Autopoiesis and Darwinism
Escobar J. M. (2012) Autopoiesis and Darwinism. Synthese 185: 53–72. Available at http://cepa.info/2792
Table of Contents
2. Darwinism for dummies
3. Autopoiesis: the organization of the living
3.1 Organization and phenomenology of the living systems
3.2 The organization of the living systems
3.3 The phenomenology of living systems
3.4 Conceptual challenges
4. 4. Conceptual problems for the theory of autopoiesis
4.1 Realism and constructivism
4.2 Self-referentiality and empirical testability
4.3 Autopoiesis as a demarcation criterion
5. Concluding remarks
The purpose of this paper is to offer a critical approach to the theory of autopoiesis in order to see how it challenges mainstream Darwinism. In the first part of the paper, I characterize Darwinism from the concepts of natural selection, heredity, reproduction, and evolution. This characterization is absolutely schematic, and I hope not controversial at all, since my aim is to provide a general background for the discussion of the rest of the paper. The second part presents the main tenets of the theory of autopoiesis, also paying special attention to the concepts of natural selection, heredity, reproduction, and evolution. The third and final part considers some criticisms that have been directed against the theory and suggests some new ones. As I said, my intention is to offer a critical approach, so that I pretend to assess neither autopoiesis nor Darwinism. The assessment, it seems to me, would be a matter of scientific debate – not properly of philosophy. Therefore, given that my approach attempts to be a conceptual clarification, my contribution to the contemporary debate about Darwinism is twofold. On the one hand, I show that conceptually autopoiesis constitutes an important challenge to Darwinism, but on the other, I also show that some fundamental aspects of the theory appear to be both epistemologically and empirically problematic, which perhaps helps to understand why autopoiesis is not widely accepted in mainstream Darwinism.
Key words: Autopoiesis, Darwinism, evolution, reproduction, natural selection, natural drift, fitness
The theory of autopoiesis, proposed by the Chilean biologists Humberto Maturana and Francisco Varela in the early 1970s, constitutes an attempt to offer a non-Darwinian approach to biological problems. Its main methodological assumption is that in order to deal with such problems in a proper way, it is first necessary to distinguish two questions: What is a living system, and what kind of historical processes can a living system generate? The first question would refer to what they call the organization of living systems, whereas the latter would refer to what they call the phenomenology of living systems, which includes things like ontogeny, phylogeny, reproduction, and evolution.
My aim in this essay is to show that this distinction is meaningful, and therefore, that the theory of autopoiesis constitutes an important conceptual challenge to contemporary Darwinism, for it entails a radical change of perspective regarding the treatment of biological problems. However, I also show that the theory still faces some empirical and epistemological problems that make a wider acceptance of it a very difficult matter.
I will begin the next section with a brief characterization of Darwinism. My purpose there is to offer a background against which the challenges of the theory of autopoiesis appear more evident. In the following section, I present the main tenets of the theory of autopoiesis and its main biological consequences. In the final section, I discuss some previous criticisms directed to the theory and propose others that, it seems to me, compromise its plausibility. Even so, my intention is not to assess the theory. I do not attempt to show that it is better or worse than Darwinism. I think that the problems to which I will point are not necessarily insurmountable, but I also think that they are serious problems that still require more conceptual clarification by the supporters of autopoiesis. So, I consider the assessment not as a philosophical, but rather as a biological, scientific task. The most that a philosophical approach like this can offer is to provide a conceptual clarification that could be useful when the scientific assessment be performed.
2. Darwinism for dummies
My major goal in this section is not to be controversial. So, the schematic characterization of Darwinism that I will outline here attempts only to highlight some elements that may easily be recognized as Darwinian by either supporters or critics of it. They constitute the Darwinian background against which the theory of autopoiesis initially emerged. In short, my attempt is to describe what sort of Darwinism Maturana and Varela had in mind when they proposed their theory.
I will focus on four of those elements. However, to begin with, let me remind us what Darwinism is not about. Darwinism is not a theory about the origin of life, nor is it properly a theory about what a living being is. It rather assumes that there are living beings, and tries to explain how they have evolved in different environmental conditions and through long periods of time, producing in that way the huge variety of living beings that we nowadays find on the Earth. In order to provide such an explanation, Darwinism employs concepts like natural selection, heredity, reproduction, and evolution.
The concept of natural selection is indeed the most important one in Darwinism. However, what the precise ontological status of natural selection is appears to be a really problematic issue even since Darwin’s days. Darwin was always ambiguous when he talked about natural selection. He called it a force, a power, an agency, and an agent, which seems to imply that he saw natural selection as a sort of entity in the world. Nonetheless, he also called it a principle, which at least lets us ask whether he was referring to an ontological or an epistemological principle. He also upheld positive and negative roles for natural selection at the same time, in the sense that natural selection would be what produces diversity as well as what eliminates the unfit organisms. In the twentieth century, this ambiguity was somehow ignored by means of the more general view of natural selection as a mechanism with a positive role. It is the mechanism that produces, among other things, the adaptations of organisms to their environment.
The mechanism of natural selection chooses those biological variations that increase the fitness of an organism in a given environment. These variations can be of different types, according to the kind of Darwinism that one defends. They can be, for instance, genetic variations, organismal variations, or even species variations. This generates the problem of on what variations natural selection properly acts, or in other words, the problem of whether there exist different levels of selection. For my purposes here, it is enough to say that the variations in general serve as the raw material on which natural selection acts in order to produce the fittest organisms with reference to a given environment. In this sense, fitness is a characteristic gradually reached through different generations.
These variations appear due to several reasons. Reproduction is the most obvious one, for Darwinism takes for granted that the individuals of a population are always different, thereby neither the offspring with respect to its parents, nor the individuals among the offspring with respect to each other, are ever exactly the same. Other causes of these variations are mutations, genetic drift, and genetic recombination. The important point here is that some of these variations are heritable – namely, the variations that involve a genetic character, for they are the variations that transmit the information that will make the phenotypic and even the behavioral development of the present and future generations possible.
When we take all these elements into account, the picture at which we arrive is something like this. We find a lot of living beings displaying a lot of differences (variations) on the Earth. However, the environment that these living beings inhabit cannot support all of them. Therefore, some have to die, whereas others can survive. Natural selection acts on the differences (variations) between those living beings and selects those better fitted to that environment. The offspring of those organisms selected inherit, among other things, the variations that helped their parents to survive, and with them, they face new selective pressures. When this process occurs for a number of generations, producing descent with modification, hence different branches in the tree of life, we say that an evolutionary process has taken place. In this sense, evolution occurs by means of natural selection, which continuously chooses those organisms with the better chances for survival. So, according to Darwinism, evolution is a process with a random as well as an oriented side. Variations always arise randomly, but natural selection directs the process via the selection of those variations that contribute to the fitness of the organism.
From the viewpoint of the theory of autopoiesis, the problem with this picture of the evolution of living beings is that insofar as it lacks a theory of what a living being is, it cannot ensure that this is an accurate description of the evolutionary process.[Note 1] In other words, from that viewpoint, only if we know what a living being is can we be allowed to claim that there is, for instance, a mechanism called natural selection acting on it, or that there is something called information that is transmitted from parents to offspring. Consequently, as we will see in the next section, according to this theory, the first problem that biology has to solve is not how living beings evolve, but rather what living beings are.
3. Autopoiesis: the organization of the living
In this section I will proceed as follows. I will first make the argument for the methodological distinction between the organization and the phenomenology of living systems explicit. Then, I will explain what that organization is according to the theory of autopoiesis. Finally, I will show what phenomenology follows from it according to this theory. To do this, I will rely solely on primary sources, since my aim is to clarify the original formulation of the theory rather than other clarifications or its developments.[Note 2]
3.1 Organization and phenomenology of the living systems
Maturana and Varela are emphatic in their argument in favor of a strong distinction between the organization and the phenomenology of the living systems. As I said before, organization refers simply to what a living system is, whereas phenomenology refers to the historical processes that a living system can generate. Nonetheless, in order to avoid misunderstandings, I will follow their own terminology.
The argument is simple, and commonsensical to some extent: Since the evolution of living systems depends on their reproduction, and since reproduction depends on there being a living system to be reproduced, it follows that the entire phenomenology of a living system depends on the prior constitution of that living system. In other words, a living system must always appear before it can both reproduce and contribute to the generation of an evolutionary network. So, the constitution of a living system has logical as well as operational precedence over the phenomenology associated with it. For that reason, the argument continues, we must not confuse the organization with the phenomenology of the living systems, for both fields involve their own particular problems. In other words, the theory of autopoiesis maintains that there is something like a basic taxonomy concerning the fields having to do with biology. One field, the one devoted to issues related to the organization of the living beings, tries to solve problems such as what a living system is and, with respect to this, what the relevance of issues like structure, relations of components, and properties of components is. On the contrary, the other field, the one devoted to issues related to the phenomenology of living beings, tries to solve problems such as to what degree the changes that a living system may undergo depend on the environment or on the living system as such.
This argument has a couple of consequences that deserve to be highlighted. On the one hand, the first question – i.e., what is a living system? – becomes always more fundamental than the second – i.e., what is the phenomenology of the living systems? The reason is that, as we just saw, in order for living systems to be able to reproduce and evolve, it is necessary that there first exist living systems, and therefore, the solution to the second question theoretically depends on the solution to the first. This means that an appropriate understanding of historical processes like reproduction and evolution requires a previous understanding of what living systems are, given that if one does not understand this, then one cannot be really sure that those historical processes actually take place in the way in which one thinks that they do. So, the explanation of the first question has theoretical precedence with respect to the explanation of the second, but never the contrary.
As I said, this is a methodological distinction. So, the theory of autopoiesis does not imply that there is only one way to deal with biological problems. It rather says that some procedures are better than others for dealing with such problems, and proposes what seems to be the best procedure. And the best procedure would be: Understand first what a living system is, for from a theoretical viewpoint, the understanding of its phenomenology relies completely on that. So, this methodological distinction does not entail that Darwinism is necessarily wrong, because it does not begin by following this procedure. It could still be the case that the answer to the first question agrees with a Darwinian answer to the second question, since theoretical precedence does not directly require historical precedence as well. The important point is that, according to the theory of autopoiesis, a decision concerning the truth of theories about the phenomenology field is possible only based on a more fundamental theory about what living systems are. Theoretically, the starting point would always be the understanding of what a living system is. Once this is achieved, the next step would be the understanding of what sort of phenomenology follows from that definition. But of course, it could be the case that the historical steps do not match the theoretical steps.
Another consequence is that the facts associated with the second question are not constitutive elements of the facts associated with the first question, and therefore, no aspect of the phenomenology defines the living system as a unity. Put in another way, the existence of a living system does not necessarily imply that, for instance, the living system will reproduce or generate an evolutionary process. On the other hand, for the theory of autopoiesis, it is not due to reproduction and evolution that a living system is what it is. Living systems are not defined by the historical processes that they can generate, but rather these historical processes are (at least partially) defined by what living systems are. In short, given any living system, the historical processes may or may not take place, for they are nothing more than accidental outcomes related to the living system. In this sense, “the establishment of a unity defines the domain of its phenomenology” (Maturana and Varela 1980: 97; Varela 1979: 31). A living system can undergo only the transformations that do not compromise its existence as such, for otherwise, it would become something else.
This distinction, as I claimed above, is meaningful. By this, I refer not to some technical sense of the term in the tradition of Wittgenstein and the Vienna Circle, but to the most familiar sense of something that is significant and worth taking into account. As I have insisted, the distinction entails a theoretical precedence of the first question with respect to the second, and this is significant because it seems to be completely natural, part of our common sense, to accept that living beings can be involved in historical processes only if they exist first. In other words, the existence of living beings constitutes the ontological as well as epistemological condition of possibility of any biological historical process. Ontological, because no matter what a biological historical process is, its basis must be some kind of living being. Epistemological, because we can think that a historical process is biological only if it relies on some kind of living, biological being. So, the distinction is significant and worth taking into account because it both makes some of our intuitions about living beings explicit, and gives them a more rigorous expression.
Darwinism, however, does not take account of this distinction, and that is why we may find a first challenge to it here. As I have repeated, it does not necessarily follow from the theory of autopoiesis that Darwinism is wrong. But it does follow that if it is right, this is somehow due to a matter of luck, because a supporter of the theory of autopoiesis could argue that Darwinism relies on a methodological misunderstanding. It attempts to explain the phenomenology of living systems without explaining first what makes that phenomenology possible, both ontologically and epistemologically. The problem with such an approach would be that since the living system defines the domain of its phenomenology, it is possible that the Darwinian looks at a phenomenology that does not in fact correspond to the living system. So, it is possible that evolution, for instance, could take place in a way completely different from that proposed by the Darwinian, and the only way to determine this would be to first clarify what a living system is, for only thanks to this clarification would we be able to know whether a living system can actually undergo the kinds of transformations predicted by Darwinism. In other words, only thanks to this would we be able to evaluate different evolutionary theories in order to determine which one is better. Therefore, the challenge consists in the request of a theoretical justification for the Darwinian practice of dealing with the latter question without first solving the former. And of course, it would not be enough to reply that it does not matter, because Darwinism works, since the problem is precisely to determine why it works. After all, it is always possible to find very efficient practices with no theoretical justification at all, and it seems plausible to claim that we do not want Darwinism to be one of these.
So, the options for Darwinism seem to be either to deny that the distinction is meaningful, or to provide a justification for the Darwinian practice.
3.2 The organization of the living systems
I now turn to the theory of autopoiesis’s solution for the first question. To see this solution, we must begin by considering a distinction that is central for it: the distinction between organization and structure. Organization would refer to “the relations that define a system as a unity, and determine the dynamics of interaction and transformations which it may undergo as such a unity” (Maturana and Varela 1980: 77).[Note 3] Structure would, on the other hand, refer to “the actual relations which hold between the components which integrate a concrete machine in a given space” (Maturana and Varela 1980: 77).[Note 4] Thus consider, for instance, what happens with a computer. We can identify three different things in a computer: First, its components, for example, its processor, hard disk, and memory; second, its operating system; third, a particular configuration of both the components and the operating system. Suppose that one has four computers with different components. They may differ concerning the distribution of these components in the computer case, or even concerning the capacity of their hard disks or the speed of their processors. Nonetheless, all of them could work with the same operating system. In this sense, the operating system would be what organizes those components in order to work in a certain specific way. The operating system would be the organization, whereas the components organized in that specific way, _that is, the particular configuration of both the components and the operating system in each of the four computers, would be the structure. So, the same organization could be realized in many different structures.[Note 5]
Another point that we must take into account is the distinction between allopoietic and autopoietic machines. Allopoietic machines would be those machines that “have as the product of their functioning something different from themselves” (Maturana and Varela 1980: 80). On the other hand,
::An autopoietic machine is a machine organized (defined as a unity) as a network of processes of production (transformation and destruction) of components that produces the components that: (i) through their interactions and transformations continuously regenerate and realize the network of processes (relations) that produced them; and (ii) constitute it (the machine) as a concrete unity in the space in which they exist by specifying the topological domain of its realization as such a network. _(Maturana and Varela 1980: 78–79, their italics)
A final point is that Maturana and Varela employ terms like ‘systems’ and ‘machines’ to refer to living beings. In other words, for them, ‘living being’, ‘living system’, and ‘living machine’ are equivalent terms. That is the reason why I used the terms ‘living being’ and ‘living system’ interchangeably before. As a consequence of this, two further features of these definitions must be underlined. On the one hand, the distinction between organization and structure applies not only to living systems, but to any system in general. So, no matter what system one considers, it will always be possible to distinguish its organization from its structure. On the other hand, neither the definition of allopoietic machine, nor the definition of autopoietic machine necessarily includes the notion of living machine. That explains why “living” never appears in the list of characteristics that identify autopoietic machines, whereas other characteristics like autonomy and individuality always do. In short, this last distinction has theoretical priority with respect to the definition of living system as such.
From all this, we finally arrive at the definition of living system, i.e., the answer to our first question:
If living systems are machines, that they are physical autopoietic machines is trivially obvious: they transform matter into themselves in a manner such that the product of their operation is their own organization. However we deem the converse is also true: a physical system if autopoietic, is living. In other words, we claim that the notion of autopoiesis is necessary and sufficient to characterize the organization of living systems. (Maturana and Varela 1980: 82, their italics)
A system is a living system if and only if it is a physical autopoietic machine. One important implication of this definition is that one could ask whether there are autopoietic machines that are not physical. Put in another way, the definition does not clarify what we should understand by physical. For example, are geometrical relations physical as well? Are social, linguistic, emotional, political, or economical relations physical? In general, are only certain relations physical? As far as I can tell, Maturana and Varela provide no clarification of this issue, which, as we will see below, could be the cause of several problems for the interpretation of their theory. For the time being, I want to stress other consequences of this definition.
On the one hand, it constitutes a demarcation criterion between living and nonliving beings. As Maturana and Varela claim,
to the extent that the nature of the living organization is unknown, it is not possible to recognize when one has at hand, either as a concrete system or as a description, a system that exhibits it. Unless one knows which is the living organization, one cannot know which organization is living. (Maturana and Varela 1980: 83)
On the other hand, the definition stipulates conditions for the production of living systems (ibid). The reason is that for a given system to be a living system, its components must maintain the kind of relations referred to in the definition, which can be further detailed in this way:
Relations of constitution that determine that the components produced constitute the topology in which the autopoiesis is realized. Relations of specificity that determine that the components produced be the specific ones defined by their participation in the autopoiesis.Relations of order that determine that the concatenation of the components in the relations of specification, constitution and order be the ones specified by the autopoiesis. (Maturana and Varela 1980: 88)
So, in principle, it could be possible to produce living systems at will, which has already been tried with different results (Guiloff 1981; Luisi 2003). In short, the definition not only establishes the conditions for recognition, but also for production and manipulation of living systems.
Likewise, as the reader might have already noticed, the definition does not deal with any particular sort of components. This means that the fact that living systems as we know them on Earth are constituted by a certain class of components should not lead us to conclude that autopoietic organization could not be realized in a different class of components. In other words, components do not have to meet any particular condition in order to enter into a living system, except the condition of contributing to the realization of the living system’s autopoiesis. This is another way of saying that autopoietic organization, as with any other organization, can be realized in a variety of structures as long as those structures are able to maintain the autopoietic organization constant. In short, the structures of the physical autopoietic systems that we find on Earth are an accidental, and not a necessary, feature of living systems.
3.3 The phenomenology of living systems
I have argued that it does not necessarily follow from the theory of autopoiesis that a Darwinian explanation of living systems’ phenomenology is automatically wrong. To the best of my knowledge, this claim is made neither in the primary, nor in the secondary literature. However, it is a claim that, it seems to me, agrees with Maturana and Varela’s approach to the problem, for they are completely aware that “the assertion that physical autopoietic systems are living systems requires the proof that all the phenomenology of a living system can be either reduced or subordinated to its autopoiesis” (Maturana and Varela 1980: 88; see also Maturana and Varela 1987a: 56). So, after providing a theory of what living systems are, their next step is to provide a theory for the most relevant aspects belonging to their phenomenology. They focus on ontogeny, reproduction, and evolution.
The first element that we must consider regarding the phenomenology field is that it deals exclusively with what they call historical phenomena. They define this notion thus: “[E]ach time in a system that a state arises as a modification of a previous state, we have a historical phenomenon” (Maturana and Varela 1987a: 57, their italics; see also Maturana and Varela 1980: 102–103). Notice that historical phenomena refer to states that arise as modifications of previous states. These modifications are only structural, for a modification in the organization of the living system is never possible. A modification in the organization would imply that the living system is not anymore what it was. In short, modification in organization means death.[Note 6] So, what happens in a historical phenomenon is that the living system undergoes a series of modifications in its structure, modifications that always maintain its organization constant.
They treat ontogeny as “the history of the structural transformation of a unity. Accordingly, the ontogeny of a living system is the history of maintenance of its identity through continuous autopoiesis in the physical space” (Maturana and Varela 1980: 98). Then, ontogeny is the result of the interactions between the living being and its environment. However, that structural transformation would not be determined by the environment, but only triggered by it, for any structural transformation in the living system would be determined by its own structure. The same applies to the environment with respect to the living system. Put another way, the perturbations caused by the environment on the living being – and vice versa – never entail a particular transformation in the living system. In the face of some perturbation, the living system tries to compensate for it, and to do so, it transforms its structure in such a way that the compensation can take place. If the living system cannot produce the structural compensations, then it cannot realize its autopoiesis, so it dies. Therefore, ontogeny involves a constant structural coupling between environment and living system. In this “ongoing process” of structural coupling, “environment and unity [living system] act as mutual sources of perturbation, triggering changes of state” (Maturana and Varela 1987a: 95–99). In Maturana and Varela’s own terminology, this structural coupling is but a way of saying adaptation, thereby “every ontogeny as an individual history of structural change is a structural drift that occurs with conservation of organization [i.e., autopoiesis] and adaptation” (ibid: 102–103, their italics).
A second historical phenomenon explained by the theory of autopoiesis is reproduction. In its more general sense, reproduction is the phenomenon in which “a unity undergoes a fracture that results in two unities of the same class” (ibid: 61, their italics). So, with respect to living systems, it appears as a complication of its autopoiesis:
[S]ince living systems are characterized by their autopoietic organization, reproduction must necessarily have arisen as a complication of autopoiesis during autopoiesis, and its origin must be viewed and understood as secondary to, and independent from the origin of the living organization. (Maturana and Varela 1980: 100)
This complication means that the living system suffers a modification in its structure as a consequence of trying to maintain its organization constant due to some perturbation. And the only way to do so is by fracturing its structure, which entails that the reproducing and the reproduced living systems are never exactly the same (ibid: 101; Maturana and Varela 1987a: 65–66).
Given that the theory of autopoiesis is completely epigenetic, informational notions would imply some form of preformationism. Thus, the reproductive process, according to Maturana and Varela, involves no informational notion at all. What is reproduced is not a code, genetic or otherwise, but an organization, therefore “there is no transmission of information between independent entities” in this process (Maturana and Varela 1980: 102). Heredity means “the transgenerational conservation of any structural aspect in a linage of historically connected unities” (Maturana and Varela 1987a: 68, my italics). Nucleic acids or genes would be some of, but not the only, structural aspects conserved during reproduction, and certainly their role would not be to transmit certain information transgenerationally, for there exists no information to be transmitted. In other words, these components play no crucial role regarding reproduction and heredity. As any other component in the living system, their role is determined by the maintenance of the autopoietic organization.
We finally arrive at the phenomenon of evolution. As in the case of ontogeny, evolution refers to a historical process in which the same organization is realized in a series of related structures. It differs from ontogeny in that this kind of realization depends on an intermediate historical process: reproduction. So,
evolution is the history of change in the realization of an invariant organization embodied in independent unities sequentially generated through reproductive steps, in which the particular structural realization of each unity arises as a modification of the preceding one (or ones) which, thus, constitutes both its sequential and historical antecedent. Consequently, evolution requires sequential reproduction and change in each reproductive step. Without sequential reproduction as a reproductive process in which the structural realization of each unity in the sequence constitutes the antecedent for the structural realization of the next one, there is no history; without change in each sequential reproductive step, there is no evolution. In fact, sequential transformations in a unity without change of identity constitute its ontogeny, that is, its individual history it is an autopoietic unity. (Maturana and Varela 1980: 103)
But what are the details of this historical process? To solve this question, Maturana and Varela proposed their theory of evolution by means of natural drift.
As happens with ontogeny and reproduction, this theory asserts that the environment does not determine any specific modification in the living system – nor does the living system determine any in the environment. The modifications are structurally determined by the living system as a response to the perturbations caused by the environment. In this sense, once again, the environment only triggers the modifications. On the other hand, as we saw, ontogeny presupposes an ongoing process of structural coupling (adaptation) between the environment and the living system. The same applies to evolution: The evolutionary process presupposes structural coupling (adaptation) as well. However, it is worth noticing that structural coupling (adaptation) is not something that the living system acquires gradually, but rather what guarantees that the living system does not disintegrate. So, evolution is not “a process in which there is an environmental world to which living beings adapt progressively, optimizing their use of it” (Maturana and Varela 1987a: 115). Adaptation (or structural coupling) is a condition for evolution, and not a result or an outcome of it.
Therefore, evolution depends on three elements: reproduction, conservation of autopoiesis, and conservation of adaptation (structural coupling). Reproduction generates structural variation, but each instance of structural variation occurs in a process of continuous conservation of both autopoiesis and adaptation. In this sense, no living system is fitter than others, for “as long as they are alive, they all met the requirements for an uninterrupted ontogeny” (ibid: 114). Consequently, differences in fitness become only a descriptive notion used by the observer, not a constitutive element of the evolutionary process, given that “comparisons about efficiency belong to the realm of the observer’s descriptions; they are not directly related to what happens in the individual histories of conservation of adaptation” (ibid). Something similar happens with natural selection: It is only a descriptive notion used by the observer, not a mechanism in the evolutionary process: “the word ‘selection’ denotes the observer’s understanding of what occurs in each ontogeny, even though this understanding arises from his comparative observation of many ontogenies” (ibid: 101). Put in another way, the observer beholds a number of living systems in a first moment and a different number in a second moment. From the comparison of these numbers, the observer concludes that an instance of natural selection has taken place.
Hence, since natural selection is not a mechanism, and since there is no degree in fitness distinguishing some living systems from others, the conclusion is that evolution must have a different cause. This cause, which Maturana and Mpodozis (2000) call “mechanism,” is natural drift:
[E]volution is a natural drift, a product of the conservation of autopoiesis and adaptation. […] there is no need for an outside guiding force to generate diversity and complementarity between organism [i.e., living system] and environment. Nor is that guiding force needed to explain the directionality of the variations in a lineage, nor is it the case that some specific quality of living beings is being optimized. [Living systems] are composed of harmoniously interconnected parts that are a product not of design but of a natural drift. (Maturana and Varela 1987a: 117, their italics)
So, in the process of structural drift (ontogeny), a complication with structural fracture occurs (reproduction) and a new living system appears with a different process of structural drift (a different ontogeny). This sequence is repeated several times, always with conservation of autopoiesis and adaptation, but never in terms of a mechanism optimizing the performing of the different living systems with respect to the environment. When the observer looks at this historical phenomenon, she may conclude that some living beings have been selected, for not all of them have been able to conserve their autopoiesis and adaptation. But what has in fact happened is just a process of natural drift:
[T]he natural drift will follow only the courses that are possible at each instant, often without any great variation in the appearance of the organisms (phenotype) and often with many ramifications, depending on the kinds of organism-environment relations that are conserved. Organisms and environment vary independently: the organisms at each reproductive stage and the environment according to a different dynamics. From the encounter of these two variations will emerge phenotypic stabilization and diversification as a result of the same process of conservation of adaptation, and autopoiesis depending on when the encounter takes place: stabilization when the environment changes slowly, diversification and extension when it changes abruptly. (ibid: 109)
3.4 Conceptual challenges
I said above that the theory of autopoiesis’s emphasis on the distinction between two questions implies a methodological challenge for Darwinism. The answers to these questions as such imply a different challenge. The most obvious reason for this is that they offer an explanation of the living system and its phenomenology doing away with core Darwinian notions like information, coding, natural selection as a mechanism, and degrees of fitness. However, I want to argue that in addition to this, the real challenge lies in another place.
We have seen that the common feature of the explanations of the phenomena of ontogeny, reproduction, and evolution is that whereas the environment’s perturbations trigger modifications in the living system, their actual occurrence is structurally determined by the living system itself. In this sense, a single mechanism is enough to explain the whole phenomenology of living systems. Or as some would like to claim, we have here a unified theory of the whole phenomenology of living systems. Whether this is the case with Darwinism is hard to decide, since it is not completely clear yet what the proper formulation of issues like the levels-of-selection problem, the degree of genetic determination in the processes of ontogeny and phylogeny, or the questions about the mode and tempo of evolution raised by neutralist and punctuationist approaches are. I am talking in this case about the conceptual formulation, or in other words, I am referring to the lack of consensus regarding even the proper way to understand and frame such issues in order to know how to perform the pertinent empirical tests to decide between the different options. On the contrary, it is certainly clear what the mechanism generating the phenomenology of living systems is according to the theory of autopoiesis.
If we add to this mechanism the denial of those notions mentioned above, the conceptual challenge to Darwinism becomes evident. The theory of autopoiesis attempts to provide an explanation of the phenomenology of living systems in which core Darwinian notions turn out to be mere descriptive notions used by the observer to deal with biological phenomena, not notions with actual empirical referents. From the point of view of the theory of autopoiesis, this could help to clarify why the ontological status of things like natural selection and information are so problematic – because they are notions without any ontological ground beyond the discursive domain of the observer. On the contrary, the mechanism of the theory of autopoiesis would have easily recognizable empirical referents like perturbations, structure, and modifications.
So, my claim is that, in a way similar to what happened with the first challenge, the option for Darwinism here seems to be, first, to make its notions precise and unambiguous, and second, to undermine the mechanism proposed by the theory of autopoiesis by means of a discussion of the conceptual and empirical problems involved in this theory.
4. 4. Conceptual problems for the theory of autopoiesis
Though in this section I discuss some epistemological and empirical problems regarding the theory of autopoiesis, my aim will not be to directly reply to the challenges that I pointed out in the previous section. Instead, I attempt to show that due to these problems, it is somehow understandable why the conceptual challenges have not been really taken into account by mainstream Darwinism. I will concentrate on four problems: realism and constructivism, explanatory circularity, empirical testability, and autopoiesis as a demarcation criterion. In the treatment of such problems, I will consider some previous objections and propose some new ones.
4.1 Realism and constructivism
It is indeed easy to agree with Zolo (1990) that Maturana and Varela are evidently ambiguous regarding the epistemological background that supports the theory of autopoiesis. They repeatedly claim that everything said is said by an observer, meaning that it is impossible to begin from an absolutely neutral epistemological position. For them, the observer can never avoid seeing the world from his or her own point of view, which implies that there is no such thing as a completely objective observation. The observer always constructs his or her own reality; thereby the question about what reality is in itself is meaningless. The only meaningful question is what reality is for a given observer. This radical epistemological constructivism becomes problematic, however, because the theory of autopoiesis does not attempt to be a theory of what a living system is for a given observer, but rather a theory about what a living system is for any possible observer. _For that reason, it is rather perplexing that Maturana and Varela forget their realist commitments, even criticizing realism as such, and focus mainly on the constructivist ones. After all, the theory assumes not only a radical epistemological constructivism, but also a radical epistemological realism, and of course, it is doubtful that both epistemological options can turn out to be compatible. So, though I agree with Zolo that this is a very important issue, I think that the theory still faces other critical, and perhaps more serious, problems.
4.2 Self-referentiality and empirical testability
I think that Scheper and Scheper (1996) completely miss the point when they maintain that the theory of autopoiesis is circular. They seem to confuse the explanation of a circular phenomenon with the circularity of an explanation. However, though it is evident that the theory of autopoiesis deals with circular phenomena, its explanations are not circular. An example of this is that, as we saw earlier, the definition of living system is not included in the definition of autopoietic machine. On the contrary, the latter definition is what makes the arrival to the former possible.
Nonetheless, it is beyond discussion that the theory relies on a manifest instance of self-referentiality. On the one hand, in order to recognize living systems, we have to check whether an entity fulfills the criteria established by the notion of autopoiesis. But on the other, the only way in which we can recognize a given entity as a living system is to assume that those are the only possible criteria, for autopoiesis is a necessary and sufficient condition to characterize living systems. In other words, we are able to recognize living systems because _we have first defined what a living system is.
The self-referentiality is clear. However, it seems to me that this should not be confused with a tautological character of the theory, as some appear to have done (Scheper and Scheper 1996; Zolo 1990). It is true that if the theory were a tautology, it would have no real empirical content, no matter that some or even all of its terms could have empirical referents. But it is not true, as Scheper and Scheper seem to suggest, that if the theory is empirically untestable, then it must necessarily be a tautology, and therefore, that it has no empirical content at all. I rather defend the thesis that the theory is empirically untestable, but that even so, it has empirical content. Hence, it is not a tautology. Once again, self-referentiality should not be confused with something like the tautological character of a theory.
As we have seen, autopoiesis attempts to be a definition of living systems. As we have also seen, the definition speaks about physical things like physical autopoietic systems, physical structures, components and relations between components, and so on. All these physical things occur in physical space. Thus, the theory speaks about things that are, at least in principle, empirically perceptible. It is not a formal theory, but an empirical one. However, it cannot be empirically tested because autopoiesis was proposed to play in biology the role of concepts like matter in physics, a planet in astronomy, or a chemical reaction in chemistry. We can recognize that a given entity is a material entity, a planet, or a chemical reaction because _we have first defined what material entities, planets, and chemical reactions are. But we can provide no empirical test of our theories about material entities, planets, and chemical reactions given that what those theories do is precisely to tell us what kind of entity certain entities are. Moreover, it makes no sense to ask how to empirically test these theories. They are self-referential theories with no possible empirical test. However, they enable us to do empirical tests at other levels. To borrow a Lakatosian terminology, these theories belong to the hard core of the research program.
So, autopoiesis not only lets us identify a given entity as a living or nonliving system, but also lets us know the conditions that an experiment must satisfy to produce a living system. The production part of the last sentence is subject to empirical testing, and that is why some efforts to produce living systems have already been done, though based, of course, on the definition of living system provided by the theory. In the same way, one can only produce a chemical reaction if one has previously defined what a chemical reaction is. However, the important point for my purposes here is that the explanation of the phenomenology of living systems should also be subject to empirical testing. A Darwinian can easily design experiments in which different populations of the same sort of organism face different selective pressures, and from this, predict that each population will follow different evolutionary paths. If this does not happen, one may conclude that something does not work with the theory. But can a supporter of the theory of autopoiesis do something similar concerning her explanation of the phenomenology of living systems? I am not very sure that she can.
Consider, for example, to take only the most relevant case, the natural drift theory. In the face of the Darwinian experiment just mentioned, the supporter of the theory of autopoiesis can doubtless reinterpret all the details in the Darwinian explanation in terms of her natural drift theory. So, if we criticize the theory by claiming that it just offers a restatement of well-known terms like natural selection, fitness, adaptation, and evolution, she could argue that what we have here is in fact an obvious instance of empirically equivalent theories: both theories would be able to explain the same evidence. Therefore, no matter who designs the experiment, a decision between the two theories cannot be taken based only on it. However, this would be a really bad move on the part of the supporter of the theory of autopoiesis, for the question is whether she can in fact both design the experiment and put forward some predictions at the same time. Once again, I am not very sure that she can. The reason is that whatever the result may be, that result would have been already predicted by the natural drift theory. Let us see why.
We saw above that according to Maturana and Varela , phenotypic stabilization and diversification emerge from the encounter of the environment and the living system: “stabilization when the environment changes slowly, diversification and extension when it changes abruptly” (Maturana and Varela 1987a: 109). However, if we follow the natural drift theory, we can never be sure whether that is actually the case, given that if the environment changes slowly and, even so, there is diversification, the natural drift theory can still explain this. It would be a situation in which the environmental changes, though too slow, originated some perturbations in the living system that triggered some modifications. These modifications, finally realized due to the structural determination of the living system, appear to be abrupt only in the descriptive domain of the observer, for slowness and abruptness are not operative notions in the domain of the autopoietic unity. I would not go as far as saying that the natural drift theory is just an ad hoc theory, as Valenzuela (2007) maintains, for it was proposed not to fix problems in the theory concerning the organization field, but to provide explanations concerning a completely different field, the phenomenology field. However, I certainly think that the incapacity to independently explain notions like stabilization and diversification show that there is no way to contradict the natural drift theory. The most precise prediction that it makes is that living systems drift, which is too vague to be really considered as a prediction, for any observation would be favorable to it.
It seems obvious to me that similar criticisms may easily be raised regarding the other explanations having to do with the phenomenology of living systems. So, let me finish by saying that although I believe that self-referentiality is not a serious problem for the theory of autopoiesis – though I agree that some precisions about this self-referential character of the theory could be provided – the theory does lack empirical testability, not properly regarding the answer to the first question, but to the second. This seems to confer further support to my earlier remarks that the answer to the first question does not imply any particular answer to the second. For that reason, autopoiesis and Darwinism are not necessarily incompatible, as the primary and secondary literature have frequently taken for granted. An approach integrating both perspectives could in principle be tried, and perhaps that is why some efforts in this sense have been proposed (Krohn and Küppers 1989; Weber and Depew 2001).
4.3 Autopoiesis as a demarcation criterion
We saw above that, according to Maturana and Varela, autopoiesis constitutes the necessary and sufficient condition to characterize living systems. If so, one would expect that in the face of a list of entities, the application of the criteria proposed by the theory of autopoiesis would be enough to distinguish the living from the nonliving beings in that list. In this sense, autopoiesis would work as a demarcation criterion that would easily let us recognize which entities in the physical world are alive, and which are not. This aspect of the theory, as we also saw, was emphasized by the earliest version of it, and in fact remains one of its strongest aspects according to its supporters, who offer lists of entities classified as living and nonliving beings depending on whether those entities fulfill the autopoiesis criteria (Luisi 2003; Varela 2000).
Nonetheless, there is still a serious problem with this idea of autopoiesis as a necessary and sufficient condition. As has happened with other biological theories in the past, the theory of autopoiesis has been applied to other domains in the social and human sciences such as sociology, law, political science, and psychology, domains where people refer to autopoietic systems that would not properly be living systems. The plainest example of this sort of system would be a society, whose functioning would clearly display the features of an autopoietic system.
Varela (1981, see also his preface in Maturana and Varela [Note 1973] 1994) openly rejected the plausibility of this extension of the theory, whereas others speak about it as a “metaphorical” extension of it (Mingers 1989; Luisi 2003). This shows that a consensus has not been reached yet concerning the plausibility of applying autopoiesis to fields different from biology. However, if autopoiesis does constitute the necessary and sufficient condition to characterize living systems, then this should not be a problem at all. If a society is an autopoietic system, it must be a living system, for otherwise we would be assuming that there are other conditions that must be taken into account in order to characterize living systems. In short, if we cannot decide whether an entity like a society is a living system according to the criteria offered by the theory, then autopoiesis does not constitute the necessary and sufficient condition to characterize living systems. We must appeal to other criteria as well.
This problem, it seems to me, arises from the lack of precision concerning the notion of the physical, and specifically, of physical space in the primary literature. An example of this is an affirmation like this: “there is no restriction on the space in which an autopoietic system may exist. The physical space in which living systems exist is only one of many” (Maturana 1981: 22–23). This suggests that it is possible to talk about non-physical autopoietic machines, i.e., autopoietic machines that are not living beings. However, the question is in which space those autopoietic machines realize, or in other words, the question is whether space is used here in a merely metaphorical sense, or if Maturana in fact speaks about the existence of spaces different from the physical one. If the latter is the correct option, what is the ontology of those spaces? Moreover, if that is correct, the difficulty is why only the autopoiesis realized in the physical space generates living systems, whereas in other spaces, the autopoietic machines generated are not living machines.
A similar problem has to do with the identification of organisms (and other autopoietic systems of second and third order, such as an organ or the nervous system, for instance) as physical autopoietic systems. They are indeed living systems, but are they also autopoietic systems? Maturana and Varela never agreed about this point, but once again, if autopoiesis is a necessary and sufficient condition, this should not be a problem at all. The application of the criteria proposed by the theory would be enough to settle the issue.
So, there is a deep ambiguity concerning the status of autopoiesis as the necessary and sufficient condition to characterize living systems. The actual use of the notion of autopoiesis in the primary and the secondary literature suggests that autopoiesis is certainly considered as a necessary, but indeed not as a sufficient, condition for the characterization of living systems (Luisi 2000). Someone could claim that this weak version of the theory is preferable to the strong one. However, the main implication of this perspective is that it becomes really hard to accept that this weak autopoiesis constitutes a genuine demarcation criterion between living and nonliving systems. It would, at most, be a plausible indication to draw such a line. But if so, we still lack a solution for the first question, and since this question has theoretical precedence over the second, then the theory concerning the phenomenology of living systems still requires the same theoretical justification that Darwinism does. In short, in order for the theory of autopoiesis to truly work, autopoiesis must be the necessary and sufficient condition to characterize living systems. But how it can be a necessary and sufficient condition is something that no one really knows.
5. Concluding remarks
As I said in the introduction, my purpose has been to assess neither Darwinism nor the theory of autopoiesis. My purpose, on the other hand, has been to show that conceptually, the theory of autopoiesis appears as a serious challenge to Darwinism, since it constitutes an attempt to totally change the orientation of the biological sciences as a whole. From this viewpoint, perhaps everything in biology still makes sense only in light of evolution, but only because evolution first makes sense in light of autopoiesis. On the other hand, I have aimed to show that the theory of autopoiesis has its own conceptual problems, which makes a complete acceptance of it very difficult.
I think that the second aspect of my purpose does not contradict the first part, i.e., it is not the case that, because the theory of autopoiesis presents the problems that I have pointed out, one should not consider it as a serious conceptual challenge to Darwinism. After all, Darwinism itself has suffered a series of improvements since Darwin’s initial formulation of his theory, and one could expect that something similar will happen with respect to autopoiesis. But even if it turns out to be irremediably wrong in the end, it seems to me that the theory of autopoiesis has done at least two things: namely, to show that Darwinism could plausibly be taken as arbitrarily incomplete due to the lack of an answer for the first question, and to suggest that it is possible and desirable to provide explanations integrating the answers to both questions. The main conceptual challenge lies, then, in being an option whose rejection cannot be based on the argument that it is wrong because it contradicts orthodox Darwinian grounds, as Valenzuela 2007) sometimes argues, because what is at stake is precisely the plausibility and verisimilitude of those grounds. Evidently, the theory of autopoiesis is not wrong because, for instance, it does not consider natural selection as a mechanism. However, Darwinism would have very difficult times if natural selection were not a mechanism. Then, no matter whether the theory of autopoiesis is wrong, the Darwinian must clarify what it means to say, for example, that natural selection is a mechanism. Is this a mere metaphor taken from the physical sciences, or should it be read at face value? If so, how do we determine the truth-value of a statement like “natural selection is a mechanism with a positive role in evolution?” Or if, on the other hand, natural selection is understood as a force, should it be perhaps understood in terms of a Newtonian relation between mass and acceleration? That is the serious challenge of the theory of autopoiesis – to ask Darwinism for a clarification of some of its core methodological and theoretical notions and assumptions.
It should also be noticed that a conclusion that one can draw from my discussion of the problems of the theory of autopoiesis is that contrary to what some claim or imply, Darwinism and autopoiesis are not necessarily competitive paradigms, and therefore, they are not necessarily incompatible. If so, autopoiesis could maybe be taken as that lacking notion that Darwinism needs to settle some contemporary debates around it, as some have already tried to do. But if someone thinks that it is not, then the challenge is to show why. Of course, the answer cannot be given only by pointing to the problems in the explanations concerning the phenomenology of living systems, given that the main issue is whether Darwinism is compatible with the answer to the first, and not properly with the answer to the second question. Neither would it be enough to point to the problem of the necessary and sufficient character of the notion of autopoiesis, for perhaps Darwinism could work with a weak version of it – though, as I have argued, the theory of autopoiesis as such cannot. So, to deal with the problems and challenges of the theory of autopoiesis, it is mandatory to provide further scientific accounts and clarifications of notions from both Darwinism and autopoiesis, something that goes beyond the limits of this essay. My purpose was simply to contribute a conceptual clarification of the theory of autopoiesis, which does not necessarily rule out the possibility of future improvements of it.
I want to thank all the participants in the seminar “Darwin in the 20th century” (Spring 2009, University of Notre Dame) for a very helpful and useful session devoted to the theory of autopoiesis. Particularly, I want to thank Pr. Grant Ramsey, Anne Peterson, and Manuela Fernández Pinto for their comments and suggestions in previous versions of this essay. I also want to thank two anonymous referees for very constructive observations on different points of my arguments.
Camus: A. (2000). Evolution in Chile: Natural drift versus natural selection, or the preservation of favoured theories in the struggle for knowledge. Revista Chilena de Historia Natural, 73, 215–219.
Etxeberria, A. (2004). Autopoiesis and natural drift: Genetic information, reproduction, and evolution revisited. Artificial Life, 10, 347–360.
Guiloff, G. D. (1981). Autopoiesis and neobiogenesis. In Zeleny (Ed.) Autopoiesis: A theory of the living organization (pp. 118–125). New York: North Holland.
Krohn, W., & Küppers, G. (1989). Self-organization: A new approach to evolutionary epistemology. In K. Hahlweg & C. A. Hooker (Eds.), Issues in evolutionary epistemology (pp. 151– 170). Albany: State University of New York Press.
Luisi: L. (2003). Autopoiesis: A review and a reappraisal. Naturwissenschaften, 90, 49–59.
Maturana, H. R. (1981). Autopoiesis. In Zeleny (Ed.) Autopoiesis: A theory of the living organization (pp. 21–33). New York: North Holland.
Maturana, H. R. (1995). La realidad: ¿objetiva o construida? Fundamentos biológicos de la realidad (Vol. 1), Fundamentos biológicos del conocimiento (Vol. 2). Madrid: Anthropos.
Maturana, H. R., & Mpodozis, J. (2000). The origin of species by means of natural drift. Revista Chilena de Historia Natural, 73, 261–310.
Maturana, H. R., & Varela, F. J. (1994). De máquinas y seres vivos. Autopoiesis: La organización de lo vivo. (Santiago: Editorial Universitaria).
Maturana, H. R., & Varela, F. J. (1980). Autopoiesis and cognition: The realization of the living. Boston: Reidel.
Maturana, H. R., & Varela, F. J. (1987a). The tree of knowledge. Boston: Shambala.
Maturana, H. R., & Varela, F. J. (1987b). El árbol del conocimiento. Santiago de Chile: Editorial Universitaria.
Mingers, J. (1989). An introduction to autopoiesis: Implications and applications. Systems Practice, 2, 159–180.
Mingers, J. (1995). Self-producing systems: Implications and applications of autopoiesis. New York: Plenum.
Scheper, W. J., & Scheper, G. C. (1996). Autopsies on autopoiesis. Behavioral Science, 41, 1–12.
Valenzuela, C. Y. (2007). Within selection. Revista Chilena de Historia Natural, 80, 109–116.
Varela, F. J. (1979). Principles of biological autonomy. New York: North Holland-Elsevier.
Varela F. J. (1981) Describing the logic of the living: The adequacy and limitations of the idea of autopoiesis. In Zeleny (Ed.) Autopoiesis: A theory of the living organization (pp. 36–48). New York: North Holland.
Varela, F. J. (2000). El fenómeno de la vida. Santiago de Chile: Dolmen Ensayo.
Varela, F. J., Maturana, H. R., & Uribe, R. (1974). Autopoiesis: The organization of living systems, its characterization and a model. BioSystems, 5, 187–195.
Weber, B. H., & Depew, D. J. (2001). Developmental systems, Darwinian evolution, and the unity of science. In S. Oyama: E. Griffiths, & R. D. Gray (Eds.), Cycles of contingency: Developmental systems and evolution (pp. 239–253). Cambridge, MA: MIT Press.
Zeleny, M. (Ed.). (1980). Autopoiesis, dissipative structures, and spontaneous social orders. Boulder, CO: Westview Press.
Zeleny, M. (Ed.) (1981). Autopoiesis: A theory of the living organization. New York: North Holland
Zolo, D. (1990). Autopoiesis: Critique of a postmodern paradigm. Telos, 23, 61–80.
Notice that I am talking about evolution and natural selection concerning living beings exclusively. So, whether similar processes occur with respect to other sorts of entities is something that does not concern us here, and therefore, it is irrelevant for my purposes to determine whether one can formulate a much more comprehensive theory of evolution by means of natural selection dealing with entities different from living beings.
For the primary sources, see Maturana (1981), Mingers (1995), Maturana and Mpodozis (2000), Maturana and Varela (1973, 1994, 1980, 1987a,b), Varela (1979, 1981, 2000) and Varela et al. (1974). For some secondary sources, the reader can consult Camus (2000), Etxeberria (2004), Luisi (2003), Mingers (1989, 1995) and the articles in Zeleny (1980, 1981).
“Organization denotes those relations that must exist among the components of a system for it to be a member of a specific class” (Maturana and Varela 1987a: 47, their italics).
“Structure denotes the components and relations that actually constitute a particular unity and make its organization real” (Maturana and Varela 1987a: 47, their italics).
Notice that this is only an illustrative example, so I am not implying that the computational metaphor can be applied to the case of autopoiesis. I am not saying that it is possible to distinguish between something like software and hardware regarding life according to the theory. My point is rather Maturana and Varela’s own point, namely, that one can always identify components, structure, and organization in any system of our universe, living or otherwise, and that structure would be a particular set of components organized in a particular way. I will insist on this in a moment.
This does not entail, however, that organization is the same as being alive, for, as we saw before, it is always possible to identify the organization in any system of our universe; it does not matter whether it is alive or not.
Found a mistake? Contact corrections/at/cepa.infoDownloaded from http://cepa.info/2792 on 2016-06-15 · Publication curated by Alexander Riegler