Varieties of Causal Closure

Barbara Montero

barbaramontero@gsu.edu
Quoted from: http://barbara.antinomies.org/papers/VarietiesofCausalClosure

To know that the physical world is causally closed, it is often thought, is to be well on one’s way to solving a number of great philosophical problems. For example, is the mind physical? If the physical world is causally closed, then since the mind causally affects the body, it appears that it must be. Does God exist? If the physical world is causally closed, then it seems that a nonphysical God who causally affects the physical world cannot exist. Can we, as physical beings, have knowledge of abstract entities? If the physical world is causally closed, then if the abstract realm is nonphysical and knowledge requires causal interaction, it seems that we cannot. The principle of physical causal closure thus appears to be a great philosophical panacea. But as with other apparent panaceas, it is well worth looking into how much of the cure is due to the placebo effect and how much real philosophical work is being accomplished.

           

1. Causal closure principles

What exactly is the causal closure of the physical? The relevant notion of closure is not the one we apply when talking about a set being closed. To talk about a set A being closed under a function f is to say that whenever a is in A, then f(a) is in A. For example if f(x)=x² and A is the set of positive even numbers, then A is closed under f since if a is positive and even, then a² is also positive and even. The notion of closure relevant to the above philosophical arguments is a distinct notion. A set is closed under a function when applications of that function do not bring us from inside to outside of the set, while it is typically thought that the physical world can be closed under causation even if a causal chain takes us from the physical to the nonphysical, that is, even if a physical cause, such as a punch in the gut, has a nonphysical effect, such as moving one’s soul into a state of pain. The focus, rather, as the above examples involving God, abstracta and the mind should make clear, is typically (though not always exclusively) on the other direction: closure puts restraints on how one lands inside the set, as it were, rather than on how one leaves it.

The notion of causal closure that is most often employed in arguments for physicalism is, roughly, that every physical phenomenon has a physical cause. But, as stated, this principle is too weak in one way and too strong in another. It is too strong in as much as there may be physical phenomena that are not caused at all—for example, the big bang would seem to be physical yet it is not clear that it has a cause—and we certainly would not want to say that those phenomena have a physical cause. And it is too weak since it leaves open the possibility that a nonphysical cause could be necessary for the physical effect, that the physical antecedents of my bodily movements, for example, are merely partial causes of these movements. To address this, causal closure is often taken to be the principle that every physical phenomenon that has a cause has a sufficient physical cause, that is, a physical state of affairs that alone suffices to determine the effect. However, even this is too strong since we do not want to require a sufficient physical cause of a phenomenon that does not itself have a sufficient cause. So let us take causal closure (CC) to be the following:

CC: every physical phenomenon that has a sufficient cause has a sufficient physical cause.[1]

There is a sense in which quantum effects fall outside the scope of CC; for in a sense, quantum effects, such as the radioactive decay of a particle, do not have sufficient causes since there does not seem to be any state of affairs that fully determines them. However, if the quantum world falls outside the scope of CC, CC becomes vacuously true of a large part, perhaps the whole part, of the world. This problem is typically addressed by interpreting “sufficient cause” as a cause that either fully determines its effect or a cause that fully determines the chances of its possible effects.[2] Yet if quantum events are really just purely random and not caused or determined in any sense, then the quantum world would still fall outside the scope of CC. For the purposes of argument, let us grant that this problem can be solved; for as we will see, even granting this, formulating a useful and defensible notion of causal closure is still no simple task.

In addition to CC, one also finds a stronger notion of causal closure in the literature. Strong Causal Closure (SCC) states:

SCC: Physical effects have only physical causes.

While CC allows for the possibility of nonphysical causes of physical effects, requiring only that the physical cause is sufficient (e.g. as long as there is a sufficient physical cause of my bodily movement, CC does not exclude the possibility of this movement also being caused by a nonphysical soul), SCC excludes nonphysical causes of physical effects altogether.[3]

Finally, causal closure is sometimes expressed not as a one-way restraint on possible causes, but as a two-way restraint. For example, at one point Jaegwon Kim says that physical causal closure amounts to the claim that, “no causal chain will ever cross the boundary between the physical and the non-physical.”[4] Two-way causal closure (TCC), thus states that:

TCC: There is no causal interaction between the physical and the nonphysical.

Here we have restrictions both on certain types of causes given certain types of effects and on certain types of effects given certain types of causes, making it much stronger than both CC and SCC, neither of which rules out physical to nonphysical causation. For according to TCC, physical/nonphysical causal interaction (i.e. causation in either direction) is ruled out entirely.

Since TCC implies SCC, which implies CC, an argument for TCC would be of great use. However, TCC is stronger than is needed to argue for physicalism, if one accepts that the ontologically questionable phenomena (God, nonphysical minds, etc.) have physical effects. If God, for example, has physical effects then one could apply SCC, which would lead to the conclusion that God is physical. Of course, it is usually assumed that the physical also affects at least some of these phenomena, in particular, the mental (broken bones, for example, cause pain). But it is much more difficult to argue that physical causes do not have nonphysical effects than it is to argue that physical effects (that have sufficient causes) have sufficient physical causes.[5] Nonphysical effects of physical processes, it is often thought, could, for all we know, exist—just as invisible causally impotent fairies could, for all we know, exist. This is why even many physicalists take epiphenomenalism (the view that the mental is caused by, but has no causal effect on, the physical) to be possible, though perhaps profligate. Such epiphenomenal effects would be of no consequence to our physical theory of the world. And so, let us ignore the question of whether there are physical causes that have nonphysical effects. As we will see, arguing for either CC or SCC is difficult enough.

 

2. From causal closure to physicalism

Let us turn to the use of the notion of causal closure in arguments for physicalism with respect to the mind. The general form of these arguments is typically something like this. If physicalism is false, the mind is like a ghost in a machine, a ghost who flips various switches in the machine, causing our physical bodies to move. However, we have good reason to believe that all of these switches are flipped on or off by other physical parts of the machine. And since it is absurd to think that the switches are doubly flipped by both the machine and the ghost, we should conclude that there is no ghost in the machine, that the mental causes of our bodily movements are themselves physical parts of the machine.

Putting CC to work, we can state this argument more precisely as:

CAP: The causal argument for physicalism:

1) The physical world is causally closed (CC).

2) There is mental to physical causation.

3) There is no systematic causal overdetermination (i.e. if p has a sufficient cause, it will not be the case that, systematically, it will have another cause).

-----------------------------------------------------------

Thus: The mental is physical.[6]

Strictly speaking, however, the argument is not valid: the conclusion that the mental is physical does not follow since, as is often pointed out, the argument captures only those mental phenomena that are causes of physical events; left out are those mental phenomena that are, like the invisible fairies, causally impotent. But even this does not sufficiently narrow the domain if causal closure is interpreted as CC: that every physical phenomenon that has a sufficient cause has a sufficient physical cause. Given CC, we can only conclude that at least those mental phenomena that are causes of physical events, which themselves have sufficient causes, are physical. For if there are physical events that have only partial causes, then the above argument does not rule out nonphysical mental causes of these physical events. However, while physicalism does not strictly follow, the move from the precise conclusion of the argument—that at least those mental phenomena that are causes of physical events, which themselves have sufficient causes, are physical—to the conclusion that the mental is physical is fairly harmless since many would accept both that there are mental phenomena that are causes of physical phenomena, which themselves have sufficient causes, and that it would at least be odd for just this subset of the mental to be physical.

Now, what are we to say about the status of the premises? Epiphenomenalism is inconsistent with mental to physical causation (premise 2) and while, as noted, many take epiphenomenalism to be profligate, it has its share of fans.[7] Moreover, occasionalism and pre-established harmony are inconsistent with mind-body causal interaction, entirely. Arguments against premise 3 are typically found in the context of defending nonreductive physicalism, for it is sometimes claimed that systematic overdetermination is excluded only when the causes are independent.[8] However, even the possibility of overdetermination by independent causes has been defended.[9] But let us focus on premise 1, the causal closure of the physical, a premise that is usually taken as inviolable.[10]

3. The argument for causal closure from the success of science

Support for CC is often thought to come from the great success physics, or science in general, has had in accounting for the causal structure of the world.

 

The argument for CC from the success of science:

1) If CC were false, then physics would be necessarily incomplete.[11]

2) Physics is not necessarily incomplete.

-----------------------------------------

Thus: CC is true.

 

Some may reject the second premise of the argument, claiming that a complete physics is not even in principle possible. For example, one might argue that science is necessarily an ongoing process, that the nature of physics, and science in general, is turning over its earlier results. Moreover, if being a complete physics means being a theory of absolutely everything including basic arithmetic truths, a complete physics—if such a theory is formalizable—is, by Gödel’s incompleteness theorem, provably not possible.

While I cannot say whether physics is necessarily incomplete, I do think that there is good reason to reject the first premise: a completable physics does not imply CC, at least, not when the physical is understood in the way it must be understood in order to formulate a debate over the mind-body problem. Many accept the first premise since they understand CC as Frank Jackson explains it: “that the physical sciences, or rather some natural extension of them, can in principle give a complete explanation for each and every bodily movement, or at least can do so up to whatever completeness is compatible with indeterminism in physics.”[12] Since support for CC is thus thought to come from the tremendous success of the physical sciences, its denial is thought to amount to the view that there are phenomena that will permanently elude scientific explanation. In Kim’s words, if the physical world were not causally closed, then “to explain some physical events you must go outside the physical realm and appeal to nonphysical causal agents and laws governing their behavior!”[13] And thus to deny causal closure, he claims, is to accept that we cannot have a complete physics of even the physical world, as he says, “complete physics would in principle be impossible, even as an idealized goal.”[14]

If one takes the physical to be, loosely, whatever can be accounted for by physics, as Jackson and Kim do, then to reject CC is to reject the in principle possibility of a complete physics, that is, a physics capable of identifying the sufficient causes of all physical events (that have sufficient causes). However, there are good reasons not to define the physical in this way. Physics is in the business of investigating the most fundamental aspects of the world, whatever they may be. And, given the great surprises that physics has brought us in the past, it would be rash to place a priori restrictions on the type of phenomena that it can investigate. If there is good evidence that the world contains fundamental mental properties, then physicists would or at least should try to figure out how these properties work. Moreover, if we define the physical over physics, physicalism becomes a thesis that need not conflict with dualism since dualism could be true even if the mental were accountable for by physics, as long as this physics incorporates the mental as a fundamental feature of the world.[15] For example, dualists (and panpsychists) can accept that something like Wigner’s hypothesis (i.e. that pure acts of consciousness are required in order to account for the collapse the wave function) could be part of future or final physics. This would be a situation in which physics accounted for consciousness (in as much as it accounts for any other fundamental feature of the world) yet consciousness would be a fundamental feature of the world. Thus, taking the physical to be whatever physics, or more broadly, science, will account for does not provide us with an understanding of the physical that can be used to formulate a real debate, that is, a debate with more than one side. [16]

So the argument from the success of science succeeds if it employs a notion of the physical that does not ground a debate over the nature of the mind. Moreover, it fails when it employs what seems to me to be, for the purposes of the mind-body problem, the only useful notion of the physical, and that is, the fundamentally nonmental. For what is at stake in the debate over the mind-body problem is not whether physics or science in general can provide the fundamental building blocks of the mental, but rather whether the mental is a fundamental feature of the world. Thus, when our concern is with the mind-body problem, the “causal closure of the physical” should be understood as the causal closure of the fundamentally nonmental, and physicalism, with respect to the mental, as the thesis that the mental is reducible to (or composed of, or otherwise determined in some sense) the nonmental. But when it is so understood it is clear that premise 1 fails. For physics could be complete (or completeable) even if the fundamentally nonmental were not causally closed, since if fundamentally mental properties exist, they could be accounted for by physics.

Some may still think that at least something like this argument supports CC; for some may still have a vague nagging feeling that rejecting CC is somehow being antiscientific. But rejecting CC pits one against science no more than accepting the incompleteness theorem pits one against mathematics. So neither the particular argument, nor that vague nagging feeling that rejecting CC is antiscientific is well grounded. Science can be as successful as you like and CC could still be false.

4. The argument for physicalism from the causal closure of the fundamentally nonmental

A causal argument for physicalism, however, can be formulated in terms of the fundamentally nonmental. Transposing “physical” with “fundamentally nonmental,” the argument that we are concerned with can be represented thus:

CAF: The causal Argument for fundamental nonmentalism:

1) The fundamentally nonmental world is causally closed.

2) There is mental to fundamentally nonmental causation.

3) There is no systematic causal overdetermination.

---------------------------------------------------------------------

Thus: The mental is, at its most fundamental level, entirely nonmental.[17]

Should we accept the causal closure of the fundamentally nonmental? We should note that the type of causal closure that is relevant here is the causal closure, not just of the nonmental, but of everything that is nonmental at its most fundamental level. For the conclusion ought to be, not that the mental is nonmental—an outright contradiction—but that, depending on what you think is necessary for a full blooded physicalism, the mental is ultimately decomposable, or reducible to, or analyzable in terms of fundamentally nonmental phenomena.

Leaving aside the possibility of panpsychism, it seems that empirical evidence at least points to the fundamentally nonmental as causally relevant, or perhaps even causally necessary for fundamentally nonmental effects. For example, there are no known cases of bodily movements without underlying bodily antecedents. Indeed, that bodily effects of mental causes require bodily antecedents is something for which even Lucretius thought there was evidence. Moreover, he thought that this evidence was reason to accept physicalism. In his words, “when [the mind and soul] is seen to push the limbs, rouse the body from sleep, and alter the countenance and guide and turn about the whole man, and when we see that none of these effects can take place without touch nor touch without body, must we not admit that the mind and the soul are of a bodily nature?”[18] But while the premise of this argument seems compelling, the conclusion does not seem to follow (even with the additional claim that there is no systematic causal overdetermination). To say that bodily effects of mental causes cannot occur without fundamentally nonmental causes leaves open the possibility that fundamentally mental causes are required as well. This is why today’s causal argument for physicalism is formulated in terms of the physical being necessary and sufficient for the relevant effect rather than being just necessary.

An argument for the causal sufficiency of the fundamentally nonmental is an important step in the causal argument for physicalism since without it dualists have a response to CAF. For unless there is reason to think that the fundamentally nonmental is causally sufficient, dualists can argue that the physicalist perspective is correct as far as it goes, but may be only a partial view of the mental. The dualist can accept that certain bodily antecedents of my bodily movements are components of the cause of these movements, but deny that they are sufficient causes. He, thus, might claim, much as Leibniz did with respect to the whole of philosophy, that physicalism is largely correct in what it affirms, but not in what it denies; in as much as it states that the physical (i.e. the fundamentally nonmental) is causally relevant, it is correct; in as much as it states that the nonphysical (i.e. the fundamentally mental) is causally irrelevant, it is incorrect. Physicalists, then, to continue the Leibnizian analogy, would be like the individuals in a town who make various pronouncements about the town from where they are standing. One will claim that the town has a bakery and a church; another will claim that the town has many vast fields, and so forth. All can be correct as long as they do not add the extra claim, “and there is nothing more to the town.” To defend their position, physicalists must show that there are good reasons to take the fundamentally nonmental to be causally sufficient.

5. The physiological argument for Strong Causal Closure

            There is an argument, the argument from physiology, that is sometimes taken to show that we do have good reason to think that the fundamentally nonmental world is causally closed, to show not only that the physicalist perspective is correct as far as it goes, but that it is correct, period.[19] Moreover, it is an argument not just for CC, but for SCC, and one that apparently does not rely on a notion of the physical that is tied to whatever is scientifically acceptable, but rather takes the physical to be the fundamentally nonmental. And so, if successful, it could ground an argument for a physicalism with some vim and vigor.

The argument is straightforward. As David Papineau explains it, we have good reason to accept the causal closure of the fundamentally nonmental since the developments of physiology in the 20^th century have provided us with good reason to think that there are no basic or fundamental mental forces that act on the body. In his words, the discovery of DNA along with great improvements in neurophysiology,

. . .made it difficult to go on maintaining that special forces operate inside living bodies. If there were such forces, they could be expected to display some manifestation of their presence. But detailed physiological investigation failed to uncover evidence of anything except familiar physical forces [i.e. forces that are not fundamentally mental]. In this way, the argument from physiology can be viewed as clinching the case for completeness.

To put it simply, there are no nonphysical causes of physical phenomena (SCC) because after extensive investigation, none have been found. Papineau focuses on physiology since he thinks that if there were causal, fundamentally mental phenomena, they would turn up under physiological investigation. But one need not restrict oneself to physiology since no type of investigation has revealed the existence of fundamentally mental phenomena. If the fact that physiological investigation has failed to turn up evidence of the fundamentally nonmental is evidence in favor of SCC, then the fact that no other investigations have revealed any fundamental mental forces can only help the case. Of course, we still do not have a conclusive argument for SCC; for all we know, fundamental mental forces may be revealed tomorrow. Moreover, there are arguments that are intended to show that such forces exist. However, if these arguments can be shown to be flawed, scientific induction should lead us to accept that fundamentally mental forces do not exist. As Papineau states it, to deny this would be to reject “a premise which, by any normal inductive standards, has been fully established by over a century of empirical research.”[20]

If this simple argument based on the lack of evidence for the fundamentally mental is successful, it eliminates any need to argue for premise three of CAF, that there is no systematic causal overdetermination.[21] The existence of causal overdetermination does not pose a threat to physicalism, if there are no nonphysical causes.[22] Rather, given that there are no nonphysical causal forces, physicalism follows simply from the additional premise that the mental causally affects the physical. Indeed, one might even interpret Papineau as arguing for physicalism directly, that is, as arguing that physiology has shown that there is nothing nonphysical, period. But he does not seem to be willing to go so far: he takes the physiological considerations to show that physiology has failed to reveal any nonphysical forces, which presumably shows that only those mental phenomena that are forces are physical and so leaves open whether there are nonphysical phenomena that are causally impotent.

Does this argument, then, give the physicalist what she needs? Not according to Carl Gillett and Gene Witmer, who claim that this version of the causal argument for physicalism is no less prey to what has come to be known as Hempel’s dilemma than the argument for physicalism based on a definition of the physical in terms of what physics tells us about the world.[23] Hempel’s dilemma states that physicalism is not a well-formulated thesis.[24] For on the one hand, if we define the physical in terms of current physics we are left with a theory we cannot defend; for according to such a view if a new particle is discovered next week it will not be physical. However, on the other hand, if we define the physical in terms of future physics we are left with a theory that is impossibly vague; for who knows what future physics will bring.[25]

As Gillett and Witmer see it, a similar problem arises for the physiological argument for physicalism. On the one hand, today’s best physiological theory does not show that all bodily movements are determined, ceteris paribus, by preceding physiological conditions and laws since today’s physiology is not complete.[26] On the other hand, it is difficult to predict, with great confidence, what future physiology will bring. Moreover, if future physiology incorporates basic mental properties, then the strong causal closure of the physiological is consistent with dualism (even assuming mind-body causal interaction). Thus, if the physiological argument proceeded by claiming that physiology has identified (ceteris paribus) the sufficient causes of all bodily movements, Gillet and Witmer would be right: this argument would most likely be no better than arguments for causal closure that take the physical to be whatever is accountable for by physics.

However, in order to justify the causal closure of the fundamentally nonmental, we need not argue that current physiology has identified the sufficient causes of bodily movements, or that future physiology will do so. Rather, we can argue that the recent physiology shows that there is good reason to think that there are no mental forces. One horn of Hempel’s dilemma punctures definitions of the physical in terms of current physics since current physics is, most likely, neither a true nor complete account of the fundamental features of the world. However, the physiological argument can be stated in terms of current physiology, or rather current physiology as well as the physiology of the past two centuries, and still escape this horn. For even if current physiology has not identified the sufficient causes of bodily movements, it still could give us reason to accept that there are sufficient fundamentally nonmental causes of physical effects, if it provides reason to think that there are no fundamentally mental forces at all. And this is precisely what Papinaeu thinks the lessons of physiology, which has never revealed anything fundamentally mental, along with good scientific induction should lead us to accept.

           The physiological argument, moreover, does not seem prey to what I take to be the most serious objection to defining the physical over physics, or science in general, namely, that such definitions of the physical turn physicalism into a thesis that dualists can accept (yet, certainly, a condition of adequacy for any definition of the physical is that it makes physicalism incompatible with dualism and other antiphysicalist positions). For the argument claims that we have good inductive evidence for the view that the mind is fundamentally nonmental, which is not something dualists can accept.

But defining the physical so as to avoid Hempel’s dilemma and present a position that is opposed to dualism is just the first step in arguing for physicalism. We still need to ask whether the physiological argument succeeds: does physiology actually provide strong evidence for the causal closure of the fundamentally nonmental? And here I think the answer is not entirely clear, since it is not clear that lack of evidence for the fundamentally mental amounts to evidence that the fundamentally mental does not exist.

To be sure, there are cases where lack of evidence for p seems to provide evidence that p does not exist. For example, we reject the idea that ghosts exist since we have never found anything that could be identified as a ghost. But this alone would probably not suffice to show that ghosts do not exist, if it weren’t the case that we also have a pretty good understanding of what actually causes those bumps in the night that scare people into thinking their houses are haunted.[27] Similarly, one might claim that failure to find fundamental mental forces does not suffice to show that they do not exist unless we also have a fairly good understanding of what fundamentally nonmetal force actually causes us to cry out when in pain, and so forth. But this should not pose a great problem, for while we certainly do not have a complete nonmental account of what we take to be mental causes, we have a good start.[28]

The reason why physicalists cannot rest here is that there are also clear cases where failure to find evidence for p does not count as evidence against the existence of p and it may be that the physiologist’s failure to find evidence for the fundamentally mental is more similar to these cases than to the situation illustrated above. For example, the fact that I have no evidence for the existence of a cat who spends each night in the alley behind my home, is not evidence that there is no such cat since I have only walked through the alley during the day. And if the physiologist, in investigating only the bodily causes of bodily movements, is in a similar position, her lack of evidence for fundamentally mental forces does not amount to evidence that such forces do not exist since in investigating only the bodily causes of bodily movements, fundamentally mental causes would never be revealed.[29] Moreover, anti-physicalists might argue that not only would physiological investigation miss the fundamentally mental, but that another type of investigation reveals it: first person experience. For when I introspect I am acquainted with mental forces, which, at least to all appearances, seem irreducible to anything else. Given these considerations, it seems we must conclude that the argument from physiology falls at least somewhat short of providing us with a compelling reason to accept causal closure.[30]

Of course, since the physiological argument is an argument for Strong Causal Closure (SCC)—the principle that states that physical effects have only physical causes, or as we have now understood it, that fundamentally nonmental effects have only fundamentally nonmental causes—even if it or any other argument fails to support SCC, SCC is more than we need to argue for physicalism. Moreover, while we found the argument from the success of science for the weaker principle that every physical phenomenon that has a sufficient cause has a sufficient physical cause to be flawed, it may very well be other arguments for CC that are not similarly flawed, and, in a strong enough dose, can help to cure the mind-body problem. But addressing the question of whether such medicine exists must wait for another occasion.