John Collier

Almost fifty years ago Wilfrid Sellars described two competing ways of imagining the world, the Manifest Image and the Scientific Image. The Manifest Image is an idealization of common sense aided by critical philosophy, whereas the Scientific Image is the product of our best science. The methodologies of the two images are very different: the Manifest Image deals with experience and looks only at relations among bits of experience and analysis of experience into the relations that must lie behind it, whereas the Scientific Image is grounded in explanations of experience, typically causal explanations. This need not be a problem if the two images are compatible. Sellars argued, however, that the Manifest Image implies continuity, but the best science of the time told us that the world is made up of discrete subatomic particles and discrete transitions between quantum states, making the two incompatible. Although Sellars noted that future science might show that the world is continuous, he did not follow this up. Science in the last fifty years has given much more evidence for continuity in the world from complexity studies and Quantum Mechanics, so perhaps the two images can be reconciled after all

The paradigm of Laplacean determinism combines three regulative principles: determinism, predictability, and the explanatory adequacy of universal laws together with purely local conditions. Historically, it applied to celestial mechanics, but it has been expanded into an ideal for scientific theories whose cogency is often not questioned. Laplace’s demon is an idealization of mechanistic scientific method. Its principles together imply reducibility, and rule out holism and emergence. I will argue that Laplacean determinism fails even in the realm of planetary dynamics, and that it does not give suitable criteria for explanatory success except within very well defined and rather exceptional domains. Ironically, the very successes of Laplacean method in the Solar System were made possible only by processes that are not themselves tractable to Laplacean methodology. The results of some of these processes were first observed in 1964, and violate the Lapacean requirements of locality and predictability, opening the door to holism and nonreducibility, i.e., emergence. Despite the falsification of Laplacean methodology, the explanatory resources of holism and emergence remain in scientific limbo, though emergence has been used somewhat indiscriminately in recent scientific literature. I make some remarks at the end about the proper use of emergence in its traditional sense going back to C.D. Broad

Against standard approaches to evolution and ethics, this book develops the idea that moral values may find their origin in regularly recurring features in the cooperative environments of species of organisms that are social and intelligent. Across a wide range of species that are social and intelligent, possibilities arise for helping others, responding empathetically to the needs of others, and playing fairly. The book identifies these underlying environmental regularities as biological natural kinds and as natural moral values. As natural kinds, moral values help to provide more complete explanations for the selection of traits that arise in response to them. For example, helping in an aquatic environment is quite different than helping in an arboreal environment, and so we can expect the selection of traits for helping to reflect these underlying environmental differences. With the human ability to name, talk, and reason about important features of our environment, moral values become part of moral discourse and argument, helping to produce coherent systems of moral thought. Combining a naturalistic approach to morality with an equal emphasis on moral argument and truth, this book will be of interest to philosophers and historians of biology, theoretical biologists, comparative psychologists, and moral philosophers.

A system is autonomous if it uses its own information to modify itself and its environment to enhance its survival, responding to both environmental and internal stimuli to modify its basic functions to increase its viability. Autonomy is the foundation of functionality, intentionality and meaning. Autonomous systems accommodate the unexpected through self-organizing processes, together with some constraints that maintain autonomy. Early versions of autonomy, such as autopoiesis and closure to efficient cause, made autonomous systems dynamically closed to information. This contrasts with recent work on open systems and information dynamics. On our account, autonomy is a matter of degree depending on the relative organization of the system and system environment interactions. A choice between third person openness and first person closure is not required.

Kuhn's incommensurability thesis has generally been interpreted by friends and foes alike so as to preclude direct rational communication across revolutionary divides in science. In this paper, a weaker form of incommensurability is sketched which allows eventual comparison of incommensurable theories, but is consistent with Kuhn's model of science. Incommensurability occurs whenever the knowledge or ability to translate from the language of one theory to that of another is lacking. It can be resolved by acquiring the necessary knowledge or ability.

Goodman’s account of the ‘grue’ paradox stands at a crossroads in the history of twentieth century epistemology. Published in 1954, Fact, Fiction, and Forecast is a reaction to the logical empiricist views that held sway in the first half of the last century and anticipates many of the conventionalist and/or relativist moves popular throughout the second half. Through his evaluation of Hume’s problem of induction, as well as his own novel reformulation of it, Goodman comes to reject a number of the fundamental parts of logical empiricism. In particular, Goodman argues that the formal epistemic methods the logical empiricists wanted to rely upon are insufficient. This leads him to turn towards conventionalism, which is the basis of his rejection of the objective view of knowledge. In his conventionalism he is a precursor of writers such as Hanson, Kuhn and Feyerabend, who, seeing the inadequacy of formalism, sought to fill the gaps it left in epistemic methodology with scientific values, tacit knowledge and linguistic practices. Making use of Susan Haack’s distinction between three different kinds of foundationalism, we analyse Goodman’s position and its relation to the views of the logical empiricists. The two are reflections of each other in that, where the logical empiricists are foundationalists in their views on empirical evidence, methods used and objective criteria of justification, Goodman consistently opts for the corresponding coherentist options. As it turns out, this choice of options makes it impossible for Goodman to deal with the very objections he raises against the logical empiricists. In adding the new problem of induction to the old one, Goodman adds a problem which clearly can not be resolved by a purely syntactic solution. The charge Goodman raises is that logical empiricists cannot determine which predictions are sound and which are not. We argue that his method of identifying acceptable predictions based on the use of entrenchment is, in the end, arbitrary and only postpones the underlying problem.

In these notes I want to address some issues concerning self-organization that seem to me to apply generally from the micro-physical through the biological and social to the cosmological. That is, they are a part of the general theory of self-organization. I prefer to distinguish the theory of selforganization from the analysis of the concept of self-organization (which Maturana claims is oxymoronic, since there is no self that organizes1). General usage gives us something to which the term 'self-organization' refers. We can set aside the question of whether or not selves can really do such a thing until we know what it is they are supposed to do.2 This approach also allows the possibility that self-organization does not pick out a single natural kind, but may refer to a range of things that are grouped together by a Wittgenstein style “family resemblance”

Typically, we think of both artificial and natural computing devices as following rules that allow them to alter their behaviour (output) according to their environment (input). This approach works well when the environment and goals are well defined and regular. However, 1) the search time for appropriate solutions quickly becomes intractable when the input is not fairly regular, and 2) responses may be required that are not computable, either in principle, or given the computational resources available to the system. It may seem that there is no way to deal with these conditions, but if we think of systems as dynamical non-equilibrium autonomous entities, there are ways to deal with the unexpected and irregular by taking advantage of self-organising and self-preserving capacities of such systems. A generalised force acting on a system far from equilibrium will cause the system to reorganise itself in the direction of the generalised force in such a way as to minimise its effects (Nicolis and Prigogine, 1977), but there can be unpredictable effects in different generalised directions in the system’s phase space. In order to preserve system integrity, these effects must be damped or used for further self-reorganisation, possibly starting a cascade effect that leaves the system in a substantially different state in which it can handle further instances of this sort of information. This model is similar to and extends the theoretical model of accommodation and assimilation of Piaget, derived from his observations of the development of intelligence in children

Wilfrid Sellars described his Manifest Image and Scientific Image as idealizations of our common sense and scientific views of the world, including our own special role in the world as humans. If, as Sellars suggested, there is an irreconcilable conflict between these images, it may not be possible to reconcile science with common sense. The Scientific Image, as we have inherited it, has a strong reductionist element that seems to imply that things are not really as they appear to common sense. Although some amount of discrepancy between common sense and science is not a problem, since we know from experience that we may be mistaken about nearly any given thing, a systematic clash between science and common sense presents problems for the acceptance of one, the other, or both views

Both natural and engineered systems are fundamentally dynamical in nature: their defining properties are causal, and their functional capacities are causally grounded. Among dynamical systems, an interesting and important sub-class are those that are autonomous, anticipative and adaptive (AAA). Living systems, intelligent systems, sophisticated robots and social systems belong to this class, and the use of these terms has recently spread rapidly through the scientific literature. Central to understanding these dynamical systems is their complicated organisation and their consequent capacities for re- and self- organisation. But there is at present no general analysis of these capacities or of the requisite organisation involved. We define what distinguishes AAA systems from other kinds of systems by characterising their central properties in a dynamically interpreted information theory.

We find symmetry attractive. It is often an indicator of the deep structure of things, whether they be natural phenomena, or artificial. For example, the most fundamental conservation laws of physics are all based in symmetry. Similarly, the symmetries found in religious art throughout the world are intended to draw attention to deep spiritual truths. Not only do we find symmetry pleasing, but its discovery is often also surprising and illuminating as well. For these reasons, we are inclined to think that symmetries are informative. On the other hand, symmetries represent a kind of invariance under transformation, i.e., redundancies. Redundancy, in turn, implies that the information content of a symmetrical structure or configuration is less than that of a similar nonsymmetrical structure. Symmetry, then, entails a reduction in information content. These considerations present us with somewhat of a paradox. On the one hand, many symmetries that we find in the world are surprising, and surprise indicates informativeness. On the other hand, the surprise value of information arises because it presents us with the unexpected or improbable, but symmetries, far from creating the unexpected, ensure that the known can be extended through invariant transformations. How can this paradox be resolved? Rhythmic entrainment is the formation of regular, predictable patterns in time and/or space through interactions within or between systems (resulting in symmetry). The result of entrainment is a simplification of the entrained system. It is the complement to symmetry breaking. Entrainment can be either forced or spontaneous, with the spontaneous form being uncontrollable. It results from processes that are called self-organising. Interestingly, spontaneous entrainment require much less power to form and maintain. It is also the source of levels in systems. I finish with some observations for social systems

Complex systems are dynamic and may show high levels of variability in both space and time. It is often difficult to decide on what constitutes a given complex system, i.e., where system boundaries should be set, and what amounts to substantial change within the system. We discuss two central themes: the nature of system definitions and their ability to cope with change, and the importance of system definitions for the mental metamodels that we use to describe and order ideas about system change. Systems can only be considered as single study units if they retain their identity. Previous system definitions have largely ignored the need for both spatial and temporal continuity as essential attributes of identity. After considering the philosophical issues surrounding identity and system definitions, we examine their application to modeling studies. We outline a set of five alternative metamodels that capture a range of the basic dynamics of complex systems. Although Holling’s adaptive cycle is a compelling and widely applicable metamodel that fits many complex systems, there are systems that do not necessarily follow the adaptive cycle. We propose that more careful consideration of system definitions and alternative metamodels for complex systems will lead to greater conceptual clarity in the field and, ultimately, to more rigorous research.

Function refers to a broad family of concepts of varying abstractness and range of application, from a many-one mathematical relation of great generality to, for example, highly specialized roles of designed elements in complex machines such as degaussing in a television set, or contributory processes to control mechanisms in complex metabolic pathways, such as the inhibitory function of the appropriate part of the lac-operon on the production of lactase through its action on the genome in the absence of lactose. We would like a language broad enough, neutral enough, but yet powerful enough to cover all such cases, and at the same time to give a framework form explanation both of the family resemblances and differences. General logic and mathematics are too abstract, but more importantly, too broad, whereas other discourses of function, such as the biological and teleological contexts, are too narrow. Information is especially suited since it is mathematically grounded, but also has a wellknown physical interpretation through the Schr dinger/Brillouin Negentropy Principle of Information, and an engineering or design interpretation through Shannon's communication theory. My main focus will be on the functions of autonomous anticipatory systems, but I will try to demonstrate both the connections between this notion of function and the others, especially to dynamical systems with a physical interpretation on the one side and intentional systems on the other. The former are based in concepts like force, energy and work, while the latter involve notions like representation, control and purpose, traditionally, at least in Modern times, on opposite sides of the Cartesian divide. In principle, information can be reduced to energy, but it has the advantage of being more flexible, and easier to apply to higher level phenomena

We argue that living systems process information such that functionality emerges in them on a continuous basis. We then provide a framework that can explain and model the normativity of biological functionality. In addition we offer an explanation of the anticipatory nature of functionality within our overall approach. We adopt a Peircean approach to Biosemiotics, and a dynamical approach to Digital-Analog relations and to the interplay between different levels of functionality in autonomous systems, taking an integrative approach. We then apply the underlying biosemiotic logic to a particular biological system, giving a model of the B-Cell Receptor signaling system, in order to demonstrate how biosemiotic concepts can be used to build an account of biological information and functionality. Next we show how this framework can be used to explain and model more complex aspects of biological normativity, for example, how cross-talk between different signaling pathways can be avoided. Overall, we describe an integrated theoretical framework for the emergence of normative functions and, consequently, for the way information is transduced across several interconnected organizational levels in an autonomous system, and we demonstrate how this can be applied in real biological phenomena. Our aim is to open the way towards realistic tools for the modeling of information and normativity in autonomous biological agents.