Ideas from 'The Metaphysics of Causation' by Jonathan Schaffer [2007], by Theme Structure

[found in 'Stanford Online Encyclopaedia of Philosophy' (ed/tr Stanford University) [plato.stanford.edu ,-]].

green numbers give full details    |     back to texts     |     unexpand these ideas


5. Theory of Logic / E. Structures of Logic / 1. Logical Form
Logical form can't dictate metaphysics, as it may propose an undesirable property
                        Full Idea: Logical form should not have the last word in metaphysics, since it might predicate a property that we have theoretical reason to reject.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 1.3.1)
                        A reaction: These kind of warnings need to be sounded all the time, to prevent logicians and language experts from pitching their tents in the middle of metaphysics. They are welcome guests only,
7. Existence / D. Theories of Reality / 8. Facts / b. Types of fact
There is only one fact - the True
                        Full Idea: It can be argued that if all facts are logically equivalent, then there is only one fact - the True.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 1.1)
                        A reaction: [he cites Davidson's 'Causal Relations', who cites Frege] This is the sort of bizarre stuff you end up with if you start from formal logic and work out to the world, instead of vice versa.
26. Natural Theory / C. Causation / 1. Causation
In causation there are three problems of relata, and three metaphysical problems
                        Full Idea: The questions about causation concern their relata (in space-time, how fine-grained, how many?) and the metaphysics (distinguish causal sequences from others, the direction of causation, selecting causes among pre-conditions?).
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], Intro)
                        A reaction: A very nice map (which has got me thinking about restructuring this database). I can't think of a better way to do philosophy than this (let's hear it for analysis - but the greatest role models for the approach are Aristotle and Aquinas).
Causation may not be transitive; the last event may follow from the first, but not be caused by it
                        Full Idea: It is not clear whether causation is transitive. For example, if a boulder roll's towards a hiker's head, causing the hiker to duck, which causes the hiker to survive, it does not seem that the rolling boulder causes the survival of the hiker.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 1.2)
                        A reaction: Maybe survival is not an event or an effect. How many times have I survived in my life? We could, though, say that the hiker strained a muscle as he or she ducked. But then it is unclear whether the boulder caused the muscle-strain.
There are at least ten theories about causal connections
                        Full Idea: Theories of causal connection are: nomological subsumption, statistical correlation, counterfactual dependence, agential manipulability, contiguous change, energy flow, physical processes, property transference, primitivism and eliminativism.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 1.3.1)
                        A reaction: Schaffer reduces these to probability and process. I prefer the latter. The first two are wrong, the third right but superficial, the fourth wrong, the fifth, sixth and seventh on the right lines, the eighth wrong, the ninth tempting, and the last wrong.
26. Natural Theory / C. Causation / 4. Naturalised causation
Causation transcends nature, because absences can cause things
                        Full Idea: The main argument for causation being transcendent (rather than being immanent in nature) is that absences can be involved in causal relations. Thus a rock-climber is caused to survive by not falling.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 1.1)
                        A reaction: I don't like that. The obvious strategy is to redescribe the events. Even being hit with a brick could be described as an 'absence of brick-prevention'. So not being hit by a brick can be described as 'presence of brick prevention'.
Causation may not be a process, if a crucial part of the process is 'disconnected'
                        Full Idea: One problem case for the process view of causation is 'disconnection'. If a brick breaks a window by being fired from a catapult, a latch is released which was preventing the catapult from firing, so the 'process' is just internal to the catapult.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.1.1)
                        A reaction: Schaffer says the normal reply is to deny that the catch-releasing is genuinely causal. I would have thought we should go more fine-grained, and identify linked components of the causal process.
A causal process needs to be connected to the effect in the right way
                        Full Idea: A problem case for the process view of causation is 'misconnection'. A process may be connected to an effect, without being causal, as when someone watches an act of vandalism in dismay.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.1.1)
                        A reaction: This is a better objection to the process view than Idea 10377. If I push a window with increasing force until it breaks, the process is continuous, but it suddenly becomes a cause.
Causation can't be a process, because a process needs causation as a primitive
                        Full Idea: It might be that if causation is said to be a process, then a process is nothing more than a causal sequence, so that causation is primitive.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.1.2)
                        A reaction: This again is tempting (as well as the primitivist view of probabilistic causation). If one tries to define a process as mere chronology, then the causal and accidental are indistinguishable. I take the label 'primitive' to be just our failure.
26. Natural Theory / C. Causation / 5. Direction of causation
At least four rivals have challenged the view that causal direction is time direction
                        Full Idea: The traditional view that the direction of causation is the direction of time has been challenged, by the direction of forking, by overdetermination, by independence, and by manipulation, which all seem to be one-directional features.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 1.3.1)
                        A reaction: Personally I incline to the view that time is prior, and fixes the direction of causation. I'm not sure that 'backward causation' can be stated coherently, even if it is metaphysically or naturally possible.
Causal order must be temporal, or else causes could be blocked, and time couldn't be explained
                        Full Idea: Reasons for causal order being temporal order are that otherwise the effect might occur but the cause then get prevented, ..and that they must be the same, because the temporal order can only be analysed in terms of the causal order.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.2)
                        A reaction: If one took both time and causation as primitive, then the second argument would be void. The first argument, though, sounds pretty overwhelming to me.
Causal order is not temporal, because of time travel, and simultanous, joint or backward causes
                        Full Idea: Reasons for denying that causal order is temporal order are that time travel seems possible, that cause and effect can be simultaneous, because joint effects have temporal order without causal connection, and because backward causation may exist.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.2)
                        A reaction: The possibility of time travel and backward causation can clearly be doubted, and certainly can't be grounds for one's whole metaphysics. The other two need careful analysis, but I think they can be answered. Causation is temporal.
26. Natural Theory / C. Causation / 6. Causation as primitive
Causation is primitive; it is too intractable and central to be reduced; all explanations require it
                        Full Idea: Primitivism arises from our failure to reduce causation, but also from causation being too central to reduce. The probability and process accounts are said to be inevitably circular, as they cannot be understood without reference to causation.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.1.2)
                        A reaction: This is very tempting. The primitive view, though, must deal with the direction problem, which may suggest that time is even more primitive. Can we have a hierarchy of primitiveness? To be alive is to be causal.
If causation is just observables, or part of common sense, or vacuous, it can't be primitive
                        Full Idea: The three main objections to causation being primitive are that causation can't be anything more than what we observe, or that such a primitive is too spooky to be acceptable, or that primitivism leads to elimination of causation.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.1.2)
                        A reaction: [summarised] I don't like the first (Humean) view. I suspect that anything which we finally decide has to be primitive (time, for example) is going to be left looking 'spooky', and I suspect that eliminativism is just Humeanism in disguise.
26. Natural Theory / C. Causation / 7. Eliminating causation
Causation is utterly essential for numerous philosophical explanations
                        Full Idea: Causation can't be eliminated if it is needed to explain persistence, explanation, disposition, perception, warrant, action, responsibility, mental functional role, conceptual content, and reference. It's elimination would be catastrophic.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.1.2)
                        A reaction: [compressed list] I think I am going to vote for the view that causation is one of the primitives in the metaphysics of nature, so I have to agree with this. Most of the listed items, though, are controversial, so eliminativists are not defeated.
The notion of causation allows understanding of science, without appearing in equations
                        Full Idea: The concepts of 'event', 'law', 'cause' and 'explanation' are nomic concepts which serve to allow a systematic understanding of science; they do not themselves appear in the equations.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.1.2)
                        A reaction: This is a criticism of Russell's attempt to eliminate causation from science. It shows that there has to be something we can call 'metascience', which is the province of philosophers, since scientists don't have much interest in it.
26. Natural Theory / C. Causation / 8. Particular Causation / a. Observation of causation
If two different causes are possible in one set of circumstances, causation is primitive
                        Full Idea: Causation seems to be primitive if the same laws and patterns of events might embody three different possible causes, as when two magicians cast the same successful spell, each with a 50% chance of success, and who was successful is unclear.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.1.2)
                        A reaction: I'm cautious when the examples involve magic. It implies that the process that leads to the result will be impossible to observe, but if magic never really happens, then the patterns of events will always be different.
If causation is primitive, it can be experienced in ourselves, or inferred as best explanation
                        Full Idea: The view that causation is primitive can be defended against Humean critics by saying that causation can be directly observed in the will or our bodies, or that it can be inferred as the best explanation.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.1.2)
                        A reaction: I like both views, and have just converted myself to the primitivist view of causation! I can't know the essence of a tree, because I am not a tree, but I can know the essence of causation. The Greek fascination with explaining movement is linked.
26. Natural Theory / C. Causation / 8. Particular Causation / b. Causal relata
Events are fairly course-grained (just saying 'hello'), unlike facts (like saying 'hello' loudly)
                        Full Idea: Events are relatively coarse-grained, unlike facts; so the event of John's saying 'hello' seems to be the same event as John's saying 'hello' loudly, while they seem to be different facts.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 1)
                        A reaction: The example seems good support for facts, since saying 'hello' loudly could have quite different effects from just saying 'hello'. I also incline temperamentally towards a fine-grained account, because it is more reductivist.
Causal relata are events - or facts, features, tropes, states, situations or aspects
                        Full Idea: The standard view make causal relata events (Davidson, Kim, Lewis), but there is considerable support for facts (Bennett, Mellor), and occasional support for features (Dretske), tropes (Campbell), states of affairs (Armstrong), and situations and aspects.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 1)
                        A reaction: An event is presumed to be concrete, while a fact is more abstract (a proposition, perhaps). I'm always drawn to 'processes' (because they are good for discussing the mind), so an event, as a sort of natural process, looks good.
One may defend three or four causal relata, as in 'c causes e rather than e*'
                        Full Idea: The view that there are two causal relata is widely assumed but seldom defended. But the account based on 'effectual difference' says the form is 'c causes e rather than e*'. One might defend four relata, in 'c rather than c* causes e rather than e*'.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 1)
                        A reaction: [compressed] This doesn't sound very plausible to me. How do you decide which is e*? If I lob a brick into the crowd, it hits Jim rather than - who?
If causal relata must be in nature and fine-grained, neither facts nor events will do
                        Full Idea: Theorists who reject both events and facts as causal relata do so because the relata must be immanent in nature, and thus not facts, but also fine-grained and thus not events.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 1.2)
                        A reaction: Kim, however, offers a fine-grained account of events (as triples), and Bennett individuates them even more finely (as propositions), so events might be saved. Descriptions can be very fine-grained.
The relata of causation (such as events) need properties as explanation, which need causation!
                        Full Idea: The primitivist about causation might say that the notion of an event (or other relata) cannot be understood without reference to causation, because properties themselves are individuated by their causal role.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.1.2)
                        A reaction: Having enthusiastically embraced the causal view of properties (see Shoemaker and Ellis), I suddenly realise that I seem required to embrace primitivism about causation, which I hadn't anticipated! I've no immediate problem with that.
26. Natural Theory / C. Causation / 8. Particular Causation / d. Selecting the cause
Our selection of 'the' cause is very predictable, so must have a basis
                        Full Idea: The main argument against saying that there is no basis for selecting the one cause of an event is that our selections are too predictable to be without a basis.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.3)
                        A reaction: The problem is that we CAN, if we wish, whimsically pick out any pre-condition of an event for discussion (e.g. the railways before WW1). I would say that sensitivity to nature leads us to a moderately correct selection of 'the' cause.
Selecting 'the' cause must have a basis; there is no causation without such a selection
                        Full Idea: Another argument against the view that there is no basis for selecting 'the' cause is that we have no concept of causation without such a selection.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.3)
                        A reaction: Good. Otherwise we could only state the conditions preceding an event, and then every event that occurred at any given moment in a region would have the same cause. How can 'the' cause be necessary, and yet capricious?
26. Natural Theory / C. Causation / 8. Particular Causation / e. Probabilistic causation
The actual cause may make an event less likely than a possible more effective cause
                        Full Idea: If Pam threw the brick that broke the window, then Bob (who refrained) might be a more reliable vandal, so that Pam's throw might have made the shattering less likely, so probability-raising is not necessary for causation.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.1)
                        A reaction: That objection looks pretty conclusive to me. I take the probabilistic view to be a non-starter.
All four probability versions of causation may need causation to be primitive
                        Full Idea: All four probability versions of causation may need causation to be primitive: nomological - to distinguish laws from generalizations; statistical - to decide background; counterfactual - decide background; agent intervention - to understand intervention.
                        From: Jonathan Schaffer (The Metaphysics of Causation [2007], 2.1.2)
                        A reaction: I don't need much convincing that the probabilistic view is wrong. To just accept causation as primitive seems an awful defeat for philosophy. We should be able to characterise it, even if we cannot know its essence.