural selection is thus viewed as the general notion that unifies biology, the universal aspect of life and the basis where to ground the distinction between the mere physical world and the biological realm. Here the idea is that the cumulative effect of natural selection over generations results in a kind of natural design, that can be viewed in fact as analogous to rational, intelligent design, conferring functions to the parts of the organisms, and giving thus the appearance that those parts were made for certain purposes.

However, the term «reproduction» can also be related to self-production, or the active conservation of a self-established identity. A reproductive system is thus a system capable of self-maintenance. This second sense of reproduction focuses not on the genealogical history of the organism, or its potential for further self-replication, but on the metabolic processes that at every moment regenerate and/or modify the form of the living being and its way of being in the world. Humberto Maturana and Francisco Varela were pioneers in this second approach. They called living organisms «autopoietic machines», that is, a machine organized as a system of processes of production of components concatenated in such a way to continually reproduce the topology of the whole network of processes, and through its own activity delimiting itself physically as a unit in space.

According to Maturana and Varela a crucial distinction can be traced between autopoietic machines, on one hand, and allopoetic machines, on the other:

[…] in a man-made machine in the physical space, say a car, there is an organization given in terms of a concatenation of processes, yet, these processes are not processes of production of the components which specify the car as a unit since the components of a car are produced by other processes, which are independent of the organization of the car and its operation²⁷.

Allopoetic machines produce by their activity only something different from themselves, while in autopoietic machines there is a coincidence between operation and fabrication – the autopoietic machine fabricates itself as result of its own functioning. Maturana and Varela pointed out that as a consequence of their autopoietic way of being, a living system is autonomous, that is to say, characterized by its own dynamics, which the environment can «irritate», affect positively or negatively, but not determine. In every interaction with what is outside, the organism responds in its own way, subordinating the changes in its own structure to the conservation of its autopoietic organization.

By «autopoietic organization» it is meant simply that the constitution of the system is the result of a specific way to concatenate internal processes, in which the overall result is the maintenance of the conditions of existence for the system as a whole, as a unity. This circular concatenation of processes allows the emergence of an individuality, a processual, precarious identity that persists through time despite a never ending flux of material composition and all kinds of deformations caused by its necessary interaction with the environment. The living system is a concrete unity, dynamically stable, maintained by a constant material and energetic flow, and has a history of interaction with the environment which both enables and threatens its existence as a distinct entity.

About the same time the theory of autopoiesis was being formulated, the mathematical biologist Robert Rosen expressed some ideas very similar to those developed by Maturana and Varela. Rosen also highlights the difference between organization and structure, and denies the hypothesis that what is essential to life can be recovered by the accumulation of purely structural investigations (what he calls «the hypothesis of reductionism»)²⁸. In particular, Rosen observes that the lifespan of a cell exceeds considerably that of their components, and that there is, for each functional component, a natural tendency to decay, as the physical structures spontaneously degrade to the point that they are no longer able to perform their proper function within the system. The system as whole, however, keeps functioning by continuously repairing and reproducing its components. The question of «who repairs the repairers» promptly announces the threat of an infinite regress. Rosen’s solution is to bend the chain of components onto itself, including the reparative activity in metabolism itself; to break that infinite regress, Rosen folds back the linear hierarchy into a loop. In Rosenean parlance, the system becomes now «closed to efficient causation», expressing thus a distinct circular character. The system as a whole expresses a self-repairing property due to its circular organization. As noted by Letelier et al (2003):

Rosen’s main result is the demonstration that the synergy of metabolic and repair actions can imply, under some circumstances, self-replication in the sense of self-production (or self-maintenance) of the complete metabolic network²⁹.

In a way similar to autopoiesis, in Rosen’s MR-systems «the closed-loop hierarchy produces an integrated whole that has an identity that can be perturbed by, but remains distinct from, its ambience»³⁰.

It is always possible to analyze separately the operation of a particular segment of metabolism just assuming the operation of the parts concerned as something given, but in the context of the organism, and its whole metabolic network, it is necessary to remember that all these parts are also always a result of metabolism – ultimately, each component is produced by and through the combined activity of other components. It is to highlight that fact that Rosen makes a point of stating that organisms are categorically distinct from machines because they are «closed to the efficient cause», a property that can only be expressed by impredicative models. Organisms are closed in the sense that every efficient cause, which for Rosen means the function that controls a material transformation, is itself in turn produced within the organism. The contrast with manmade conventional machines is obvious, since in these, generally, each component is separately produced and subsequently put together in an ordered, but external, fashion. Not only the mechanisms that produce them are exterior to the artifact being produced, but also the very operation of the artifact has, in general, nothing to do with the repair/maintenance or replacement of components – the artifact activity has nothing to do with its production process³¹.

Varela’s notion of organizational closure is thus related to Rosen’s notion of a system being «closed to efficient causation». Although the two models have been developed independently, both seem to have a similar goal: to emphasize the importance of circularity in modeling the phenomena of life and to affirm a concept of organism based on circular organization. Both Rosen and Varela are mainly interested in the basic question of a general theory of biology: what makes living things alive? What do all organisms share that separate them from inanimate matter? Or, as Rosen poses the question: what is life itself? Investigating living systems from a relational perspective, both Varela and Rosen come to the conclusion that the answer to the question of what is life itself lies in its circular organization: organisms are natural realizations of «strange loops»³².

Hofstadter coined the term «strange loop» to make reference to situations in which in a series of stages that constitute a cycling-around, a succession of shifts in level that feels like an upward movement in a hierarchy turns out to give rise to a closed cycle: in violation of the seeming hierarchy, one winds up where one started³³. That is exactly what happens when Rosen creates an impredicativity in order to avoid the infinite regress, by folding back the hierarchy into a loop³⁴. Varela, as well as Rosen, stresses the necessity of both hierarchy and circularity, resulting in a seemingly paradoxical structure that defies atomic analysis³⁵. This type of system seems to resist modeling by conventional mathematical tools. Rosen (1991) uses Category Theory in order to accommodate impredicative loops in a mathematical formalism, Varela (1975) uses Spencer Brown’s calculus of indications to develop a «calculus for self-reference»³⁶, and, more recently, Chemero and Turvey (2008) suggested hyperset theory as a tool to model autonomy and complex systems with circular organization³⁷.

The contrast between Maturana and Varela, on one hand, and Rosen is that the later puts a lot more focus on the internal functional differentiation of the system itself and emphasizes that we are dealing not just with a mere material cycle, but a circularity in terms of control conditions of metabolic transformations. Whereas the autopoietic model is more focused in the spatial enclosure with membranes, Rosen is more preoccupied with the logical organization, expressed in the formal language of category theory.

Rosen, however, is keen to draw attention to his conclusion that the relational description of machines does not allow causal closure, and the circular description is at the very core of what makes life distinct from non-life. So at least for the late Rosen, an organism is not a machine at all. Although Matura and Varela call their systems «autopoietic machines», in order to explicitly fulfill a «non-animistic criteria», they are so different from conventional machines in their autonomy, plasticity and circular organization that it is possible to question to what extent the use of the term is really useful or adequate.

The similarities between Rosen’s and Varela’s models are now amply recognized, by Roseans and Vareleans alike³⁸. But both approaches, being purely relational, emphasize organizational demands while paying almost no attention to material and thermodynamical conditions. Autopoiesis was explicitly conceived as an abstract machine, independent of the nature of the materials³⁹, and Rosen’s diagrams convey the loopy form of those systems, but say nothing about the nature of constraints and chemical process that are actually able to realize this distinct form⁴⁰. That is not the case with the more recent efforts of Kauffman, Mossio & Moreno and Deacon, where the emphasis also lays on the far-from-equilibrium and thermodynamically open nature of this kind of system. Being interested not only on synchronic analysis, but also in the actual historical genesis of autonomy within the material world, and struggling to give a diachronic account of how material systems came to express this peculiar circular organization, they offer a bottom up approach, which builds from physical self-organizing process up to biological, and cognitive, complexity.

Kauffman’s effort is to give a naturalized account of agency, explaining how can autonomous systems, that «act on their own behalf», exist in nature and evolve from inanimate matter. Noting that constraint and work are related terms, since work is constrained release of energy and in general (although not always) it takes work to build constraints, Kauffman defines organism as an autocatalytic set that does at least one work-constraint cycle. Living beings thus couple spontaneous and nonspontaneous processes to build constraint on the release of energy and use the resulting work to reproduce their own constraints.

The basic idea of a work-constraint cycle is also found in Deacon, who, following Kauffman, stresses that «besides being a product of work, constraint is also a precondition of work», and that «the maintenance, reconstruction and reproduction of dynamical constraints is a core characteristic of life»⁴¹. What they all share is the insight of organisms as dissipative systems and the emphasis on a kind of continuity between the simple self-maintenance of physical self-organizing processes, characterized by the spontaneous emergence of dynamical, precarious constraints, and life proper, with its enduring, propagational logic. The circular form must be realized in far from equilibrium conditions, which imply that organisms, by being organizationally closed, are also necessarily thermodynamically open.

Although it is true that both machines and organisms are fundamentally characterized by their organization, whose physical realization involves material structures that constrain underlying processes, the nature of these structures differs profoundly from one case to the other. While the machine components are often rigid and inert, in organisms they are typically precarious, malleable, plastic. The order and stability expressed by organic matter in its form do not result from stable materials, in contrast to machines, where the constraints are built to resist deformation and to be relatively insensitive to thermodynamical fluctuation. Rather, the organic dynamics is based on non-equilibrium chemical processes, generating regularity and order by self-organization and self-assembly. Organisms are dynamical systems out of equilibrium – as dissipative systems they are constantly renewing their material composition and requiring energy input to maintain their constraints. In order to produce order spontaneously, they must be thermodynamically open⁴².

Terrence Deacon rightly calls attention to the fact that what distinguishes organisms from other dissipative systems is that in the case of living beings there is not merely a process of self-organization, but rather «a reflexively organized constellation of self-organizing processes»⁴³. The self-organization characteristic of the organism as a whole is distinctively of second order kind, a meta-self-organization: «The constraint maintaining propagation logic of the organism is in a sense a higher-order self-organizing dynamic among component self-organizing processes»⁴⁴. For Deacon, it is the synergetic arrangement of self-assembly and non-equilibrium self-organizing processes that gives rise to the «teleodynamical features» of living beings, their «immanent purposiveness». The correlated co-production of constraints ensures the perpetuation of this holistic co-dependency, resulting in what Deacon calls (making reference to Varela) a «virtual self»⁴⁵.

Due to their precarious nature, biological structures are always being built and degrading, and a functional component persists not because it is structurally rigid, but rather because it is incessantly produced and reproduced by the collective activity of the other components. It is at this level of reciprocal constraint generation that we find the unique circular pattern of life. Material structures that play the functional role of organizational constraints reinforce and remake each other, thereby maintaining the identity of the system as a whole, in what Mossio & Moreno denominated «organizational closure»:

The main idea is that biological systems are able to maintain themselves by constituting a web of structures exerting mutual constraining actions on their boundary conditions, such that the whole web is collectively self-maintaining. The mutual dependence between a set of constraints is what we call organizational closure⁴⁶.

Being built on soft constraints, organisms have to be organizationally closed to persist through time for longer than their constituent components, and this organizational closure in turn demands that they are necessarily material-thermodynamically open systems, engaged with the environment in order to feed their own process of self-fabrication in far from equilibrium conditions. Machines, on the other hand, do not express organizational closure. Machines don’t have a metabolism, and their parts are not produced internally, each component being fabricated separately and arranged during the assembling process. The organic components, by contrast, are integrated and interdependent from the outset. The rigid nature of machine components give them a self-subsistent aspect, they can be easily separated from the rest of the machine, and the process of analysis, of dividing a machine into pieces, doesn’t modify at all their essential nature: they remain basically the same whether within or without the rest of the system. However, in the case of organisms, components tend to cease to exist, in a sense they lose their “essence”, when removed from the context of the overall organic activity, within which they are continuously being reproduced and repaired. While machines are assembled from pre-existing components that have independent existence, in living beings the existence of each component depends on the context afforded by the other components, in a tangled web of mutual dependence and reciprocal causation.

4. Natural purposes

As several authors have recently called attention, this view of life based on the idea of closure resembles the Kantian notion of “natural purpose”. In the second half of the third critique, reflecting about the question of purposiveness in nature, Kant draws a distinction between external, relative, teleology, that which we find in mechanical artifacts, and internal, intrinsic, teleology⁴⁷. Organisms, for Kant, are organized and self-organizing beings, both causes and effects of themselves.

Following that line of thought Kauffman coined the term «Kantian wholes» to refer to such natural systems as organisms, where the whole exists for and by means of the parts and the parts for and by means of the whole⁴⁸. As Juarrero highlights, reclaiming Kant’s original terminology for contemporary uses (specifically for dealing with complex systems, offering a naturalistic account of agency), a «natural purpose» is an object in which

a member is not only a means but also an end; it both contributes to the whole and is defined by it. No machine exhibits this kind of organization, for the efficient cause of a machine lies ‘outside’ the machine in its designer, and its parts do not owe their existence to each other or to the whole⁴⁹.

In the last two and a half decades, biologists and philosophers of biology have shown an increasing interest in the Kantian approach to organisms, reigniting a philosophical discussion about the concept of natural purpose and its place in a naturalist project. Juarrero (1999)⁵⁰, Weber & Varela (2002)⁵¹, Moss (2003)⁵², Thompson (2007)⁵³, Kauffman (2000)⁵⁴, Moreno & Mossio (2010)⁵⁵ and Deacon (2012)⁵⁶ – all refer back to Kant and his pioneering use of self-organization as conceptual tool to understand purposiveness in natural systems and what distinguishes organisms from artificial machines. What unites all those authors is the shared intellectual aim to naturalize, rather than simply eliminate, teleology.

As we have seen, modern thought, in order to understand living beings as fully natural parts of a thoroughly non-teleological physical reality, initially replaced the intrinsic purpose of Aristotelian type by the external purpose derived from a transcendent creator. The tendency to see the teleology of organisms as merely relative, derived, not intrinsic to the organism itself, persists even with the emergence of evolutionary thought, to the extent that the process of natural selection is interpreted as analogous to the action of an intelligent designer (as in Dennett and Dawkins, for example). Kant, in contrast, proposes a conception of the organism as a self-organizing entity, in general lines quite similar to the contemporary theory of autopoiesis – as Varela himself eventually recognized⁵⁷.

Kant, however, found himself faced with the dilemma, named by him as the «antinomy of teleological judgment», between the demands of Newtonian physics, according to which all natural objects should be understood in purely mechanistic terms, which are integral to Kant’s own conception of nature, and our experience in dealing with the biological realm, that seems to force us to think in terms of final causes. Kant himself was quite pessimistic regarding the possibility of the human mind to explain «organized beings» (natural purposes) according to «mere mechanical principles of nature».

On the one hand, the phenomena of life leads us to think the idea of a «natural purpose», in which the parts are only possible by reference to the whole, and whose organization has nothing analogous in the rest of nature:

An organized being is thus not a mere machine, for that has only a motive power, while the organized being possesses in itself a formative power, and indeed one that it communicates to the matter, which does not have it (it organizes the latter): thus it has a self-propagating formative power, which cannot be explained through the capacity for movement alone (that is, mechanism)⁵⁸.

On the other hand, the very idea of an «organized and self-organizing» being, that is «both cause and effect of itself», is, at least according to Kant, beyond comprehensibility. For Kant, the very idea of a material vitality, implied in a realist, ontological interpretation of natural purpose is in fact a non-sense, since it contradicts the very essence of matter:

the possibility of a living matter (the concept of which contains a contradiction, because lifelessness, inertia, constitutes its essential characteristic), cannot even be conceived⁵⁹.

The dilemma is rooted in the utter incompatibility between a concept of life that involves intrinsic teleology and a concept of nature that does not allow for the self-organization of matter. However, as stated by Thompson⁶⁰:

This dilemma no longer seems compelling. Our conception of matter as essentially equivalent to energy and as having the potential for self-organization at numerous spatiotemporal scales is far from classical Newtonian worldview. In particular, the physics of thermodynamically open systems combined with the chemistry and biology of self-organizing systems provides another option that is not available to Kant: life is an emergent order of nature that results from certain morphodynamical principles, specifically those of autopoiesis.

Autopoiesis theory thus offers a contemporary naturalized version of the Kantian notion of natural purpose⁶¹. The notion of organizational closure allows us to explain the continued existence of a system as a function of its own activity. The function of a particular component is found on its particular contribution to the persistence of a totality that produces it and on which it depends and relies for its own existence. Accordingly, purpose may be naturalized without reference to an external artifice, not even an analogy with the process of conscious design. Indeed, authors such as Varela actually put in question the analogy between evolution and engineering and the popular image of natural selection as a kind of designer (a reified «mother nature»), even if a blind one. Reversing the path of modernity, Varela rediscovers naturalistic teleological thinking, making purpose immanent once again in the real material system circularly arranged, and negating altogether the external teleology of a transcendent designer as a good model to think biology.

Contemporary science went beyond the vision of matter as essentially inert and passive, developing at least the foundations of a science of self-organized complex systems. In particular, a thermodynamic theory of organizational closure seems to accommodate all the determinations of «natural purposes», in so far as it characterizes life as a network of processes which produces its own components and also the topological boundary that defines it as an individual entity: being and doing are one and the same process in this self-reproducing dynamics. In such a network the parts mutually produce each other precisely in accordance with Kant’s definition of natural purpose, so the whole network can be viewed as cause and effect of itself. The result is a self-perpetuating totality that emerges out of local processes, but is at the same time the condition, the presupposition of these processes.

By moving in that direction, however, we seem to be back to something similar to the naturalism with final cause so dear to Aristotelian thought. We are thus forced to recognize that Aristotle was not completely wrong, and that his defense of teleology inherent in the material world, particularly in respect to organized and self-organizing beings, that is to say, organisms, is not absurd or inconsistent with an updated and enriched form of naturalism. As noted by Moss⁶²:

A sense of similarity between Aristotle’s hylomorphic understanding of soul and much more recent descriptions of self-organization dynamic systems is not entirely accidental. […] the idea that epigenesis was achieved by self-organizing movements driven by internal orientation toward an adapted form was entirely consistent with his metaphysics[…] Aristotle, by contrast, and epigenesists ever since, have endeavored to explain life-forms not as artifacts designed from without but as self-organizing, ‘autopoietic’, ends-unto-themselves.

It is worth noting, however, that nothing of what was discussed above requires us as theoretical commitment to any comprehensive teleology. Just as the primitive mechanicism of modern physics sought to impose on the entire universe an impoverished notion of matter, extrapolated from the observation of simple systems, Aristotelian hylozoism extrapolated in the opposite direction, generalizing the particular experience of living beings to the cosmos as a whole. But the fact that teleology is a natural reality, that is to say, is objectively real in the world, does not mean that it must be the reality of the whole of nature. That some material objects allow naturally explanation by final causality does not imply that all natural objects have a final cause. Natural purposes, Kantian wholes, are quite particular types of material systems that emerge from a specific organization of physical-chemical process, and it is completely plausible that this type of organization is only manifest in a very limited range of phenomena.

5. Conclusions

The perception that living beings are something distinctively different from inert matter and the attempt to explain this difference in scientific terms has been a constant theme in the history of the human inquiry on nature. One possibility consisted in postulating a difference in substance: life would be the manifestation of a special sort of substance, the so-called elan vital, which is absent in ordinary physical objects, and the distinctive features of the living were to be explained in terms of the special properties of this substance. Eventually, technology advanced to the point where the material constitution of both living and inanimate matter could be examined in minute detail, and no difference in substance was ever found. That same investigation, however, revealed living beings as extremely complex systems, composed of heterogeneous parts, each seeming to express a definite purpose and involved in very intricate relationships with other parts. So, the difference should be in the arrangement: living beings are organized physical systems, as opposed to sand and rock which, even though composed by heterogeneous materials, cannot be said to be organized. But manmade machines are also organized systems, which can be appropriately described in terms of components with specific functions and relationships among those components. Of course, men have never been able to construct a machine which resembled, not even closely, the complexity of the simplest living organism. In spite of that, the machine paradigm eventually became dominant, establishing as a general conviction that there is no essential difference between machines and organisms, a mere difference in degree of complexity, which is to be inevitably reduced and perhaps eliminated as technology advances.

The quest for the ultimate facts that would separate living beings from inanimate matter was eventually replaced by the quest for the ultimate facts about the organism as an organized physical system, under the machine paradigm. The machine should be taken not only as a metaphor, but as a model for the living. That movement inaugurated what could be called an analytical research program in biology: the living organisms were to be understood in terms of components parts with independently assessed properties, and relationships among those parts. This strategy was extremely successful as witnessed even today by the increasingly fast development of molecular biology, and the progressive identification of ever more specific components and processes involved in metabolism. The main goal in this line of research is to achieve a complete understanding of the physicochemical operation of a living organism, in the same way that the operation of a machine can be completely understood by a thorough examination of all its parts.

Maturana and Varela, and Robert Rosen, begin all their discussions by emphasizing that what is most relevant about the living organism is to be found in its organization. More precisely, they first claim that in order to understand the phenomenon of life biology should adopt the same type of relational description already applied to machines, where components are presented in terms of the abstract functions they support, and the system as a whole is described in terms of a set of relations among those components. They go so far as to postulate a principle of independence between organization and physical structure: specific material details are unimportant in the sense that several material structures can typically instantiate the same properties and functions that characterize the components. As a natural consequence, they propose to that what is special about life cannot be found by examining the particular composition and properties of its material structure, but rather by attempting to have a glimpse of what is essential about its overall organization. It is in this context that they indicate that the specificity of living organisms resides on the special type of relations in which their components are involved: relations of mutual production and interdependence. Every component in a living organism has a natural tendency to decay, but the long lived existence of the system is made possible by a permanent activity of self-maintenance or self-production, where components repair and produce other components. Their position can be restated in a few words as follows: not operation but construction is the key to understand life. Those very same ideas can also be found in the work of Robert Rosen.

Now, it seems to us that the fact that living beings must be explained in terms of processes of material construction does intertwine organization and material structure again. If correct, this observation would have the consequence that a purely analytical approach to biology is not viable. Living beings are entities which only exist as such while embodied as material assemblage of interdependent components. In a situation where the existence of those components is a given, that is, their construction and disintegration lie outside the scope of the dynamics under investigation (e.g., when we are trying to understand the operation of a machine), abstracting away materiality is not a real issue and the analytical approach is unproblematic. But, in the case where the components are not stable entities, being constantly subjected to processes of construction and disintegration, it is not clear any-more if they are the sort of thing that can be properly described as standing in fixed, well-defined relationships with each other. Here, materiality must be taken into account because, since the components are composite material structures, their specific properties and functions, and the relations they can participate in, are determined by their specific material form which, in turn, is the result of the processes of material construction which constitute the activity or operation of the living organism. Now we can see more clearly the challenge posed by the living organism to the traditional mode of analytical explanation. The basic and tacit assumption of this method is that we can always explain the operation of a system in terms of properties and relations among well-defined components. But, in the case of the living organism, these very properties and relations are the (indirect) result of the operation of the system, and so we cannot take the first for granted in order to explain the later. In other words, we cannot find the basic elements which are required to support the analytical explanation.

Any adequate explanation of the life process will necessarily have to consider the living organism as an organized system of material aggregates, whose specific properties and functions depend on their particular form, which are subject to disintegration but are constantly repaired and replaced by the activity of material construction of the system. Moreover, if this is indeed the case, the analytic method is not enough if we want to understand life. This, of course, does not imply that the analytic method is of no use in biology. Mechanical explanations can be used unproblematically to explain the operation of specific metabolic processes (e.g. the citric acid cycle) by assuming the existence of stable components and relations among them. We only have to give up the idea that if we obtain sufficient explanations of this kind we can put all of them together to have an explanation of life itself. Last, but not least, we emphasize that the abandonment of the machine metaphor in anything weakens naturalist commitments, and actually may be a crucial step to truly enrich our conception of nature and open the path towards a fully, and compelling, naturalization of teleology and normativity, with important consequences to the study of evolutionary processes and cognition. As Lewontin & Levins once argued⁶³:

The program of Harvey and Descartes to reveal the details of the bête machine has worked. The problem is that the machine metaphor leaves something out, and naive mechanistic biology, which is nothing physics carried on by other means, has tried to cram it all in at the expense of a true picture of nature.

1. As noted by Lenny Moss: «Nature as a whole for Aristotle was lifelike – conceptually modeled not by the example of inertness but rather by the example of living activity». L. MOSS, What Genes Can’t Do, MIT Press, Cambridge 2004, p. 7.
2. H. JONAS, Das Prinzip Leben, Suhrkamp, Frankfurt 1977.
3. L. MOSS, What Genes Can’t Do, cit., p. 7.
4. R. ROSEN, Essays on Life Itself, Columbia University Press, New York 2000, p. 266.
5. L. MOSS, What Genes Can’t Do, cit., p. 7.
6. G. KAMPIS, Self-modifying systems in biology and cognitive science. A New Framework for Dynamics, Information and Complexity, Pergamon Press, Oxford 1991, p. 192.
7. As stressed by Kampis: «[…] machines are decomposable in the sense that they are built from stable and separately accessible parts, which have separately knowable properties. Much as the machine is made up from these separate parts, so is the understanding of the machine made up from pieces of understanding of its atomistic parts». G. KAMPIS, Self-modifying systems in biology and cognitive science, cit., p. 186.
8. H. MATURANA & F. VARELA, Autopoiesis and Cognition. The realization of the living, D. Reidel Publishing Company, Dordrecht 1980.
9. «In saying that living systems are ʻmachinesʼ we are pointing to several notions that should be made explicit. First, we imply a nonanimistic view, which should be unnecessary to discuss any further. Second, we are emphasizing that a living system is defined by its organization, and hence that it can be explained as any organization is explained, that is, in terms of relations, not of component, properties». F. VARELA, Principles of biological autonomy, Elsevier North Holland, New York 1979, p. 7.
10. Ivi, p. 9.
11. R. ROSEN, Some relational cell models: the metabolism-repair systems, in Foundations of Mathematical Biology, Academic Press, New York 1973.
12. R. ROSEN, Life Itself. A Comprehensive Inquiry into Nature, Origin and Fabrication of Life, Columbia University Press, New York 1991.
13. Ivi, p. 14.
14. Ivi, p. 140.
15. F. VARELA, Principles of biological autonomy, cit., p. 7.
16. «Relational biology itself is as mechanistic a theory as any reductionist structural theory»
    R. ROSEN, Some relational cell models: the metabolism-repair systems, in Foundations of Mathematical Biology, New York 1973.
    «By adopting this philosophy, we are in fact just adopting the basic philosophy that animates cybernectics and systems theory […]. This is, I believe, nothing more and nothing less than the essence of a modern mechanicism». F. VARELA, Principles of biological autonomy, p. 7.
17. R. ROSEN, Life Itself, cit., p. 183.
18. W.R. ASHBY, Principles of the self-organizing system, in Principles of Self-Organization: Transactions of the University of Illinois Symposium, ed. H. Von Foerster and G. W. Zopf, Jr. Pergamon Press, London 1962.
19. M. POLANYI, Life’s irreducible structure. «Science», LIV (160), 1968, p. 1308.
20. R. ROSEN, Causal Structures in Brains and Machines, «Int. J. General Systems», LIV (12), 1986, p. 107.
21. H.H. PATTEE, Physical theories of biological co-ordination, «Quarterly Reviews of Biophysics», LIV (4, 2 & 3), 1971, p. 255.
22. S.A. KAUFFMAN, Investigations, Oxford University Press, Oxford 2001.
23. M. MOSSIO & A. MORENO, Organisational Closure in Biological Organisms. «Hist. Phil. Life Sci.», LIV (32), 2010, p. 269.
24. T. DEACON, Incomplete nature: How mind emerged from matter, Norton & Company, New York 2012.
25. Nicholson (2013) points roughly to the same direction. He acknowledges some elements of similarity between organisms and machines – both are physical systems that function in accordance to natural laws, consume energy and transform part of it into work, are hierarchically structured and internally differentiated, admit relational description and can thus be modeled in terms of causal relations between interacting components, and are fruitfully characterized in teleological terms – but stresses that all those similarities are ultimately superficial, and the distinctive features of organism are fundamentally different from those of machines. Above all, «organisms are intrinsically purposive, whereas machines are extrinsically purposive». Nicholson notes that in a machine the parts are causally independent, and temporally antecedent, to the whole, whereas in an organism neither is the case: the parts exist in a relation of collective interdependence – «the organism maintains its autonomy as a whole by constantly regulating, repairing, and regenerating its parts». He also draws attention to the peculiar «transitional structural identity» displayed by the organism – the materials change, yet the organization remains: «While a machine always consists of the same material components […] an organism naturally maintains itself in a state of continuous flux in which there is a permanent breaking down and replacement of its constituent materials». D.J. NICHOLSON, Organisms ≠ Machines, «Studies in History and Philosophy of Biological and Biomedical Sciences», LIV (44-4), 2013, p. 669.
26. R. LEWONTIN & R. LEVINS, Biology under the influence. Dialectical essays on ecology, agriculture, and health, Monthly Review Press, New York 2007, p. 222.
27. H. MATURANA & F. VARELA, Autopoiesis and Cognition. The realization of the living, cit., p. 79.
28. R. ROSEN, Some relational cell models: the metabolism-repair systems, in Foundations of Mathematical Biology, ed. Robert Rosen, Academic Press, New York 1973.
29. J. LETELIER, M.L. CARDENAS, A. CORNISH-BOWDEN, From L’Homme Machine to metabolic closure: Steps towards understanding life, «Journal of Theoretical Biology», LIV (286), 2011, p. 100.
30. S. W. KERCEL, The Endogenous Brain. «Journal of Theoretical Biology», LIV (3-1), 2004, p. 61.
31. Following Rosen‘s analysis, Hofmeyr concludes: «the defining difference between a living organism and any non-living object should be that an organism is a system of material components that are organized in such a way that the system can autonomously and continuously fabricate itself». J-H. S. HOFMEYR, The biochemical factory that autonomously fabricates itself: A systems-biological view of the living cell, in Systems Biology: Philosophical Foundations, ed. F.C. Boogerd, F. Bruggeman, J.-H.S. Hofmeyr and N.V. Westerhoff. Elsevier, Amsterdam 2007, p. 217.
32. «Strange loop» is a concept coined by the cognitive scientist Douglas Hofstader, but Varela, already in 1984, also used it in an article to express that with the phenomena of autonomy we enter in a “world of strange loops”: «A loop is completed whereby two levels are collapsed, intercrossed, entangled. At this point, what we wanted to hold in separate levels is revealed as inseparable, our sense of direction and foundation seems to falter, and a sense of paradox sets in». F. VARELA, The Creative Circle: Sketches on the Natural History of Circularity. in The Invented Reality. How do we know what we believe we know? (Contributions to constructivism), ed. P. Watzlawick. W.W. Norton & Company, New York 1984, p. 309. Deacon also mentions it, noting that biomolecules exhibit «process-dependent properties in the sense that they are reciprocally producers and products, means and ends, in a network of synthetic pathways. […] But in this case, this hierarchic ontological dependency is tangled in what Douglas Hofstader has called ‘strange loops’». T. DEACON, Incomplete nature: How mind emerged from matter, p. 178.
33. D. HOFSTADTER, What is it like to be a strange loop?, in Self-representational Approaches to Consciousness, ed. U. Kiregel and K. Williford, MIT Press, Cambridge 2006, p. 465.
34. Cf. R. ROSEN, Essays on Life Itself. Columbia University Press, New York 2000; S.W. KERCEL, The Endogenous Brain, «Journal of Theoretical Biology » LIV (3-1), 2004, p. 61; J-H.S. HOFMEYR, The biochemical factory that autonomously fabricates itself: A systems-biological view of the living cell, in Systems Biology: Philosophical Foundations, ed. F.C. Boogerd, F. Bruggeman, J.-H.S. Hofmeyr and N.V. Westerhoff. Elsevier, Amsterdam 2007, p. 217.
35. Varela attributes the reluctance to concede a central role to circularity to the influence of what he calls «Fregean viewpoint», which he associates with the theory of types in logicism and the atomism of well-founded sets theory: «The mental picture is that of a tree with roots and branches. But this view is awkward for describing whole systems, where the picture is more that of a closed network with roots and branches intertwining […]. No type distinctions are possible in such a network». F. VARELA, Principles of biological autonomy, p. 167. Deacon et al also relate the circularity of life with circularity in the context of linguistic or formal systems: «this is analogue of self-reference, a logical type violation, and it is not surprising that this feature is even the defining characteristic of reflexive reference in language».
    T. DEACON, J. HAAG & J. OGILVY, The Emergence of Self. in In Search of Self: Interdisciplinary Perspective on Personhood, ed. J.W. van Huyssteen & E.P. Wiebe. William B. Eerdmans Publishing Co., Gran Rapids 2010, p. 333.
36. F. VARELA, A Calculus for self-reference, «Int. J. General Systems» LIV (2), 1975, p. 5.
37. For Chemero & Turvey what all those models, including Kauffman ‘s auto-catalytic sets, share is «having every function be a product of a system, having loopy hyperset diagrams that terminate only with raw materials». They are closed to efficient cause and open to material cause. A. CHEMERO & M. TURVEY, Autonomy and hypersets, «Biosystems» LIV (91), 2008, p. 320.
38. Kercel, a Rosean, states: «In processes of life and mind, Rosenesque complexity is equivalent to autopoiesis. Its distinguishing feature is a hierarchical closed-loop of causal entailment». S.W. KERCEL, The Endogenous Brain. «Journal of Theoretical Biology» LIV (3-1), 2004, p. 61. But also Evan Thompson, a long time Varela ‘s collaborator: «Like Maturana and Varela, Rosen aims to give a precise account of the organization of life, and although they never mention each other in their writings, there are deep affinities between their theories. Unlike Maturana and Varela, however, Rosen presents a rigorous argument for distinguishing between organisms and machines. An intriguing feature of this argument is that it is precisely what Maturana and Varela would call the circular and self-referential organization of the living that distinguishes organisms from mechanisms and machines». E. THOMPSON, Mind in Life: Biology, Phenomenology, and the Sciences of Mind, Belknap Press, Cambridge 2007, p. 141.
39. «For Maturana and Varela, autonomous systems are defined by the abstract property of operational closure, leaving aside material and energetic requirements». X. BARANDIARAN & A. MORENO, Adaptivity: From Metabolism to Behavior, « Adaptive Behavior» LIV (16-5), 2008, p. 325.
40. As Chemero & Turvey observes: «From our point of view, Rosen’s metabolism-repair systems and Maturana and Varela’s autopoietic systems are valuable as characterizations of an abstract property, while Kauffman’s work is valuable for connecting this abstract property to chemical processes». A. CHEMERO & M. TURVEY, Autonomy and hypersets. «Biosystems», LIV (91), 2008, p. 320.
41. T. DEACON, Incomplete nature: How mind emerged from matter, cit., p. 262.
42. «This property of causal closure in  ‘soft material automata ‘ (as opposed to the rigid or fixed structure of relationships in traditional man-made machines) involves high rates of energy dissipation, so it requires the continuous production of work by the system. Thus, living systems, which are continuously and literally fabricating themselves, can only maintain their organization in far from equilibrium conditions by being material-thermodynamically open». A. MORE-NO, K. RUIZ-MIRAZO, X BARANDIARAN, The impact of the paradigm of complexity on the foundational frameworks of biology and cognitive science, in Handbook of the Philosophy of Science. Volume 10: Philosophy of Complex Systems, ed. C. Hooker. North Holland Elsevier, Amsterdam 2009, p. 325.
43. T. DEACON, J. HAAG & J. OGILVY, The Emergence of Self, in In Search of Self: Interdisciplinary Perspective on Personhood, ed. J. W. van Huyssteen & E. P. Wiebe. William B. Eerdmans Publishing Co., Gran Rapids 2010, p. 319.
44. Ibidem.
45. T. DEACON, Incomplete nature: How mind emerged from matter, cit., p. 311.
46. M. MOSSIO & A. MORENO, Organisational Closure in Biological Organisms, «Hist. Phil. Life Sci.», LIV (32), 2010, p. 269.
47. «In such a product of nature each part is conceived as if it exists only through all the others, thus as if existing for the sake of the others and on account of the whole, i.e., as an instrument (organ), which is, however, not sufficient (for it could also be an instrument of art, and thus represented as possible at all only as an end); rather it must be thought of as an organ that produces the other parts (consequently each produces the others reciprocally), which cannot be the case in any instrument of art, but only of nature, which provides all the matter for instruments (even those of art): only then and on that account can such a product, as an organized and self-organizing being, be called a natural end».
    I. KANT, Critique of the Power of Judgment, tr. P. Guyer & E. Matthews, Cambridge University Press, Cambridge 2000, p. 245.[
48. See, for instance: G. LONGO, M. MONTÉVIL, S KAUFFMAN, No entailing laws, but enablement in the evolution of the biosphere. arXiv:1201.2069 2012.
49. A. JUARRERO, Dynamics in Action. Intentional Behavior as a Complex System, MIT Press, Cambridge 1999, p. 47.
50. A. JUARRERO, Dynamics in Action. Intentional Behavior as a Complex System, MIT Press, Cambridge 1999.
51. A. WEBER & F. VARELA, Life after Kant: Natural ends and the autopoietic foundations of biological individuality. «Phenomenology and the Cognitive Sciences», LIV (1), 2002, p. 97.
52. L. MOSS, What Genes Can’t Do, MIT Press, Cambridge 2004.
53. E. THOMPSON, Mind in Life: Biology, Phenomenology, and the Sciences of Mind. Belknap Press, Cambridge 2007.
54. S. A. KAUFFMAN, Investigations, Oxford University Press, Oxford 2001.
55. M. MOSSIO & A. MORENO, Organisational Closure in Biological Organisms. «Hist. Phil. Life Sci.», LIV (32), 2010, p. 269.
56. T. DEACON, Incomplete nature: How mind emerged from matter, Norton & Company, New York 2012.
57. A. WEBER & F. VARELA, Life after Kant: Natural ends and the autopoietic foundations of biological individuality, «Phenomenology and the Cognitive Sciences» LIV (1), 2002, p. 97.
58. I. KANT, Critique of the Power of Judgment, cit., p. 246.
59. Ivi, p. 265.
60. E. THOMPSON, Mind in Life, cit., p. 140.
61. Although we don’t have the space to develop this point further here, we would like to at least remark briefly that, more than Kantian, this position seems to have a distinctive ʻGerman Idealismʼ flavor to it. Hegel in his philosophy of nature, while cheering Kant for the resumption of the notion of internal teleology, already present in Aristotle but lost in modern philosophy, is also a critic of what he sees as ambivalence and hesitation in the Kantian position. With the current growing interest in the Kantian concept of Naturzweck the Hegelian critique of Kant’s hesitation becomes relevant and surprisingly contemporary. Building upon Kant, Hegel offered a conceptualization of life based on «inner purposiveness», «assimilation», «self-referring negative unity», and self-determination as self-limiting – a philosophical account of living beings as incomplete wholes. But unlike Kant, Hegel never understood this circular organization (being both cause and effect of itself) as beyond comprehensibility, nor saw any incompatibility between life and matter. In his Philosophy of Nature Hegel even identified life with chemical process circularly arranged. In fact, some biologists and philosophers who, like Thompson, try to rescue Kant’s third critique end up looking more like Hegel, in the sense that what they really want is not only a mere regulative principle, but the proper German idealist insight that the Kantian notion of life can gain an ontological interpretation and life can thus serve as a starting point for questioning the traditional mechanical image of nature. What Kauffman refers as «Kantian whole» can thus more properly be called «Dialectical wholes». This view of life as natural end, a metabolic and circular conception, was basically inherited by the philosophy of dialectical materialism, which in turn had a concrete impact, via the work of Marxist (or quasi-marxist) biologists, usually the main critics of the machine metaphor and very much engaged in the development of theoretical biology; one just has to think of J.B.S. Haldane, Joseph Needham, J.D. Bernal, Oparin, Conrad Waddington, Richard Levins and Richard Lewontin. The considerable influence of Marxist ideas on the anti-reductionist camp of biological science is a story still waiting to be told.
62. L. MOSS, What Genes Can’t Do, cit., p. 8.
63. R. LEWONTIN & R. LEVINS, Biology under the influence. Dialectical essays on ecology, agriculture, and health, cit., p. 222.