Autogens mark the transition from maximum entropy production to constraint production and preservation, and from orthograde processes characterized by self-simplification (morphodynamics) to the orthograde processes exemplified by self-preservation and correlative complexification (teleodynamics). This transition from morphodynamic to teleodynamic organization can be described as an emergent transition, in the same sense as the transition from thermodynamic to morphodynamic organization is emergent. Each of these transitions is characterized by the development of an orthograde disposition that is contrary to what preceded and gave rise to it.

Thus morphodynamic organization emerges due to the interaction of opposed thermodynamic processes (e.g., perturbation and equilibration), and it results in constraint amplification rather than constraint dissipation (i.e., increase in entropy). a.n.a.logously teleodynamic organization emerges due to reciprocally organized morphodynamic processes, and results in constraint stabilization rather than constraint amplification, and entropy ratcheting rather than entropy production. In this respect, autogen formation exemplifies the defining feature of an emergent phase transition-the appearance of a new form of orthograde organization.

Autogen dynamics also demonstrates that the chasm between thermodynamics and living dynamics cannot be crossed in a single step, but requires an intermediate morphodynamic bridge: a synergistic arrangement of non-equilibrium self-organizing processes. The evolutionary processes that result are in this way two emergent levels removed from simple thermodynamic processes. Teleodynamics emerges from morphodynamics emerges from thermodynamics.

Being able to provide a complete description of an extremely simple molecular system, capable of self-reproduction and susceptible to natural selection, puts us in a unique position to reflect on the emergent status of living and evolutionary process and their relationships to other natural processes. The plausibility of autogens provides a simple model system for exploring the requirements of ententional properties in general. This is an important first step on the way to an augmented physical theory adequate to explain living and mental processes, as well as non-living processes. In simple terms, it is the scaffolding from which to build an emergent rather than an eliminative methodology for studying all consequence-organized phenomena, from metabolism to mental representation (though this in no way suggests that they are equivalent).

Theoretically, this a.n.a.lysis also helps to distinguish between self-organized and selection-based features of complex adaptive systems, and shows how they are intrinsically linked. The generality of this type of dynamical relationship suggests that the specific molecular features of life on Earth may be far less universal than might otherwise have been guessed, and that autogenlike processes may be present in forms and planetary conditions with very different features than ever were present on Earth. Beyond explaining the linked contribution of self-organization and Darwinian selection to phylogenetic evolution, this a.n.a.lysis may also shed light on their interaction in other biological and even non-biological processes, such as epigenesis, neural signal processing, and language evolution. Self-organizing processes can arise in many dynamic systems and can be const.i.tuted by many substrates. The dependence of evolutionary processes on self-organizing processes is not necessarily confined to molecular processes. This a.n.a.lysis should be general. Evolutionary dynamics should emerge spontaneously in any domain where a.n.a.logous reciprocal self-organizing conditions are met.

The emergence of teleodynamic processes includes more than merely consequence-organized phenomena. Even within a teleodynamically organized system as simple as an autogen, we can identify component relationships that are the minimal precursors of many end-organized features we a.s.sociate with life.

For example, although the higher-order orthograde dynamic that characterizes autogens provides their robustness to perturbation and their maintenance of integrity, this is not in itself sufficient to const.i.tute something we could justifiably call autonomous individuality. Robustness to perturbation is one of the defining features of any orthograde dynamic. Thus systems in thermodynamic equilibrium resist being driven further from that state, and morphodynamic (self-organizing, dissipative) systems near their most "relaxed" attractor dynamics will also tend to resist disruption of their characteristic regularity. Because autogens are const.i.tuted both by thermodynamic and morphodynamics processes, they inherit both of these forms of resistance. But they don't merely resist perturbation. In the face of catastrophic perturbation, the physically dissociated components nevertheless retain their systemic ident.i.ty sufficiently well to be able to rea.s.sociate into identically organized unit structures. This indicates that these components are to some extent present because of their contributions to a higher-order whole. It is in this respect that self and function are interdependent concepts that are intrinsically defined with respect to consequences.

Although the concept of a function is also sometimes applied to inanimate systems-such as when describing the gravitation of the Sun as functioning to counter the expansive effect of fusion-this is largely a metaphoric use, treating some physical relationship as though it were organized to accomplish that end. Of course, with the exception of man-made devices, inanimate non-living systems are not organized for the sake of achieving a given consequence. But autogens are. For example, it is appropriate to describe the self-a.s.sembling container of an autogen as functioning for the maintenance and perpetuation of the autocatalytic set's capacity to cycle, and the autocatalytic set can likewise be described as functioning for the sake of supporting the self-a.s.sembly process. So, in Kant's terms, each of these component processes is present for the sake of the other. Each is reciprocally both end and means. It is their correlated co-production that ensures the perpetuation of this holistic co-dependency.

Functions are normative to the extent that they fail or succeed to achieve some end. The autocatalysis, the container, and the relationship between them are generated in each replication precisely because they are of benefit to an individual autogen's integrity and its capacity to aid the continuation of this form of autonomous individual. Although one could say that, like the gravitation and fusion forces in the Sun, the component processes in a first spontaneously generated autogen just happened to accidentally co-occur to produce this metastable form, one couldn't say this about the replication of this organization in succeeding autogens. Unlike accidental correlations, the organization that is created anew via autogen replication occurs precisely because it had this consequence. But to repeat: it was the teleodynamic organization that had this consequence, and not merely some collection of interreacting molecules, because these are replaceable while the organization is not. The ident.i.ty and the beneficiary is not a thing but a dynamical form.

So, even these simple molecular systems have crossed a threshold in which we can say that a very basic form of value has emerged, because we can describe each of the component autogenic processes as there for the sake of the autogen integrity, or for the maintenance of that particular form of autogenicity. Likewise, we can describe different features of the surrounding molecular environment as "beneficial" or "harmful" in the same sense that we would apply these a.s.sessments to microorganisms. More important, these are not merely glosses provided by a human observer, but intrinsic and functionally relevant features of the consequence-organized nature of the autogen itself.

Adaptive functions are, however, more than just elements of an ent.i.ty that respond to the ent.i.ty's environment. They embody in their form and dynamic potential-as in a photo negative-certain features of this environment that, if present, will be conducive or deleterious to the persistence of this complex dynamic. The presence of these conditions may or may not obtain. Thus adaptations may be appropriate or inappropriate to a given context, to the extent that the consequence with respect to which they are organized may or may not be achieved. This is another indication of functional normativity: the possibility of dysfunction.

In the case of dysfunction, the correspondence relationship between internal organization and extrinsic conditions no longer exists. In a crude sense, then, we can describe this as an erroneous prediction based on a kind of physical induction from past instances. It is in this respect-of possible but fallible correspondence-that we can think of an adaptation as embodying information about a possible state of the world. And like more obvious forms of representation, this projection can be in error; there may be nothing in the immediate environment to which it corresponds. So, in this very basic sense, autogens could be said to represent their environment, in roughly the same sense as a shoeprint could be said to represent a shoe.

We are far from a full explanation for the sorts of teleological processes experienced at the level of human consciousness, and a considerable distance from what is found in the simplest living forms on Earth. The "proof of principle" is in this regard quite minimalistic. Nevertheless, it is a definitive exemplar of crossing the fundamental gap that separates the mechanical world from the functional and normative world.

In summary, then, in this simple dynamical molecular system we can discern the minimal precursors of function, adaptation, teleology, valuation, and even the dim antic.i.p.ation of information about the environment and a self with respect to which all this matters. Each of these attributes is implicitly ententional, and yet their emergence can be precisely understood without the need either to attribute them to mysterious or weird forms of causality or to argue that there isn't any fundamental threshold crossed at this point. This demonstrates that, at least in principle-and ultimately that's all that matters-real teleological and intentional phenomena can emerge from physical and chemical processes previously devoid of these properties.

Now that we have thoroughly explored this simple teleodynamic system, and have demonstrated how it can emerge from less convoluted morphodynamic processes, and ultimately from thermodynamics, we are in a position to begin to reframe the way we understand the physics of consequence-organized processes (i.e., Aristotle's final causality). At a minimum, this a.n.a.lysis demonstrates that there can be no simple one-to-one mapping of teleodynamic relationships to mechanistic relationships. The link between basic mechanics and ententional processes and relationships must necessarily be bridged by an intervening level of morphodynamic processes. The persistent failures to map living dynamics and cognitive processes directly to simpler physical processes, and thus to reduce mental to mechanistic processes, was always doomed to fail, though not because of some fundamental discontinuity or dualism. The supervenient relationship between them is indeed necessary, but it is both doubly indecomposable and doubly negative. It is doubly indecomposable because it is based on two emergent transitions, each defined by the intrinsic generation of higher-order holistic constraints, that cannot be decomposed to any lower-level components or relationships. It is doubly negative because each level of dynamics emerges from the interaction of lower-level constraints, which are themselves absential properties. So, in effect, each is the expression of a form of higher-order absence generated by relationships between lower-order absences. This doubly negative, doubly absential character of teleodynamic processes is almost certainly one reason that we find them so mechanistically counterintuitive.

The a.n.a.lysis of this additional level of emergent transition also enables us to identify a common logic for unambiguously defining emergence more generally. An emergent dynamical transition is signaled by a change in the topology of the phase s.p.a.ce of probable dynamical trajectories. Using the term that we have applied to asymmetries of probable changes of state, each emergent dynamical transition involves the appearance of a new mode of orthograde attractor logic. Thus the transition from a simple thermodynamic regime to a morphodynamic regime is marked by the emergence of orthograde tendencies toward highly regularized global constraints that run counter to the constraint-dissipation orthograde tendencies of the underlying thermodynamic processes.

Emergence is, in effect, defined by a polarity reversal in orthograde dynamics with ascent in scale. Thus the orthograde signature of thermodynamic change is constraint dissipation, the orthograde signature of morphodynamic change is constraint amplification, and the orthograde signature of teleodynamic change is constraint preservation and correlation. The polarity reversal that defines the emergence of teleodynamics from morphodynamics is what characterizes life and evolution. A fit or interdependent correspondence between constraints in different domains is the essence of both biological adaptation and the relationship characterizing representational relationships. So this provides the first bridge across the "epistemic cut" that has been the dividing line between the two sides of the Cartesian dilemma-the no-man's-land that has divided both science and metaphysics into seemingly incommensurate universes.

So far this a.n.a.lysis has been mostly confined to processes in the range from molecular thermodynamics to the very simplest lifelike processes. Emergent dynamical transitions are, however, ubiquitous in nature, and although they are necessarily simpler at lower levels of scale, they can be correspondingly far more complex at levels of scale where multicelled organisms, brains, and human social phenomena emerge. Nevertheless, there is a general principle exemplified by the simplest autogenic system, described here, that applies at all higher levels of emergent dynamics: all teleodynamic processes must be const.i.tuted by reciprocally synergistic morphodynamic relationships, and all morphodynamic processes must be const.i.tuted by competing homeodynamic processes. Although higher-order teleodynamic processes may exhibit properties that are more elaborate than those exhibited by basic autogenic systems, they must arise by a recapitulation of this same hierarchic emergent dynamic logic, even if the components are themselves teleodynamic systems. Teleodynamic systems can interact homeodynamically; homeodynamic relationships between teleodynamic systems can produce morphodynamic relationships; and synergistically reciprocal morphodynamic relationships const.i.tuted by interacting teleodynamic systems can produce higher-order teleodynamic relationships.

With each such emergent transition, there will be a characteristic new level of orthograde geometry of causality, but the generation of each emergent transition must necessarily depend on this homeo-morpho-teleo emergent logic. Like basic autogentic systems, higher-order teleodynamical systems will also be self-creating, self-maintaining, self-reproducing individuated systems; but they may also exhibit emergent teleodynamic properties not exhibited by their lower-order teleodynamic components. Terming such second- and third-order teleodynamic systems autogens and describing their properties as autogenic would therefore be too restrictive and falsely reductive. So I will call such higher-order teleodynamic systems teleogens, in order to designate both their individuality and their capacity to generate additional forms of teleodynamic processes. We will survey some of the properties of such higher-order emergent teleodynamic relationships (such as sentience) in the final chapters of this book, and discuss the implications of homeodynamic, morphodynamic, and even teleodynamic processes const.i.tuted by the interaction of lower-order teleodynamic systems, as is found in ecosystems, complex organisms, brains, and even social systems. But before we can apply this logic to ever more complicated systems, we need to reflect on the generality of the a.n.a.lysis beyond this basic model system, which we have used to exemplify this third realm of dynamics.

1 1.

WORK.

What is energy? One might expect at this point a nice clear, concise definition. Pick up a chemistry text, a physics text, or a thermodynamics text, and look in the index for "Energy, definition of," and you find no such entry. You think this may be an oversight; so you turn to the appropriate sections of these books, study them, and find no help at all. Every time they have an opportunity to define energy, they fail to do so. Why the big secret? Or is it presumed you already know? Or is it just obvious?

-H. C. VAN NESS, 1969.

FORCED TO CHANGE.

The theory of emergent dynamics that we have outlined in the previous three chapters does not in any way conflict with the basic principles of physical dynamics. Indeed, it is based almost entirely on well-established pre-twentieth-century physical theory. However, there is an interesting way that it can amplify and broaden these principles to extend into organizational and ententional realms that have until now remained mostly disjointed from the physical sciences. In this chapter, then, we will reexamine the common notions of energy, power, force, and particularly work (from which these other concepts are abstracted), in an effort to understand how they can be generalized and reformulated in emergent dynamic terms.

It took the geniuses of Galileo and Isaac Newton to show definitively that, neglecting the effects of friction, objects moving in a straight line will maintain their velocity and direction of movement indefinitely, so long as they are not affected by gravity, impeded by collision, or otherwise forced off course. And as a result, constant rectilinear motion is now understood to be equivalent to being at rest. It tends to persist as a spontaneously stable pattern of change. A thermodynamic system in equilibrium is also in continuous change, even though at a macroscopic level it appears to be unchanging. In this respect, it is a.n.a.logous to an object in constant undisturbed movement. Extending the a.n.a.logy, we can likewise compare the interactions of differently moving objects to the interactions between thermodynamic systems with different specific heat or energy levels. The differences in relative momentum when objects interact result in accelerated changes in their motions. The differences in the total heat or energy content between different thermodynamic systems at equilibrium can translate into changes to both if brought into interaction. These correspondences form part of the bedrock on which cla.s.sical physics was built, and later found even more subtle expression in relativistic and quantum theories.

How much things change from what would have occurred spontaneously is a reflection of the amount of work exerted to produce this change. So long as contragrade change persists, work is involved, and thus it can acc.u.mulate over time and distance. The amount of change also reflects the amount of energy exchanged during the transition from the prior condition to the changed condition in the interaction. Energy is related to the capacity to do work, irrespective of whether it is exhibited in the collision of two ma.s.ses or held in the electrical potential compressed into the covalent bonds of a hydrocarbon molecule. As we have discovered, however, the concept of energy is remarkably difficult to pin down. This is in part because it is a notion abstracted from the concept of work. In one sense, it could be considered merely a way of balancing the books. It is what has not changed in any physical change, spontaneous or not.

Work, on the other hand, is intuitively more tractable, as it is directly reflected in the extent of non-spontaneous change that results. And, unlike energy, the capacity to do work does not remain constant. This is what we intuit when water has drained downhill and we can no longer use its motion to turn a waterwheel, or when heat is fully transferred between two systems so that they are at equilibrium. During the falling of water, we say that energy is transferred from the movement of the water to the turning of the wheel; and during the exchange of heat, we say that energy is transferred from the hotter to the cooler container. After reaching ground level, the water can do no more work; and after reaching equilibrium, no more work can be extracted from the thermodynamic system. We say in this case that there is no more "free energy," which is also to say that there is energy that is no longer free to be transferred in such an interaction. Of course, some of the energy implicit in the elevated water could become "freed" were we to find a way to get it to flow to an even lower elevation, and some of the energy of a thermodynamic system at equilibrium would again become free if that system were subsequently placed in contact with an even cooler system. Clearly, at some point, there is no place lower for water to flow to, and no cooler system to take on some of the heat of a thermodynamic system. The latter condition is called absolute zero (and not surprisingly matter begins to act strangely near this temperature). This ability to "free" and "trap" energy tells us that, whatever it is, energy is not in itself the source of change. Rather (as we discussed in chapter 7), it is its availability for "movement" from place to place or transition from form to form that provides the potential for contragrade change. This is why constraint on this "movement" is so central to notions of causality.

Moreover, the generation of higher-order emergent dynamics also depends on work at a lower order. What we will discover in this chapter is that with each emergent transition, a novel capacity to do work emerges as well. And a new mode of work introduces new causal possibilities into the world. So, to understand the emergence of novel forms of causality, we need to explain the emergence and nature of these higher-order forms of work.

EFFORT.

Newton's precise a.n.a.lysis of the concept of mechanical work was only the beginning of the physical a.n.a.lysis of this explanation for change in state. To make all forms of physical work consistent and to explain their interconvertability, it was necessary for nineteenth- and twentieth-century scientists to discover how to equate this Newtonian conception with the change-producing capacities of heat engines, chemical reactions, electromagnetic interactions, and nuclear trans.m.u.tations. By the third decade of the twentieth century, this project was largely complete. But behind this towering achievement of theoretical physics is a broader conception of work, one that is generally ignored in the natural sciences because it is a.s.sumed to be merely a colloquial a.n.a.logue to this more technical understanding. Indeed, the technical a.n.a.lysis of work was long preceded by this more generic conception that we use to describe a vast array of effortful enterprises.

The question that we will address here is whether these other not-exactly-energetic conceptions of work can also be brought into a more precise formal relationship with the physical science notion. In other words, to highlight just one exemplar, can the work of conceiving of these thoughts, transforming them into sentences, and selecting the most appropriate words to express them be understood to be as real and measurable as the work of an internal combustion engine? I am not merely inquiring about the metabolic support for the neural and muscular biochemical reactions involved, though these are relevant, but the higher-order work of forming and interpreting the concepts involved-that which makes daydreaming effortless but metabolically equivalent problem solving difficult. Though often they are not energetically equivalent, I will argue that it is not the energy use that makes a difference, but the manipulation of the content of these thoughts-something that we have argued is not physically present.

This is not a merely academic problem to be solved. Measuring the amount of work needed for a given task is often an important factor in determining the minimum requirements for success, or for determining the relative efficiency of different ways of achieving the same goal. In the a.s.sessment of physical tasks, for example, engineers need to routinely calculate how much work must to be done to move objects around on a factory floor, to lift a heavy object to a certain height, or to accelerate a vehicle to a given velocity. But the sort of work that we are often interested in a.n.a.lyzing is not always simply physical in this sense. Perhaps the most common uses of the term work refer to activities that do not neatly fit into this physical schema at all, but involve making difficult decisions, a.n.a.lyzing unknown causal effects, and exploring mysteries.

We even tend to describe our occupations as work. We ask strangers, "What kind of work do you do?" and talk about "driving to work" in the morning, where work is either considered a cla.s.s of human activity or even the location of that activity. We consider it work to keep one's home clean, to organize and manipulate food and cooking utensils to prepare a meal, or to convince colleagues of a counterintuitive theoretical idea. What these more colloquial notions share in common with the Newtonian conception of work is the superficial implication that the activity being described makes things happen that wouldn't come about without it, or else that would happen unless work is done to prevent it. In general terms, then, we can describe all forms of work as activity that is necessary to overcome resistance to change. Resistance can be either pa.s.sive or active, and so work can be directed toward enacting change that wouldn't otherwise occur or preventing change that would happen in its absence. This also means that work may be activity pitted against some other form of competing work, as when law enforcement officers work to counter the work of criminals.

This more generic use of the concept of work is also employed to describe activities merely requiring mental effort, in which the linkage with physical activity and energetic processes can be quite obscure. We know that producing a novel software routine that is able to accomplish some computational task, or solving a difficult crossword puzzle, or conceiving and organizing the steps necessary to efficiently construct a garment, all take work. But it is not at all obvious how to measure or compare these forms of work to each other or to the general physical concept of work.

Nevertheless, we are often adept at coming up with relative measures of mental work. Difficult jigsaw puzzles take more work to solve than easy ones. This might be simply because they have more parts. And they can be more difficult if the pieces are all turned upside down so that there is no picture to suggest which pieces are likely adjacent to one another (a constraint that reduces the amount of work required?). To make it even more difficult, some puzzle makers cut the pieces into very nearly the same shape, so that constraints of shape incompatibility are made unavailable. These complications do not necessarily require more physical manipulation of the pieces (although this might also follow), but almost certainly they will require more "mental manipulation." This simple example indicates that numbers of optional arrangements of things that aren't desirable or functional, and the difficulty of distinguishing among them, contributes to this notion of the amount of mental work required.

How might this relate to the more familiar notion of mechanical work? Though some mechanical work is involved in moving puzzle parts from place to place, and more movements are necessary to discover the correct fittings by trial and error in many-component problems than those with fewer components, for the most part this is not how we intuitively estimate the amount of work that will be required. Probably the best estimator is some measure of the number of operations that will likely be required to reach a solution. Of course, each operation-even each mental operation-requires energy, since these too are not likely to happen spontaneously. But this is often the most trivial source of resistance to be overcome.

In writing this book, for example, the finger work of typing that is necessary to input text into the computer is trivial in comparison to the work required to conceive of and express these ideas. This agrees with the intuition that the re-a.n.a.lysis and reformulation of otherwise widely accepted ways of thinking-particularly the effort to craft a convincing critique and alternative explanation-and that which is involved in discovering how best to communicate ideas that are counterintuitive or alien or otherwise go against received wisdom, is particularly difficult work, though it may be no more energetically demanding that walking a mile. This suggests that the sources of resistance that are the focus of the work to be done also include many tendencies not generally considered by physicists and engineers; for example, tendencies of thought that contribute to the difficulty of changing opinions or beliefs.

One might be tempted to object that the family resemblances between these various ententional notions of work and the physicist's concept of work are merely superficial, and that the use of the same term to refer to a form of employment or a creative mental effort is only metaphorically related to the Newtonian notion. But if there is a deeper isomorphism linking them all, there could be a great benefit in making sense of this connection. It is becoming increasingly important to discover how best to measure and compare all these diverse forms of work, especially in an era in which vast numbers of people spend their days sorting and a.n.a.lyzing data, organizing information in useful ways, and communicating with one another about how they are doing it. If it were possible to identify some unifying principles that precisely express the interdependencies between the physicist's conception of work and the computer programmer's experience of work, for example, it might have both profound scientific and practical value. This is not just because such knowledge could help to a.s.sess the relative efficiency of management strategies, aid Wall Street agencies in discerning optimal advertising campaigns, or even contribute to political efforts to manipulate public opinion, but because work is the common denominator in all attributions of causal power, from billiard ball collisions to military coups to the creative outputs of genius.

More generally, what we mean by causality and what we mean by work are deeply interrelated. One of the main reasons scientists and philosophers still argue about the kind of causality that const.i.tutes our ability to initiate goal-directed activities is that we can't figure out how to link this mental form of work to the physical forms of work that are also necessarily involved. To finally cut through the tangle of confusions that surround the mysteries of mental agency and the efficacy of representations, then, we first need to develop a general theory of work: one that explicitly demonstrates the link between the ways that both energy and ideas can introduce non-spontaneous change into the world.

AGAINST SPONTANEITY.

The anthropologist and systems thinker Gregory Bateson is well known for his relentless effort to expose the widespread fallacy of describing informational relationships in biology and the human sciences using energetic metaphors. In an effort to define information in its most general sense, and to distinguish it from energy, he described it as "a difference that makes a difference."2 This makes explicit a conception of information that is central to Claude Shannon's Mathematical Theory of Communication, and to which Bateson added an implied cybernetic aspect by virtue of the double meaning of "making a difference." We will return to the problem of defining information in the next chapters, but this way of talking about difference is somewhat ironically also relevant to the concept of work.

Bateson was trying to distill the essence of the logic of information and control theory by highlighting the fact that according to this theory, information was merely a measure of variety (e.g., of letters or signal patterns) or difference, and not some "thing." He was emphasizing the fact that a difference is an abstract relationship, and as such behaves quite differently from material substances and their interactions. As an example, he points out that a switch is neither within nor outside an electric circuit. It mediates a relationship between events outside and those inside the circuit. When the switch is thrown by an external difference in some feature (e.g., a rise in temperature) it creates an internal difference in the circuit (e.g., breaking the circuit and cutting power to the furnace), which in turn causes an external difference (e.g., a drop in temperature), and so on.

Bateson was reacting against the misleading metaphorical use of energetic concepts to talk about informational processes, such as show up in concepts like the "force" of ideas, the "power" of ideology, or the "pressure" of repressed emotions (due to impeding the flow of libido in Freudian theories of neurosis). He was struggling against an entrenched substance terminology (a.n.a.logous to the eighteenth-century conceptions of phlogiston and caloric), which obscured the critical differences between physical principles and those beginning to be articulated by the infant fields of information theory and cybernetics. This misleading conflation of energy with information often leads us to treat information as though it is a physical commodity; a kind of stuff that one can acquire, store, sell, move, lose, share, and so on. Indeed, as this list makes obvious, this is precisely the colloquial understanding of the concept. Bateson's point suggests that, as in the case of energy, progress could only be made when this was replaced by a dynamic relational conception of information. This was finally achieved (though incompletely) in the 1940s. In the next chapter, we will rea.n.a.lyze the concept of information in some detail, and both explain this insight and explore how it still falls short of a full conception of information. But for now, it is sufficient to recognize that this substantializing tendency is similar to the substance conceptions of energy that dominated the eighteenth and early nineteenth centuries. So, although Bateson's phrase captures a number of important features that characterize information, and which show it to be different from mere stuff, much is left ambiguous as well. In fact, it doesn't quite disambiguate energy and information, as Bateson had intended.

Recently, a colleague (Tyrone Cashman) recounted a discussion he had some years ago with the influential systems ecologist Howard Odum. Cashman was attempting to explain the distinction that Bateson was making between energy and information by the use of his epigrammatic phrase "a difference that makes a difference." But Odum objected that his phrase did not in fact uniquely demonstrate this distinction, because it could equally well be applied to the concept of energetic work: a difference in the distribution of energy in one system that can be used to produce a difference in the distribution of energy in another. This objection is well taken. As we saw above, this is a quite accurate abstract definition of the concept of mechanical or thermodynamic work. Was Bateson mistaken? Is this a poor definition of information?

On the one hand, I have to agree with Odum that this epigram does not do a very good job of picking out the distinguishing feature of information processes that make them different from energetic work. On the other hand, if it is nevertheless a useful characterization of information (which I think it is), then this parallelism suggests something quite interesting. It suggests that the generation of information might also be understood as a form of work, perhaps related to, though not merely, energetic work. So, comparison to the development of the energy concept might offer clues about the kinds of misconceptions that tend to arise when a.n.a.lyzing information. This will be the topic of chapters 12 and 13; but before embarking on this issue in greater depth, we can for now explore the implications for a technical conception of work that is as precise, but more generalizable than just what we describe with Newtonian physics and thermodynamics; generalizable even to processes as diverse as order creation, information production, and decision making.

How might this Batesonian conception, treated as a description of physical work, point the way to a precise general conception of work? Consider how it might apply to a physical process. When energy is transformed from one form to another, it is a difference that is being transferred from substrate to substrate (a gradient of non-equivalence, an asymmetry of distribution, say, of molecular momenta), but there is inevitably some resistance involved in this transfer. Systems "resist" being shifted away from a state of equilibrium (though as will become clear in a moment this resistance is not a simple concept). This is also implicitly captured in Bateson's phrase, since it implies that the second difference is compelled-or made-to come into existence by the first difference. The difference that is "made" depends on a difference that is provided as a given. The implication is that the new difference that is created in this process would not have occurred had the first difference not existed. So another way to describe work, using this Batesonian characterization, is that it involves something that doesn't tend to happen spontaneously being induced to happen by something else that is happening spontaneously. In the dynamical terms introduced in chapter 7, we can describe work as the organization of differences between orthograde processes such that a locus of contragrade process is created. Or, more simply: Work is a spontaneous change inducing a non-spontaneous change to occur. With this first approximated generic definition, we can now begin to unpack the logic that links energy, form, and information.

Isaac Newton had already provided a precise definition of mechanical work prior to the nineteenth century. The importance of Joule's experiment a few centuries later was to show that there was a precise relationship between this accepted notion of mechanical work and the generation of heat. In both cases, work was being defined with respect to a change of something that would otherwise tend to stay the same. But the relationship between mechanical and thermodynamic work is deeper and more thoroughly interrelated than merely parallel.

Recall the fact that thermodynamic properties are macroscopic reflections of Newtonian dynamics at the molecular level. From a Newtonian perspective, each collision of molecules in an ideal gas involves a minute amount of mechanical work, as the colliding molecules are each altered from their prior paths. Even at equilibrium there is constant molecular collision, and thus constant work occurring at the molecular level. Indeed, the overall energy of the average collision does not differ, whether the system is at equilibrium or far from it. This incessant "Brownian motion" is the means by which-even at thermal equilibrium-the molecules in a drop of ink dripped into water are eventually diffused throughout the solution. Without this molecular work, there can be no change in state of any kind. But notice that while it is possible to get work from a system that is in a state far from equilibrium, as it spontaneously develops toward equilibrium, this capacity rapidly diminishes, even if the total amount of molecular level work remains constant throughout. And at equilibrium, the vast numbers of collisions and the vast amount of microscopic work that is still occurring produce no "net" capacity for macroscopic work. So, although the potential for macroscopic work depends upon an incessant process of microscopic work, macroscopic work doesn't derive from it. Rather, macroscopic work depends on microscopic work being distributed in a very asymmetric way throughout the system. This shows us that the two levels of work-microscopic-molecular and macroscopic-thermodynamic-are not directly correlated. Microscopic work is a necessary but not sufficient condition for macroscopic work.

Any interaction with this system that shifts it from equilibrium will also be the result of changes in these micro collisions. Speeding up a subset of molecules that are in contact with a heat source, or slowing down some molecules in contact with a cold surface, both produce the capacity for thermodynamic work. Irrespective of adding or removing energy from the system, it is the degree of spatially asymmetric difference in average molecular velocities that matters. Ultimately, the capacity of the perturbed system as a whole to be tapped to perform work at the level above that of molecular collision is a consequence of the distributional features of the incessant micro work, not the energy of the component collisions, which as a whole can increase, decrease, or remain unchanged. In other words, in thermodynamics the macro doesn't simply reduce to the micro, even though it is dependent upon it. The macroscopic form of the distribution is the critical factor.

So also, like a ma.s.s in rectilinear motion, a state of incessant change can also be a stable state in thermodynamics. This suggests another interesting a.n.a.logy between Newton's notion of work and the thermodynamic conception of work. In Newtonian terms, a ma.s.s can only be perturbed from rest or from a linear trajectory by the impositions of an extrinsic force, such as by interaction with another ma.s.s with different values of velocity and direction, or under the influence of some field of force, like gravity. In other words, its resistance to change is reflected in the amount of work required to produce a given change. Resistance is also characteristic of thermodynamic systems. A thermodynamic system at equilibrium can only be driven away from its dynamically symmetric basin by being coupled to a second system with a different value of some system variable, such as temperature or pressure. When two systems with different equilibrium values are coupled, stability gives way to change, as the coupled system changes toward a new equilibrium point. The transient phase, during which the now-combined larger thermodynamic system changes to a new global equilibrium state, is thus a.n.a.logous to the brief period during which colliding objects in a Newtonian world are being accelerated or decelerated due to their interaction.

In the real world, even Newtonian interactions are not instantaneous, especially if the colliding objects are elastic. Elastic effects underscore the thermodynamiclike basis of even Newtonian interactions, since the elastic rebound of real colliding solid objects involves an internal asymmetric destabilization, in the form of compression of some molecular distances, followed by re-equilibration as both objects' internal energies redistribute. Of course, the a.n.a.logy becomes increasingly stretched (not to make a pun) at this point, because unlike the Newtonian a.n.a.logue-in which the objects' internal states return to where they were before collision and the objects permanently decouple-interacting thermodynamic systems do not have such a neat distinction between internal state and external relational features, such as momentum. Whatever the source of resistance and stability, however, the change toward the new equilibrium values and the new dynamical stability is spontaneous. This means that the intrinsic pattern of spontaneous change is itself also the source of a system's resistance to change. Thus, two thermodynamic systems which are either at different equilibria or are both undergoing spontaneous change at different rates can be considered contragrade to one another. Because of this, they will do work on one another if they become coupled.

This allows us to propose an even more general definition of work: it is simply the production of contragrade change. This way of describing work with respect to a spontaneous tendency of change shows us that the possibility of doing work to change the state of things is itself dependent on the relationships between processes that do not require work. In other words, differences in spontaneous processes of change, and the resistance of these to deviation from the specific parameters of that change, are the source of work. Or to put it succinctly, contragrade processes arise from the interaction of non-identical orthograde processes.

So, somewhat paradoxically, interactions between systems' different spontaneous tendencies are responsible for all non-spontaneous changes. Given that composite systems, with inherently iterative dynamics, display statistical asymmetries due to variations in their component interactions, a given composite system will generally have quite different asymmetric spontaneous tendencies than a second system. This is particularly likely in cases where the substrates of this interaction are of radically different form (e.g., interactions between light radiation and thermal motion).

In previous chapters we borrowed the distinction between efficient and formal causes from Aristotle, to argue that the capacity to produce non-spontaneous change could be loosely a.n.a.logized to Aristotle's efficient cause and the conditions that produce spontaneous change could be loosely a.n.a.logized to his formal cause. We now are in a position to be more explicit about this comparison.

First, let's recap what we have concluded about orthograde processes. If the global distribution of lower-order (micro) work is not symmetric in a thermodynamic system then it will tend to change in an asymmetric direction, to symmetrize this distribution, and will resist any tendency to reestablish a new asymmetry. It is this distributional feature, and the statistical asymmetry of interaction possibilities at the micro level, irrespective of the total work occurring at that lower level, that is responsible for the directionality of change. This property of the whole is effectively a geometric property-both of the spatial distribution and of the statistical distribution of work at a lower level. This distributional basis is why it makes sense to think of this property as a kind of formal cause. But formal constraints and biases do no work. They do not bring about change away from what would occur spontaneously. Yet as the foundation of any asymmetries that may arise between systems, they become the basis for work at a higher level.

Now consider the capacity for work at these two levels. The non-correlation of relative molecular movements is the basis for the work done by molecular collisions within a gas, but it is the global and statistical distributional character of this lower-level work that is responsible for higher-level orthograde properties. a.n.a.logously, the non-correlation (i.e., non-equivalent values) of the orthodynamic properties of linked thermodynamic systems (e.g., creating a temperature gradient) is responsible for higher-level work. So, without micro work, there can be no orthograde tendency to change or resist change at a higher level and no capacity for work at a yet higher level. In this regard, orthograde and contragrade processes provide the necessary conditions each for the other, but at adjacent levels in a compositional hierarchy.

If we think of work as the a.n.a.logue of Aristotle's efficient causality, then, we can treat efficient and formal cause as interdependent and inseparable counterparts to one another. But there is another interesting distinction that must be added, relating to how they inversely react to the symmetry relationships that are their basis: their symmetry with respect to time. Both Newtonian interactions and the transformations produced by thermodynamic work can be run in reverse with approximate symmetry. For example, the movie of a simple billiard ball collision run in reverse does not appear odd, and a heat engine can be run in reverse to produce refrigeration. Yet a billiard ball break that scatters the b.a.l.l.s previously racked into a triangle and the diffusion of a drop of ink into a gla.s.s of water would appear quite unnatural if shown in reverse. We can describe this in abstract terms as follows: when one asymmetry is transformed into another asymmetry, they are symmetric to one another in their respective deviations from symmetry, which is to say that they are interconvertable; but an asymmetric state transformed to a more symmetric state involves a fundamental change in symmetry, and so the two states are not interconvertable. But because the production of reversible contragrade change-work-is always based on lower-order irreversible orthograde processes, the reversibility of work is never completely efficient. There must always be some loss in the capacity to do work with each transformation. To put this in the form of a mnemonic pun: efficient cause is never 100 percent efficient.

In summary, then, we have identified what appears to be a quite general principle of causality that shows how work and the constraints that make it possible are to be understood in terms of levels of scale and supervenient organization. More generally, this dissects the logic of physical causality into two component aspects, roughly compared to Aristotle's notions of efficient and formal causality. It also distinguishes their complementary roles in the production and organization of change, and shows how they complement one another at adjacent levels in a supervenient hierarchy.

TRANSFORMATION.

Normally, for human purposes (and for living organisms in general), we focus on only one direction of this interactive effect. That is, we are typically interested in changing the spontaneous status of some one thing in particular, to make it more useful, and so endeavor to bring some other influence to bear to accomplish this. The change this reciprocally imposes on the other system's spontaneous tendency is often not of any consequence to us and tends to be ignored. So, for example, in the familiar case of a simple internal combustion engine used to raise a weight off the ground (see Figure 11.1), we take advantage of the spontaneous expansion of the ignited fuel to overcome the spontaneous inertia of the vehicle or the weight, though in the process, the rate of expansion of the gas is considerably impeded compared with what would have occurred in the absence of this coupling. The raising of the weight, the slowing of the expansion of the gas, and the constraint that limits where it is able to expand are all different than they would have been in the absence of this coupling. But the expansion of the gas, though slowed by contragrade action, still proceeds in an orthograde direction, while the change in the position of the weight proceeds in an entirely contragrade direction due to the imbalance in the forces involved. For this reason, we typically describe work being performed on the weight by the exploding gas and not the other way round, even though the slowing of the rate of this expansion is also contragrade and const.i.tutes an equal and opposite amount of work.

FIGURE 11.1: The left diagram schematically depicts the logic of thermodynamic work in which one physical system (A), which is changing in an orthograde direction (in which the reduction of free energy and the increase of entropy are depicted as an arrow from higher potential to lower potential), is coupled to another system (B) via constraints that cause the second system to change in a contragrade direction (depicted as a reversal of A). A familiar example of this relationship is depicted on the right where the exploding air-fuel mixture in a cylinder is constrained to expand in only one direction, and this is coupled to a simple mechanical device that raises a weight.

Within the mechanical and thermodynamic realms, of course, there are as many diverse forms of work as there are heat engines. And this variety is only a fraction of the possibilities, which are as diverse as the possible kinds of substrates and couplings that can be realized. Ultimately, every transformation of energy from one form to another involves work as we have defined it so far. In the transformation, organization matters. How the interactions are constrained is a critical determinant of the nature of the work that results, because ultimately all such transformations involve a change in the dimensions and degrees of freedom (i.e., mode of dynamics and constraints) while the total energy remains unchanged. This inevitably requires work, because it is a process of restructuring constraints.

Over the past few centuries we have learned to build all variety of devices that can convert energy from one form to another. Thus, a temperature gradient can be transformed to generate mechanical work in a heat engine, an electric current can be transformed to generate a temperature gradient in a refrigerator, light energy can be transformed into electric current in a photovoltaic cell, and all these processes can be organized to run in the other direction as well. All involve the production of contragrade changes, by coupling these otherwise largely uncoupled domains.

In the case of a heat engine, these domains are separated by the radical differences in scale between microscopic molecular motions on the one hand and large-scale mechanical motions on the other. But the distinctions can be even more qualitatively extreme, and as a result, the limitations on possible interactions can be quite restrictive. In order to overcome such natural part.i.tioning, it often takes highly specific materials organized in precise forms. This substrate-based limitation on the possibility of interaction is the basis for the diversity of forms of work that can be generated, given the appropriate mediating dynamical linkage. So, for example, a photovoltaic cell requires metals in which electrons (and their complementary absences, called "holes") are easily displaced by light to mediate the transformation of light energy into the energy of an electric charge gradient, and the immense compression force of the Sun's gravitation is required to transform nuclear ma.s.s into light and heat, via nuclear fusion.

The real value of this abstract conception of work, however, is not that it provides another (perhaps intuitively more transparent) way to describe forms of work that are already well understood in contemporary physics, but that it provides an abstract general characterization that can even be applied outside these mechanical realms. What is provided by this approach-and is not at all obvious from the physics alone-is the possibility of extending an a.n.a.lytically precise concept of work into domains that we would not identify with the physics of energetic processes: namely, the domains of form generation and semiotic processes. In other words, this abstract characterization can be applied to all three of the levels of dynamics that we have been exploring in this book. On this basis, we can begin to frame a theory of morphodynamic work and of teleodynamic work.

Such formulations are possible because at each level of dynamics there are cla.s.ses of orthograde tendencies-conditions of broken symmetry that tend toward greater symmetry-which correspondingly define cla.s.ses of "attractor" states a.n.a.logous to thermodynamic equilibrium. Thus, the coupling of morphodynamic systems (or teleodynamic systems) can, a.n.a.logously, bring their distinct and often complex orthograde tendencies into interaction, causing contragrade tendencies to emerge from their net differences. This opens the door to an emergent capacity to generate ever more complex, unprecedented forms of work, at progressively higher-order levels of dynamics, thereby introducing an essentially open-ended possibility of producing causal consequences that wouldn't tend to arise spontaneously. That is, we can begin to discern a basis for a form of causal openness in the universe. To frame these insights in somewhat more enigmatic and cosmic terms, we might speculate that whereas the conservation laws of science tell us that the universe is closed to the creation or destruction of the amount of possible "difference" (the ultimate determinate of what const.i.tutes ma.s.s-energy) available in the world, they do not restrict the distributional possibilities that these differences can a.s.sume, and it is distributional relationships which determine the forms that change can take.

Having said this, we must keep in mind that the relationship between these different dynamical paradigms of work is complex. In one sense, they are but a.n.a.logues of one another, and it is important that they not be confused. They involve very different substrates and conceptions of what is spontaneous or not, very different conceptions of what const.i.tutes the differences that generate spontaneous change, and quite distinct notions of what const.i.tutes a locus or system and how they may be linked together. And yet they are also more than mere a.n.a.logues of one another. They play the same functional role at each of dynamical level, and more important, they are hierarchically interdependent and nested in the same way as are these three levels of dynamics. Teleodynamic work is dependent on morphodynamic work is dependent on thermodynamic work. At each level there is a cla.s.s of orthograde and contragrade tendencies in which contragrade change can only be effected by pitting orthograde processes against each other.

The change of a thermodynamic system toward a state of increased entropy, if totally unperturbed by extrinsic influences, is the cla.s.sic and most basic example of orthograde change. In the real world, of course, nothing is ever completely isolated, and no physical system remains forever unperturbed by external influences (though it is often argued that this is true of the whole universe, even if it is finite). This means that thermodynamic orthograde change is often resisted, modulated, or reversed, depending on the relative degree and duration of interaction with other contrary constraints or processes. In any context, orthograde processes will continue until they reach a state in which there is symmetry in the probable directions of change, or until the supportive conditions change. This dynamical terminus of an orthograde process is its attractor, which may or may not be a quiescent state.

In chapter 8, we extended the concept of orthograde dynamics to include intrinsically asymmetric forms of dynamics that arise in persistently far-from-equilibrium contexts. Such morphodynamic change is comparable in form to the orthograde asymmetry described by the second law in three important respects: (1) it exhibits a highly probable, characteristic, intrinsically generated asymmetric bias in the way that global properties change; (2) the direction of this change converges toward a common attractor irrespective of initial conditions; and (3) this asymmetry is dynamically supervenient on a balanced (i.e., symmetric) lower-order contragrade dynamic (work).

Recall that thermodynamic orthograde processes depend on incessant lower-order molecular interactions whose contragrade effects (work) are time symmetric, and in an isolated system the total energy of the system (e.g. measured as its specific heat) remains constant3 throughout the orthograde change (and in this sense is balanced), though entropy increases. This incessant lower-order contragrade activity guarantees constant change from state to state, but the asymmetric directionality of the trajectory of global change is not a direct consequence of this work. The constant lower-order contragrade activity determines constant change, but not its directionality. This global asymmetry has a "formal" origin, because it develops within the biased geometry of the s.p.a.ce of possible trajectories of change.

FIGURE 11.2: Countercurrent exchange demonstrates how formal constraints can be harnessed to do thermodynamic work. By causing media like coolant liquids (in a heat exchanger), or blood and environmental water (in fish gills), to flow in opposed directions, the asymmetries created can locally drive the system far beyond pa.s.sive thermodynamic equilibrium (e.g., pa.s.sive diffusion or parallel flow, above), so long as the movement continues. Though most naturally occurring countercurrent processes involve fluids and heat transfer, or chemical diffusion (as in fish gills and kidneys), this is a generalizable relationship. It can apply to any process involving entropy increase/decrease, including information processes.

An interesting example of the relationship between formal constraints and thermodynamic work is provided by processes that involve counter-current diffusion (also often described as counter-current flow; see Figure 11.2). This is a common mechanism found in the living world for driving systems beyond the point of thermodynamic equilibrium using oppositely directed fluid flow. It is characteristic of the ways fish gills extract oxygen from water, kidneys extract metabolites from blood, sea turtles cool themselves, and many birds regulate their core body temperature via blood flow in their legs. It is also an important trick used in engineering applications, such as in cooling systems for nuclear reactors and desalination processes.

Similarly, morphodynamic orthograde processes are also dynamically supervenient on incessant lower-order stable contragrade dynamics: the balance (symmetry) between incessant extrinsically introduced forms of destabilization, which introduce constraints, and the incessant spontaneous (orthograde) dissipation of these constraints. This const.i.tutes a dissipative system, in Prigogine's terms. An asymmetric trend toward amplification and propagation of constraints to higher levels of scale can result if these constraints are not dissipated as quickly as they are introduced. This asymmetry is not, however, simply an elaboration of this lower-order symmetric dynamic, but rather again reflects global formal biases of the available trajectories of global property change. The asymmetry arises under these conditions as constraints compound non-linearly. This occurs because any dynamical option that is impeded from occurring due to the introduction of extrinsic constraint cannot lead to increased dynamical variation via any further contragrade interactions that the constrained region of the system has with others. Thus, as long as new extrinsic constraint is introduced faster than it is dissipated (e.g., due to incessant disruption, as in heating), subsequent stages of change will exhibit progressively reduced ranges of variation. In other words, they will self-simplify and become more "orderly."

Teleodynamic orthograde processes are more complex because they dynamically supervene on morphodynamic processes. Nevertheless, these general principles still apply with respect to their lower-order dynamical support. Teleodynamic patterns of change emerge from the contragrade interactions between morphodynamic processes. Thus in the model autogenic system described in the last chapter, two morphodynamic orthograde processes with entirely different attractor dynamics-autocatalysis and self-a.s.sembly-interact, and as a result do work with respect to one another. Importantly, this interaction occurs in both the thermodynamic and morphodynamic domains. a.s.suming that both processes are thermodynamically orthograde in supportive conditions, each can only be sustained if there is constant work to maintain the thermodynamic imbalance that supports this asymmetry. In the case of each morphodynamic process, both autocatalysis and self-a.s.sembly require continuously maintaining locally high levels of substrate molecules. This could, for example, be extrinsically provided in laborator