Abstract and Keywords
A core tenet of the emerging field of research in living machines is that biological entities that live and act—organisms—have much in common with certain kinds of man-made entities—machines—that can display autonomous behaviour. But underlying this parallel, which is made even more persuasive by observing the life-like behaviour of many of the artifacts described in this book, are a host of critical, and still only partially answered questions. Perhaps the most fundamental of these is “what is life?” This section of the Handbook of Living Machines delves into this core question, exploring some of the most fundamental properties of living systems, such as their capacity to self-organize, to evolve, to grow, to metabolize, to self-repair, and to reproduce. This introduction provides a brief discussion about the nature of life, seen from a systems perspective, followed by summaries of each of the contributed chapters in this section.
A core tenet of the emerging field of research in living machines is that biological entities that live and act—organisms—have much in common with certain kinds of man-made entities—machines—that can display autonomous behaviour. But underlying this parallel, which is made even more persuasive by observing the life-like behaviour of many of the artifacts described in this book, are a host of critical, and still only partially answered questions. Perhaps the most fundamental of these is “what is life?”
The chapters in this section of our handbook delve into this core question, exploring some of the most fundamental properties of living systems, such as their capacity to self-organize, to evolve, to metabolize, to grow, to self-repair, and to reproduce. But perhaps the best way to begin is with the answer given to the question of life by the biologist Ludwig von Bertalanffy, one of the sharpest and most forward-thinking minds of the twentieth century.
Von Bertalanffy (1932, 1969) argued that although an organism can be viewed as an aggregate of a very large number of processes that can be described and explained in terms of their physics and chemistry, this description alone does not get to heart of the question of what it means to be alive or not. What makes an organism alive rather than dead requires a different kind of explanation, and it is the presence of order1. He went on to explain that order comes in different forms and he illustrated this by reference to the very idea that is at the heart of our handbook—that is of the “living machine” (Bertalanffy, 1969). From the clockwork machines of the seventeenth century, through the thermodynamic machines of the days of steam, to the self-regulating machines of cybernetics, he saw patterns of order emerging at different scales, and from very different chemical and physical substrates, in carefully designed artificial systems. The machine metaphor, he suggested, allows us to look again at organisms and to see them as made up of lots of little molecular machines all intricately coupled and working to support and maintain one another.
But von Bertalanffy also saw problems in the analogy between the organism and the machine, or at least unanswered questions. One of these is the problem of origin. Since organisms are evolved and machines are made, how can theories about machines help unravel the mystery of the origin of life? In particular, can machines make themselves? We will see in the chapters of this book that they are beginning to and that the answer to the origins question may be that it was living machines all the way back to the dawn of life.
(p.52) A second problem for von Bertalanffy concerned regulation. He was aware, following Turing, that many complex processes can be resolved into a series of steps (an algorithm) that can be reproduced by an automaton. Nevertheless, he worried that the number of steps that would underlie such processes as are required for life might be too immense to be computable. With the benefit of more than sixty years of advances in computing it is now easier to be reassured on this point, particularly whilst the trend towards faster computation at smaller and smaller scales continues. Perhaps as importantly though, the science of Living Machines is also showing that the principles underlying the regulation of life can be abstracted, and that it should be possible to recreate the dynamic patterns that von Bertalanffy saw to be so critical to natural living systems without the need for all of life’s intricately rich biochemistry.
Finally, von Bertalanffy wondered about the fundamental characteristic of the organism as a system engaged in the continuous exchange of components, the process we call metabolism. If metabolism is what creates the organism then higher-level descriptions of the organism as a machine, such as its cybernetic capability, do not explain life. Rather, order is already needed to bring the organism into existence, and to allow it to maintain its existence2, before we can explore all these further interesting analogies with our man-made automata.
This brings us to the core idea that von Bertalanffy considered to underpin life—the notion of the organism as an open system. An open system, as compared with a closed one such as a chemical reaction inside a test-tube, is one that continuously exchanges material with its environment, and so succeeds, at least temporarily, in thwarting the second law of thermodynamics—the tendency towards entropy or maximum disorder. Living things, in staving off entropy, achieve and maintain a steady state that is equilibrium-like but that is far from a true equilibrium. In order to maintain this state they do work, and in working, all of the marvellous properties that we see in living systems such as behaviour and cognition can come about.
The idea of living things as open self-sustaining systems has further evolved over the past century. The physicist Erwin Schrödinger (1944) popularized the notion that living organisms extract order from the environment by feeding on “negative entropy”, while the chemist Ilya Prigogine emphasized the irreversibility of the dynamical processes that sustain life, describing living systems as structures that thrive on a throughput of energy that then dissipates into the environment (Prigogine and Stengers 1984). A formal definition of a living machine, intended to capture the self-generating nature of life and the ability of living organisms to acquire and distribute the materials they need to stabilize themselves within a boundary, was proposed by Humberto Maturana and Francisco Varela (1973/1980) in their notion of an autopoietic machine. Whilst there have been various attempts to create such machines as computational models (e.g. Agmon et al. 2016; McMullin 2004), it is much harder to build a physical device that instantiates these ideas; with the exception of some of the machines described in this volume, most of today’s robots enforce a separation between fixed physical hardware and mutable software such that robotic embodiment is typically very different from autopoietic embodiment (Ziemke 2004).
The chapters in this section of the book further explore these fundamental questions of the essence of life and some of the steps that are being taken to understand them through building living machines.
We begin with the notion of emergent order itself, which in recent decades has increasingly been described using the term self-organization. Wilson (Chapter 5) explores the fundamental (p.53) nature of self-organization, its relationship to chaos, and its role in the evolution of biological organisms.
Next we look at the question of metabolism and steps towards the development of living machines that generate their own energy by harvesting and metabolizing food. Specifically, Ieropoulos and colleagues (Chapter 6) explore the symbiotic relationship between a microbial fuel cell and the robotic body in which it is embedded, and the lessons that can be learned from biological homeostasis about designing and building self-sustaining living machines.
Moses and Chirikjian (Chapter 7) investigate the key challenge of machine reproduction beginning with von Neumann’s design for a cellular automaton, called the Universal Constructor, which was devised both to allow machine replication and to support open-ended growth. Moses and Chirikjian track the history of attempts to build physical instantiations of such a system up to and including recent modular robotic systems inspired by cell molecular biology.
Biological organisms are products of the twin processes of evolution and development, and contemporary systems biology increasingly recognizes the interdependence of these two in giving rise to the complexity and variety of forms that we find in nature. Artificial evolution is widely explored in synthetic systems; however, attempts to combine evolution and development—artificial evo-devo—are fewer and far between. Prescott and Krubitzer (Chapter 8) trace some of the key ideas in the modern biological synthesis, particularly in relation to the evolution and development of the nervous system, and show how these are being investigated to allow evo-devo for living machines.
Living systems that move and grow are the focus of Mazzolai’s contribution to our handbook (Chapter 9), focusing particularly on what we can learn from the study of plants as living machines that operate in diverse and often hostile environmental conditions. Mazzolai explains how the biology of plants, and particularly of their root systems, is inspiring the design of a new class of plant-inspired actuator systems.
Vincent (Chapter 10) addresses the broad topic of biological materials noting their composite nature, their durability, and multifunctional capabilities. Vincent considers how we can learn from biology to manufacture low-energy, high-performance materials that can self-assemble.
The adaptability of living systems arises in part from their capability to self-monitor and adapt or self-repair in response to change or damage. Bongard (Chapter 11) discusses how living machines can use self-models to understand their own morphology, adapt to change, learn from others, or metamorphose from one body plan to another.
We conclude this section by returning to the core question of the nature of life, and to the differences between living and man-made systems that intrigued von Bertalanffy and many others. Building on Darwin’s theory of evolution, von Neumann’s Universal Constructor, and theories of self-organization, Deacon (Chapter 12) attempts a general theory of evolution, intended to straddle both the organic and inorganic domains and to resolve the challenges left aside by Darwin’s theory of natural selection—the origin of organisms that are capable of self-replication, a question that is explored empirically, via the physics and chemistry of self-replicating molecular open systems, by Mast et al. (Chapter 39) later in this book. In setting out this theory, Deacon rejects the notion that organisms are “mere mechanisms” and seeks to reinject a notion of agency into our definition of living systems. Specifically, he argues that living entities are both self-specifying and self-determining, and that they create their own inherent teleology or purpose, towards which their dynamics evolves.
If Deacon is right then the developers of living machines may need to embed more of the core “formative powers” of biological systems into their artifacts in order to bring them truly to life. (p.54) Beyond this, however, I would follow Maturana and Varela in being sceptical of further efforts to move the goal-posts:
To the extent that the nature of the living organization is unknown, it is not possible to recognize when one has at hand, either as a concrete synthetic system or as a description, a system that exhibits it. Unless one knows which is the living organization, one cannot know which organization is living. In practice, it is accepted that plants and animals are living but their characterization as living is done through the enumeration of their properties. Among these, reproduction and evolution appear as determinant, and for many observers the condition of living appears subordinated to the possession of these properties. However, when these properties are incorporated in a concrete or conceptual man-made system, those who do not accept emotionally that the nature of life can be understood, immediately conceive of other properties as relevant, and do not accept any synthetic system as living by continuously specifying new requirements. (Maturana and Varela, 1973/1980, p. 83)
Agmon, E., Gates, A.J., Churavy, V., and Beer, R.D. (2016). Exploring the space of viable configurations in a model of metabolism–boundary co-construction. Artificial Life, 22(2), 153–71. doi:10.1162/ARTL_a_00196
Bertalanffy, L. v. (1932). Theoretische Biologie: Band 1: Allgemeine Theorie, Physikochemie, Aufbau und Entwicklung des Organismus. Berlin: Gebrüder Borntraeger.
Bertalanffy, L. v. (1969). General system theory: foundations, development, applications. New York: Braziller.
Köhler, W. (1924). Die physischen Gestalten in Ruhe und im stationären Zustand. Braunschweig: Vieweg.
Maturana, H., and Varela, F. (1973/1980). Autopoiesis: the organization of the living. Dordrecht, Holland: D. Reidel Publishing.
McMullin, B. (2004). Thirty years of computational autopoiesis: A review. Artificial Life, 10(3), 277–95.
Prigogine, I., and Stengers, I. (1984). Order out of chaos: Man’s new dialogue with nature. New York: Bantam Books.
Schrödinger, E. (1944). What is Life? Cambridge, UK: Cambridge University Press.
Ziemke, T. (2004). Are robots embodied? In: C. Balkenius, J. Zlatev, H. Kozima, K. Dautenhahn, and C. Breazeal (eds), Proceedings of the First International Workshop on Epigenetic Robotics: Modeling Cognitive Development in Robotic Systems. Lund University Cognitive Studies, 85. Lund: LUCS.
(1) Von Bertalanffy also emphasized the wholeness of living entities, influenced, in part by the notion of Gestalt (from Köhler (1924)). The emphasis on the whole being more than sum of the parts permeates the multi-scale approach in biology that von Bertalanffy helped to establish.