The Atlas of Reality
Page 98
FURTHER PROBLEMS WITH EXPLANATIONIST ACCOUNTS OF CAUSAL DIRECTION There are a few additional Explanationist accounts of causal direction. For example, one might appeal to the fact that, at least in our local universe, the process of the formation of black holes seems to be irreversible. However, all Explanationist accounts must appeal to certain global features as the ground of causal direction, since none of the local features that ground causal explanation, whether logical deducibility, the truth of counterfactual conditionals, or the raising of probability, are intrinsically irreversible. There are two objections to all such global accounts of causal direction.
1. Inverted time thought experiments. Each of these accounts is vulnerable to inverted time thought experiments. Since the fundamental laws are (barring a couple minor exceptions) time-reversible, there is nothing inherently impossible about a universe in which entropy is decreasing or dispersed order converges (light waves or water ripples converging often to a single point) (see Huw Price 1996). It's true that the vast majority of states with low entropy move toward greater entropy, but, since the high-entropy states are more probable by an astronomically high factor, the high-entropy to low-entropy transitions will occur with the same probability as those in the opposite direction. Reichenbach is, in effect, guilty of the statistical fallacy of ignoring the base rate (ignoring the far greater probability of the high-entropy states).
2. Small world thought experiments and the intrinsicality of causal direction. In addition, all such accounts, and even accounts like Dowe's that appeal to exotic, irreversible events, have the consequence that the causal direction of a process is not intrinsic to that process. Whether, in a particular case, E caused E′ or E′ caused E will depend on what is going on in other, perhaps quite remote, parts of the universe. In fact, in some cases, the causal direction of the process will never be settled so long as the history of the universe continues, since events in the remote future might reverse the predominate direction of entropy increase or order dispersal or whatever else we fix on to define temoral order.
This problem can be illustrated by Tooley's small world thought experiments (Tooley 1987, 1990). It would seem to be possible for the history of the universe to be very simple, consisting only of a handful of causal interactions. In such a universe, there may be no change in entropy and no asymmetric dispersal of order whatsoever (and no kaon decay, either). Nonetheless it seems that there could still be a causal order to the events.
27.1.1.3 Objection 3: Inclusion of Causally Irrelevant Factors.
Causal Explanation also seems to include causally irrelevant factors as causes. This objection challenges the sufficiency of Explanationist definitions of causation. That is, the objector attempts to show that pairs of propositions can satisfy Explanationist conditions without being genuine cases of causal explanation.
There are two cases to consider, namely, conjunctions (compound propositions formed by ‘both’ and ‘and’) and disjunctions (compound propositions formed by ‘either’ and ‘or’).
1. Addition of irrelevant conjuncts. Suppose that A is a genuine causal explanation of the truth of B. Let C be an irrelevant truth. Explanationist accounts wrongly count the conjunction A&C as a causal explanation of B.
If A, together with the laws of nature, entails B, then so will A&C.
If A raises the probability of B, relative to some background condition, and C is causally irrelevant, then A&C will also raise the probability of B.
Suppose it is true that if A were not the case, then B would not be the case. Let's suppose that C is some causally irrelevant truth such that, if A&C were not the case, then A would not be the case. (In other words, to use Lewis's semantics, A is false in all of the closest worlds in which A&C is false.) On those suppositions, it will be true that if A&C were not the case, then B would not be the case, wrongly making A&C a cause of B.
2. Addition of irrelevant disjuncts. The version of the causal relevance objection that exploits irrelevant disjuncts applies only to the Probability-Raising Theory. Once again, suppose that A is a causal explanation of the truth of B. Let C be an irrelevant truth. If A raises the probability of B, relative to some background condition, and C is causally irrelevant, then (A or C) will also raise the probability of B.
The upshot of the foregoing is that Causal Explanationism faces serious challenges. We turn now to Causal Connectionism, to test whether it fares better.
27.2 Causal Connectionism
27.2.1 Truthmakers and the relata of the causal connection
Causal Connectionists posit a category of thing, events (or states or conditions), members of which stand in the fundamental relation of causation. Recall again (1) and (2):
(1) Mary's kicking the ball caused the ball's entering the goal.
(2) The ball entered the goal because Mary kicked the ball.
Connectionists propose that we are to understand (2) in terms of (1), not vice versa. The fundamental truth is (1). We can appropriately state something like (2) because the two events mentioned in (1) stand in the relation of causation. Causal Connectionists provide a causal explanation of a truth by accurately describing part of the actual causal structure of the world. But, what part of the world's causal structure? To explain the ball's entering the goal, we must describe a part of the causal structure of the world that is somehow relevant to the truth of the proposition that the ball entered the goal.
The most natural proposal would be to turn to the idea of truthmakers, as discussed in Chapters 2 and 3. We causally explain the truth of p by providing information about the causes of the truthmaker of p. On this view, truthmakers are the sorts of things that stand in the causal relation to each other. If that is right, then we could in principle provide a causal explanation of any truth that has a truthmaker, provided that the truthmaker itself has a cause.
Therefore, Causal Connectionism provides new grounds for Truthmaker Theory (2.1T/2.1A.1T/2.1A.1A.1T). In addition, it provides us with grounds for extending the truthmaker principle to various categories of true propositions. Any true proposition for which we can give a causal explanation must have a truthmaker. Ideally, each such truth will have exactly one truthmaker. Given this connection, any reason for extending the scope of causal explanation becomes a reason for extending the truthmaker principle.
Suppose, for example, that every contingent truth has a causal explanation. If so, then every contingent truth has a truthmaker. Similarly, if every truth attributing a finite attribute or every true proposition attributing a beginning to some condition has a cause, then every such truth also has a truthmaker.
27.2.1.1 A Linguistic Argument for Causal Relata: Davidson's Theory of Adverbs.
Donald Davidson (1967) provides us with a further, linguistic argument for Causal Connectionism. We'll focus here on one piece of linguistic data that is best explained by positing the existence of events as truthmakers of certain propositions, namely, the use of adverbs. Consider (7) and (8):
(7) John walked.
(8) John walked quickly.
Clearly, the truth of (8) entails the truth of (7). John cannot walk quickly without walking. The best linguistic theory of adverbs would provide a simple and uniform explanation of this logical entailment. Davidson (1967) offers the following theory. Action verbs like ‘walked’ introduce an existential quantification over events. Consequently, the logical form of (7) and of (8) is best captured by (9) and (10):
(9) There is a walking, and it is by John.
(10) There is a walking, and it is by John, and it is quick.
Statement (9) now follows evidently from (10) by the rule of simplification (from A&B, infer A). The alternative theory would have to suppose that adverbs like ‘quickly’ represent functions from one set (the set of those who V) to another set (the set of those who V quickly). We would then have to introduce a new linguistic axiom to the effect that anything that V's quickly also V's. In fact, an infinite number of axioms of this kind would have to be added. For ex
ample, we would have to add an axiom to the effect that those who V quickly and gracefully also V quickly and V gracefully in order to capture the fact that (11) logically entails (7) and (8):
(11) John walked quickly and gracefully.
Davidson's theory captures all of these implications without the need to add additional axioms. The logical form of (11) can be seen in (12):
(12) There is a walking, and it is by John, and it is quick, and it is graceful.
In addition, Davidson's theory handles tense easily. The past tense of (7)–(12) can be captured simply by adding a clause to the effect that the walking is past.
Davidson's theory gives us good reason to extend the scope of Truthmaker Theory to include all atomic truths containing an action-verb, like ‘walks’, since any such sentence can meaningfully be modified by adding an adverb. The verb ‘to be’, in contrast, does not naturally accept adverbial modification. In English, we can say ‘John is wise’ but not ‘John is wisely’. We can sometimes modify a verb phrase that contains ‘to be’, however. We can say ‘John is effortlessly wise’, for example. This suggests that the truthmaker principle can be applied to statements of condition or state as well as to action-statements.
Jaegwon Kim (1973, 1976) proposed a theory of events that can also be taken as a theory of truthmakers. Kim proposed that there is an event whenever there is a particular that bears a property at a time. This corresponds to the proposal that every statement of the following kind has a truthmaker that is (at least potentially) a cause or effect:
(13) Particular A has property P at time t.
Of course, Kim's proposal raises several metaphysical questions: what are particulars, properties, and times? Presumably, we should interpret the Kimian theory as applying only to fundamental particulars, properties, and times.
If events or causal relata are truthmakers, then each theory of truthmakers corresponds to a theory of the relata of causation. As we saw in earlier chapters, there are three principal accounts of the nature of truthmakers: tropes (both modules and modifiers), nexuses (between universals and particulars), and states of affairs (constituted by universals and particulars).
27.2.1.2 Fine-Grained Causal Explanation and Truthmakers.
The phenomenon of fine-grained causal explanation provides further support for extending the truthmaker principle to all true statements involving the predication of fundamental properties. In explaining things causally, we often appeal, not only to what actions have been done, but also to how they are done. For example, L.A. Paul (2004) has argued that we can explain Susan's writing slowly by appealing to her broken arm, and her writing insightfully by appealing to her intelligence. The insightfulness of her writing can have effects that are not shared by the slowness of her writing. Thus, the causal structure involves more than just actions, where these actions are individuated coarsely (by, for example, the main verb occurring in the corresponding true statement).
(14), (15) and (16) can have different causes and effects:
(14) Susan wrote an article.
(15) Susan wrote an article insightfully.
(16) Susan wrote an article slowly.
If so, and if causal relata are truthmakers, then we must suppose that these three truths have different truthmakers. For example, the truthmaker for (15) must include some connection (by way of a trope, nexus or state of affairs) between Susan's writing and the property of intelligence. Neither (14) nor (16) require such a connection.
27.2.2 Objection to Causal Connectionism: Negative causation
The weightiest objection to Causal Connectionism concerns the phenomenon of negative causation. Causal Connectionism entails that causation is always a matter of some real relation between things (events or truthmakers). However, there are many cases in which we appeal to absences and other negations, like holes, shadows, privations, and inaction, to explain real phenomena. In those cases, there doesn't seem to be any real entity to bear the causal relation to the effect.
Causal Connectionists can respond in either of two ways. First, they can deny the reality of negative causation altogether. Second, they can attempt to provide a Connectionist account of negative causation. This accounts can take one of two forms. Either the account attempts to accommodate real negative causation without introducing absences or other negative entities or the account posits the real existence of negative things. We'll consider each of these responses in turn.
27.2.2.1 Is Negative Causation Real?
Schaffer (2004) has provided a number of compelling examples of negative causation:
Death by heart attack. People who die as a result of a heart attack die because their hearts are not pumping blood, which results in a shortage of oxygen in the bloodstream, which in turn leads to the failure of the cells to perform certain activities essential to the maintenance of life.
Muscle contraction. The contraction of muscles is caused by the transmission of an electrical signal through the nervous system, which in turn involves the successive activation of synapses between serial nerve cells. The activation of a synapse occurs as the result of the removal of certain chemical obstacles to the movement of ions across the gap.
Firing of a gun. Pushing the trigger removes an obstacle to the expanding of a spring, which triggers the firing of the bullet.
Bombs can be designed in which pushing a button removes the shielding of one element from another within a bomb, resulting in an explosion.
Such examples can be multiplied. Droughts can cause famine by depriving plants of the water they need. As Schaffer argues, these are central and not peripheral examples of what we mean by ‘causation’.
27.2.2.2 Can Connectionists give an Account of Negative Causation?
It would seem, then, that Connectionists must provide some sort of account of negative causation. There are two avenues to explore: posit negative entities, such as real absences, to serve among the causal relata, or provide a disunified or disjunctive account. A disunified theory would have different implications for positive and negative causation. Positive causation would involve a connection between two entities, while negative causation would not.
a. Disunified theories
A disunified theory of causation would have to be Connectionist about positive causation and Explanationist about negative causation. Phil Dowe defends such a view in Physical Causation (2000). All such accounts face an obvious challenge: if Explanationism is good enough for negative causation, why isn't it good enough for positive causation as well? The danger is that the arguments that Connectionists offer against the adequacy of Explanationism in the case of positive causation will have exact counterpart arguments against Explanationist theories of negative causation. It is relatively easy to produce cases of symmetric overdetermination or preemption involving negative causes. One's dying for lack of water can be preempted by one's dying for lack of oxygen, for example (see Koons 2003).
It would be hopeless to deny entirely that absences play any significant role in causation. However, Powerists (4.4A.3) can plausibly make a distinction between the role of positive things and their causal powers, on the one hand, and the role played by mere absences, on the other. Whenever an absence does play a role in causing a change or initiating a new process, it always does so in collaboration with some positive entities possessing appropriate active and passive causal powers. For example, when a gun is fired, the absence of the obstacle does play a role, but only in the presence of a spring or other mechanism with the active power of producing the appropriate motion in conjunction with the absence. Even when death is produced by an absence of water or oxygen, there is always a living organism with the capacity to produce, in those circumstances, an unliving corpse.
It would, therefore, be tempting for Connectionists to reduce the causal role of the absence to the fact that a certain result can be deduced from the proposition that something is absent, when that proposition is combined with propositions about the laws of nature and about the states of the other, pos
itive participants in the causal interaction. Even if absences don't exist, there do exist propositions to the effect that certain things are absent, and we can always consider what follows logically from such propositions, either alone or in conjunction with others. Connectionists could distinguish between causes and enabling conditions, arguing that absences can never be causes in the strict sense, but only conditions that enable something else to act as a cause.
However, this maneuver comes at a significant cost, since it threatens to deprive Connectionists of the argument from causal linkage against Explanationism. This problem emerges when we focus on the other form of negative causation: prevention, or causation of absences. Consider a case of two, symmetrical potential causes C1 and C2 of some event E, and suppose that the effect E is an absence, the absence of an event of type K. It still seems to make sense to ask, on each particular occasion, whether it was C1 alone, C2 alone, or C1 and C2 jointly that produced the absence? However, we will always be able to give a causal explanation of the proposition that no event of type K occurred that refers to both C1 and C2. How then can we distinguish the three possibilities?
The answer to this problem might lie in hypothesizing that effects always consist in some change or absence of change in a persisting patient. When we prevent some event from happening, this always involves our preventing some thing, the patient, from changing in certain ways. We could now suppose that the continuing existence of any patient is a temporally extended process of a certain kind. (We explain more about what this means in Chapter 28.) The causal linkage between the agent and the absence of change in the patient consists in some powerful state of the agent's being an integral part of the process of the patient's continued, unchanging existence. If C1 causes the absence and C2 does not, then some state of C1's power will be literally a part of the patient's continued and unchanging existence, while no such state of C2 will be contained in that process of continuing existence.