A semiotical reflection on biology,
living signs and artificial life.

by Claus Emmeche

ABSTRACT:
It is argued, that theory of signs, especially in the tradition of the great philosopher Charles Sanders Peirce (1839-1914) can inspire the study of central problems in the philosophy of biology. Three such problems are considered: 1. The nature of biology as a science, where a semiotically informed pluralistic approach to the theory of science is introduced. 2. The peculiarity of the general object of biology, where a realistic interpretation of sign- and information-concepts is required to see sign-processes as immanent in nature. 3. The possibility of an artificial construction of life, hereby discussed as a conceptual problem in the present form of the artificial life project and its implied definition of life.

KEY WORDS:
philosophy of science, biology, semiotics, sign, information, Peirce, teleology, artificial life, semiotic realism.

The final version of this paper was published in: Biology and Philosophy 6: 325-340, 1991.

What's in a name? that which we call a rose,
by any other word would smell as sweet. [1]

Introduction

In our century, many theories of sign and language in disciplines such as philosophy of language, linguistics and the French semiotical tradition have been afflicted by a kind of narrowness in theoretical perspective, that tends to exclude the possibility of natural signs, which are not interpreted by human beings. This is due to a very basic concept, namely the concept of the sign, the extension of which is established within our social and cultural domain. As a consequence, the signs of living nature -- the genetic code, the communication of cells in the body, or the signals of animals -- are understood in a nominalistic way, merely as concepts in the minds of scientists (which they also are, of course). Talking about a relation between the natural world and the world of scientific language can, according to one of today's most prominent semioticians in Europe,[2] only be understood as a question of "intersemiosity", i.e., as a question concerning the relation between different semiotics or sign systems, all situated within the social sphere. Certainly, such a version of semiotics has a "problem of the external world", excluding the referent of the sign from its horizon -- Nature seems to be reduced to culture-dwellt sign systems.

It is among other things in this ontological respect that the semiotics of Charles S. Peirce can bring a radically different understanding of the nature of the sign.[3] A sign, according to Peirce, is "something which stands to somebody for something in some respect or capacity" (Peirce, 1955:99; CP. 2.228), i.e., the sign is constituted by a triadic relation between a representamen (the primary sign, i.e., the "physical" carrier of the sign), an object, and an interpretant. The relational character of the sign and its definition and the irreducibility of the object-pole of the sign is most evident in this formulation:

"A sign, or Representamen, is a First which stands in such a genuine triadic relation to a Second, called its Object, as to be capable of determining a Third, called its Interpretant, to assume the same triadic relation to its Object in which it stands itself to the same Object." (Peirce, 1955:99-100).

Peirce's concept of sign demands a treatment that is beyond the scope of this paper. Let us clarify with a few examples. Printing ink in a characteristic form on a piece of paper may be a primary sign (i.e., the representamen) of a word (i.e., an object in the system of language) and in a specific context this sign determines a reader (an interpreter) to think of a concept or imagine a detail of an event (i.e., the intrepretant), and in doing so establishing a complex relation between sign and signification similar to an aspect of some conventional meaning of the sign in that language system. Notice the distinction between the organism as an interpreter, and the effect of the sign upon that organism, i.e., the interpretant. The shrieking sound of a vervet monkey can be a representamen of a snake determining another monkey to hide in a tree, the act of hiding being the dynamic interpretant of the sign. A sequence of nucleic acid nucleotides sometimes functions as a genetic sign for a specific array of amino acids, resulting in an "object", i.e., a protein, produced in a specific context. In another situation, this very protein may act as a primary sign or as an immediate interpretant. Moreover a sequence of nucleic acids can be the object of attack by enzymes. This is not to suggest that the triadic concept of sign-action -- or the whole semiotics of Peirce -- in any way exhausts the complexities of molecular biology, ethology or linguistics, or that one can dispense with more specific theories in order to comprehend concrete phenomena of nature; the point is metatheoretical: there is nothing in the peircean semiotics which restricts sign-actions to culture or human practice. The concept of the sign in Peirce is of very general nature (and related to his metaphysical categories of Firstness, Secondness and Thirdness, cf. Esposito, 1980), its semantic extension is broad, and nonetheless its definition and theoretical-semantic intension is highly specific.

This comprehensive understanding of signs should make it possible to make a renewed assessment of the conceptual foundations of present research into the phenomena of life and thought. If one is to do research in biology, psychology or cognitive science, and wants to grasp the aspect of nature or thought one attempts to compute or simulate, and certainly, if one thinks that such understanding should include a world-historical or evolutionary perspective, one promptly comes up against basic problems with respect to the conceptual quality of sign, information, thought, and similar "heavy" categories.

Incontrovertible as it seems, Peirce's ideas should be used to think with; thus slipping through the Scylla & Charybdis of a great many Peirce-scholarships: Scylla is the blasé state of belief, that Peirce is only actual as an interesting forerunner of present ideas in science and philosophy -- Peirce anticipated Popper's falsificationism, the evolutionism in astrophysical cosmology, he extended and improved the philosophy of signs thus founding a modern semiotic tradition, he was the father of pragmati(ci)sm, he conceived new ideas in logic and mathematics, he forestalled more recent interpretations of the concept of probability, etc. (and these statements seem to be true) -- these ideas, however, are worked out in detail, reinterpreted and expounded by more recent authors, and therefore Peirce has nothing new to tell. Charybdis is the servile state of belief, that it is possible, in a thorough-going Peirce-philology, to reconstruct a profound and coherent system of philosophy, logic, semiotics and theory of science, so that the grail of insights in Peirce's thought are just lying there, waiting to be read out of the piles of fragmentary information. As an academic petty industry this neo-peirceism may take new goods to the market, as an isolated project, however, it degenerates into exegesis and trifles. Both readings conceal the fact that Peirce has still something more to tell us, an extra insight and an extended perspective, yet demanding a constructive capitalising on his ideas. They should be brought into a clinch with vital currents in science, philosophy and culture of today. As an example, his general ideas actualise new interpretations of the chronically foundational problems of molecular biology and evolutionary theory. Thus the concept of biological information, often thought of as a mere metaphor alien to theoretical biology (Stuart, 1985), or as a member of a class of mere intentional ascriptions with no ontological significance (Nagel, 1977; Rosenberg, 1985), can now be interpreted within a richer ontology allowing for a natural evolution of sign-relationships (Hoffmeyer & Emmeche, 1990; Esposito, 1980; Anderson et al., 1984). However, his specific thoughts on the molecular nature of the "slimy" substances of life is of course restricted by the state of art in biology of that time.

I shall sketch three kinds of problems, where the Peircean insights seem to be striking as well as clarifying. The first relates to the theory of science, the second to biology in general, and the third concerns the possibility of artificial life, artificial intelligence, and by implication, the nature of cognition. As to the first, I'll point to the trivial fact that any science (whether natural science or cultural sciences as anthropology or linguistics) is concerned with the management of signs in the form of data and concepts, and can thus be studied from the perspective of semiotics. On this general level, semiotics is one among other perspectives on science such as the sociological, the historical or the logical perspective. As to the second point I'll suggest that Peirce's naturalistic concept of the sign helps to clarify what determines the peculiarity of the biological domain compared with research into the physical or other non-biological systems. To the extent that biology is the study of the living organism it is, from the semiotical point of view, the study of possible interpreters of signs. Man, however, is the only organism that have managed to make artificial sign processing systems (e.g., a word processor). My third point is that the question about the reality status of artificially produced behaviour in computational media can be approached from the perspective of semiotics.

Science as a semiotic process.

As is the case with philosophy, the scientific project is born and bred by a love for knowledge. Yet this project is just so curious, that it will not stop at mere reflection upon other peoples ideas. It will grasp and understand the world, and such an understanding can be used for many purposes. One can manipulate, sell and buy, make exchanges, win knowledge about the script of Nature and perhaps even insight into the limits to possible knowledge and the limitations of the way scientific knowledge is used. Science is a variety of sciences, that cannot be put into one generalized form, and in any science a multitude of interests takes part, which, just as feelings, can be very mixed. In the meta-disciplines such as the philosophy, sociology and history of science, the scientific activity is made an object of empirical or analytic studies, which have revealed many structures and dynamics of science as a collective enterprise. On the individual level, one may speculate if some kind of compassion or feeling (e.g., a "feeling for the organism", see Keller 1983), not only the feeling of fascination or curiosity, is among the driving forces or motive powers of science. Peirce expressed the thought, which may seem hyperbolic to many, that love is the prime mover of evolution,[4] and there may be some truth in it, even for the evolution of the sciences.

Sometimes a branch of science can contribute to an understanding of what science as a general enterprise is, and even the multiplicity of sciences is a good guarantee against ever reaching a complete and deathly agreement as to that question. The social and historical sciences can analyse some factors steering the development of the exact sciences, and ecology may stimulate an understanding of the limitations of the science of economy and its analysis of man's productive relationship to nature, but as a rule, the particular fields of knowledge preserve their relative autonomy. As with other cultural activities, self-reflection in science and in scholarship are determined by contradictory ideas that stick to a given period about society, man and nature. Thus the theory of science is multifaceted (often much more than some professional philosophers have ever dreamt), and both the theory of signs following Peirce and pragmaticism are compatible with that theory of science, that I, in good semiotic manner, have spliced together from others, and shall sketch here.

As any rat knows, science is a rather concocted type of social, rational, systematic sign-activity, going on in particular institutions organised for the purpose of knowledge-production. Therefore a science can be analysed with respect to different dimensions, among which primarily a) the theoretical, b) the historical, and c) the social dimensions are pertinent -- especially when made an object for a) philosophy of science, b) history of science and c) sociology of science. Let us consider biology as a sole example. When asking what is this thing called biology, we should try to answer on all levels -- a very complicated task, indeed.

a). With respect to philosophy of science, one can propound a generalised sign-relation for the theoretical dimension of biology (and, per analogy, any other science): Biological theory can thus be understood as a sign-supported mediation between (1) the very inner logical structure of a complex of ideas that are claimed to be a more or less general theory of living systems (i.e., biological theory as general representamen -- the field of conceptual structure), (2) the empirical field of biological systems, as constituted by the very scientific activity itself (biological theory as representing a general object), and (3) the producers and interpreters of the theory as the collective of actual agents of institutionalised bioscience (biological theory as structure of signification for a general interpretant - the sociosemiotic field). Philosophers of science often take their only stand on the logical structure of theories, hoping to discern the theoretical activity exclusively by this approach. But the theoretical dimension of biology is, via the sociosemiotic field inter alia, connected with the other dimensions of science. Biological theory can thus be understood semiotically as a triadic relation between (1) a particular conceptual structure (for example the logical structure of the theory of natural selection by variants with different hereditary determined fitness), (2) the field of observations, measurements, perceptions and concepts about something real or imagined, often roughly epitomised as "empiric" or "data" or naively conceived as "brute facts" about nature (for example what Darwin thought to see on Galapagos, or a reptile skeleton, or Brontosaurus "as it really was"), and (3) the meaningful broader interpretation of the empirical object that biologists in a given psychological, social and historical frame construct in the light of the theory as a symbol system representing the dynamics of the object (for example the biologist S.J.Gould's critical opinion of the neodarwinian explanation of the origins of species). This triad applies to a relatively high level of abstraction (general biological theory), but also in the singular scientific activity Nature's particular signs ("data" as for example what Peirce called indexical sinsigns) are mediated through those theories and that experimental praxis unfolded by the collective of concrete human interpreters. In the humanities the sign-relations with respect to the empirical field has a very different character, because the "object" itself is a cultural sign system of symbolic and often linguistic mediated processes; however, the basic triadic structure for the theoretical dimension of these Geisteswissenschaften is the same. Does this add anything to our understanding of biology, one may ask? On the general level, it represent the theoretical dimension of biology as a science among other sciences. In order to add specifically to our understanding of the structure of say, the clonal selection theory in the discipline of immunology, one has to apply a more specific analysis.

b). With respect to the history of science, the question of identifying a scientific biology in the modern sense can be posed within different theoretical frames. An interesting example comes from Foucault's comprehensive archeology of science. He argued that the phenomenon of "life", or its essence as a scientific problem, could not have appeared until relatively late in history, after a deep epistemological rupture with the complex of mere classificatory problems in classical natural history.

"To establish the great, unflawed table of the species, genera, and classes, natural history had to employ, criticise and finally reconstitute at new expense a language whose condition of possibility resided precisely in that continuum [of species] (...) Up to the end of the eighteenth century in fact, life does not exist: only living beings. These beings form one class, or rather several classes, in the series of all the things in the world; and if it is possible to speak of life it is only as of one character (...) in the universal distribution of beings." (Foucault, 1970:160).

An epistemological break implies a mental restructuralisation in the way signs of nature are perceived and interpreted by human beings in the actual cognitive process of their translation into the symbol systems of theories. Without regard to the validity or virtues of Foucault's perspective, the very question about the historical dimension of science is obviously a semiotic one, too.

c). With respect to the sociology of science, Peirce seemed to have been foreign to positions like some of the recent relativistic accounts of scientific knowledge (e.g., Barry Barnes and David Bloor of the so-called Edinburgh school). The difference may be smaller than originally thought, and one can probably unite Peirce's pragmaticism including his ontological conception of signs (that I will call semiotic realism) with a theory of science, that cannot escape being informed by the sociology of knowledge. Science is a social construction, as well as it is under an obligation to seek and tell something true about the world. The mathematical metaphor expressing Peirce's view -- that in the long run the scientific community through hypotheses and experiments will converge to the truth but never reach it, as a curve converging to its asymptote -- may seem naive in the light of the whole discussion following Kuhn, where "truth" was made paradigm-relative. But one must remember, that Peirce's fallibilistic conception of science as a trial and error process does not implicate any automatics in the progress of knowledge, or any form of positivism with respect to the apprehension of the nature of theories. A scientific hypothesis is a general assertion, which always remains underdetermined by a finite number of observations. In the short run there is no particular reason to believe that a given hypothesis, chosen e.g., because of its falsifiability, is true. So one can legitimately investigate a range of other factors in the whole context of discovery and understanding related to a specific discipline or a given paradigm, which determine the construction of hypothesis. It is obvious, that also the "technical" interest in knowledge is a co-determining constituent in the general development of the natural sciences, as for instance Habermas has argued; this constituent, however, cannot be reduced to a mere "interest in technical application" of theory, and is not exclusively determining the natural sciences.

Likewise, one cannot reduce Peirce's theory of science to some instrumentalistic or "vulgar-pragmatistic" conception of truth as practical usefulness. The pragmaticism of Peirce emphasises an intimate connection between knowledge and techne in the meaning of action or praxis, but this is in relation to the human action in e.g., the experimental activity. Peirce's anti-instrumentalism and his pragmaticism are compatible and work well with the realism as with the idea of convergence. These ideas can be applied, recursively so to say, to the theory of science itself, to generate a new idea about the convergence, in theory of science, towards many truths about the multiple sciences and the plenitude of scientific communities.

Biology consists of many disciplines, and can be conceived of as a relatively autonomous network of areas, that in all three dimensions has mutual relations as well as connections to other disciplines, both in the natural, social, and linguistic sciences and in other humanities.

With respect to the theoretical dimension of biology, it is to a considerable extent the semiotic character of concepts about biological information and coding, that determines the peculiarity of the biological domain. In this domain (approximately as in the humanities) the objects of knowledge -- the various phenomena of life -- reveal an immanent semiotic activity; Nature is full of living signs!

Signs in living nature

Biology is a natural science, indeed - a "physical science", as Peirce called it. His attempt to classify the particular sciences did not end up with some nice final result, but there is no doubt, that Peirce also regarded biology as a partly "psychical science" (as the other "psychical" human and social sciences). This is, among other things, due to the fact that

"Mind has its universal mode of action, namely, by final causation. The microscopist looks to see whether the motions of a little creature show any purpose. If so, there is mind there. Passing from the little to the large, natural selection is the theory of how forms come to be adaptive, that is, to be governed by a quasi purpose. It suggests a machinery of efficiency to bring about the end - a machinery inadequate perhaps - yet which must contribute some help toward the result. But being governed by a purpose or other final cause is the very essence of the psychical phenomenon, in general." (CP. I.269 [1902]).

Although biological disciplines are mainly dependent on chemical and physical concepts, as opposed to the humanities, the phenomena studied in biology are not a-psychical, or non-semiotic. Peirce actually speculated on the possibility, that certain physical processes, as the non-conservative and irreversible ones, were of a psychical nature. The dividing line between the human (so-called "behavioral") and the natural sciences can thus be partly dissolved; there are no such substantial differences, as have traditionally been claimed. The crux of the matter is differences in approach of the research and whether it focuses on effective causation, or final or teleological causation as a dominating theme. Neither exclude the application of semiotic concepts in the physical sciences or non-semiotic concepts in the psychical sciences.[5]

It is, firstly, possible to reconstruct a sketch of a natural history of signs, that firmly establishes the human language and other sign-actions in a natural historical perspective (by emphasising the continuity in the evolution of different sign systems from the genetic to the linguistic through the world history), and at the same time, avoids to reduce the complexity and specificity of the language-mediated cultural signs, that has to be analysed in the light of specific linguistic and other theories. A natural history of signs is compatible with, and an important element of, a general evolutionary theory. Secondly, both nature's and culture's signs can with advantage be conceived of realistically, i.e., as real existing relations, acting in the world, and not just as the delicate products of mind -- this semiotic realism is in absolutely accordance with Peirce's own evolutionary metaphysics. (For a more detailed account of the semiotics of nature and biological information, see Hoffmeyer and Emmeche, 1990).

Even if you at first comprehend semiotics as a purely non-ontological affair, only concerned with philosophy of language (Clarke 1987) or with epistemology, one is led towards realism by a kind of self-reflective movement: Semiotics as epistemology is the insight that perception, cognition, recognition (Erkenntnis), the relation between experience and its objects, is always sign-mediated, and that we never have a total immediate access to the rose ... only the name or the signs are left. (Some of these are, indeed, more concrete, sensuous and even sensual, than others, and in this sense more "immediate", but they are nevertheless signs). This insight leads us, via the discovery of analogous sign- and relation-structures in our ancestry, the living nature (e.g., the sign system of the genetic code) to an extension of the very concept of experience within an evolutionary frame. Even though we are forever thrown upon signs, we share this condition with all living beings. Subjectivity as a mode of being is not just a property of the human subject, but exists "dissolved" in smaller dose in all living cells, according to Peirce. Even a cell can thus be seen as functioning as a non-personal interpretant of the signs embodied in its genetic material, regulating its metabolism. Without explicit or implicit reference to the concepts of codes, information, or signs, one could hardly understand these relations. Even at the metabolic level they would in fact appear biologically meaningless.

Peirce comprehended these things when he told that a lump of protoplasm feels (CP. 6.133, 6.264 [1892]), yet how better by far a position of knowledge we are holding today, when we aspire to understand the real web of sign processes in evolution and ontogenesis! What goes on in a cell is not just "mere chemistry" - if one by that means simple effective causation. The "script" of the genetic code, which Roman Jakobson (1973:50ff) compared in detail with the formal structure of human language, is an instance of the general phenomenon, that one in Nature finds a complementarity of dyadic (physical, like action and reaction) and triadic (semiotic) processes. It might be termed kinesis and semiosis : certainly a living organism is subjected to the kinesis of the physical laws of motion, but it will only be more fully described and understood in the perspective of semiosis, as a sensuous being, depending upon signs from its environment and its inner processing of information. Kinesis and semiosis can be seen as instances of Peirce's metaphysical categories of Secondness and Thirdness.

One can point at other codes in the biological domain in addition to the genetic one. There is a metabolic code, regulating the "physiology" of the organism; there is an immune code in the sense of the sign structure of the cellular defence system of the body; and there is -- probably on several levels -- a neural code, carried by the nerve network of the body, and which in the brain is the substrate of higher mental processes. With respect to the human being, we are confronted here with a limit on the type of codes that biology, in its present state of art and outlook, can tell anything about: accordingly, psychology, linguistics, anthropology and other cognitive sciences must take over, to investigate the linguistic, social, and more "spiritual" codes. To describe the biological codes thoroughly is still an enormous scientific challenge, because the "epigenetic code" for example (steering the embryo's processes of growth, differentiation and morphogenesis) is still very poorly understood. Biological phenomena on various levels of organization are thus semiotic to the extent that they include processes, which today are described (within different theoretical traditions and vocabularies) as involving coding, information processing, communication, sign exchanging and interpretation.

Realising the relevance of general semiotics to the analysis of sign systems in living nature, it is also appropriate to recognise a certain inner connection between biology and semiotics.

It is a fundamental task of biology to deal with the richness of purposeful structures and purposive processes related to every organism. An organism is hardly a biological object, if it does not exhibit that teleology in the form of end-directedness with respect to survival and reproduction, maintenance of an inner dynamic process structure (homeostasis), and a selectively response to perturbations from the environment. These phenomena belong so to speak to the quintessence of the concept of organism. As hinted at above, an organism can be seen as a complex sign in the sense of being a nodal point for semiotic processes (structure and genesis of various sign systems). The organism is a spatio-temporal relational node in a web of evolutionary and ecological processes, which is not merely energetic, but involve exchange of signs. From a semiotic point of view, these sign systems are central for an account of the organism's telos, i.e., in understanding the self-organizing processes within the organism and how it is connected to the environment in a functional and biologically meaningful way. But as an organism is a complex sign (i.e., an organism is mediating a complex system of semiosis), so is the sign itself an organism (the sign has in a specific sense an organismic aspect).[6] This can be best examplified by saying that the sign and generally semiosis, sign processes, show a form of teleology. The sign is an element of an indivisible triad: sign (representamen), object and interpretant. As T. L. Short has formulated it:

"Nothing is an object which is not [at least potentially, C.E.] signifiable; nothing is a sign that is not interpretable as signifying some object; and nothing is an interpretant that does not interpret something as signifying an object. An object need not be signified and a sign need not be interpreted; but they are what they are in virtue of potential signs and potential interpretants, respectively. These potentialities are more than mere possibilities. Something is a sign in virtue of a ground -- or relation of sign to object -- that would justify a particular interpretation of it. Significance, then, is grounded interpretability. But an interpretant can only be grounded or justified in relation to some goal of interpretation." (Short 1982: 285)

Thus the interpretant can only be grounded in relation to some goal of interpretation - that I will call the perspective or horizon of the interpretation. Related to any ground and any sign is an actual or possible goal of interpretation; thus signification is in this sense grounded interpretation. Therefore, the process of sign interpretation (semiosis) is basically teleological. It does not in general or necessarily demand any conscious end or goal, but an end, which concerns an actual (eventually non-personal) interpretation of the sign as part of new semiotic processes. One can argue, that the sign manifests intentionality (Short 1981). Not as in Brentano, for whom the phenomenon is bound and restricted to human consciousness. The intentionality of signs must as suggested be seen as based on the very semiotic activity: It is implied by the quality of semiosis as Thirdness. Something becomes a sign (and therefore a Third) by standing for something else to someone or something. An interpretant (which may be an idea, a thought, some effect upon an animal, a cell, a non-personal being) has, at least to some degree, intentionality because it interprets signs, and a sign is intentional to the extent it is related to one of its potential or actual interpretants.

Does this mean that a single bacterial cell shows intentionality? Feeling a little puzzled by such a consequence one must emphasise that the phenomenon of intentionality in conscious human beings is by no means the same as the sort of generalized intentionality of signs inherent in semiotical processes on all scales of complexity including rather simple organisms. As indicated above, one must on the one hand recognise the continuity of end-directed and interpretative behaviour within a natural history of signs (the "teleological continuum" as Burks (1988) puts it) and on the other hand distinguish the different levels of organization and make use of specific theories within each level. And this is a matter of empirical research.

Artificial life and artificial intelligence.

Peirce's ideas may contribute not only to the development of theory of science and in the analysis of biosemiotic phenomena, but also to the areas of cognitive science, artificial intelligence (AI), and the latest hot-spot within computer-dependent research disciplines "Artificial Life", which could all benefit from Peirce's insights. Research in artificial life, "AL" (official designation of 1987) originates in cybernetics, automata theory and general systems research, and is, of cause indebted to classical and connectionist research in artificial intelligence. As within AI one attempts via computers to create synthetical systems, that display some of the same characteristics as non-artificial living systems. Before trying to do so, one should resurrect Peirce to ascertain the critique he would put forward to the current version of the project. He would probably be found in New Mexico at Los Alamos' Centre for Nonlinear Studies, in fierce discussion with Christopher G. Langton, one of the leading figures within the AL world.

Langton claims, that one can separate "the logical form" of an organism from its material base, and that its "aliveness" (its capacity to live and reproduce itself) will be a property of the form, not the matter (see Langton, 1989, p.11). According to the argument, this is exactly why you can synthesise life -- not just a model, but genuine life -- in a computer: first, out of the biological and chemical knowledge, you abstract those logical principles that apply to the whole of life. Next, the principles must be formalised, such that they can be implemented by a computer program. In this context one utilises the so-called bottom-up method, where relatively few and simple rules (concerning for instance how living cells react to each other) suffice. Once instantiated in the machine and running the program, interesting dynamical patterns and forms emerge out of the interactions between the multiple (simulated) cells. It is this "emergent" or collective behavior, rather than the individual artificial cell or computing unit, that Langton claims to be synthetic life.

At issue is not the building of a mechanical organism or robot, able to react purposefully to particular problems on the basis of general steering principles (the so-called "top-down" method in classical research in AI). Langton's point -- which we might call the "strong version" of artificial life -- is, that life can be studied abstractly with no attention to the material levels, since life is essentially an assembly of functions, which can be realised in the medium of the computer quite as well as in the biochemical medium of known life (Peirce's slimy protoplasm; DNA, proteins etc., in modern biology). On the level of individual cells there is a difference in kind: a natural cell is not the same as the one simulated in the computer (in fact, a cellular automaton is a mathematical object). These exist only as information structures in the computer. But -- and this is according to Langton the decisive point -- on the level of the collective behaviour of many parts, the interplay between natural cells in an organism and the interaction of "cells" in a computer are two instances of the same phenomenon: pattern-creating interaction (between cells, but also the relations between whole organisms in an ecosystem could be modelled in a similar way). Life is form and can be synthesised as such.

Peirce would have two comments to Langton: First, Langton has a point, since it is in fact possible to extend the perspective of theoretical biology, in order to let its object be not only life as it is, but also life as it could be. Life could have evolved in a different way on earth because of chance events, as it could have (and probably is) evolved in other ways on other planets. Moreover, we still could, in principle, be able to discover those quasi-logical regularities and lawfulness, that hold for all life. For this enterprise, semiotics may be a useful tool. (Peirce's view may be reinforced from the fact that when the mathematician John von Neumann in the forties analysed the general logic of self-reproduction, he was very close to Peirce's position).

Peirce's second comment to Langton is critical, though. Strong AL conceals a false dualism, viz. the claim that one can describe life functionally aloof from its material level. Form and matter do not stand to each other as two foreign modes of being, as Langton seems to presuppose. If the research program to study "life as it could be" is not connected to any study of possible material environments for life and its possible self-organisational material support, in other words the study of how forms of life and semiosis can evolve from the material and less complex, it is doomed to degenerate, perhaps into manufacture of computer games, which has little or nothing to do with biology. It is especially important to understand the constraints that harness the evolution of Nature's biological, chemical, and physical habits. This is present in what Peirce called synechism, his idea of continuity. Life does by no means exist as pure Thirdness or mere "universes of information", that have taken up residence in an accidental material; Thirdness keeps and encompasses Firstness and Secondness -- semiosis presupposes kinesis -- and less metaphysically one could say that evolution exposes a mutual interplay between analog and digital codes.

The patterns and complex behaviours that emerge within the parallel computational medium of the AL-models may well function as interesting simulations of real biological systems (which is the claim of a moderate or "weak" version of AL), but such patterns are "realized" in a non-biotic device that has no intrinsic teleology. A computer, as any other tool, may function very purposively (a computer running a program for the development of a self-reproducing data structure is a very useful instrument for implementing formal theories of life), yet it does not have a life of its own. A robot may have such an autonomy only to the extent that we abstract from its artificial genesis, its human history so to say. To understand special aspects of the construction of an organism, we may temporarily abstract from the ecology and evolution of the species in question, but we know, that we only get a part of the history of its natural generation. Nature's own "realization" of organismic form proceeds within a historical context, where form and substance are concepts of the same reality, relative to the observed level of organization. Form in abstraction explain neither the meaning of a word, the intention of a cognitive behaviour, or the end-directedness of a simple organism creating a niche for survival and reproduction, converting simple resources into organized structure.

These remarks does not exclude that AL-research can provoke or evoke a better understanding of general biological principles, but it is quite important to understand the different levels of life, including the material, in their internal connections. The point that life is not medium-independent, as a "multiple realization argument" for a computational account of life would suggest, does not mean that life could not in principle be realized in other material substrates (on other planets for example). The questions about what kind of molecules in what kind of prebiotic environments can allow for the emergence of other forms of life must be a part of a more general biology.

The project of AL can thus be criticised on biological, semiotical, as well as metaphysical reasons. Something equivalent holds for the attempts to build "artificial" intelligence (AI) or understand human cognition as rule conducted manipulation of symbols (computation) in the head. This project too suffers from a dualistic conception of the human condition.

Within AI, one of the really big mysteries is how a machine (let us say a hypothetical intelligent general problem solver) that can only process its symbols "syntactically" -- by algorithmic procedures with no concern for our interpretation of their meaning -- could ever come to just a slightest approximate meaningful "understanding" of the content of those problems, we might pose to the machine. How do the physical machine symbols achieve any meaning? If the machines do not "mean" anything with their strings of symbols, and never will, if it is only the users, that ascribe meaning to their inputs and outputs, how then can they ever be intelligent? Intelligence -- which can constitute an advantage if one is to translate Shakespeare into Danish, write a meeting report or cover the Eastern European news -- must include the capability to deal with complex contexts in a meaningful way. One has termed it "the mystery of the original meaning" (cf. Haugeland, 1985). If the meaning of computers is only lend (ascribed by human beings), they merely mean what we interpret them to mean, and they will never be able to understand anything by themselves. And that is unbearable to an AI-researcher, to whom there is principally no difference between a man and a machine, because the very human being is a symbol-manipulating machine.

Having Peirce in mind, I will remark two things. First, the concept of symbol within AI seems too narrow, and too dyadic: symbols are conceived of as mere physical marks, whose ability to represent thus exclusively results from the human subject. One should notice, however, that when the symbols enter into a complex whole system, and are interpreted by other symbolic structures in machines or organisms -- which are not just purely formal structures, but each possessing their own specific materiality (their "dynamic mode", as Pattee (1977) would say), and which eventually can learn to represent their "self" and their relation to the outer world by symbolic or non-symbolic sign relations -- one cannot a priori exclude, that the triads being mediated in this way are mindful and meaningful in the fullest sense of the word. The problem with the original meaning is arch-semiotical, and it would be fatal to believe that it can be solved exclusively by way of formalisation within the particular domain of linguistic signs. The meaning for an organism or a human being is something that extends far beyond a linguistic description.

Secondly, this means, that cognitive science must leave the logicism and the tight management-rationalism, that have afflicted at least the classical AI research. One must comprehend thinking as something which encompasses more than linguistic mediated signs, namely those signs, that stand in both inner and outer relation to the richness of iconic, indexical and other semiotic processes in nature. When Peirce said, in the famous aphorism, that "man is a sign" (cf. CP. 5.310-317), it would be a logocentric reduction of human nature, if Peirce had not precisely those exceedingly broad and partly phenomenologically based concepts of sign and thought, he in fact had. What Peirce likely meant, was, that compared with the social level (eventually with the society of seekers after truth), the individual human being is to be likened with a sign in its relation to a composite living system of signs, like language or nature itself. Man is, just as little as the sign, something in itself -- therefore a critique of AI, that takes a pure subjectivistic humanism as its starting point, will hardly be a sufficient critique.

It is true, we should "keep cool" and avoid reducing logic to psychology or biology (or vice versa). But we are left with a cognitive science, philosophically split between subjective and objective concepts of information, which too well mirrors the pervading dualism of our culture. In order to transcend dualism, it is appropriate to accept a Peircean perspective on sign and information as something, which is not purely mental or purely material. Peirce did not try to reduce logic to something else. Contrarily, he threw out a fascinating sketch of a broader logic, that may help us to understand the logical structure, not only in arguments of philosophers, but as the logos in the psychical, biological and physical reality surrounding us as a part of the cosmos.

Acknowledgements

I would like to thank Peder Voetmann Christiansen and Jesper Hoffmeyer for introducing me to the philosophy of science and semiotics. Many thanks also to Jonathan Trent and Siro Trevisanato their valuable comments to an earlier version of the paper. I am especially grateful for the help of Thomas L. Short, who thoroughly reviewed the paper with big patience and insight, and whose articles have been an indispensable inspiration.

Notes

Abbreviations as CP. 3.123 refer to volume number and paragraph in Peirce (1931-1958).

[1] Romeo and Juliet, act II, scene II, William Shakespeare.

[2] Algirdas Julien Greimas, see Greimas & Courtés (1979). This tradition, which goes back to F. de Saussure, has been incisively characterised as "end[ing] up making sign systems self-contained, largely conventional structures, referring to none but immanent objects. Extra-linguistic or extra-semiotic reality remains at best ineffable" (T.L.Short, pers. comm.) In the last decade, semioticians as Jean Petitot has revised French semiotics in the perspective of a more realistic conception of universals.

[3] I must instantly admit my skepticism towards a philosophy of science in the form of letting philosophical "first principles" or authorities define methodological prescripts of theoretical and experimental science, or in any way trying to tell the scientists what to do or think. I will not "sell" Peirce as a key to science, but rather point at the possible relevance of his theory of signs to some metatheoretical problems, such as the nature of science, the general concepts in biology, the problem of teleology, etc.

[4] Love in the sense of the Greek Agape, different from the Eros of desire, the attraction to another person, is a collective devotion, concern, and spirit of community. "Evolution" should here be taken in its widest, metaphysical sense. As a biologist, I would not nominate any candidate, whether love, God, or Natural Selection, as "prime movers" of evolution.

[5] The question about the relation between different types of causation is complicated. One may assert that Peirce's final causation can be reduced to effective causation for cybernetic and computational reasons (Burks 1988). Like many reduction claims, this is an argument for "in principal" -- not in practice -- reduction (and it leaves computation or "logical mechanism" that should account for final causes rather undefined; computation as such does not have causal powers). Or one may more pragmatically assert that the functionalist viewpoint of evolutionary biology dealing with formal and final causes and the structuralist viewpoint of developmental biology dealing with material and efficient causes simply complement each other (Rieppel 1990).

[6] Note that I am not intending to reduce the concept of sign to the concept of an organism, or vice versa, only to point at some inner connections between our conceptual understanding of semiotic and biotic phenomena.

References.

Anderson, M., Deely, J., Krampen, M., Ransdell, J. Sebeok, T.A., Uexküll, T.v.: 1984, `A semiotic perspective on the sciences: steps toward a new paradigm', Semiotica 52(1/2), 7-47.

Burks, A.W.: 1988, `Teleology and logical mechanism', Synthese 76, 333-370.

Clarke, D.S., Jr.: 1987, Principles of Semiotic, Routledge & Kegan Paul, London.

Esposito, J.L.: 1980, Evolutionary Metaphysics. The Development of Peirce's Theory of Categories, Ohio University Press, Ohio.

Foucault, M.: 1970, The Order of Things: An Archaeology of the Human Sciences, Tavistock, London. [orig. Les Mots et les Choses, Gallimard, Paris, 1966].

Greimas, A.J. & Courtés, J.: 1979, Sémiotique. Dictionaire raisonné de la théorie du language, Hachette, Paris.

Haugelang, J.: 1985, Artificial Intelligence: The Very Idea, The MIT Press, Cambridge, Mass.

Hoffmeyer, J. & C. Emmeche: 1990, `Code-duality and the semiotics of nature.' in M. Anderson and F. Merrell (eds.), On Semiotic Modelling. Mouton de Gryter, Berlin (in press).

Jakobson, R.: 1973, Main Trends in the Science of Language, George Allen & Unwin, London.

Keller, E. F.: 1983, A Feeling for the Organism. The Life and Work of Barbara McClintock, Freeman, San Francisco.

Langton, C.G. (ed.): 1989, Artificial Life, (Santa Fe Institute Studies in the Sciences of Complexity, vol.VI), Addison-Wesley Publishing Co., Redwood City, Calif.

Nagel, E.: 1977, `Teleology revisited', The Journal of Philosophy, 74(5): 261-301.

Pattee, H. H.: 1977, `Dynamic and linguistic modes of complex systems', Int. J. General Systems 3, 259-266.

Peirce, C. S.: 1931-1958, Collected Papers of Charles Sanders Peirce, (Eds.: C.Hartshorne, P.Weiss & A.Burks), Harvard University Press, Cambridge, Mass.

Peirce, C.S.: 1955, Philosophical Wrightings of Peirce, (ed. J. Buchler), Dover Publications, New York.

Rieppel, O..: 1990, `Structuralism, functionalism, and the four Aristotelian causes', J. Hist. Biol. 23(2), 291-320.

Rosenberg, A.: 1985, The Structure of Biological Science, Cambridge University Press, Cambridge.

Sebeok, T.A. (ed.): 1986, Encyclopedic Dictionary of Semiotics, Vol.1-3, Mouton de Gruyter, Berlin.

Skagestad, P.: 1979, `C.S.Peirce on biological evolution and scientific progress', Synthese 41, 85-114.

Short, T.L.: 1981, `Semiosis and intentionality', Transactions of the Charles S. Peirce Society, 17(3), 197-223.

Short, T.L.: 1982, `Life among the legisigns', Transactions of the Charles S. Peirce Society, 18(4), 285-310.

Stuart, C.I.J.M.: 1985, `Bio-informational equivalence', J. theor. Biol. 113, 611-636.

*