- QUESTIONS & ANSWERS
- Q: To What Extent Has Autopoiesis Been Related to Information Technology (IT)?
Your humble servant / editor is always glad to receive communications!
Thanks for Issue 1: Keep up the good work. We really are listening out here! Below is my much belated introduction of myself, for your next issue.
Cheers, Barry [McMullin -- see his personal introduction below].
Thanks ... for the first Autopoiesis Observer! [It]... was quite an achievement.
-- Hans-Erik [Nissen]
FROM THE EDITOR:
In the earlier "prospectus", I suggested that personal / professional introductions would be a good thing to include in The Observer, both as: (1) material "orienting" readers to interests and work in this area, and (2) a means of generating more of a sense of "community". Barry McMullin of Dublin City University has submitted the following personal / professional introduction to "kick off" this line of introductions. I encourage everyone to submit similar self-referential descriptions to help us get this forum going.
I leave the form, the contents, and the degree of contact information to your discretion. -- R.
- Research Interests:
- Autonomous (Autopoietic) Systems, Artificial Darwinism, Artificial Life, Cellular Automata, Mobile Robots, Real Time Software.
- Graduated with the degree of Bachelor of Electrical Engineering from University College Dublin, in 1980. Sponsored by the Irish Electricity Supply Board to carry out postgraduate research into electricity distribution system protection, 1980--82. Awarded the degree of Master of Engineering Science by University College Dublin, 1982; thesis title: "Digital Distance Relay". Employed as a design engineer by the Electricity Supply Board, 1982--83. Employed by Hyster Automated Handling Limited, Dublin, in the period 1983--87, as a project leader, and, subsequently, design manager; responsible for the development of real time control software for large systems of industrial robotic vehicles. Employed by Dublin City University (then NIHED) as a Lecturer from 1987 to date. Concurrently engaged in research leading toward the degree of Ph.D. with the Department of Computer Science, University College Dublin; thesis submitted in October 1992, entitled "Artificial Knowledge: An Evolutionary Approach".
- Work on "Autopoietic Theory":
- My interest in Autopoietic Theory arises from work I did during my Ph.D., in field of Artificial Life (I do hate that phrase, but I fear we're stuck with it...). Specifically, I re-examined a system proposed c. 1976 by John Holland for investigating the spontaneous origin of "life" in a highly simplified (one dimensional even!) "chemical soup". Holland's work had been purely theoretical; I did experimental computer studies of his predictions. Turned out the system behaves in ways more complex than allowed for in Holland's theory---and doesn't do the things he hoped. An interesting, but kind of negative result. (A summary report of these results was carried in the proceedings of the first European Alife conference, ECAL-91).
My "global" objective, of which this was a part, was the realisation of a spontaneous growth in automaton complexity via a fundamentally Darwinian mechanism. Not as easy as you might think (Tom Ray notwithstanding). Without ever having read my way into Autopoietic Theory, the failure of Holland's system to behave as expected, and some other straws in the wind (such as my re-evaluation of von Neumann's seminal work---that bit is summarised in my contribution to the Dublin workshop, last August) gave me a vague feeling that I was missing something important. When I then eventually did do some serious reading on Autopoietic Theory, it seemed to me that the idea that "production precedes reproduction", and that the kinds of "automata" I was studying lacked any proper autonomy, neatly captured a large part of what I was dissatisfied with.
At this point I have some ongoing work to explore these issues in much more detail with one of my students (Hyder Aswad). The initial objective of this work is to revisit the original simulation work on Autopoiesis done by Varela and colleagues, and then elaborate it in two directions: firstly to see if the autopoietic phenomenology can be realised in even simpler systems (I have a conjecture that it can be achieved even in a one dimensional framework, loosely inspired by the Holland system mentioned above), and secondly to explore what happens when the framework is made more complex and extensive (so that significant populations of moving, interacting, reproducing, autopoietic entities can be studied). A longer term objective is to formulate mathematical tools for analysing autopoiesis at least in some such restricted class of model universes. I expect this work to go on over the next two to three years---but, as always, that is contingent on someone giving me money to keep going in this relatively unfashionable area.
I also have some work going on in mobile robotics, quite similar to the sorts of things Inman Harvey and his colleagues are doing in Sussex. But, at this stage, I cannot yet see how it is going to be possible to relate this to Autopoietic Theory, even though I would quite like to!
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Barry McMullin, School of Electronic Engineering, ++ Dublin City University, Dublin 9, IRELAND. ++ E-mail: McMullinB@DCU.IE Phone: +353-1-704-5432 FAX: +353-1-704-5508 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
While there are some IT researchers familiar with / interested in autopoietic theory, their interests cannot be attributed to much in the way of clear-cut allusions in the basic literature. I've seen some unfortunate instances where widely-read conclusions have been developed on the basis of The Tree of Knowledge alone (cf. West & Travis, mentioned below). On the other hand, I've also seen a number of passing references to autopoiesis in papers which seem to have served as ornamentation rather than integral structure. My own field is computing and cognitive science, and my preferred fundament in addressing cognition is autopoietic theory. Having looked at some length for connections between autopoietic theory and IT, I offer the following overview. -- Randy
IT Theory & Autopoiesis: An Overview
First one may ask to what extent information technology (specifically computers) is addressed in the background literature. While a number of themes may leap off the pages of (e.g.) Principles of Biological Autonomy at those schooled in (e.g.) human-computer interaction, artificial intelligence, etc., one must search long and hard to find any direct reference to computers. The only such direct discussion comes in an early paper (Maturana, 1970a). This paper (a gem, by the way, delivered at a workshop on cognition and artificial intelligence) directly (but all too briefly) addresses the implications of autopoietic theory and computers. Maturana states computers:
"...differ from living systems not in the principles used for their function, but in their reference. The organization of ...[computers]... is such that they always have an alloreferential  domain of interactions, that is, the relevance of their functioning is determined by how they satisfy the desires of their maker." (p. 21)
 "Alloreferential" is an earlier term for what Maturana & Varela (1980) would later call "allopoietic" -- defined by an external ascription of purpose.
With specific reference to artificial intelligence (AI), Maturana notes:
- the apparent goal is to devise an "...artificial cognitive system."; and
- such artificial systems fall into two general categories: those which "...describe their interactions in our terms, that is, recognize what we recognize..." and "...those that will make descriptions in their own terms, but which we have to map afterward into ours." (Ibid.)
Attempts to apply autopoietic theory to information technology can be divided into three categories:
- application of the theory as a general paradigm;
- application of the formal (systemic) aspects to structure software form and/or function (artificial cognitive systems of both sorts noted above); and
- application of the phenomenological aspects to inform the design of IT applications. I shall review these in turn.
- AUTOPOIESIS AS AN IT PARADIGM:
An example of addressing autopoiesis as an IT paradigm is West and Travis (1991a;1991b), who briefly discuss autopoiesis as a potential alternative for the "computational metaphor" central to AI. Unfortunately, this discussion, based solely on Maturana and Varela (1987), cannot be recommended on the basis of several key errors:
- West & Travis claim the cell is the theory's "conceptual base" (1991b, p. 76), when in fact they (and other organisms) are the examples from which much of the theory was extrapolated.
- They claim (Ibid., end note 6) "autopoiesis" is not a coined word based on its Greek roots. This is patently false (cf. Maturana & Varela, 1980; Zeleny, 1980a).
- They use the biological context of autopoietic theory's origins to warrant ascribing equivalences to dissipative structures, which are only partially comparable (cf. Jantsch, 1980; 1981).
- They insinuate (Ibid., p. 77) causal connections between levels in a hierarchy of biological systems which Maturana (e.g., 1975) specifically denies.
- They mischaracterize structural coupling as specifically pertaining to interactions among units composing a metacellular system (Ibid.).
- They misrepresent (Ibid., Figure 6) higher-order structural coupling as a coupling of relations between two autopoietic systems. This concept refers to either
- (a) recursive coupling within the cognitive domain of an individual system (cf. Maturana, 1978; 1988) or
- (b) coupling involving a higher order (subsuming) system of which an autopoietic system is a component (cf. Maturana & Varela, 1980; Varela, 1979a, Chapter 7).
- They contrast the combination of autopoiesis/dissipative structures against AI theories (e.g., Minsky's "society of mind"), stating (p. 78) Minsky's model is deterministic, but autopoiesis is not because dissipative structures are not. This completely overlooks the structural determination of autopoietic systems.
West & Travis conclude by claiming (1) advocates (apparently meaning advocates of autopoiesis and the other theories they equate with it) assume intentional design of a simulated mind must fail owing to a further assumption any such intelligence would have to be engineered and (2) such advocates overlook the possibility of creating an intelligence by means other than engineering.
SOFTWARE "INSPIRED" BY AUTOPOIESIS AND RELATED PRINCIPLES:
There have been some attempts to devise IT applications displaying autopoietic characteristics. Zeleny (1980b; Zeleny & Pierre, 1976) outline and report on a self-organizing simulation program as an illustration of autopoiesis (really only self-organization). The computer program does not strictly exhibit autopoiesis itself, but provides the medium in which self-organizing patterns of graphic objects continually emerge and evolve.
Another example is Nissan's (1987) attribute autopoiesis -- an application of autopoietic terminology to describe "organic growth" in a metarepresentation of a knowledge-based systems' (KBS') knowledge base. However, Nissan's allusion to autopoiesis should be considered at best simply figurative and at worst inappropriate. Nissan addresses autopoiesis as a synonym for "organic development" or "evolution", apparently based on the theme of historicality (structural ontogeny). At one point, he refers to autopoiesis as equivalent to decomposition (p. 244); later, he states autopoiesis is a "...process of growth at the attribute level" (p. 246). At no point does he characterize his knowledge base in terms (e.g., organizational closure; self-production) of an autopoietic system.
More recently, autopoiesis has been a subject of interest for researchers in AI vision and artificial life (McMullin, 1992), but its influence to date has mainly been as a source of inspiration. Barry McMullin has sent The Observer a substantial note outlining some of the questions and issues surrounding the application of autopoietic theory to this sort of software application. The next issue of The Observer will present Barry's contribution, thus setting the stage for what I hope will be a vigorous examination of these issues.
Varela, Thompson and Rosch (1991) illustrate their enactive approach to cognitive studies through the robotics work of Brooks (1986; 1987; 1989a; 1989b). Brooks explicitly abandons representations in programming his robots, subdividing control among subsystems specialized according to activities. In doing so, he lets the world serve as its own model, and configures the robot to explore the world (through direct action) rather than traversing some abstract data structure. This approach does not explicitly rely on autopoiesis per se, but it does provide an interesting example of IT geared to structural coupling with its task domain rather than logical coupling to a "knowledge representation". In effect, Brooks is abandoning AI's prior reliance on "alloreferentiality" for Maturana's (1970a) second class of artificial systems -- those describing their interactions in their own terms -- in that their behavior (a "bottom-up" emergent phenomenon) is construable as coherent only in the eyes of an observer (cf. Varela, Thompson & Rosch, 1991, p. 211).
THE THEORY'S "PHENOMENOLOGICAL ASPECTS" USED TO INFORM IT DESIGN
This leads to the third type of application -- using autopoietic theory (especially the phenomenological aspects) to evaluate or inform IT design. Winograd & Flores (1986) easily adopt Maturana and Varela in addition to the others whose orientations they embrace. Subsuming the perceiver within the act of perception links autopoietic theory with Heidegger. Similarly, interaction as structural coupling is congruent with Wittgenstein's vision of endlessly creative language games. Finally, the constructivistic nature of interpretation propounded in autopoietic theory is consonant with Gadamer's hermeneutics. However, Winograd and Flores do not make any detailed comparative analyses of Heidegger, Gadamer and/or Wittgenstein versus autopoietic theory, and treat them as a fortuitous juxtaposition of views. Varela (1979a), on the other hand, draws several parallels between his own views and Gadamer's hermeneutics. An even more detailed exposition of links to phenomenology (especially Merleau-Ponty) is Varela et al. (1991).
For all their advocacy of Maturana, Winograd and Flores' exposition did not lead to direct application of autopoietic principles to IT systems design. They claim The Coordinator (tm) system (Flores & Ludlow, 1981; Flores et al., 1988) was inspired by Maturana's version of interaction as mutual coordinations of action -- a claim also made for the CHAOS project (De Cindio, De Michelis & Simone, 1988). In the case of The Coordinator, the IT artifact was claimed to provide a medium in which actors could couple. The problem was this coupling was subordinated to a specific type of orientation (coordination of prospective actions) within a specific framework (the "conversation for action") based on a decidedly rigid and non-phenomenological theory (Searle's speech acts). Beyond this figurative allusion, no principles from autopoietic theory are directly embodied in the resulting applications.
Others following in the path of Winograd and Flores (e.g., Ehn, 1988; Boedker et al., 1988) have followed only in terms of Wittgenstein's language games, Searle's speech acts, and/or Heidegger's "breakdowns"; they have not made any effort to address or incorporate any elements of autopoietic theory. Robinson & Bannon (1991, p. 230) call The Coordinator "...the one major attempt to embed this framework in systems...", but note the many criticisms leveled at this application. Insofar as The Coordinator does not represent a "faithful" application of autopoietic theory, this criticism does not contradict their suggestion (Ibid.) that further investigation into Maturana and Varela's structural coupling might be useful in illuminating means for addressing different "...ontologies in action..." (Ibid.).
Autopoietic theory has had little impact on IT applications to date. Its invocation as a theoretical paradigm has either been figurative and flawed (Nissan, 1987; West & Travis, 1991b) or limited to recantation of principles without further development (e.g., Winograd & Flores, 1986). The most direct applications to IT artifacts have concerned simulation of self-organizing principles (e.g., Zeleny, 1980b) or non-representational controllers for robots (cf. Brooks). Varela, Thompson & Rosch (1991, 212-213) claim such pragmatic, evolutionary applications could serve as the proving ground for enactive AI -- an example of the second application category in which the IT artifact itself is manifesting an enactive history. This is the manner in which the emerging artificial life community will probably utilize autopoietic theory. I would classify my own work as falling into the third category, since I wish to apply the phenomenological aspects of autopoietic theory to inform the manner in which humans can interact with (and through) IT systems.
Boedker, S., J. Knudsen, M. Kyng, P. Ehn, and K. Madsen, Computer support for cooperative design, in CSCW 88, Portland OR: ACM Press, 377-394.
Brooks, R., Achieving artificial intelligence through building robots, MIT Artificial Intelligence Laboratory AI Memo no. 899, May 1986.
Brooks, R., Intelligence without representation, MIT Artificial Intelligence Laboratory Report, 1987.
Brooks, R. [1989a], A robot that walks: Emergent behaviors from a carefully evolved network, MIT Artificial Intelligence Laboratory AI Memo no. 1091, February 1989.
Brooks, R. [1989b], A robust layered control system for a mobile robot, IEEE Journal of Robotics and Automation, RA-2 (1989), 14-23.
De Cindio, F., G. De Michelis, and C. Simone, Computer-based tools in the language/action perspective, in Speth, R. (ed.), EUTECO '88: Research in Networks and Distributed Applications, Amsterdam: Elsevier, 1988, 243-258.
Ehn, P., Work-Oriented Design of Computer Artifacts, Stockholm: Arbetslivscentrum, 1988.
Flores, F., and J. Ludlow, Doing and speaking in the office, in Fick, G., and R. Sprague (eds.), Decision Support Systems, Issues and Challenges, New York: Pergamon Press, 1981, 95-118.
Flores, F., M. Graves, B. Hartfield, and T. Winograd, Computer systems and the design of organizational interaction, ACM Transactions on Office Information Systems, 6, no. 2 (April 1988), 153-172.
Jantsch, E., The unifying paradigm behind autopoiesis, dissipative structures, hyper- and ultracycles, in Zeleny (1980a), 81-88.
Jantsch, E., Autopoiesis: A central aspect of dissipative self-organization, in Zeleny (1981), 65-88.
Maturana, H. [1970a], Neurophysiology of cognition, in Garvin, P. (ed.) Cognition: A Multiple View, New York: Spartan Books, 1970, 3-24.
Maturana, H., The organization of the living: A theory of the living organization, International Journal of Man-Machine Studies, 7 (1975), 313-332.
Maturana, H., Biology of language: The epistemology of reality, in Miller, G., and E. Lenneberg (eds.), Psychology and Biology of Language and Thought: Essays in Honor of Eric Lenneberg, New York: Academic Press, 1978, 27-64.
Maturana, H., and F. Varela, Autopoiesis and Cognition: The Realization of the Living, Dordrecht: D. Reidel, 1980.
Maturana, H., and F. Varela, The Tree of Knowledge: The Biological Roots of Human Understanding, Boston: Shambhala, 1987.
Maturana, H., Reality: The search for objectivity or the quest for a compelling argument, Irish Journal of Psychology, 9 (1988), no. 1, 25-82.
McMullin, B. (ed.), Autopoiesis and Perception (Proceedings of the workshop with ESPRIT BRA 3352), Dublin: Dublin City University, 1992.
Nissan, E., Knowledge acquisition and metarepresentation: attribute autopoiesis, in Ras, Z., and M. Zemankova (eds.), Methodologies for Intelligent Systems, New York: Elsevier, 1987, 240-247.
Robinson, M., and L. Bannon, Questioning representations, in ECSCW 91: Proceedings of the 2nd European Conference on Computer Supported Cooperative Work, Dordrecht: Kluwer,1991, pp. 219-233.
Varela, F. [1979a], Principles of Biological Autonomy, New York: Elsevier (North-Holland), 1979.
Varela, F., E. Thompson, and E. Rosch, The Embodied Mind : Cognitive Science and Human Experience, Cambridge, MA: MIT Press, 1991.
West, D., and L. Travis [1991a], The computational metaphor and artificial intelligence: A reflective examination of a theoretical falsework, AI Magazine, Spring 1991, 64-79.
West, D., and L. Travis [1991b], From society to landscape: Alternative metaphors for artificial intelligence, AI Magazine, Summer 1991, 69-83.
Winograd, T., and F. Flores, Understanding Computers and Cognition, Norwood NJ: Ablex, 1986.
Zeleny, M. (ed.)[1980a], Autopoiesis, Dissipative Structures, and Spontaneous Social Orders, AAAS Selected Symposium 55, Boulder CO: Westview Press, 1980.
Zeleny, M. [1980b], Autopoiesis: A Paradigm Lost?, in Zeleny (1980a), 3-44.
Zeleny, M. (ed.), Autopoiesis: A Theory of Living Organization, New York: North Holland, 1981.
Zeleny, M., and N. Pierre, Simulation of self-renewing systems, in Jantsch, E., and C. Waddington (eds.), Evolution and Consciousness, Reading MA: Addison-Wesley, 1976, 150-165.