If one was trying to describe "What is a living thing?" one could enumerate a list of characteristics, such as "moving" or "reproducing." This does not seem fruitful, as most of the lists can be violated, and the very characteristics we use in the list are from our own limited cognitive and epistemological background. Much of the future of cognitive science is predicted in the strange work of Maturana: the focus on embodiment, dynamic systems, anti-representationalism. It even has strictly computational ramifications, for example,
The organization of a machine (or system) does not specify the properties of the components which realize the machine as a concrete system, it any specifies the relations which these must generate to constitute the machine or system as a unity(Autopoiesis and Cognition)
Maturana and in particular Varela press very heavily not only the point that to describe a process one is necessarily using linguistic terminology that laden with meaning given to it by the outside observer and not necessarily the process being observed. Therefore in the spirit of scientific explanation and their own epistemological limitations as humans, they strive to use a language that describes the process without a fixed reference point outside the process, i.e. to describe the process without using a third-person objective viewpoint unless their theory can explain how an observer comes to be. This does tend to make their language tortured and often sound strangely postmodern.
The observer is a living system and an understanding of cognition as a biological phenomenon must account for the observer and his role in it(Biology and Cognition)
The main reason this is important is because to the system any change is just a change, and not necessarily coming from inside or outside the system from the point of view of the system. What is required to be "in" the system are unclear except in the case of autopoietic systems: the component must have a role as a process that maintains the system.
An observer beholding an autopoietic system s a unity in a context that he also observes, and which he describes as its environments, may distinguish in it internally and externally generated perturbations, even though these are intrinsically indistinguishable for the autopoietic system itself.(Autopoiesis and Cognition)
In brief, an autopoietic system is a system whose processes are directed at maintaining itself as a unity. Less briefly, it can be defined as:
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 which: (i)through their interactions and transformations continuously regenerate and realise the network of processes (relations) that produced them; and (ii) constitute it (the machine) as a concrete unity in the space in which they )the components) exist by specifying the topological domain of its realisation as such a network(Autopoiesis and Cognition)
A unity allows for some notion of identity, even if the subject of the identity is changing continually (for according to Maturana, an autopoietic system is after all maintaining itself through processes!).This notion of identity is immanent, only realized in the concrete reality of the organism's existence, unlike much previous work in philosophy. We'll return onto identity, which is a bit more complex than unity, later.
The notion of identity remains for Kant forever transcendental(Autopoiesis and Cognition) Maturana would have done better to say "unity" instead of "identity" here!
Inputs and outputs of the system are self-contained within the autopoietic unity. This unity allows it to be
changed in a manner relevant to its maintenance, and it enters into a new interaction without loss of identity(Biology and Cognition)
Only processes that actually maintain the unity are considered, all other processes are allopoietic, that is having inputs and outputs.
Autopoietic machines do not have inputs and outputs. They can be perturbed by independent events and undergo internal structural changes which compensate these perturbations. If the perturbations are repeated, the machine may undergo repeated series of internal changes which may or may not be identical(Autopoiesis and Cognition)
In other words, thinking in terms of "inputs" and "outputs" is just not really applicable to autopoietic systems. In essence, two unities have some form of physical connection, and in reaction they may each adjust their organisation.
When an autopoietic unity has a perturbation from "the environment" (i.e. any process outside the autopoietic system), it reacts to maintain its structural integrity as an autopoietic system.
There may be two types of changes
conservative changes in which only the relations between the components change; or they may innovative changes in which the components themselves change
Example: A human being can be considered an autopoietic system since its main function can be considered maintaining its human state. This state is carefully regulated by chemical balances in the body, maintaining functioning organs, and so to maintain its state as "human" a human breathes, eats, sleeps, and so on. When a human is no longer able to maintain his state as an autopoietic unity, he dies. For example, when the unity of human encounters the unity of a bullet through the heart!
Thought: Even if we accept third-person objectivity, an autopoietic state may have no clear list of properties since the process may be in a continual state of change from one moment to another, and so no list of properties is applicable over the entire autopoietic cycle. However, it might have invariances (possibly constructed as properties) that do not change either through its entire state (such as breathing for humans - since I would assume even the primitive foetus breathes) or change in an orderly fashion throughout the cycle of a poetic system (such as Piaget's stages of infant development).
Possible Example: Documents such as the AWWW give a descriptive notion of the Web, while the Web itself is defined per se not by those descriptions but the actual processes (such as sending HTTP GET and POST) embodied in concrete computers sending electrically-charged bytes around. The second-order patterns (first-order being the very concrete sending of a particular byte across the wire) that make up the Web exist and must be maintained for the Web to continue, and so the Web propagates itself by enveloping ever further mediums (Images in Mosaic, sound files, etc.) in its framework and by maintaining itself against possibly destructive fragmentation of its processes (such as HTML fragmentation, improper URI schemes, and so on). Even if the actual servers and networks are replaced, as long as the relationships among the processes holds, the Web will continue..
Somewhat surprisingly for a biologist Maturana comes out against a strict neural determinism, stating that
the neuron cannot be considered as the functional unit of the nervous system; no neuron can have a fixed functional role in the generation of conduct if it must be continuously changing its participation in it(Biology and Cognition)
So, the neuron is not itself per se important, and so reducing to the neural level leaves out the heart of the action - that the neuron is only of interest due to its interaction with the environment. However, to a neuron it has no idea if it's interacting with part of the "body" (unity) or with part of the environment.
Two interactions that produce the same state of relative activity are identical for the nervous system, no matter how different they may be in the cognitive domain of the observer
A good point, but here a hole begins to open up in Maturana's autopoetic systems. First, with the last point in mind, it becomes impossible to tell between the inside and outside of a system, ala Clark's Extended Mind hypothesis. In fact, one wonders if one should talk about systems at all, but just machines interconnecting with each other through multitudinous inputs and output.
Objection: "Autopoietic systems" are always "interacting" with other systems, and it's unclear how one can say a particular component is inside or outside the system. Now, some parts of the autopoietic system are clearly tied to maintenance of the system. Yet, these are often considered to be inputs and outputs. For example, food is important for the maintenance of the human autopoietic system. Yet, food can be considered a definite input. The food itself, once inside, changes into all sorts of things (like protein) that later are definitely needed to maintain autopoiesis. However, before consumption a particular bit of food is probably not directly considered part of the system!
Another objection to autopoietic systems is that is an over-reaction, not only does Maturana assume autopoietic systems "have no inputs" (which is clearly false and thus undermines the whole concept autopoietic systems), but he also gives too much primacy to the system:
For in each interaction it is the nervous system's structural state that specifies what perturbations are possible and what changes trigger them (Biology and Cognition). This problematic because an autopoetic system is clearly more flexible than Maturana seems to point out, and so its not just that the pre-existing system bumps into another one and changes, it's that the pre-existing system expands (even if ever so briefly) into connection with another system. Therefore the change in the system is dependent on the other system as much as it is on the pre-exisiting autopoietic system. For example, you give a poisoned apple to a sheep designed to make it sick (damage it, but let the poor sheep maintain autopoiesis!). The sheep does not have much choice in getting sick, and this is clearly the result of the sheep temporarily absorbing into its own system another component (an input even!) that it has direct connection with, one that is crucial to its autopoiesis.
The domain of all the interactions in which an autopoietic system can ever without loss of identity is its cognitive domain.(Autopoiesis and Cognition)
This is a much broader definition of cognition than is usually associated with cognitive science. The cognitive realm of bacteria is anything that bacteria bumps into without being destroyed.
Maturana also seems to foresee much of the extended mind hypothesis, but is torn between the autopoietic system being opened or closed. For example, a sort of neural determinism crops up often, as they define operational closure as a
closed network of changes in relations of activity between its components(The Tree of Knowledge). Again, it's unclear why this is "closed", in particular giving that
The possibility of the enlargement of the cognitive domain is unlimited; it is ahistorical process.(Biology and Cognition) At what point does the unity grow and fall apart? Maturana cannot fully escape the trap of using linguistic description, for it seems somewhere there is a declarative definition of a system in terms of the "networks of relations and their production".
Although he does this without anyone noticing, he then equates cognition with behavior:
The range of possible behaviour, however, is determined by its structure. This structure specifies its realms of interaction.(The Tree of Knowledge) Indeed, Maturana sees the
The operational closure of the nervous system tells us that it does not operate according to either of the two extremes,: it is neither representational nor solipistic.(The Tree of Knowledge) While they are trying to paint an admirable viewpoint, it is unclear what they mean by "representational." If one denies representationalism, even if one upholds some common ground of possible interactions between self-contained and self-maintaining systems instead of pure solipsism, Maturana is really left with no recourse.
However, he does seem to point to the nervous system as the original extended mind, in that it allows a vast multitude of cells to be connected and to allow the internal structure of the organism to be very complex. The nervous system allows organism to expand cognition in two ways:
The first, and most obvious, through expanding the realm of possible states of the organism that arise from the great diversity of sensory and motor patterns which the nervous system allows for and which is the key to its participation in the operation of the organism....the second is through opening new dimensions of structural coupling for the organism, by making possible in the organism the association of many different internal states with the different interactions in which the organism is involved.(The Tree of Knowledge)
For Maturana, the autopoiesis is primarily established on the individual level and so, contrary to much folk biological thinking, the individual must come before the species:
Reproduction requires the existence of a unity to be reproduced, and it is necessarily secondary to the establishment of such a unity; evolution requires reproduction and the possibility of change(Autopoiesis and Cognition) To return to his argument against species, Maturana states:
We have shown however, that these arguments are not valid to justify the subordination of the individual to the species, because the biological phenomenology is determined by the phenomenology of individuals, and without individuals there is no biological phenomenology whatsoever.(Autopoiesis and Cognition)
For computer scientists, Maturana usefully differentiates between reproduction and replication, since
self-reproduction takes place when a unity produces another one with a similar organization to its own, through a process that is coupled to the process of its own production.(Autopoiesis and Cognition) while in replication
aA system which successively generates unities different from itself, but in principle identical to each other So a
copy takes place whenever a given object or phenomena is mapped by means of some procedure unto a different system, so that an isomorphic object or phenomenon is realized in it. In the notion of copy the emphasis is put on the mapping process. This is topic that is mentioned in passing but it would be useful to return to at some point.
It's obvious that two autopoietic systems can interact systematically over long periods of time and this is called structural coupling:
Whenever the conduct of two or more unities is such that there is a domain in which the conduct of each one is a function of the conduct of the others, its is said that they are coupled in that domain. Coupling arises as a result of the mutual modifications that interacting unities undergo in the course of their interactions without loss of identity.(Autopoiesis and Cognition) For example, the relationship between myself and the state of the internal memory of the computer is definitely a coupled. I don't know exactly how the computer is "structurally adjusting" to my typing in terms of electrically charged bits on a memory chip, but we interact often, and interactions with the computer often change my behavior - for example, I discover something on the Internet that is of particular interest, such as great web-page on how to eat healthy foods. And this changes my concrete behavior.
Through continual coupling a new autopoietic system can form:
If the autopoiesis of the component unities of a composite autopoietic system conforms to the allopoietic roles that through the production of relations of constitution, specification, and order define an autopoietic space, the new system becomes in its own right an autopoietic unity of second order. So complex cells form from simple cells with organelles, multi-cellular systems form by combining cells, and my body forms through the inter-relations of many multi-cellular organs, skin, and so on. This second-order autopoiesis seems to be definitely limited to interactions of directly physically connected systems.
Therefore, now we are getting to the heart of the qualia problem and phenomenology:
A phenomenological domain is defined by the properties of the unity or unities that constitute it, either singly or collectively through their transformations and interactions.(Autopoiesis and Cognition)
Each phenomenology is irreversibly separate (almost solipistic!) until one realises that when unities share components, then they share phenomenologies:
Two phenomenological domains intersect only to the extent that they have common generative entities, that is, only to the extent that the unities that specify them interact; otherwise they are completely independent and, obviously, they cannot generate each other without transgressing the domains of relations of their respective specifications.(Autopoiesis and Cognition) It is then framework that we begin to open a hole in Maturana big enough to reconstruct a notion of information and representation. Indeed, it seems the key lies in the fact that certain entities can become part of the autopoietic system, even if only temporarily, and these entities can be shared - one is getting to the notion of information-sharing between organisms, constructed on a very concrete physicalism. However, this physicalism gets interesting if the shared entity was a digital image and text shared over the Web, one that could be endlessly replicated and so shared among many autopoietic systems such as humans.
Maturana is heavily against information, since he views it as the idea that a third-party description is "controlling" the autopoietic process:
Communicative and linguistic interactions are intrinsically not informative, organism A does not and cannot determine the conduct of organism B because due to the nature of the autopoietic organization itself every change that an organism undergoes is necessarily and unavoidably determined by its own organization..
notions such as coding and transmission of information do not enter in the realization of a concrete autopoietic system because they do not refer to actual processes in it(Autopoiesis and Cognition) and
the notion of coding is a cognitive notion which represents the interactions of the observer, not a phenomena operative in the observed domain(Autopoiesis and Cognition)
There is something communicated, and what is communicated is an integral part of that which travels in the tube. Thus, we usually speak of the "information" contained in a picture, an object, or more evidently, the printed world. According to our analysis, this metaphor is basically false. It presupposes a unity that is not determined structurally, where interactions are instructive, as though what happens to a system is determined by the perturbing agent and not by its structural dynamics...saying does not ensure listening. The phenomenology of communication depends on not what is transmitted, but on what happens to the person who receives it.(The Tree of Knowledge)
Objection: If one considers information to be the net of invariances in some component that can be shared among autopoietic systems, and encoding another set of invariances for autopoietic systems to easily assimilate other components in their system, then one can create a notion of information that is compatible with autopoiesis. This allows the interdependence between information and structure, so that while saying does not imply listening, if one is listening, one has the possibility of sharing what the other person is saying, and this influences, and builds, both phenomenologies.
In this coupling,
the autopoietic conduct of organism A becomes a source of deformation for organism B, and the compensatory behavior of organism B acts, in turn, as a source of deformation of organism A...and so on recursively...then these couplings can be described
as meaningful in context of the coupled behavior and are labelled
communicative interactions. What is interesting here is how these interactions then create new interactions, that can be further coupled with, and so on ad infinitum, or recursively. Here is where langauge and thinking make their appearance.
Accordingly, thinking is a mode of operation of the nervous system that reflects functionally its internal anatomical projection (possibly multiply) onto itself. The idea espoused here is very much a forerunner of the concepts used by Brian Smith, such as his definition of the Knowledge Representation Hypothesis and the implementation of reflection. In essence, thinking and language allow us to have structural couplings with ourselves.
Note that these third-order couplings are not rare, but very common, as in many systems the prime third-order couplings are
social interactions and
sexual reproduction(The Tree of Knowledge)
However, when it is recognised that language is connotative and not denotative, and that its function is to orient the orientee within his cognitive domain without regard for the cognitive domain of the orienter, it becomes apparent that there is no transmission of information through language (Biology and Cognition) However, I would point out that there are structural similarities between autopoietic systems, and so there are structural similarities in their cognitive domain, and so we may assume language can help two systems orient towards the same domain.
They precede to use the same ideas to explain culture:
By cultural behaviour we mean the transgenerational stability of behavioral patterns ontogenically acquired in the communicative dynamics of a social environment and
When we describe words as designators of objects or situations in the world, as observers we are making a description that does not reflect the condition of structural coupling in which words are ontologically established coordinations of behavior(The Tree of Knowledge)
They also use the same ideas to explain the creation of the observer within their framework:
The key feature is that language enables those who operate in it to describe themselves and their circumstances through the linguistic distinction of linguistic distinctions(The Tree of Knowledge) and
What an observer does is precisely this: he makes linguistic distinctions of linguistic distinctions(The Tree of Knowledge)
Finally, language is used to explain consciousness, making consciousness the creation of a primarily linguistic narrative that combines the disparate parts of autopoietic system into an ongoing entity:
Consciousness can be explained as the maintenance of "an ongoing descriptive recursion which we call the "I." It enables us to converse our linguistic operational coherence and our adaptation in the domain of language(The Tree of Knowledge)
Maturana also, at least in "Autopoiesis and Cognition," is very careful to refer to the systems as being composed of machines, and not necessarily what is traditionally considered "biological." As such, he leaves ample room for AI, even foreseeing the approach of scientists like Rodney Brooks and robots:
Accordingly, the full explanation of the organization of the nervous systems (and of the organisms) will not arise from any particular observation or detailed description and enumeration of its parts, but rather like any explanation, from the synthesis, conceptual or concrete, of a system that does what the nervous system (or the organism) doese.(Biology and Cognition) He even goes as far as noting that these systems can be definition be constructed!
However, if our characterisation of living systems is adequate it is apparent that they could be made at will. What remains to be seen is whether such a system has already been made by man, although unwittingly, and with what consequences.(Autopoiesis and Cognition)
Almost every Maturana book ends on a note of how his thinking provides a sort of scientific grounding for what could broadly be construed as a humanistic politics, and against transcendence, in particular the transcendent use of "natural selection" within biology.
The social history of man shows a continuous search for values that explain or justify human existence, as well as continuous use of transcendental notions to justify social discrimination, slavery, economic subordination and political submission of the individuals, isolated or collectively, to the design or whim of those that pretend to represent the values contained in those notions.(Autopoiesis and Cognition)