- #1
- 2,285
- 3
As we have seen in previous chapters, many of the central theoretical difficulties we have about phenomenal consciousness touch on the issue of causation in some way or another. Central among these is the problem of how our p-conscious experiences are related to our physical behavior, specifically our knowledge claims about the existence and character of p-consciousness. If we reject physicalism but do not wish to contradict our physical understanding of the world, it seems we are forced into some sort of epiphenomenalism where p-consciousness is relegated to being a superfluous feature of nature that plays no role in the world's causal dynamics. Epiphenomenalism is not a satisfactory answer to the question of how p-consciousness figures into the world's causal mesh, but it is not a trivial matter to reject it outright, since it seems we only arrive upon it by rejecting other views that are likewise unsatisfactory. Let us now consider this matter in more detail by reviewing the premises that build up to epiphenomenalism, and the possible responses to the epiphenomenalist argument. Here is how Rosenberg frames the argument (pg. 130):
As Rosenberg eloquently puts it, "This argument creates a prism that refracts the different ways of placing consciousness in the natural order." If we deny premise (1), we wind up with either a reductionism that claims that p-consciousness can be reduced to physical phenomena, or an eliminativism that claims that p-consciousness does not even need to be explained because, in some sense, it does not really exist. We have already seen rejoinders to these positions in chapters 2 and 3. If we deny premise (2), we get nonreductive physicalism, which holds that p-consciousness and the brain are linked by some sort of primitive metaphysical necessity. However, it is not clear that such primitive metaphysical necessities can be at once meaningful and effective; a full critique of nonreductive physicalism and primitive metaphysical necessity is presented in chapter 3. Denying premise (3) commits us to either interactionist dualism or some sort of brute emergence supplemented with downward causation. However, there seems to be no empirical support on offer for either view, and they both seem to conflict with our theoretical understanding of physics. If we reject none of the premises, we wind up at some kind of epiphenomenalism or parallelism, whose problems we have already discussed.
One route that has not yet been well explored is denial of premise (4), perhaps because it has been overlooked or perhaps because it has seemed too obvious to be worth challenging. Denying premise (4) would amount to claiming that, although a completed physics would afford us with a descriptively adequate chatacterization of the physical world's dynamics, it would still not be a complete account of the world's causal structure. This is precisely the line of investigation we will pursue in the coming chapters. To this end, we must engage in detailed considerations of precisely what causation is, and what relationship it bears to physical theory.
Problems with Hume's View
A good place to start in our philosphical investigation of causation is with the question of whether anything like relations of causal dependency, constraint, or production really exist among objects and events, or whether these are just human inventions that we project onto the world. The latter position was held by philosopher David Hume. According to Hume, objective causal connections among objects and events in the world do not exist. Rather, we humans merely project the notion of causal connections onto nature as a kind of psychological habit in order to account for the world's apparent regularities. So, for instance, a Humean view of causation might say that a dropped ball is not caused to fall to the ground by gravity or anything else; rather, falling objects just happen to have followed a pattern of regular behavior up till now, and this behavior has not been causally constrained by anything in any objective sense. A dropped ball may just as well float upwards as fall to the ground, then, as there is nothing 'forcing' it to do the latter. Let us call this view of causation the conventionalist view.
The conventionalist view is surely counterintuitive; nonetheless, it has historically held appeal for some thinkers, particularly empiricists, since it dodges complicated metaphysical and epistemological issues of the nature of causal connections and how we could come to know anything about them. However, in recent years the conventionalist view has begun to fall out of favor, as substantial critiques have been leveled against it. For instance, it is not clear why we call only certain regularities 'causal'; it seems we cannot be realists about scientific truth or use physical laws as explanitory of observed regularities in nature if we are to be conventionalists; and, contrary to the Humean view, we cannot arbitrarily imagine any pattern of cause and effect holding between objects and events because in many cases, we define these things at least partially in terms of causal properties (e.g., 'father' or 'electron').
In response to these serious objections, conventionalists have developed more sophisticated regularity theories that have "paralleled Ptolemy's astronomy, adding epicycle after epicycle to a poorly conceived theory" (pg. 133). Rather than engage these revisions, Rosenberg aims to knock down conventionalism at its roots. He presents a metaphysical problem and an epistemological problem for conventionalism, both of which arise only from the single premise that real causal connections do not exist among objects and events in the world. As such, these seem to be problems for any theory that denies the existence of real, objective causal connections in nature.
The Metaphysical Problem: The Unity of the World
The world is unified in the sense that there are many phenomena which are part of it, and that these phenomena can be meaningfully compared and related to each other within it, e.g. along the dimensions of space, time, mass, motion, etc. What natural condition allows for the observed unity of the world? What allows things to be meaningfully related, and what is the condition upon which we can say something is or is not in our world? It is intuitive to appeal to causation and causal closure to account for this unity. For instance, we might say that all events causally descended from the Big Bang are those which reside in our world, and all other events lie outside our world; likewise, we might say that it is the causal connections among events that allow them to be meaningfully compared and related.
But conventionalism sweeps the rug out from under this proposal. According to conventionalism, causal facts are constituted just by the regularities we observe in the world, and so in this sense, unity precedes causation for a conventionalist. Thus, conventionalism cannot appeal to causation as the unifying condition of the world without falling prey to circular reasoning, and must find some other way to account for the world's unity.
In fact, things appear even worse for conventionalism when we consider that such a view seems to readily imply a completely disunified world at the outset, where all events are fundamentally isolated from each other and 'blind' to each other's existence. What prevents us from regarding each such atomic event as its own self-contained world, rather than a component of a much larger one? What establishes an overarching temporal framework in which these events are ordered? Rosenberg calls this the Humpty Dumpty problem: given such a disjoint set of events, how can we theoretically put them back together again to resemble the world we actually live in?
A conventionalist might be tempted to appeal to a reified spacetime as a natural structure that establishes an inclusion condition for the world and the temporal succession of events. But appealing to the arrow of time to establish the latter is not an option for a conventionalist, as the time dimension has no inherent direction beyond that which is specified by causality. To salvage the appeal to spacetime, a conventionalist could assert that time achieves direction extrinsically, e.g. as a consequence of the lawfulness of increasing entropy as one proceeds away from the time dimension's low entropy extremity, as described by the second law of thermodynamics. But if such a lawful distribution of properties confers directionality to time, there is no principled reason why the same sort of thing could not occur for any of the spatial dimensions. Thus, the conventionalist's account of time's extrinsic directionality is not effective, as it only raises substantial further questions: Why do the extrinsic conditions that give time a direction hold, and why don't analogous conditions hold for any of the spatial dimensions? What is so special about time?
The Epistemic Problem: Solipsism of the Present Moment
Consider again the Humean notion of events. All events in the world, according to Hume, are free of any causal constraint; they are insular phenomena that are completely independent of and unaffected by each other, not unlike Leibniz's windowless monads. What happens when we consider the mind as just such a monad or complex of monads, without supposing the existence of Leibniz's benificent God to direct their behavior? Rosenberg argues that we lapse into a profound skepticism where the only thing we can claim to know is the present moment.
The basic reasoning behind the argument is that 1) if events do not causally constrain each other, then they cannot share information, and 2) if an epistemic agent EA cannot gain information about a set of events E, then EA cannot know about E. Given the Humean premise that present events do not causally constrain future events, it follows that we cannot know the future. Given the additional assumption that future events do not constrain present events (and thus that present events do not constrain past events), it follows that we cannot know the past. Given that perceptual events at time T are about external events that occurred at some time T-k in the past, for some measurable interval k, it follows that perception is about past events; and as we cannot know the past, it follows that perception cannot give us knowledge of external events.
Thus, we have arrived at a radical and untenable skepticism where we cannot know anything about the future or the past, and we cannot even rely on perception to give us knowledge about the external world. We have been forced into a kind of solipsism where the only thing that escapes our global skepticism is knowledge of our own minds at the present moment. If our mental events are correlated at all with any external events, it is only by improbable coincidence, and so we still could not consider ourselves to know about these external events.
After considering the metaphysical and epistemic problems for conventionalism, the view has no legs left to stand on. After all, Hume initially proposed the view in order to avoid problematic metaphysical and epistemological questions that arise from being realists about causal constraint, but being anti-realists about causal constraint only leads to far more severe metaphysical, epistemological, and explanitory difficulties. We are left with a picture of a fractured, disunified world (or a unified world whose unity we cannot explain) in which we must succumb to global skepticism; furthermore, we cannot explain the world's observed regularities and cannot account for the successes of science. If these objections hold, it is clear that we must reject conventionalism and be realists about causal constraint.
As Rosenberg eloquently puts it, "This argument creates a prism that refracts the different ways of placing consciousness in the natural order." If we deny premise (1), we wind up with either a reductionism that claims that p-consciousness can be reduced to physical phenomena, or an eliminativism that claims that p-consciousness does not even need to be explained because, in some sense, it does not really exist. We have already seen rejoinders to these positions in chapters 2 and 3. If we deny premise (2), we get nonreductive physicalism, which holds that p-consciousness and the brain are linked by some sort of primitive metaphysical necessity. However, it is not clear that such primitive metaphysical necessities can be at once meaningful and effective; a full critique of nonreductive physicalism and primitive metaphysical necessity is presented in chapter 3. Denying premise (3) commits us to either interactionist dualism or some sort of brute emergence supplemented with downward causation. However, there seems to be no empirical support on offer for either view, and they both seem to conflict with our theoretical understanding of physics. If we reject none of the premises, we wind up at some kind of epiphenomenalism or parallelism, whose problems we have already discussed.
One route that has not yet been well explored is denial of premise (4), perhaps because it has been overlooked or perhaps because it has seemed too obvious to be worth challenging. Denying premise (4) would amount to claiming that, although a completed physics would afford us with a descriptively adequate chatacterization of the physical world's dynamics, it would still not be a complete account of the world's causal structure. This is precisely the line of investigation we will pursue in the coming chapters. To this end, we must engage in detailed considerations of precisely what causation is, and what relationship it bears to physical theory.
Problems with Hume's View
A good place to start in our philosphical investigation of causation is with the question of whether anything like relations of causal dependency, constraint, or production really exist among objects and events, or whether these are just human inventions that we project onto the world. The latter position was held by philosopher David Hume. According to Hume, objective causal connections among objects and events in the world do not exist. Rather, we humans merely project the notion of causal connections onto nature as a kind of psychological habit in order to account for the world's apparent regularities. So, for instance, a Humean view of causation might say that a dropped ball is not caused to fall to the ground by gravity or anything else; rather, falling objects just happen to have followed a pattern of regular behavior up till now, and this behavior has not been causally constrained by anything in any objective sense. A dropped ball may just as well float upwards as fall to the ground, then, as there is nothing 'forcing' it to do the latter. Let us call this view of causation the conventionalist view.
The conventionalist view is surely counterintuitive; nonetheless, it has historically held appeal for some thinkers, particularly empiricists, since it dodges complicated metaphysical and epistemological issues of the nature of causal connections and how we could come to know anything about them. However, in recent years the conventionalist view has begun to fall out of favor, as substantial critiques have been leveled against it. For instance, it is not clear why we call only certain regularities 'causal'; it seems we cannot be realists about scientific truth or use physical laws as explanitory of observed regularities in nature if we are to be conventionalists; and, contrary to the Humean view, we cannot arbitrarily imagine any pattern of cause and effect holding between objects and events because in many cases, we define these things at least partially in terms of causal properties (e.g., 'father' or 'electron').
In response to these serious objections, conventionalists have developed more sophisticated regularity theories that have "paralleled Ptolemy's astronomy, adding epicycle after epicycle to a poorly conceived theory" (pg. 133). Rather than engage these revisions, Rosenberg aims to knock down conventionalism at its roots. He presents a metaphysical problem and an epistemological problem for conventionalism, both of which arise only from the single premise that real causal connections do not exist among objects and events in the world. As such, these seem to be problems for any theory that denies the existence of real, objective causal connections in nature.
The Metaphysical Problem: The Unity of the World
The world is unified in the sense that there are many phenomena which are part of it, and that these phenomena can be meaningfully compared and related to each other within it, e.g. along the dimensions of space, time, mass, motion, etc. What natural condition allows for the observed unity of the world? What allows things to be meaningfully related, and what is the condition upon which we can say something is or is not in our world? It is intuitive to appeal to causation and causal closure to account for this unity. For instance, we might say that all events causally descended from the Big Bang are those which reside in our world, and all other events lie outside our world; likewise, we might say that it is the causal connections among events that allow them to be meaningfully compared and related.
But conventionalism sweeps the rug out from under this proposal. According to conventionalism, causal facts are constituted just by the regularities we observe in the world, and so in this sense, unity precedes causation for a conventionalist. Thus, conventionalism cannot appeal to causation as the unifying condition of the world without falling prey to circular reasoning, and must find some other way to account for the world's unity.
In fact, things appear even worse for conventionalism when we consider that such a view seems to readily imply a completely disunified world at the outset, where all events are fundamentally isolated from each other and 'blind' to each other's existence. What prevents us from regarding each such atomic event as its own self-contained world, rather than a component of a much larger one? What establishes an overarching temporal framework in which these events are ordered? Rosenberg calls this the Humpty Dumpty problem: given such a disjoint set of events, how can we theoretically put them back together again to resemble the world we actually live in?
A conventionalist might be tempted to appeal to a reified spacetime as a natural structure that establishes an inclusion condition for the world and the temporal succession of events. But appealing to the arrow of time to establish the latter is not an option for a conventionalist, as the time dimension has no inherent direction beyond that which is specified by causality. To salvage the appeal to spacetime, a conventionalist could assert that time achieves direction extrinsically, e.g. as a consequence of the lawfulness of increasing entropy as one proceeds away from the time dimension's low entropy extremity, as described by the second law of thermodynamics. But if such a lawful distribution of properties confers directionality to time, there is no principled reason why the same sort of thing could not occur for any of the spatial dimensions. Thus, the conventionalist's account of time's extrinsic directionality is not effective, as it only raises substantial further questions: Why do the extrinsic conditions that give time a direction hold, and why don't analogous conditions hold for any of the spatial dimensions? What is so special about time?
The Epistemic Problem: Solipsism of the Present Moment
Consider again the Humean notion of events. All events in the world, according to Hume, are free of any causal constraint; they are insular phenomena that are completely independent of and unaffected by each other, not unlike Leibniz's windowless monads. What happens when we consider the mind as just such a monad or complex of monads, without supposing the existence of Leibniz's benificent God to direct their behavior? Rosenberg argues that we lapse into a profound skepticism where the only thing we can claim to know is the present moment.
The basic reasoning behind the argument is that 1) if events do not causally constrain each other, then they cannot share information, and 2) if an epistemic agent EA cannot gain information about a set of events E, then EA cannot know about E. Given the Humean premise that present events do not causally constrain future events, it follows that we cannot know the future. Given the additional assumption that future events do not constrain present events (and thus that present events do not constrain past events), it follows that we cannot know the past. Given that perceptual events at time T are about external events that occurred at some time T-k in the past, for some measurable interval k, it follows that perception is about past events; and as we cannot know the past, it follows that perception cannot give us knowledge of external events.
Thus, we have arrived at a radical and untenable skepticism where we cannot know anything about the future or the past, and we cannot even rely on perception to give us knowledge about the external world. We have been forced into a kind of solipsism where the only thing that escapes our global skepticism is knowledge of our own minds at the present moment. If our mental events are correlated at all with any external events, it is only by improbable coincidence, and so we still could not consider ourselves to know about these external events.
After considering the metaphysical and epistemic problems for conventionalism, the view has no legs left to stand on. After all, Hume initially proposed the view in order to avoid problematic metaphysical and epistemological questions that arise from being realists about causal constraint, but being anti-realists about causal constraint only leads to far more severe metaphysical, epistemological, and explanitory difficulties. We are left with a picture of a fractured, disunified world (or a unified world whose unity we cannot explain) in which we must succumb to global skepticism; furthermore, we cannot explain the world's observed regularities and cannot account for the successes of science. If these objections hold, it is clear that we must reject conventionalism and be realists about causal constraint.
Last edited:
As I said, "it is very important that one realize that there exists a need to differentiate between "what we know" and "what we think we know" before we have any method of separating the two. You are attempting to make the separation first.