Thus far in part II of the book, we have reconsidered the metaphysics of causation, rejecting notions of causal responsibility as the core explanatory target of a theory of causation and constructing a richer and more objective Theory of Causal Significance. With a new theory of causation in hand, are we now in position to tackle the subject of phenomenal consciousness? It seems we are not, because if we try to explain p-consciousness using the just the kinds of entities featured in the Theory of Causal Significance as an ontological basis, we will inevitably fall victim to the arguments from chapter 2. Recall that in chapter 2, we used the toy physics of a pure Life world to argue that physicalism (and in general, any purely schematic system) cannot account for the kind of qualitative content found in subjective experience. The Theory of Causal Significance expands on the minimal causal content proposed to govern the pure Life world, and even posits the existence of non-physical aspects of causation (see ch. 11). However, it remains the case that our understanding of causation-- both the physical (effective) and non-physical (receptive) aspects-- is merely schematic, and thus exhibits the same fundamental flaw as that of the pure Life world. It seems to be a hopeless task to try to show how our new schema could entail phenomenal/experiential content any more than other schematic systems like physics could, and so we have not made any progress on this front. (Put another way, a world completely described by the Theory of Causal Significance still admits of the logical possibility of zombies.) To address this problem in full, it will not be sufficient to add further layers of schematic elements to our theoretical system, because the same underlying flaw will remain; we need to somehow 'get under' the problem by adopting a fundamentally new and different kind of approach. The Specter of Circularity Another problem, one entirely independent from considerations of p-consciousness, also lurks in the vicinity. This problem stems from the circularilty incorporated (whether explicitly or implicitly) into the definitions of schematic systems. Rosenberg identifies two kinds of such circularities: contrastive and compositional. The elements of a schematic system S are defined in terms of contrastive circularity if each element X in S is identified by its distinctness from, and external relationships to, all other elements in S. For example, the "on" and "off" properties of a Life world are defined in such a way; "on" is defined by stipulating that it is distinct from "off" and by exhaustively describing the manner in which the presence of an "on" property at time t can make a difference to the state of the Life world at time t+1, and likewise for "off." In general, all determinable families of effective properties can be described in terms of contrastive circularity: To describe effective properties, we need only stipulate that all identified kinds of effective properties are distinct, define a range of possible values for each kind of effective property, and describe the causal relationships that obtain among them. The elements of a schematic system S are defined in terms of compositional circularity if the definition of each element X in S positively presupposes the existence of some other element Y in S. For example, in the Theory of Causal Significance, effective properties are defined as properties that contribute to constraints on the joint state of the causal nexus to which they are bound, and receptive properties are defined as properties that bind individuals into a causal nexus such that their effective properties can feel each other's causal constraint. Here, effective properties are explicitly defined in part by reference to receptive properties, and likewise, receptive properties are explicitly defined in part by reference to effective properties. The circular nature of schematic systems raises questions as to how such systems could come to exist in the first place. If the only facts that are true about two entities X and Y is that they are distinct and enter into certain relationships with eachother, we are left with a bootstrapping problem: What is the basis upon which X and Y exist? What grounds their distinctness and their characteristic relationships? It seems we cannot appeal to any part of the system itself to solve the dilemma, as that only leads to a question-begging restatement of the problem or plunges us into an endless loop (X ground facts about Y grounds facts about X grounds facts about Y...). Breaking the Circle with Carriers That said, there must be some way to sidestep this puzzle-- after all, circularly defined systems can and do exist everywhere in the world around us. For example, we have touched on how Life worlds are defined schematically in terms of circular contrasts, and yet we can observe a Life world being implemented anytime we'd like, right in front of our eyes, be it on a computer or a checkerboard or some other medium. In fact, that very notion of implementation seems to hold the answer to our riddle. A Life world can exist because there already exist other systems whose own characteristic properties can stand in for, and thus instantiate or carry, the stipulative contrasts of a Life world. For instance, on a checkerboard implementation, we have two distinct kinds of tokens (red and black checkers) whose pre-existing physical distinctness can carry the stipulated distinctness of the "on" and "off" properties; we have a checkerboard whose demarcated columns and rows can carry the Life world grid which contains the "on" and "off" properties; and we have the intentions of human operators to carry the rules that govern the changing patterns of "on" and "off" properties over time. Likewise, on a computer implementation we have computational elements whose range of possible distinct states and regular functional behavior can play the carrier role for all aspects of a Life world. These systems can function as carriers for a Life world because (1) their properties are not exhaustively defined by the Life schema, and (2) their properties are such that, when organized in the proper fashion, they can exactly mirror-- and thus embody-- the schema definitive of a Life world. Condition (1) needs to be met in order to solve the bootstrapping problem presented by circularly defined systems, as discussed above-- we need to appeal to properties outside such circular systems in order to ground their fundamental schematic facts. For instance, within a Life world, "on" and "off" are stipulated to be merely distinct, with no apparent basis for this distinctness; they are not distinct because of some deeper facts about the Life world, but rather, on the level of the Life world, their distinctness is simply taken to be fundamental. Red and black checkers can carry "on" and "off" properties because at the appropriate level of abstraction, these checkers are not merely distinct-- rather, they are distinct because they are different colors. That is to say, their distinctness is not to be taken as a fundamental fact, but rather is grounded upon further facts within the physical system of which they are a part. The specific manner in which the checkers are physically distinct gives us a means with which to ground the purely abstract manner in which "on" and "off" are stipulated to be distinct. Condition (2) needs to be met just to insure that a Life world can, in fact, be functionally instantiated by the system in question. That is, in order for a system S to instantiate the relevant system of contrasts and relationships featured in a Life world, it must be the case that S has its own internal set of contrasts and relationships that can be manipulated in such a way as to mirror those of a Life world. For instance, to implement a Life world using checkers as the carriers for the "on" and "off" properties, we must have two sets of checkers that are distinct in some way (whether it be in terms of color, or size, or whatever) in order to establish a relationship of distinctness between them that can reflect the stipulated distinctness between "on" and "off." If we were to use a single set of indistinguishable checkers, we could no longer implement a Life world, because the required distinction between "on" and "off" properties could not be carried. Saying that a Life world could not be carried in this case amounts to saying that a Life world simply could not exist, given these circumstances. At this point, it will be useful to introduce some definitions from the text. We can now summarize and generalize the above discussion, incorporating the new definitions, as follows. Properties that are intrinsic to a schematic system S present us with problems of intelligibility; it seems that such a system inevitably incorporates circular definitions on some level, which raises questions about how such circularly defined entities or properties could come to exist in the first place. In general, we can solve this problem by appealing to a wider system T that carries the conceptual schema of S by providing facts that can ground the pure contrasts found in S. In order for T to be an appropriate carrier for S, it must be the case that T has some set of properties that is extrinsic within S, such that these properties features internal contrasts capable of instantiating the more abstract internal contrasts of S. If there is no appropriate carrier for S, then S simply cannot be instantiated-- i.e., it simply cannot exist. Also, it may be the case that T itself is just another system of circular relations, in which case there could be yet another system U that carries T, and so on. Rosenberg argues that instances of such schematic systems that partially or completely incorporate circular dependencies are plentiful in the world around us, and cites computer programs, games of chess, economics, psychology, and biology as examples. In each such system, the circularity involved turns out to be harmless, because it is carried by the internal contrasts of properties extrinsic within the system. For example, in economics we have a kind of circularity existing between the definitions of goods and services on the one hand, and producers and consumers on the other; this circularity can be carried in part by things extrinsic within economy but intrinsic to psychology, such as needs, desires, and beliefs. These psychological entities, in their turn, can be at least partially defined in terms of the functional roles they play in a cognitive context, and so they, too, are at least partially circularly defined. The circular dependencies of entities on the level of psychology can be carried by the underlying system of computational dynamics implemented by neurons in the brain, which is a system of properties extrinsic within the system of high-level psychological concepts; and so on. The general pattern is that each conceptual schema featuring circular dependencies is carried by a wider system of more fundamental entities and properties, which in turn features its own circularities, which are in turn carried by yet another wider and more fundamental schematic system, etc. (Note that this process of appealing to wider and more fundamental systems to ground the pure contrasts of higher-level systems looks a lot like reductionism.) So long as we have carriers readily available, the circularities involved at each level pose no deep problem. Physics: A Unique Problem However, we are faced with a unique dilemma once our path through wider and more fundamental schematic systems finally reaches the basic level of physics. Physics, too, is a conceptual schema incorporating circular notions of its fundamental entities and properties. (Recall the chapter 2 discussion about bare differences.) For example, to be an electron is just to be distinct from the other physical properties, and to play the functional role designated to electrons in physical theory; to be gravity is just to play the functional role of gravity; and so on. Insofar as physics is yet another functionally defined schematic system heavily incorporating circularities into its definitions, physics is also in need of carriers. But this observation raises a hitherto unencountered problem: It seems there are no properties extrinsic within physics that could plausibly play the carrier role. In response to this problem, one might simply deny that physics requires carriers at all, but this denial would violate theoretical standards of uniformity and intelligibility. For every other conceptual schema, we have solved the problem of circularity by reference to carriers, and indeed, these carriers were necessary for the instantiation of such systems to occur in the first place. To suddenly abandon carriers at this point would be an unprecedented maneuver, and worse, it would leave us with just the sort of problematic, unintelligible metaphysic we sought to abolish at the outset. We would be left again to puzzle over the nature of pure contrasts and bare differences; we would have to suppose that the ground floor of reality is composed of ontologically fundamental 'differences' that do not rest on any further categorical facts, relationships without anything substantive to be doing the relating in the first place, contrasts that are in fact not contrasts between anything at all. As Rosenberg puts it, "The idea seems to melt away before the mind's eye, like an echo issuing from no originating voice" (p. 236). The unintelligibility of such a view seems overwhelming, and whatever victory we may have gained by dodging the problem of circularity is a hollow, shallow, and unjustified one. Surely we can find a more satisfactory and methodologically consistent approach than this. Another approach might be to suppose that there is, in fact, some set of properties extrinsic within physics and intrinsic to a deeper conceptual schema, and that this schema itself is carried by yet another, deeper set of properties extrinsic within it, and so on ad infinitum ("it's turtles all the way down"). Such a view is perhaps less problematic than the previous one in theory, but it runs up against an empirical barrier: It seems that Planck's constant places a fundamental limit on how fine-grained the physical structure of the world is. If this is the case, then we can only posit so many further systems of properties until we hit an underlying bedrock, a fundamental physical floor that halts any further differentiation. Given that physics requires carriers, then, what kind of carriers can do the job? The dilemma presented for the prospective carriers of physics, per the arguments above, is that they must avoid posing the problem of circularity-- the problem of 'pure' relationships-- while simultaneously not needing carriers of their own. But every carrier we've investigated thus far has not conformed to these conditions; they have all tentatively posed the circularity problem, and only resolved it by appealing to further carriers at deeper levels. So physics needs fundamentally different kinds of carriers than we've described thus far; physics needs ultimate carriers, uncarried carriers.