- #1
- 2,285
- 2
With the basic concepts of the theory of causal significance in place, an important issue to address is how physical theory fits into the picture. Does a complete physical theory already encompass all the details of a theory of causal significance, or does it only describe a subset of it? If the latter, what aspects of causation belong to the domain of physics, and what aspects are beyond its scope?
In order to better frame these questions, Rosenberg identifies three basic levels of description of the world, in ascending order of metaphysical richness. The first is the Humean mosaic, which is a rather bare description of property instantiations across space and time. If we imagine the trajectory of a ball that is tossed upwards and then falls back to the ground, a description of this event on the level of the Humean mosaic would just include facts about the ball's spatiotemporal coordinates along the trajectory and its properties (mass, elasticity, etc.) at each such coordinate. The next level of description is the nomic mosaic, which introduces counterfactual supporting natural laws to account for observed patterns in the Humean mosaic. For instance, an account of our tossed ball on the level of the nomic mosaic would describe its trajectory in terms of dynamic laws of motion. This richer account introduces modal facts above and beyond the facts about the Humean mosaic, such as, "had the ball been thrown with a different force F, it would have taken trajectory T2 rather than the observed trajectory T1." Finally, the richest level of description of the world is a complete account of the causal mesh, including facts about receptive properties and causal laws. An account on this level of description would specify the receptive connections, effective properties, and causal laws that create a constraint structure on the ball's possible effective states and spatiotemporal locations. (In this case, the constraint structure would be such that it would filter out all possible states and locations of the ball except those which correspond to its actual, observed properties and trajectory through space and time.)
The basic form of Rosenberg's main argument in this chapter is
Facts about the nomic mosaic do not entail all the facts about the causal mesh
Note that the facts about the Humean mosaic do not strictly entail the facts about the nomic mosaic. Given a Humean mosaic H, it is not the case that the facts about H alone are sufficient to fix a unique set of nomic facts, because there are multiple kinds of nomic mosaics that are logically consistent with H. For instance, there could be two sets of natural laws whose predictions are in agreement with all events that actually occur in the history of a universe, but nonetheless disagree about what would have happened had some conditions been different.
Similarly, Rosenberg argues that the facts about the causal mesh are not completely entailed by the facts about the nomic mosaic, since a single set of dynamic laws is logically consistent with multiple kinds of causal constraint structures. Revisiting the toy physics of the Life world introduced in chapter 2, he explicitly details two distinct sets of receptive structures and causal laws, each of which reproduces the same dynamic Life world law describing the state of a cell at time t as a function of its state and the state of its neighboring cells at time t-1. Rather than reproduce Rosenberg's example, I will use a somewhat simpler case study here which should suffice to demonstrate the same general principles and results.
Suppose, as in chapter 10, that there is a world whose only elementary effective property is charge, which can take on a value of either + or -. (Charge is meant to refer to a novel, imaginary effective property, rather than the property called charge in physics.) Further suppose that time is quantized in this world, and that there is a dynamic law describing the behavior of charge over time as follows: the value of an instance of charge at time t is the opposite of its value at time t-1. Call this the law of charge oscillation. In order to show that the facts about this nomic mosaic do not entail the facts about its causal mesh, we need only demonstrate two distinct sets of causal facts which are consistent with the law of charge oscillation.
One set of causal facts (call it C1) consistent with the law of charge oscillation has already been presented in chapter 10. The receptive structure of C1 consists of a regular series of overlapping, two place, asymmetric level 1 receptivities binding instances of charge at each time slice. The causal law on these receptivities states that only an odd number of +s and -s, respectively, may be bound together in the same causal nexus. So, for a temporal sequence of charge instantiations ..., +0.1, -0.2, +0.3, ... occurring at successive times t1, t2, and t3, respectively, the receptive structure looke like the following:
..., [+0.1 => -0.2], [-0.2 => +0.3], ...
The causal law described above, when operating on two-place receptivities, forces two bound instances of charge to take on opposite values. The regularity of two-place receptivities binding every pair of temporally adjacent instances of charge ensures that every instance of charge at a time t will have an opposite value from its temporal neighbors at t-1 and t+1. Thus, the causal processes described in C1 create a world that supports the dynamic law of charge oscillation described above.
We can readily draw up other sets of causal facts that are equally consistent with the law of charge oscillation, even though they feature different receptive structures and causal laws. For instance, suppose we have the following causal conditions:
Causal law on level 1 nexii: There must be an even number of + charges bound within a level 1 nexus.
Causal law on level 2 nexii: There must be an odd number of + charges bound within a level 2 nexus.
Receptive structure: There is a regular pattern of receptive connections on levels 1 and 2, such that instances of charge are bound by overlapping, three-place receptivities into level 1 causal nexii. These level 1 individuals are bound by overlapping, two-place level 2 receptivities.
The receptive structure and causal law on level 1 allow three possible effective states for level 1 individuals: [-.+.+], [+.-.+], and [+.+.-]. Independently possible dual joint states of these individuals, given that they overlap at the endpoints, are:
1. [-.+.+0.n], [+0.n.-.+]
2. [-.+.+0.n], [+0.n.+.-]
3. [+.-.+0.n], [+0.n.-.+]
4. [+.-.+0.n], [+0.n.+.-]
5. [+.+.-0.n], [-0.n.+.+]
These independently possible joint states present a prior possibility space for the effective states of the level 2 individuals. Joint state (5) is ruled out directly by the level 2 causal law, since it has an even number of + charges. Furthermore, note that level 2 nexii featuring joint states (1), (2), and (4) entail the existence of another level 2 nexus featuring joint state (5), due to the stipulated regular, overlapping level 2 receptive structure. These joint states all have at least one level 1 member individual X featuring a - charge at one of its 'endpoints,' and this individual X must be receptively bound into another level 2 individual Y, due to the regular, overlapping level 2 receptive structure existing in this world. But because of the regular, overlapping receptive structure on level 1, Y can only have joint state (5).
For instance, suppose we have a level 2 individual I2.1 with joint state (4), i.e. [[+0.1.-0.2.+0.3]1.1.[+0.3.+0.4.-0.5]1.2]2.1. Because of the stipulated regularity of overlapping receptivities at levels 1 and 2, the existence of I2.1 entails the existence of another individual I2.2 of the form [[+0.3.+0.4.-0.5]1.2.[-0.5.?0.6.?0.7]1.3]2.2. The level 1 causal law forces I1.3 to take the form [-.+.+], which in turn requires that I2.2 takes on joint state (5), but this is not permitted by the level 2 causal law.
Because joint state (4) logically entails an impossible situation given these causal conditions, it follows that it is impossible for joint state (4) to instantiate in the particular kind of level 2 individuals discussed here. Similar reasoning applies to states (1) and (2); essentially, joint states (1), (2), and (4) are inconsistent with the total constraint structure enforced by the conjunction of causal laws and receptive structures in place in this world, so joint state (3) is the only possible configuration left for level 2 individuals. Because the level 2 individuals are all forced to take on joint state (3), the instances of charge are forced to take on a regular, alternating pattern of oscillating values, in agreement with the dynamic law of oscillating charges described above.
The alternate causal conditions described here and in the text are fairly complicated, and even have an air of strangeness or superfluity, in order to demonstrate more forcefully the point that a single set of dynamic laws can be consistent with a large range of causal circumstances. The general idea is that dynamic laws describe regularities of effective state instantiations that result from the operation of causal laws on existing receptive structures. Different receptive structures and causal laws can implement the same pattern of effective state instantiations, and so there could exist substantially different worlds that nonetheless instantiate identical nomic mosaics. Thus, it is not the case that facts about the nomic mosaic entail all the causal facts, but rather it is the other way around; a complete specification of the causal facts will already subsume all the facts about the nomic mosaic.
Physics describes only the nomic mosaic
In section 9.7, Rosenberg set out a lower bound for what aspects of the causal mesh physical theory describes: He argued that physical theory at least designates the low-level effective properties, and that its physical laws at least describe the regularities of effective property instantiations. In other words, physics is at least a theory of the nomic mosaic. I summarized this argument in the thread for chapter 9 as follows:
The guiding theoretical concern of physics is to explain and predict experimental results, and by a simplicity constraint, physical theory encompasses the least set of properties needed to explain these experimental results; to posit any properties superfluous to the explanation or prediction of possible experiments would be to posit properties beyond the guiding concern of physics, and thus ultimately beyond its scope. But in principle, an exhaustive theory of the nomic mosaic would already be sufficient to predict and explain the outcome of any experiment-- such a theory would tell us, in the context of the experimental design, how the effective properties of the target phenomena systematically covary with the effective properties of the measuring devices, and ultimately, with those of our own biological senses. In other words, any prospective facts to be added to physics above and beyond the level of the nomic mosaic would come with the cost of added complexity, and without the payoff of added experimental content-- by the simplicity constraint, such facts should not belong to the domain of physics. This amounts to saying that physics is at most a theory of the nomic mosaic, and in conjunction with the lower bound introduced above, we arrive at the view that physics simply is a theory of the nomic mosaic.
Because facts about receptive connections and causal laws are facts above and beyond the level of the nomic mosaic, they are thus beyond the scope and concerns of physics as well. In short, the effective face of causation is precisely the aspect of causation studied by physics, and connectivity, the receptive face, is not entailed by the physical; physics studies only one aspect of causation, allowing the others to reside implicitly in the background because they are beyond its guiding theoretical concerns. This is not to say, however, that the receptive face of causation is entirely opaque from the physical perspective. Studying the effective aspects of causation could very well suggest facts about receptivity, if not directly and explicitly entail them (and presumably, studying the effective side in the context of a research program specifically designed to garner clues about the receptive side could be especially fruitful in a way that retrofitting receptive concepts to established physical theory is not).
Additionally, although facts about the receptive face of the causal mesh would not add experimental content to physical theory, we should not take this to be indicative of a failure of a complete theory of the causal mesh to add any empirical content to the sciences. The receptive face of causation is nonphysical, which in turn implies that not all the causal facts about the world are physical facts. If true, this implication opens up new vistas of theoretical and empirical study, and begins to lay the groundwork for scientific theories that can explain and predict nonphysical facts about subjective experience on the naturalistic basis of both physical and nonphysical causal facts.
In order to better frame these questions, Rosenberg identifies three basic levels of description of the world, in ascending order of metaphysical richness. The first is the Humean mosaic, which is a rather bare description of property instantiations across space and time. If we imagine the trajectory of a ball that is tossed upwards and then falls back to the ground, a description of this event on the level of the Humean mosaic would just include facts about the ball's spatiotemporal coordinates along the trajectory and its properties (mass, elasticity, etc.) at each such coordinate. The next level of description is the nomic mosaic, which introduces counterfactual supporting natural laws to account for observed patterns in the Humean mosaic. For instance, an account of our tossed ball on the level of the nomic mosaic would describe its trajectory in terms of dynamic laws of motion. This richer account introduces modal facts above and beyond the facts about the Humean mosaic, such as, "had the ball been thrown with a different force F, it would have taken trajectory T2 rather than the observed trajectory T1." Finally, the richest level of description of the world is a complete account of the causal mesh, including facts about receptive properties and causal laws. An account on this level of description would specify the receptive connections, effective properties, and causal laws that create a constraint structure on the ball's possible effective states and spatiotemporal locations. (In this case, the constraint structure would be such that it would filter out all possible states and locations of the ball except those which correspond to its actual, observed properties and trajectory through space and time.)
The basic form of Rosenberg's main argument in this chapter is
1. The facts about the nomic mosaic do not entail all the facts about the causal mesh.
2. Physics describes only the nomic mosaic.
3. Therefore, the causal mesh contains facts above and beyond the physical facts. (pg.223)
Facts about the nomic mosaic do not entail all the facts about the causal mesh
Note that the facts about the Humean mosaic do not strictly entail the facts about the nomic mosaic. Given a Humean mosaic H, it is not the case that the facts about H alone are sufficient to fix a unique set of nomic facts, because there are multiple kinds of nomic mosaics that are logically consistent with H. For instance, there could be two sets of natural laws whose predictions are in agreement with all events that actually occur in the history of a universe, but nonetheless disagree about what would have happened had some conditions been different.
Similarly, Rosenberg argues that the facts about the causal mesh are not completely entailed by the facts about the nomic mosaic, since a single set of dynamic laws is logically consistent with multiple kinds of causal constraint structures. Revisiting the toy physics of the Life world introduced in chapter 2, he explicitly details two distinct sets of receptive structures and causal laws, each of which reproduces the same dynamic Life world law describing the state of a cell at time t as a function of its state and the state of its neighboring cells at time t-1. Rather than reproduce Rosenberg's example, I will use a somewhat simpler case study here which should suffice to demonstrate the same general principles and results.
Suppose, as in chapter 10, that there is a world whose only elementary effective property is charge, which can take on a value of either + or -. (Charge is meant to refer to a novel, imaginary effective property, rather than the property called charge in physics.) Further suppose that time is quantized in this world, and that there is a dynamic law describing the behavior of charge over time as follows: the value of an instance of charge at time t is the opposite of its value at time t-1. Call this the law of charge oscillation. In order to show that the facts about this nomic mosaic do not entail the facts about its causal mesh, we need only demonstrate two distinct sets of causal facts which are consistent with the law of charge oscillation.
One set of causal facts (call it C1) consistent with the law of charge oscillation has already been presented in chapter 10. The receptive structure of C1 consists of a regular series of overlapping, two place, asymmetric level 1 receptivities binding instances of charge at each time slice. The causal law on these receptivities states that only an odd number of +s and -s, respectively, may be bound together in the same causal nexus. So, for a temporal sequence of charge instantiations ..., +0.1, -0.2, +0.3, ... occurring at successive times t1, t2, and t3, respectively, the receptive structure looke like the following:
..., [+0.1 => -0.2], [-0.2 => +0.3], ...
The causal law described above, when operating on two-place receptivities, forces two bound instances of charge to take on opposite values. The regularity of two-place receptivities binding every pair of temporally adjacent instances of charge ensures that every instance of charge at a time t will have an opposite value from its temporal neighbors at t-1 and t+1. Thus, the causal processes described in C1 create a world that supports the dynamic law of charge oscillation described above.
We can readily draw up other sets of causal facts that are equally consistent with the law of charge oscillation, even though they feature different receptive structures and causal laws. For instance, suppose we have the following causal conditions:
Causal law on level 1 nexii: There must be an even number of + charges bound within a level 1 nexus.
Causal law on level 2 nexii: There must be an odd number of + charges bound within a level 2 nexus.
Receptive structure: There is a regular pattern of receptive connections on levels 1 and 2, such that instances of charge are bound by overlapping, three-place receptivities into level 1 causal nexii. These level 1 individuals are bound by overlapping, two-place level 2 receptivities.
The receptive structure and causal law on level 1 allow three possible effective states for level 1 individuals: [-.+.+], [+.-.+], and [+.+.-]. Independently possible dual joint states of these individuals, given that they overlap at the endpoints, are:
1. [-.+.+0.n], [+0.n.-.+]
2. [-.+.+0.n], [+0.n.+.-]
3. [+.-.+0.n], [+0.n.-.+]
4. [+.-.+0.n], [+0.n.+.-]
5. [+.+.-0.n], [-0.n.+.+]
These independently possible joint states present a prior possibility space for the effective states of the level 2 individuals. Joint state (5) is ruled out directly by the level 2 causal law, since it has an even number of + charges. Furthermore, note that level 2 nexii featuring joint states (1), (2), and (4) entail the existence of another level 2 nexus featuring joint state (5), due to the stipulated regular, overlapping level 2 receptive structure. These joint states all have at least one level 1 member individual X featuring a - charge at one of its 'endpoints,' and this individual X must be receptively bound into another level 2 individual Y, due to the regular, overlapping level 2 receptive structure existing in this world. But because of the regular, overlapping receptive structure on level 1, Y can only have joint state (5).
For instance, suppose we have a level 2 individual I2.1 with joint state (4), i.e. [[+0.1.-0.2.+0.3]1.1.[+0.3.+0.4.-0.5]1.2]2.1. Because of the stipulated regularity of overlapping receptivities at levels 1 and 2, the existence of I2.1 entails the existence of another individual I2.2 of the form [[+0.3.+0.4.-0.5]1.2.[-0.5.?0.6.?0.7]1.3]2.2. The level 1 causal law forces I1.3 to take the form [-.+.+], which in turn requires that I2.2 takes on joint state (5), but this is not permitted by the level 2 causal law.
Because joint state (4) logically entails an impossible situation given these causal conditions, it follows that it is impossible for joint state (4) to instantiate in the particular kind of level 2 individuals discussed here. Similar reasoning applies to states (1) and (2); essentially, joint states (1), (2), and (4) are inconsistent with the total constraint structure enforced by the conjunction of causal laws and receptive structures in place in this world, so joint state (3) is the only possible configuration left for level 2 individuals. Because the level 2 individuals are all forced to take on joint state (3), the instances of charge are forced to take on a regular, alternating pattern of oscillating values, in agreement with the dynamic law of oscillating charges described above.
The alternate causal conditions described here and in the text are fairly complicated, and even have an air of strangeness or superfluity, in order to demonstrate more forcefully the point that a single set of dynamic laws can be consistent with a large range of causal circumstances. The general idea is that dynamic laws describe regularities of effective state instantiations that result from the operation of causal laws on existing receptive structures. Different receptive structures and causal laws can implement the same pattern of effective state instantiations, and so there could exist substantially different worlds that nonetheless instantiate identical nomic mosaics. Thus, it is not the case that facts about the nomic mosaic entail all the causal facts, but rather it is the other way around; a complete specification of the causal facts will already subsume all the facts about the nomic mosaic.
Physics describes only the nomic mosaic
In section 9.7, Rosenberg set out a lower bound for what aspects of the causal mesh physical theory describes: He argued that physical theory at least designates the low-level effective properties, and that its physical laws at least describe the regularities of effective property instantiations. In other words, physics is at least a theory of the nomic mosaic. I summarized this argument in the thread for chapter 9 as follows:
Effective properties are those properties that have the capacity to place causal constraints on the space of possible ways the world could be. These are precisely the kind of properties studied by physics. Physics is ultimately an empirical endeavor driven by observation, and we only observe things (either directly or indirectly) by means of the roles they play in a causal chain of events that culminates with the systematic activation of our biological senses. In other words, all phenomena that can be the objects of objective empirical study must be effective properties or complexes thereof, because such effective causal properties are precisely the things that subserve objective observation in the first place.
The guiding theoretical concern of physics is to explain and predict experimental results, and by a simplicity constraint, physical theory encompasses the least set of properties needed to explain these experimental results; to posit any properties superfluous to the explanation or prediction of possible experiments would be to posit properties beyond the guiding concern of physics, and thus ultimately beyond its scope. But in principle, an exhaustive theory of the nomic mosaic would already be sufficient to predict and explain the outcome of any experiment-- such a theory would tell us, in the context of the experimental design, how the effective properties of the target phenomena systematically covary with the effective properties of the measuring devices, and ultimately, with those of our own biological senses. In other words, any prospective facts to be added to physics above and beyond the level of the nomic mosaic would come with the cost of added complexity, and without the payoff of added experimental content-- by the simplicity constraint, such facts should not belong to the domain of physics. This amounts to saying that physics is at most a theory of the nomic mosaic, and in conjunction with the lower bound introduced above, we arrive at the view that physics simply is a theory of the nomic mosaic.
Because facts about receptive connections and causal laws are facts above and beyond the level of the nomic mosaic, they are thus beyond the scope and concerns of physics as well. In short, the effective face of causation is precisely the aspect of causation studied by physics, and connectivity, the receptive face, is not entailed by the physical; physics studies only one aspect of causation, allowing the others to reside implicitly in the background because they are beyond its guiding theoretical concerns. This is not to say, however, that the receptive face of causation is entirely opaque from the physical perspective. Studying the effective aspects of causation could very well suggest facts about receptivity, if not directly and explicitly entail them (and presumably, studying the effective side in the context of a research program specifically designed to garner clues about the receptive side could be especially fruitful in a way that retrofitting receptive concepts to established physical theory is not).
Additionally, although facts about the receptive face of the causal mesh would not add experimental content to physical theory, we should not take this to be indicative of a failure of a complete theory of the causal mesh to add any empirical content to the sciences. The receptive face of causation is nonphysical, which in turn implies that not all the causal facts about the world are physical facts. If true, this implication opens up new vistas of theoretical and empirical study, and begins to lay the groundwork for scientific theories that can explain and predict nonphysical facts about subjective experience on the naturalistic basis of both physical and nonphysical causal facts.
Last edited: