Transitivity of causation
There seems to be much debate about the transitivity of causation. I.e. the idea that if A cause B and B causes C then A causes C. Some philosophers say that it is always true, some say it is never true, and some say that it is true only in certain cases.
My question is: isn't this just a debate about the definition of 'causality'? Does it really matter which definition we accept? Can't we simply decide the definition?
My question is: isn't this just a debate about the definition of 'causality'? Does it really matter which definition we accept? Can't we simply decide the definition?
Comments (30)
Quoting clemogo
I don't think there's need for a debate on "causality". It's just the relation between cause and effect. If we are to debate over such things, we'll never be able to complete any discussion!
As for "transitivity", it refers to specific case. It's a property of logic applied to certain cases. If I am related to you and you are related to a 3d person it does not mean that I am also related to that person. Even if I tell you a joke, which you then tell to that person, I would have only indirectly make that person hear the joke, but it would not be a direct cause & effect, since I have no relation with that person. The real cause for that would be you.
A transitivity or direct/controlled cause & effect would be something like this: I kick a ball and the ball falls on a lamp and breaks it. I would be the cause and responsible for the damage, since it produced by my kicking of the ball, and thus myself.
What's more to say about it?
Things appear in consciousness and often one thing appears after another. where do they come from? Do they come from some permanent physical reality that continues to exist without consciousness? Or do they come from something beyond consciousness that has never been seen? or are they just qualia spontaneously generated by consciousness. Or some mix of those options.
Why, isn't "the relation between cause and effect" already a workable definition of causality?
What else do you need or look for?
Quoting tim wood
What questions, for example?
1. Necessary cause
2. Sufficient cause
3. Proximate cause
4. Remote cause
5. Contributory cause
That's all I can recall off the top of my head. Don't forget free will.
1. Domino Effect
2. Chain Reaction
3. Slippery slope
4. Camel's nose
The relation between cause & effect can be whatever, millions of things. From the part of either the cause or the effect. You just have to see what both terms mean. Simply put, a cause is something that produces an action, phenomenon, condition, change, etc. An effect is a change produced as a result or consequence of a cause (as described above). Using mathematic combinations you get a pretty huge number! (It would be ridiculous to give an example ...)
Now, if you ask about the "mechanics" of the cause & effect relation, and the level at which this is examined, I believe this is better explained with Physics (on which I'm far from being an expert!)
Quoting tim wood
Ah! I guess you are referring to the actual cause of the event, right? This is more interesting! :smile:
As a short answer to your question, the actual cause of the explosion was the man. The explanation is obvious, but still ... The dynamite was the apparent, direct cause of the explosion, but it could not do that by itself. Someone or something must have triggered it: a fuse, a timer device, a fire started near it, etc. However, whatever is the "direct" cause of the explausion, the actual cause is the man.
Now, on a second level, the action done by the man is itself an "apparent" cause. Because we can go further back in the cause & effect chain, and ask "What made the man do that?" Well, there can be a lot of reasons-causes. And behind every of these causes there can be a lot of reasons-causes ... to infinitum. This is actually the subject of a very interesting topic "A first cause is logically necessary", in which I got involved about a month ago. If you are interested, you can find my conclusion on this subject at https://thephilosophyforum.com/discussion/comment/622168.
Quoting tim wood
I agree. Maybe this is related to the "mechanics" of the cause & effect relation that I mentioned above ... But whatever is the case, cause-and-effect is always there. You can't escape it! :smile: Besides, this is the subject of the topic! :grin:
I explained to you the difference between "direct" case and "actual" cause. Maybe the word "actual" is the problem. But from the example(s) I gave it should be very clear. I could mabe use the word "real" or even better, "source", meaning the origination of the evenf (explosion).
When a person pays an executioner to kill someone, he is the responsible, the source, the actual murderer. It doesn't matter how the victim was killed. The executioner is just an instrument. The killing would not have occurred if the person had not ordered it. I think all this is plain enough.
As for your compalint about "I will now use for about the fifth time, but that no one yet takes on", well I took it on. You should at least acknowledge that. I assume that this, your lack of undesranding and your problem of passing your point through to people, are the reasons you have to try multiple times. With me at least, this is the case, and it is very clear.
Quoting clemogo
Your question is odd. Surely, if causality is more than an idle fantasy that we are making up here on the spot, then the question of whether causality is transitive is not independent of what we believe causality to be?
Using dynamite.
Oppeheimer would be the distal cause of the explosion, as described by transitivity. And the role of U-235 might also be relegated to that of a distal cause in a sufficiently fine-grained model of the explosion.
But not all descriptions of a process obey transitivity, in which case such descriptions aren't causal descriptions. For example, take the SEP's example in the metaphysics of causality article that purportedly refutes causal transitivity :
A large boulder roles down hill (A) causing a hiker to jump out of the way (B). The hiker's jump (B) "causes" him to survive (C). Therefore by transitivity the boulder rolling (A) causes the hiker's survival (C), which is a false conclusion.
But this isn't a valid argument, because although A --> B can be considered a true implication in the example, survival isn't definable in terms of acts of jumping alone, and so we don't have the implication B --> C, therefore we cannot use transitivity to derive A --> C.
What this example actually demonstrates is the situation (A and B) --> C.
That's understandable, due to historical disagreements and confusion in science as to how to formulate the notion, but things have rapidly changed in recent years as causal semantics has been steadily formalized, most notably in the structural equation modelling approach of Judea Pearl; the structure of a multi-variate probability distribution is factored into a set of conditional probability distributions, a subset of which are interpreted axiomatically as denoting known causal relations. With respect to these causal assumptions, the remaining correlations of the model can then be tested for the property of "cause and effect" through analytic methods and through additional interventional studies in cases where additional real-world data is required. However, the statistical quality of these non-deterministic models obscures the underlying logic of causality they employ, which is explicated more succinctly in Linear Logic, process algebras, and related semantics such as Petri Nets, monoidal categories and string diagrams Quoting tim wood
And yet their perspectives are compatible, no? Each actor is expressing the existence of a different marginal distribution conditioned upon their favourite independent variable, which are hopefully mutually consistent and can be added together into a combined model.
A good introduction is Judea Pearl's "Introduction to Causal Inference". The lesson is that causal implications cannot be derived from a statistical model without some initial causal assumptions.
Garbage causal assumptions in, garbage causal inferences out.
Quoting tim wood
Sure, and to make matters worse, intuition is often wrong with respect to logical and statistical inference. Hence the reason why formal definitions and theorem provers are useful whereby informal causal intuition is reduced to axiomatic systems, even though philosophical dilemmas remain e.g with regard to counterfactual reasoning.
What critiques would you have for these definitions (all leading up to those of “cause” and “causation”):
Here keeping things as simple as possible but no simpler, I’m hoping.
If the dynamite stick was hit by lightning, then it is safe to say that the lightning caused (was the source of) the dynamite’s explosion (as outcome, aka effect). If the dynamite stick had a fuse that was lit by a human, then it is safe to conclude that the human caused (was the source of) the dynamite’s explosion (as outcome). And so forth, depending on scenario.
Yes, there is the conceivable metaphysical possibility that no causation – as here defined – occurs. To be brief, living by upholding this metaphysical possibility to be reality in non-hypocritical manners would quickly lead to death. That (what we cognize as) outcomes have (what we cognize to be) sources is indispensable to life as we know it. Hence, if we desire to live, we will assume, if not know, that causation occurs in the world.
As to probabilistic causality, I find that it shares many merits, as well as many logical detriments, with the Buddhist notion of dependent origination. But this likely furthers the subject from the OP. Interesting topic to me though.
To my mind, this is fully contingent on the types of causes that are being contemplated. Not merely on definitions but on the type or reality we deem ourselves to live in.
In a system of causal (hard) determinism, all causes will by default be transitive without exception. In the murky realms of what nowadays gets the blanket label of causal indeterminism, things can logically vary a lot more in terms of types of causes but, in short, not all causes here will be transitive (depending on indeterministic system assumed, if any).
Still, I'll argue that regardless of further conceptualization, all causes, regardless of system, will be sources to outcomes, i.e. to effects.
Many emergency and law enforcement personnel are being taken out of commission so to speak by this virus. Spikes in crime, fire, accident and non-Covid disease-related deaths should be part of the overall pandemic syndrome (caused by collapse of basic services that lead to indirect death/injury).