INDIVIDUATION.

The central thesis of this chapter is that the core property which links the selves of even the simplest life forms with that seemingly ineffable property that characterizes the human experience of self is a special form of dynamical organization: teleodynamics. What has not been made explicit, however, is that all teleodynamic processes are implicitly individuated, that is, they are closed in a fundamental sense with respect to other dynamical features of the world. This special form of closure is reflected in the phrase "reflective individuation" used earlier, which I introduce to explicitly avoid using the prefix "self-" (as in self-maintaining, self-organizing, self-reproducing) in defining the fundamental organizational logic of self. It also provides a more accurate and more explicit unpacking of the form of this relationship by implicitly invoking a mirror a.n.a.logy. Indeed, it might even be a.n.a.logized to mirrors reflecting one another and thereby creating a way of containing a perpetually reproduced image.

Another way to look at this closure is that it is the physical a.n.a.logue of a logical type violation in logic, such as that exemplified by the famous liar's paradox: "This statement is false." The concept of logical type is basically the relationship that exists between a whole and a part, or a cla.s.s and a member of that cla.s.s. The sentence above is interpreted as a composite whole by virtue of the interpretation of the parts (words) with respect to one another. But the component phrase "This statement" refers to the whole of which it is a part and so a.s.signs the value of being false to the whole. The familiar result is that with each reading, one is enjoined to reread and reinterpret the sentence. This self-undermining statement can never be a.s.signed a final interpretation because the whole is constantly changing the reference of the parts and the parts are constantly changing the reference of the whole. The critical feature is of course that the whole is represented and reflected in the part. This is possible in a sentence because of the nature of symbolic representation. Since the reference of this phrase is not physically a.s.signed, but can apply to any object, there is nothing restricting its application to the sentence of which it is a part. In an organism, the relationships are a.n.a.logous. Each component function contributes to the continued existence of the whole and the whole is required to generate each component function. In this sense, each functional feature embodies a trace of the whole individuated organism, reflecting the coherent influence of the whole and contributing to its future coherence. This is the essence of reflexive individuation: a compositional synergy, functioning to determine its const.i.tuents in a way that both embodies and reinforces their synergistic relationship. The whole/part hierarchy thus becomes inextricably tangled.

If this tangled dynamical logic is fundamental to organism self, then it should apply equally well to forms of self at other levels too, including human minds. However, these parallels need not be simple and direct. This is because the neural teleodynamics of mind is hierarchically nested within the vegetative forms of teleodynamics that const.i.tute simple cells and multicelled animal bodies. Mental self is both a higher-order form of self, and at the same time subordinate to the organism self. To rephrase this using the technical distinctions we developed in our earlier discussion of emergence, neural teleodynamic processes dynamically supervene on what might be called the vegetative teleodynamics of the organism, and organism self dynamically supervenes on cellular self, though in different ways. Even the simplest bacterium is organized as a self, with emergence of ententional properties and possibly a primitive form of agency in the ability to propel itself (e.g., with a flagellum). However, the intentionality and subjectivity which exemplifies species with complex brains involves higher-order properties that emerge from a distinctive higher-order level of reflexive dynamics const.i.tuted by the interactions among the vast numbers of vegetative teleodynamical selves that are neurons.

This vegetative form of self exhibits some very general organizational principles, which are necessary and sufficient to explain the emergence of end-directed and informational relationships from simpler components lacking these attributes. But why must a system exhibiting teleodynamic organization be an individuated unit? This is not the case, for example, with thermodynamic or morphodynamic processes. They may be physically boundable, but it is only a contingent boundedness. Organisms also typically are materially bounded, but in addition their boundedness is intrinsic to their dynamics. It is implicit in the closed reciprocity of organism functions. As a result, physical boundedness may be ambiguous in many organisms or at certain stages in their development. The necessary interdependence of organism structures and processes implicitly determines an individuated system. This intrinsically generated individuation also determines a self/other distinction, which need not be an artifact of material discontinuity or containment within a membrane. So, while biological self and other are commonly distinguishable on either side of some physical interface-a membrane, cell wall, or skin-these are only convenient means of physically embodying this otherwise purely dynamical distinction.

SELVES MADE OF SELVES.

The primacy of dynamical boundedness over material boundedness becomes more evident as we ascend in level of biological and mental selfhood. And when we reach the level of a human subjective self, even the concept of boundary makes no sense, despite the fact that the reflexive closure of subjectivity is unambiguous.

With the evolution of ever more complex forms of organisms, the recursive complexity of self has also necessarily become more highly differentiated. Multicellularity offers a case in point. Multicelled organisms all develop from single-cell zygotes. This initial single cell self-reproduces vast numbers of genetically identical offspring cells-self copies-that at early stages of embryogenesis still retain some degree of autonomy. By remaining in contact with one another, maintaining a molecular and topological trace of their relative positions, and sending identifying molecular signals to one another, this local society of cellular twins begins to differentiate and cellular-level autonomy decreases. Each gives up progressively more of its autonomy to the growing whole embryo. Cellular-level dynamical reciprocity is thus progressively simplified, and intercellular forms of reciprocity begin to take its place. One level of individuation is sacrificed so that a higher level of compositional individuation can form. Again, what determines this transition in level of self is not the generation of an exterior skin, but rather a change in dynamical reciprocity. In this way, local cl.u.s.ters of similarly differentiated populations of cells, forming organs, begin to play an a.n.a.logous role to the various interdependent cla.s.ses of molecular processes within a cell. Developing a skin or bark or sh.e.l.l serves a different purpose. It protects the now more distributed reciprocity from extrinsic disruption.

Of course, protective encas.e.m.e.nt is a nearly ubiquitous feature of organism design-whether for viruses, bacteria, amoebas, plants, or animals-because reciprocity is always potentially disruptable by the intrusion of extrinsic influences. Dynamical reciprocity, without which there can be no teleodynamics, is therefore forced to persist in the context of mutually exclusive requirements with respect to the external non-self world. Because it is dynamical, it is dependent upon extrinsic energy and material; because it is a form of reciprocal dependency, it is dependent upon being isolated from aspects of the non-self world that might disrupt this delicate reciprocity. For this reason, the evolution of highly specialized bounding surfaces, which are also selectively permeable, is one of the most important loci of evolutionary differentiation. Whether cell membranes with their elaborate constellations of receptor and transporter molecular complexes linking inside and outside, or animal body surfaces with their mult.i.tudes of receptor types and motor organs enabling them to find food and mates or avoid predators, this interface is where the self-other distinction mostly gets negotiated. It is just not what determines self. So, although our attention tends to be drawn to these boundary interfaces when focusing on what const.i.tutes self, this interface is better understood as an adaptation to better preserve the self already there.

The critical but contingent relationship between selves and physical boundaries complicates the identification of biological self. Individual cells in your body are each physically bounded by a membrane within which the reciprocal processes essential to self-maintenance take place. But they are also critically incomplete as well. Unlike free-living cells, somatic cells have become co-dependently parasitic. Their integration into a larger self has enabled them to forgo the adaptations to the variable environments that an autonomous cell must encounter. One of the most fundamental is reproduction. Many of the cell types in your body are terminally differentiated, which means they cannot be modified to be any other kind, and many of these are non-reproductive, that is, they will never again divide. Apparently, the molecular apparatus that must be maintained in order to support this ma.s.sive metabolic transformation would be a significant impediment to the terminally differentiated function of this cell. Thus differentiated neurons and muscle cells, for example, have either redeployed some of this apparatus for other purposes or suppressed its generation altogether. They still retain one feature of organism self but have sacrificed another. The metabolic self (so to speak) of these cells is bounded in the cell membrane, but that feature of biological self which maintains this ultimate hedge against the ravages of entropy-duplication-has been shifted to the whole organism.

Of course, the locus of individuated self is also ambiguous in other biological contexts. Lichens are the result of a co-dependent symbiosis between a fungal and algal species, neither of which can exist without the other. Their reciprocal co-dependence is also guaranteed by the degradation in each species of adaptations for autonomous living. The lichen is therefore an individual defined by this reciprocal co-dependence, and not by genetic h.o.m.ogeneity. This is not really an exceptional case in biology: all eukaryotic organisms share the trait in a very basic form, including us. The mitochondria-the tiny bacterial-shaped loci of oxidative metabolism in every plant and animal cell-are the distant ancestors of a once free-living bacterial organism that evolved to become an endo-symbiont to another larger-celled form. This is not merely reflected in their characteristic beanlike structure, but also in the fact that they contain and reproduce within each cell using an enclosed circular bacterial chromosome, whose DNA sequences place all mitochondria (from plant or animal) well within the bacterial lineage. Thus we too are characterized by dual genomes with separate origins.2 Our current cellular-level individuation is in this way characterized by a synergistic reciprocity that developed between these once separately closed forms of teleodynamic unities.

Such self-components (somatic cells) integrated into higher-order self units (organisms) are each internally sustained by highly complex networks of synergistically reciprocal morphodynamic processes. They are individuated selves by virtue of the level at which there is circular closure of this morphodynamic network. This closure not only creates a distinct level of individuation but also a distinct locus of teleodynamics. I have designated each of these partially autonomous, partially dependent levels of individuated teleodynamics a teleogen, as distinct from an autogen, which is fully autonomous and simple. In the complex biological and mental worlds of the present state of life on Earth, teleogenic structures are the norm, as evolution tends to generate highly entangled forms of teleodynamic processes, given sufficient time. This is not because evolution is itself a kind of final causal process, but rather because there is no limit to how teleodynamic processes can become entangled with one another. This also leaves open the possibility that the different teleogenic units const.i.tuting a complex organism can come into conflict.

Tracing the way that higher-order forms of organism evolve was the topic of the last part of the previous chapter. There we noticed that each shift to a higher-order form of individuation was typically a.s.sociated with loss of lower-level autonomy and the serendipitous self-organization of higher-order synergies. What evolves in these cases are multiple levels of self-built-upon-self, with the higher-order reciprocities and synergies emerging as a result of the degradation of certain self-features of the lower-level component selves. Because of this, the remaining lower-level teleodynamic characteristics are those most consistent with the higher-order teleodynamics. Teleogens composed of teleogens in which lower-level degradation is present are thus common. In the case of mitochondria and their "host" eukaryotic cell, for example, both the nuclear genome and the mitochondrial genome are partially "reciprocally" degraded, so that neither can persist without the other, but for most functions they remain relatively modular and individuated in their functions. This tendency for modularity, implicit in the nature of teleodynamics, is what makes complex multi-leveled selves possible. Without it, a complexity catastrophe would be inevitable-too many components, needing to interact in a highly constrained manner in a finite time, despite vast possible degrees of freedom-setting an upper limit on the complexity of self.

Because self is defined by constraints, not by particular material or energetic const.i.tuents, it can in principle exist in highly distributed forms as well. Thus a termite colony may involve millions of individuals and many generations of turnover, but only one reproducing queen. In reproductive and evolutionary terms, only the whole functioning colony is a reproducing organism, and the "self-" of the colony is quite ambiguously bounded in s.p.a.ce or material and energetic usage, especially when vast numbers of individuals are foraging independently in the surrounding territory. Thus self, too, may become highly diaphanous at higher emergent levels, with more variable degrees of individuation and correlation with physical boundedness emerging at progressively higher-order levels of self.

These many features of teleogenic hierarchies are of course relevant to mental selves. Mental self is not a composite self in the same sense as is a multicelled organism body. Brains are composites of vast numbers of highly interdependent cells-neurons and glia (the "support cells" of the nervous system). And each neuron is extensively interconnected with other neurons. Brains evolved to enable multicelled animal bodies to move from place to place and intervene in the causality of extrinsic conditions, thereby altering the body's relationship to its local environment. Brains are in this respect part of the boundary that mediates between the teleodynamics intrinsic to the organism and the dynamics of its external world. So, in certain respects, it might be appropriate to compare brains to the specialized molecular pores, signaling molecules, and actuators (like flagellar motors) that span the membranes of cells and mediate inside/outside relationships. They are part of the boundary interface that continually creates and demarcates self. But unlike these cellular-level interface mechanisms, brains mediate the self/other distinction by using the dynamics of self itself.

NEURONAL SELF.

Despite the current focus on consciousness, mental self is subordinate to and nested within the more general form of self that is characteristic of all living organisms. This is made self-evident by the fact that although the unconsciousness of anesthesia can temporarily interrupt this experience, the continuity of self persists across such gaps, so long as the body remains alive and the brain is largely undamaged. Even where there has been profound memory impairment, as in victims of Alzheimer's disease or hippocampal damage, though ident.i.ty may be compromised, the sense of consciousness and agency still persists. Our worries about death, and our comparative unconcern with the state of unconsciousness, or even amnesia, thus provide evidence that we intuitively judge the self of Descartes' cogito to be subordinate to the self of life in general. Nevertheless, the experience of this self is the result of the way that organism self pervades and organizes neural processes.

Brains are organs which evolved to support whole organism functions that are critical to persistence and reproduction. They are not arbitrary, general-purpose, information-processing devices. Everything about them grows out of, and is organized to work in, service of the organism and the teleodynamic processes that const.i.tute it. Animal physiology is organized around the maintenance of certain core self-preservation functions on which all else depends. Critical variables-such as constant oxygenation, availability of nutrients, elimination of waste products, maintenance of body temperature within a certain range, and so forth-all must be maintained, or no other processes are possible. Sensory specializations, motor capabilities, basic drives, learning biases, emotional response patterns, and even the structure of our memories are ultimately organized with respect to how they support these critical core variables. Additionally, the reproductive capacity, which is a ubiquitous correlate of having evolved, is also part of the larger teleodynamic background that gets re-expressed as neurological self, and similarly insinuates biases into these basic neuronal processes.

So, as a critical mediator of the self/other interface, a brain must be organized around this constellation of processes that const.i.tutes the teleo-dynamics of organism self. The reciprocity and interdependence of the various physiological processes that sustain the organism thus get recapitulated as core organizing principles const.i.tuting neurological function. For the most part, however, the monitoring and regulation of these core bodily functions is automatic, and maintenance of them is relegated to the physiological a.n.a.logues of thermostats and guidance systems, that is, teleonomic mechanisms. Nevertheless, these processes must also be re-represented to a higher-order adaptive process in service of mediating action within and with respect to the environment. So the vegetative self of the organism must be triply embodied: first in the cellular-molecular relationships that form the most basic organism teleodynamics; second in higher-order automatic neuronal regulatory systems; and third in the way that changes in these regulatory processes or in the values of the underlying physiological variables alter the yet higher-order adaptive activities of the brain that mediate between the vegetative teleodynamics and the world of extrinsic possibilities. It is this third level of re-re-represented organism teleodynamics that is the substrate of the subjective self.

Neural self is further complicated by virtue of its role as inside/outside mediator. Brains evolved in animals to generate alternative virtual worlds and virtual futures. In order to be able to favorably change the contexts within which the organism finds itself, brain mechanisms must be able to model an organism's surrounding conditions and also to model the possible outcomes of acting to modify those conditions. To do this, an animal must be capable of simulating more than just the moment-to-moment changing relationships between internal dynamics and external conditions. It must also be able to predict the possible consequences of its own interventions. Of course, depending on the predictability of the environment and the complexity of the organism, this modeling capacity may need only a small constellation of relatively fixed alternative responses. But where both the environment and the organism are complex, this may require elaborate, open-ended means for generating conditional scenarios. In such complex cases, it is critical to also include the capacity to simulate the teleodynamic processes that produce these adaptive behaviors and generate these scenarios. In social species, this can reach quite convoluted levels of detail. But because of the intrinsically recursive nature of behaviors that change external conditions, even modestly complicated brains need to include some self-referential features.

Thus it is inevitable that having a brain should also entail the generation of a form of teleodynamic relationship that is partially organized with respect to itself as environment. This higher-order form of teleodynamic causal circularity creates an entirely novel emergent realm of self-dynamics. It helps to explain why being an organism with a complex brain inevitably includes a doubly reflexive organization with respect to itself. This self-referential elaboration of teleodynamic logic is therefore one level more convoluted than any autogenic process. This predictive and projective function creates a more open-ended form of individuation, which must be able to generate a diverse range of possible self-environment scenarios. This premium on the flexible generation of possible futures (and thus purposes) makes the term teleogenic particularly apt for such a level of teleodynamic process.

SELF-DIFFERENTIATION.

The development of self emerges in a process of differentiation. The self that is my entire organism did not just pop into the world fully formed. It began as a minimal undifferentiated zygote: a single cell that multiplied and gave rise to a collection of cells/selves which by interacting progressively differentiated into an embryo, a fetus, an infant, a child, and eventually an adult organism. Similarly, it is difficult to imagine one's subjective self just popping into existence fully differentiated. By the very nature of its thoroughly integrated and hierarchically organized form, it would seem to demand a bottom-up differentiation in order to arise. But if so, then this also suggests that the moment-by-moment subjective sense of self, as well, is only the most differentiated phase in a sort of micro-differentiation process happening continuously and in the s.p.a.ce of seconds.

This is consistent with introspective experience. There are times when we are only dimly aware of our memories, our intentions, or the surrounding stimuli. For example, on waking from a sound sleep, we experience a sort of ascent into differentiated awareness, and a graded "booting-up" of our more critical and goal-directed faculties. The early phases of this process involve only the incorporation of basic physiological and somatic factors into awareness, and are in this sense largely undifferentiated forms of self. Only after we are fully awake, beginning to a.s.sess our immediate surroundings, and remembering the habitual activities, demands, and plans that attend this time of day and social condition, does our mode of self become fully differentiated. But even as one experience gives way to another and one activity gives way to another in the course of awake experience, each new focus of attention and intention must differentiate anew to replace a former, more or less differentiated phase of awareness. In this sense, each change of focus is a mini-recapitulation of waking anew. There is no deep discontinuity because the least differentiated level of self changes little from moment to moment; but even so, one characteristic of severe anterograde amnesia (in which the victim cannot consolidate new memories) is a constantly recurring sense of "just now being awake for the first time."3 In summary, then, it is my hypothesis that the subjective self is to be identified with this locus of neurological teleodynamics, which is variably differentiated at various stages of life and alertness, and which in its most differentiated form can include itself as recursively represented and projected into a simulated virtual world. Because teleodynamic processes depend on lower-order morpho- and homeodynamic processes, these too must be taken into account in a full theory of neurologically generated subjective experience (a topic addressed in the next chapter). And because these are emergent dynamical processes, all rely on constant exuberant metabolic activity, and are always in some stage of differentiation or dedifferentiation that takes time to unfold.

Additionally, as is also the case with simple autogenic systems, because a teleodynamic system is self-generating, self-reinforcing, and self-similarity maintaining, it can serve as a reference dynamic against which all other dynamical tendencies and influences that affect it are contrasted. Their distinction as non-self is implicit in their tendency to initiate the generation of contragrade dynamical teleodynamic processes. This orthograde/contragrade teleodynamic distinction thus defines the dynamical boundary of self. The minimal form of this teleodynamic organization is also at the core of all neurological functions, where it can be as undifferentiated as a simple autogenic process, and likewise provides the most basic self/non-self distinction by virtue of the contragrade dynamics of adaptive physiological processes. However, this teleodynamic organization is progressively re-represented and differentiated within brain systems that are progressively more entangled with external receptors and effectors. Because of the intrinsically conservative self-similarity-maintaining nature of teleodynamic processes, each level of teleodynamic activity is an effective locus of self, and is that with respect to which otherness is implicitly marked. Each locus of teleodynamic neural activity is also the dynamical "substrate" which differentiates and "evolves" moment by moment with respect to a complex environment of sensory "perturbations," present and remembered.

In this sense, the unity of consciousness may be more mercurial than commonly imagined. Different loci of teleodynamic activity at the same level may develop in parallel and come into compet.i.tion as they differentiate and propagate to recruit additional neurological resources. Thus the differentiation of self may also involve a Darwinian component (also discussed in the next chapter). The often quite sophisticated alter egos that we find ourselves interacting with in dreams, and who can often act in unexpected ways, suggests that in this state of consciousness, there may be no winner-take-all exigency. In dreams, all action is virtual and thus need never be finally differentiated; but when awake and enjoined to behave by real-world circ.u.mstances, action depends on a winner-take-all logic to produce a single integrated action. So the unity of waking conscious experience may in this respect be a special case of a more pluralistic self-differentiation process.

THE LOCUS OF AGENCY.

Perhaps the most enigmatic feature of self is its role as agent: as the locus and initiator of non-spontaneous physical changes in the world around it. This is often confused with the age-old problem of explaining the possibility of free will in a deterministic world. However, it is different in a number of important respects. Self as agent is indeed what philosophers struggling with the so-called free will paradox should be focused on, rather than freedom from determinate constraint. Determinate causality is in fact a necessary condition for the self to become the locus of physical work. An agent is a locus of work that is able to change things in ways more concordant with internally generated ends and contrary to extrinsic tendencies.

Approaching the self-dynamics of mental agency using this same framework, we need to look to the closure of the teleodynamic constraint generation process for the locus of the capacity to do self-initiated work. For the simplest autogenic process, this closure is const.i.tuted by a complex synergy between morphodynamic processes that makes possible both the generation of constraints and also their maintenance and replication. The teleodynamics that distinguishes the agency of organisms from mere physical work is a product of this closed reciprocity of form- (i.e., constraint-) generating processes. Specific forms of work are made possible by the imposition of specific forms of constraint, and the way this channels spontaneous change, via the expenditure of energy. So this defining dynamic of organisms amounts to the incessant generation of the capacity to perform specific forms of work to alter the surrounding milieu in ways that are determined by this locus of teleodynamics, irrespective of extrinsic causal tendencies. This persistent capacity to generate and maintain self-perpetuating constraints is therefore at the same time the creation of a locus of the capacity to do self-promoting work.

Evolution can be seen as a process that has vastly complicated both the nature of these constraints and the capacity to utilize them as means of initiating specific forms of work-work that aids the persistence and further evolution of these capacities. As new forms of teleodynamics evolve, they bring new capacities to perform work into being-new options, correlated with ever more diverse extrinsic influences. In this way, as evolution has given rise to organisms that embody vast webs of constraints in their internal dynamics, and specifically with ever more diverse means of interacting with their surroundings, they have increased the ways in which they are able to impose these or complementary constraints on the dynamics of external events and relationships. So, to again describe this in the negative, the evolution of increasingly complex forms of constraints-absences-has given rise to increasingly varied ways to impose constraints on the world with respect to these internal constraints. In this sense, the source of agency can be somewhat enigmatically described as the generation of interactive constraints which do work to perpetuate the reciprocal maintenance of the constraints that maintain organism self.

This view of self-agency, defined in terms of constraints, may seem counterintuitive because of our conviction that the emergence of life and mind has increased, not decreased, our degrees of freedom (i.e., free will). Increasing levels and forms of constraint do not immediately sound like contributors to freedom. In fact, however, they are essential. What we are concerned with here is not freedom-from, but freedom-to. What matters is not some disconnection from determinate physics, but rather the flexibility to organize physical work with respect to some conserved core dynamical constraints. This is not a breakdown of causal efficacy; in fact, just the opposite. Being an agent means being a locus of causal efficacy. Agency implies the capacity to change things in ways that in some respects run counter to how things would have spontaneously proceeded without such intervention. It also implies that these influences are organized so that they support the persistence of this capacity, and specifically the persistent self-generating system that is its locus.

What we are concerned with here is not some abstract conception of causality but rather the capacity to do work, to resist or counter the spontaneous tendencies of things extrinsic to the teleodynamics that creates self. The evolution of our higher-order capacities to create and propagate ever more complex and indirect forms of constraint-from the self-organizing processes that build our bodies to the production of scientific theories that guide our technologies-has in this way progressively expanded the capacity to restrict sequences of causal processes to certain very precise options. The ability to produce highly diverse and yet precise constraints-absences-thus makes possible a nearly unlimited capacity for selves to intervene in the goings-on of the natural world.

EVOLUTION'S ANSWER TO NOMINALISM.

But where does the "freedom" come from? Clearly, it is not freedom from deterministic involvement in the world, because this would preclude the capacity to do work. The answer to this question is ultimately related to the cla.s.sic realism/nominalism problem. To see this, it is first necessary to recapitulate bits of the critique of the Realism/Nominalism debate in philosophy that we initially dealt with in chapter 6.

The cla.s.sic nominalistic view is that being a member of a general type or a categorized phenomenon, such as a whirlpool or a member of a species, does not in itself have any causal significance over and above the unique and distinct individual physical attributes of that particular instance. Exhibiting a general form does not, according to this view, have any independent causal status. These generalities and similarity cla.s.ses are merely mental constructions, due to a necessary simplification that is implicit in the nature of mental representation.

By recognizing, however, that in the physical world the a.n.a.logues to general types can be determined negatively, that is, in terms of constraints, it becomes possible to understand physical causality negatively as well. This is because constraints can propagate through physical interaction. Or to state this negatively, degrees of freedom not actualized do not tend to propagate during physical interactions. Moreover, as the discussion of work has further clarified, all non-spontaneous change (efficient cause) is a function of the coupling of constraint and constraint dissipation (i.e., the release of energy). So general tendencies, understood in this negative sense, can indeed be the loci of physical causality, because work both depends on and propagates constraints. General properties understood in this negative sense are, then, the causal determinants of other general properties.

But there is a sense in which the emergence of teleodynamics has significantly augmented this efficacy of generals; and with the evolution of the teleodynamics of brains, this mode of causality has even begun to approach a quasi-Platonic abstract form of causal influence. Ironically, the structure of the nominalistic argument for the epiphenomenality of general types provides the essential clue for making sense of this augmented notion of causal realism. This is because both living and mental processes do indeed break up the physical uniqueness of physical processes into similarity cla.s.ses, due to the way they ignore details that are not relevant to the teleodynamic processes they potentially impact. But this simplification has causal consequences.

Seeing this cl.u.s.tering by simplification as necessitating epiphenomenality turns out to be both right and wrong at the same time. It is right when it comes to providing a reliable predictor of causal properties that are present irrespective of organism discernment. It is wrong, however, when one includes organism agency as a causal factor.

Both organism adaptations and mental representations necessarily ignore vastly many physical properties of things, fail to generate distinctions that correspond to physical differences, and lump together phenomena into general types for what might be described as pragmatic reasons. This categorical fallibility is even true for scientific knowledge. It is, for example, an historical commonplace for scientific investigation to reveal that phenomena once treated as members of a common general type are in fact derived from radically different origins and have radically different causal properties. So nominalism is, in this basic sense, well supported by the ubiquity of human error. It is captured in a catch phrase attributed to the humanistic psychologist Abraham Maslow (which he intended as a criticism of therapists trained in only one tradition): "If the only tool you have is a hammer, you will tend to treat everything as a nail."4 But physical responses, perceptions, and mental categories aren't merely pa.s.sive reflections on the world; they exist to structure adaptation to the world. For this reason, the mere resemblance of an object to a perceptual cla.s.s can be what causes that object to be modified in a particular way by an animal or person.

Consider, for example, the case of the boy selecting beach stones to skip across the surface of the water. The various stones he chooses may only have a few superficial features in common, but these features are what matter. The general attributes of being of a certain size, shape, solidity, and weight, as crudely a.s.sessed by comparison to a remembered type, thereby determine the production of another specific general attribute: the fact of skipping across the water. And so these attributes, const.i.tuting a vague general type maintained in the memory of a child, have played a determinate role in the relative location of stones with respect to the water's edge. Although we noticed this loophole in our earlier discussion, it is now possible to dissect the causal structure of this aspect of agency more carefully.

This augmented efficacy of generals-even such an abstract general as a "skippable stone"-is a feature that emerges with teleodynamics. Consequently, it is not just limited to minds; it is an attribute of life itself. Since the dynamical constraints that const.i.tute teleodynamic processes are themselves not specific individuals, but effectively the recursive perpetuation of specific absences, in whatever way they become insinuated into extrinsic physical processes they interact only with respect to properties that are relevant to this teleodynamic perpetuation. Other properties are incidental to the work that results. Thus plants will grow toward artificial light as they would toward sunlight, and children (and dieters) will consume artificially sweetened candy as if it contained nutritive sugars.

Brains have elaborated this causal realism to an extreme, and minds capable of symbolic references can literally bring even the most Platonic of conceptions of abstract forms into the realm of causal particulars. To list some extreme but familiar examples, a highly abstract concept like artistic beauty can be the cause of the production of vastly many chiseled marble a.n.a.logues of the human female form; a concept like justice can determine the restriction of movement of diverse individuals deemed criminal because of only vaguely related behaviors each has produced; and a concept like money can mediate the organization of the vastly complex flows of materials and energy, objects and people, from place to place within a continent. These abstract generals unquestionably have both specific and general physical consequences. So human minds can literally transform arbitrarily created abstract general features into causally efficacious specific physical events.

THE EXTENTIONLESS COGITO.

The picture of human "selfness" that emerges from this account, then, is neither eliminative nor preformationist. It is not all or none. It is graded in its level of differentiation, both in its initial development and in its moment-to-moment dynamics. There is no ghost in the organic machine and no inner intender serving as witness to a Cartesian theater. The locus of self-perspective is a circular dynamic, where ends and means, observing and observed, are incessantly transformed from one to the other. Individuation and agency are intrinsic features of the teleodynamics that brains have evolved to generate, because of the dynamical closure, constraint generation, and self-maintenance that defines teleodynamics. However, the neurologically mediated self exhibits a higher-order form of teleodynamics than is found at any other level of life. This is because the teleodynamics of brain functions that evolved to guide animals' locomotion and their capacity to physically modify their environments inevitably must model itself. The self-referential convolution of teleodynamics is the source of a special emergent form of self that not only continually creates its self-similarity and continuity, but also does so with respect to its alternative virtual forms.

Thus autonomy and agency, and their implicit teleology, and even the locus of subjectivity, can be given a concrete account. Paradoxically, however, by filling in the physical dynamics of this account, we end up with a non-material conception of organism and neurological self, and by extension, of subjective self as well: a self that is embodied by dynamical constraints. But constraints are the present signature of what is absent. So, surprisingly, this view of self shows it to be as non-material as Descartes might have imagined, and yet as physical, extended, and relevant to the causal scheme of things as is the hole at the hub of a wheel.

16.

SENTIENCE.

. . . sentience-without it there are no moral claims and no moral obligations. But once sentience exists, a claim is made, and morality gets "a foothold in the universe."

-WILLIAM JAMES, 1897.

MISSING THE FOREST FOR THE TREES.

Do you ever worry that turning off your computer, or erasing its memory, or just replacing its operating system could be an immoral act? Many serious scientists and philosophers believe that brains are just sophisticated organic computers. So maybe this should be a worry to them. Of course, maybe turning off the machine is more a.n.a.logous to sleep or temporary anesthesia, because no data need be lost or a.n.a.lytic processes permanently disrupted by such a temporary shutdown. But erasing data or corrupting software to the point that it is unusable does seem to have more potent moral implications. If you do sometimes contemplate the moral implications of these activities, it is most likely because you recognize that doing so could affect someone else who might have produced or used the data or software. The morality has to do with the losses that these potential users might suffer, and this would be of little concern were all potential users to suddenly disappear (though "potential" users might be an open-ended cla.s.s). Aside from the issue of potential harm to users, the nagging question is whether there is "someone" home, so to speak, when a computation is being performed-something intrinsically subjective about the processing of data through the CPU of the mechanism. Only if the computer or computational process can have what amounts to experiences and thus possesses sentience is there any intrinsic moral issue to be considered.

Indeed, could your computer's successful completion of a difficult operation also produce algorithmic joy? Does your computer suffer when one of its memory chips begins to fail? Could a robotic arm in a factory take pride in the precision with which it makes welds? If possible, would such a normative self-a.s.sessment require a separate algorithm, or would it be intrinsic to successful or unsuccessful computation of the task? One might, for example, create an additional algorithm, equipped with sensors to a.s.sess whether a given task was accomplished or not. In the case of failure, it could modify the operation to more closely approach the target result next time around. In fact, this is the way many "neural net" algorithms work. One such approach, called back propagation, incrementally modifies connection strengths between nodes in a network with respect to success or failure criteria in "categorizing" inputs. In such a computation, is there anything like intrinsic feeling? Or is this just an algorithm modifying an algorithm? The normative value that comes into existence with sentience is special. No other physical process has intrinsic ethical status, because no other type of physical process can suffer. Explaining the emergence of sentience thus must at the same time be an explanation of the emergence of ethical value. This is a tall order.

When it is carefully distinguished from any specific content that is the focus of mental experience, the background "feeling of being here" is sometimes just described in terms of a distinctive quality to experience-or quale, to use the technical jargon of philosophers. This quality has a perspective, an internal and private locus, a self-reference frame with respect to which non-self content is discriminated. Following William James, I will refer to this core feature of conscious experience as sentience. The term sentience derives from the Latin, and literally means "feeling."

In what follows, I will argue that the sentience of human experience is only one highly differentiated form of a far more widespread and diverse phenomenon, and that it has its roots in far simpler processes that only dimly resemble mental experience. I will argue that this applies even to the way that organisms are sensitive to influences from their environment, but not just in the way that most material objects can be modified by interaction with other objects and forces. Because organisms are teleodynamic systems, they do not merely react mechanically and thermodynamically to perturbation, but generally are organized to initiate a change in their internal dynamics to actively compensate for extrinsic modifications or internal deficits. Feeling is in this most basic sense active, not pa.s.sive, and is a direct consequence of teleodynamic organization because of its incessant end-directed nature. Since there can be higher-order forms of teleodynamic processes, emergent from lower-order teleodynamic processes, we should not be surprised to find that there are higher-order emergent forms of sentience as well, over and above those of the simpler cellular components of the body and nervous system.

The core hypothesis of this book is that all teleodynamic phenomena necessarily depend upon, and emerge from, simpler morphodynamic and homeodynamic processes. This implies that the complex intentional features that characterize our thoughts and subjective experiences must likewise emerge from a background of neurological morphodynamic and homeodynamic processes. Moreover, these lower-order subvenient dynamical features must also inevitably const.i.tute significant aspects of our mental lives. I believe that it is impossible to even approach issues of sentience without taking the necessary contributions of homeodynamic and morphodynamic aspects of mental experience and brain function into account. Once we do so, however, we will discover new ways of asking old questions about the relationship between minds and brains, and perhaps even find ways to reintegrate issues of subjective value into the natural sciences.

Reframing the concept of sentience in emergent dynamical terms will allow us to address questions that are not often considered to be subject to empirical neuroscientific a.n.a.lysis. Contrary to many of my neuroscience colleagues, I believe that these phenomena are entirely available to scientific investigation once we discover how they emerge from lower-level teleodynamic, morphodynamic, and thermodynamic processes. Even the so-called hard problem of consciousness will turn out to be reconceptualized in these terms. This is because what appeared to make it hard was our predisposition to frame it in mechanistic and computational terms, presuming that its intentional content must be embodied in some material or energetic substrate. As a result, the vast majority of descriptions of brain function tend to be framed in terms that not only fail to make the connection between the cellular-molecular processes at one extreme and the intentional features of mental experience at the other; they effectively pretend that making sense of this relationship is irrelevant to brain function. Descriptions of psychological and neurological processes remain unequivocally on one side or the other of this cla.s.sic divide-inescapably Cartesian.

Whether we imagine that we can reduce all intentional properties of mind to the "syntactic" features of neural "computations" and avoid directly addressing issues of mental content (e.g., the computational theory of mind), or describe them only in qualitative phenomenal terms (e.g., phenomenological theories) that merely a.s.sume the existence of what we hope to explain, we still end up ignoring a fundamental step in the ladder joining material and intentional properties. And although dynamical systems approaches do a better job of accounting for the non-mechanistic features of brain function, they too are at base founded on the a.s.sumption that the content of mental representations must be physically embodied by some specific neural substrate or else be epiphenomenal.

Current techniques for correlating mental experience with brain activity using in vivo imaging technologies (fMRI, PET, and MEG) have recently enabled us to correlate local metabolic levels with experiential phenomenology (i.e., performing a mental task). But precisely because these tools only provide an a.s.say of the regional thermodynamics of neuronal signaling activity, they must ignore the necessary contribution of any supervening levels of dynamics. In previous discussions of life-and of information more generally-we have outlined principled reasons to think that there can be no simple mapping between ententional and mechanistic (teleodynamic and thermodynamic) properties, even though whatever is ententional has emerged from a statistical mechanistic base. a.n.a.logously for neurological processes, it will turn out that these ignored levels of emergent dynamical processes are essential for bridging between brain processes and mental experiences.

The ententional features of life cannot be directly mapped to specific physical substrates or chemical processes, and yet they nevertheless dynamically supervene on processes involving these physical correlates. The dynamical requirements for generating ententional phenomena turn out to be highly constrained by the need to both generate and preserve the constraints essential to maintain these processes. So the functional properties of life only emerge when homeodynamic and morphodynamic processes are organized in precisely complementary and completely reflexive ways with respect to one another. If, as I believe, an a.n.a.logous emergent dynamic infrastructure is necessary to produce any ententional property, then it must also apply to the generation of mental intentionality. Moreover, traces of this hierarchic dynamical dependency should be reflected in the very structure of the experience of perceiving, thinking, or acting. In other words, we should be able to find the signature of these emergent dynamical levels in the details of brain process and in the very quality of subjective experience. The ultimate goal of this chapter, then, will be to trace the relationships between neurally embodied forms of homeo-, morpho-, and teleodynamic processes and various aspects of mental experience.

As in the case of organism evolution, it is only by examining the dynamics of these lower-level emergent processes that we will be able to adequately explain the sentience, representation, perspective, and agency that are the hallmarks of mental experience. By reframing the problem in these dynamical terms, I believe we will discover that rather than being the ultimate "hard problem" of philosophy and neuroscience, the subjective features of neural dynamics are the expected consequences of this emergent hierarchy. The so-called mystery of consciousness may thus turn out to be a false dilemma, created by our failure to understand the causal efficacy of emergent constraints.

Before we can attempt this reframing, however, we need to get beyond the idea that we might be able to explain mental processes in computational or simple dynamical terms. This is not a trivial demand, since the computational approach to mental processes and neuronal interactions has been enormously successful at clearing away many prejudices and confusions about the mind/brain relationship. The a.n.a.lytic tools it provided heralded a veritable quantum leap in experimental and theoretical sophistication in cognitive neuroscience. And dynamical systems approaches have begun to attract serious attention as well, especially as researchers struggle to deal with the enormously complex dynamics of signal processing in a richly connected network of billions of nodes. So, before we even begin a reframing of neural and mental processes in emergent dynamic terms, it is necessary to understand why the presumed direct correspondence between neural algorithms, signal dynamics, and cognitive operations inevitably fails to even address issues of intentionality, much less the mysteries posed by sentient experience.

The point of such a critique is not to argue that we should abandon the physical modeling of brain processes-indeed, careful a.n.a.lysis of the details of brain processes is of critical importance to an emergent dynamic approach as well-but rather to overcome the desire to directly map mental phenomena onto neuronal phenomena. As the logic of emergent dynamics has repeatedly shown, the physical const.i.tuents of living process are only relevant to ententional phenomena insofar as they contribute to the generation of teleodynamic processes. To make sense of conscious intentionality, and ultimately subjective sentience, we need to look beyond the neuronal details to explore the special forms of teleodynamic constraints they embody and perpetuate. I believe that only by working from the bottom up, tracing the ascent from thermodynamics to morphodynamics to teleodynamics and their recapitulation in the dynamics of brain function, will we be able to explain the place of our subjective experience in this otherwise largely insentient universe.

SENTIENCE VERSUS INTELLIGENCE.

Do computers think? Indeed, one still sometimes hears computers being called "thinking machines." The idea that cogitating is a form of computing has a long history, stretching back to the first half of the twentieth century. In fact, the term computer used to designate a person whose job it was to do the calculations supporting the accounting needs of large companies. Slowly, over the course of the last century, mechanical aids were supplied to take the drudgery out of this work, including mechanical adding machines, slide rules, and eventually digital computers. The mathematician who ultimately began the revolution that replaced people with machines was Alan Turing. The so-called Turing test-which he suggested in a paper in 1950,2 and which now bears his name-embodies the conjecture that if mechanized operations produce indistinguishable results from human operations on the same tasks, then the two are functionally identical. In the idealized test, we a.s.sess whether a computer can produce responses that are indistinguishable from those that a normal thinking person might produce under similar circ.u.mstances-for example, by posing questions for the computer to answer. If it can, then, presumably we have no justification to claim that these are fundamentally different processes.

While superficial operational ident.i.ty may not be sufficient to prove that the two processes are ultimately identical in some deeper sense, it nonetheless is sufficient to question the a.s.sumption of difference. Thus a Turing test questioner might not be able to tell if she is being answered by an algorithm or by another person. It also provides a way to a.s.sess cognitive power, because there are good reasons to believe that algorithmic complexity and reasoning difficulty are correlated. If some task requires more operations to complete, it should take more time and resources to complete, whether by computer or person. However, whether or not one accepts this as a test of machine intelligence, it is inadequate for testing sentience. Intelligence and information-processing power are probably not irrelevant, but they appear to be a.s.sessing something quite distinct from sentience. This suggests that a very different causal architecture is likely involved in the production of intelligent versus sentient processes.

With the explosive growth of computing power in the past few decades, we have come to take for granted that quite complicated and intelligent behaviors can be simulated by computation. a.s.sume, for example, that we had the support of ma.s.sive computational power and had compiled a ma.s.sive database of human mental and behavioral responses to all manner of adaptive challenges and cognitive tasks, and with a level of minute detail never before imagined. Using these data and some powerful predictive inferential tools, it should be possible to create algorithms capable of recreating in exquisite detail the kinds of responses characteristic of people adapting to normal life situations and solving difficult cognitive tasks. In other words, it is not an unrealistic thought experiment to imagine a computational system able to produce an output that can fool even the most sophisticated Turing tester: a Turing test a.n.a.logue to Deep Blue, the IBM chess-playing computer that matched the world chess grand master (described in chapter 3). Whatever problem-solving domain such a computing system faced, it would do as well or better than any human. Of course, this context could not differ too much from the domains from which its reference data were derived. But in these conditions, this golem would act as savvy as any human counterpart; and yet, I would argue, it would be completely insentient.

This is effectively an elaboration of an argument presented by the philosopher John Searle with his well-known "Chinese room" thought experiment.3 Both help to demonstrate the difference between merely intelligent behavior and sentient (or conscious) behavior. In Searle's imaginative scenario, a man who is unfamiliar with the Chinese language sits in a room that only gives him access to pages of Chinese characters, which are slipped in through a slot. His task is to match these to an identical string of characters in a book (or ma.s.sive database) that prescribes what characters to send back out, given that specific input. To those on the outside, the characters he produces const.i.tute a very sensible response to a question posed in Chinese by the characters fed to him; but to the man inside, it is a simple pattern-matching task. Searle argues that this shows how such a simulation can be produced in the absence of any cognizance or subjective involvement with the meaning of the task. If a man can produce such behavior "mindlessly," then when a machine does it in an a.n.a.logous way, we can reasonably a.s.sume that it is not aware of what it is doing either. Of course, the man's pattern-matching behavior is accomplished consciously; but since we can more easily imagine that pattern matching can be automated without any reference to the meaning or purpose of these activities, we intuitively appreciate the force of Searle's argument.

In this sense, a machine, or any highly sophisticated algorithm plus database that captures vast quant.i.ties of useful isomorphisms between inputs and outputs required to perform a given task, might rightly be described as exhibiting intelligence, irrespective of making any claim about whether it is sentient. Intelligence is about making adaptively relevant responses to complex environmental contingencies, whether conscious or unconscious. In this sense, we can a.n.a.logously describe the increasing complexity, flexibility, and fittedness of evolved adaptations as a kind of embodied intelligence. And, of course, many of our most reliable adaptive capacities are produced with only limited, if any, awareness and forethought. The dissociability of intelligent and sentient functions does not imply that a Turingesque a.s.sessment of sentience is impossible, however. It only suggests that it would be a very different sort of a.s.sessment than a Turing test of intelligence. So how might it differ?

Consider the possibility that we could extend the "test" to give us access to the "behavior" of the const.i.tuent operations that are used to perform the task, and not just the input-output relationship. What additional insight could be gained from access to the details of how the operation was performed? Searle's Chinese room scenario implicitly suggests that this should make a difference. It is precisely by reflecting on how the process is accomplished by the man in the room that Searle argues that we know that the apparent "interpretation" of the symbols is illusory, and that it is instead accomplished without an awareness of what is being done. Our knowledge of how the man in the room is performing the task clues us in to the fact that it can be done "mindlessly," so to speak-such as by a simple computer pattern-matching algorithm and look-up table linked to a large database. But if we can discern that this process has the structure of a blind mechanism, and can confidently conclude from this that it is not a sentient process, shouldn't there also be some criterion that would allow us to determine when the task is being accomplished with awareness of what is being done?

Possibly. Critics might argue that even if we can determine when the process is mindless, this does not guarantee that we could recognize what a sentient organization of the process looks like, or that the process could ever be done otherwise. Nevertheless, it seems reasonable to a.s.sume that if we can recognize the ways that the process can be performed mindlessly, at least we should be able to eliminate cases that are non-sentient, if they are not too complicated. It could be the case, however, that there will be ambiguous examples. Actually, I think that we need not worry about the issue of sentient processes being too complex to a.s.sess. I believe that it is not so much complexity that matters, but dynamical architecture. For the same reason that we were able to judge that a system as simple as an autogen exhibits ententional properties (and thus were not forced to deal with the complexities of living organisms), I believe that we will be able to distinguish the emergent dynamical architecture which produces the intentional properties of mental processes. Not only do I believe that we can discern whether a process is sentient, but once we understand the basic criteria for making such a judgment, I believe that we will also be able to make an a.s.sessment about how sentient this process is! The challenge is, of course, to develop a clear model of this emergent dynamical architecture.

The first step will be to outline the ways that current computational and dynamical systems models of mental processes exemplify mechanistic and morphodynamic models of living processes, respectively. As with the understanding of the deep logic of life, we will find that these ways of conceiving of cognition fall short of explaining or even acknowledging the reality of intentional properties. Ignoring the emergence of teleodynamic processes, such properties can only have epiphenomenal status in these theories. I will argue that we lack a naturalistic account of sentience because of a similar failure to understand how the teleodynamics of sentience emerges from morphodynamic and thermodynamic/homeodynamic processes of nervous system function. A final challenge of this a.n.a.lysis will be to show how higher-order levels of sentience emerge from, and depend on, these lower levels.