The Nature of Causation: The Counterfactual Theory of Causation 

Just google this: RU-vid nature of causation given at Oxford in 2016

My gut would say yes. If there are a sufficient number of counterfactual, it would seem that that is a primary factor for something not being an explanation for something. For instance, a particular camper being born is not a sufficient explanation for the forest fire given how many other campers had been born as well, yet this would not result in forest fires in their cases. I think that a large part of an explanation is the probability of a cause causing the outcome. Probabilities constructed from an estimation of counterfactuals. The reason that we don't like accepting placing the bomb as a cause for Suzy's survival is the weight of the counterfactuals where she doesn't survive the bomb as opposed to the actual reality that she factually did and that the placing of the bomb is a cause of her survival. I think that explanations should be considered not only what is intelligible to a person, but also what is the sum total of the probability, based on the counterfactuals that an agent is aware of.

why bother explaining an explanation when the explanation is not true? A does not cause B. B's cause is dependent on the entire timeline of happenings in the universe. our notion of A is the cause of B is a fiction. fictions have no actual truths.

Actually, the audience questions are more keen than one may presume at first glance. She in fact misses the opportunity to discuss very deep issues about causation upon firmly correcting them. One of the most important questions the public asked is about the planting of the bomb being or not being the cause of Suzy's survival. Their intuitions are right all the way through and she being so involved in her topic is committing an epistemic mistake. She presuppose the counterfactual dependence (or the influence version of it) to explain that the enemy planting the bomb is in the causal chain to Suzy's survival, and adding transitivity of causation ,(which can be hardly objected) then she arrives to the conclusion that planting the bomb is "a" cause of Suzy's survival. But the point is that she is presupposing a model of causation to explain it. The main problem is that in that particular case, one cannot even say that the planting of the bomb is the cause of Billy's stopping the explosion. So there is a fake transitivity there she is wrongly appealing to. I'm not saying that I have a theory to explain why the planting of the bomb didn't cause the fact that Billy prevented the explosion. But the reason why planting the planting of the bomb is not the cause of Suzy's survival is indeed the former fact (taken from common sense). If we want to get a further explanation of why planting the bomb didn't cause Suzy's survival, we can say everything being equal planting a bomb would cause an explosion. Now, Billy's getting in the middle does not prevent the planting of the bomb from being a cause of a possible explosion. And this is quite in line with Lewis's counterfactual theory of causation based on possible worlds. Logically, something cannot cause both an event and its absence. So, the planting of the bomb cannot both cause a possible explosion and a possible lack of it. You have to pick one of the ramifications. The one that is more commonsensical is obviously that the bomb is a cause of a possible explosion. Therefore, blocking the transitivity to Suzy's survival. Now, as for why the planting of the bomb is not the cause of Billy's deed we have a problem, since his action takes place earlier than the explosion that should have happened had he refrained from acting. Therefore, the former argument would be useless. In the new case, I suspect that the problem is coming from the involvement of an agent with free choice. Surely, seeing the bomb there prompts Billy's action to put off the bomb but we can hardly say that such an stimulus is the cause. One can even rationalize this within the counterfactual theory by Lewis. In Lewis's theory, there is a a causal dependence from a to b iff you have both a counterfactually depending on b and not a counterfactually depending on not b. Typically, if a and b are actual events, the former is automatically true (within the theory) and you are left proving the second with the possible world framework. However, when b is an action (from a free-willed agent) that doesn't seem to work. Even though the events are actual there are possible worlds immediately closer to the actual one where Billy refrained from stopping the bomb (perhaps out of fear). So the fact that the event is actual seems to be insufficient to grant that the counterfactual from a (seeing the bomb) to b (putting it off) assume a robust truth value at the actual world.

The argument from the students is valid but not sound because they are focused mainly on the central fact of the story, where Lucy breaks the window and not the added fact that Brian would have. This extension (Brian would have) is vital to claim that the window shattering is not counterfactually dependent on Lucy's rock throwing. Because if Brian never existed, then it must be the case that the window shattering is counterfactually dependent on Lucy's rock throwing. However, Brian exists in the 'factual story' where Lucy breaks the window and will break the window if she doesn't. Therefore, the 'counterfactual story' to this one is where Lucy's rock doesn't shatter the window, but the window shatters anyway because of Brian's proceeding throw. Thus, making the 'counterfactual story' dependent on Brian. I hope this clears that up for you.

So the point that you’re missing is also the same point that they don’t understand in this example. Counterfactual’s or hypothetical and don’t actually exist in real life. In the real world what happens a.k.a. causation is a math equation with time. Meaning it is directional. And therefore because there’s always one outcome or one thing that happens in reality, counterfactual’s don’t actually exist. I appreciate the time you talk to reply to my message. Hope this helps you out.

​@@aren8798 The 'concept' of a counterfactual exists. That concept points to a reality that would have happened if not for the one we experience. We don't talk about the real world when discussing things that would be 'counterfactual' to it. The point of the speaker's story about Lucy's rock-throwing not being needed for the window to shatter comes from the simple addition of 'Brian would have shattered it.' This seemingly-negligible bit stems from a real-world observation of Brian's rock immediately following behind Lucy's. Thus making it valid and sound to say that the window shattering isn't counterfactually dependent on Lucy. Giving another scenario. If you and I are out on the range shooting the same clay target, and I shoot less than a second before you do, and it shatters, that clay target shattering isn't counterfactually dependent on my shot because your immediate shot after mine 'would have' hit it. I can say this because I witnessed your shot hit the raised dirt mound behind the target. There only needs to be one 'possible' condition met to say that the outcome of something isn't counterfactually dependent on you. If someone is murdered, we don't say that person dying is counterfactually dependent on the murderer because we know that we all have a biological clock.

What's arbitrary about it? It's just relative. Computer programs make for a perfect analogy. Certain programs can be run by (i.e. are intelligible to) certain computers but not others. Which computers can run which programs isn't arbitrary-that's fixed by facts about the computers and the programs. Still, even though it's not arbitrary, a program has no absolute intelligibility; it only has intelligibility relative to a certain kind of computer. The algorithm you'd like, I think, is really just a definition of causation. Using this to program computers to select proper causes out of the causal history would just be using causation to generate explanations. In order for any of that to work, whether on a computer or in a lecture hall, you have to be clear about your terms and what you're doing with them. Being "nit-picky" is really just avoiding sloppiness. The goal is to recognize incorrect outputs or, better yet, not to make the mistakes that yield those incorrect outputs in the first place.

maybe it has to do with the means of expression. (English) language is not precise, so when used to explain precise concepts it gets verbose and boring. The definitions of the terms seem to be very different between audience members. For example, I define "surving" as something that requires both a threat to life and escape from the threat. The discussion on whether the bomb was causal to survival, seemed unnecessary to me, because it hinged on the difference of definitions between "survived" and "lived". Programming languages require precise definitions, so the "survival" variable would explicitly compile out of "threats" and "threat_escape" variables making it easy to argue about events (causes) which contributed to "survival".

Thank you... I thought I was the only one with those thoughts.

Wasn't that the whole point? The scenario is supposed to be a counterexample to causation being counterfactual dependence. And the fact that there is causation but no counterfactual dependence means that causation cannot be counterfactual dependence.

I think there is a counterfactual dependency with Lucy's throw. When we look for counterfactual dependency we want to know if the effect would have happened anyway. But by that we mean did something else cause the window to break. Would the window have broken if Lucy hadn't thrown the rock? No, not without something else causing it instead. Edited to fix mixing up Brian and Lucy.

@@stephenlawrence4821 That is the point. Despite Lucy's throw causing the the window to smash, Brian's rock would have caused the window to smash if Lucy's did not, thus there is no counterfactual dependence on Lucy's throw. Counterfactual dependence requires that if Lucy had not thrown the rock, the window would not smash. But it would have done.

There isn't. An increase in the number of people born doesn't correspond to an increase (or a decrease) in the number of forest fires started by people. That's what would be sufficient for correlation. It's not enough that the people who do start forest fires were born. If x% of people born were destined to start forest fires or if some significant number of people started a forest fire for every n births, then there would be a correlation, but that's not the case.

Brian's throwing a rock isn't extraneous. It's essential. Without it, you're talking about a different situation. What would a proper scientific experiment without Brian be testing? I think the problem comes from two different analyses of events. The lecturer insists on treating the example as consisting of three events: (1) Lucy's throwing, (2) Brian's throwing, and (3) the window's reacting to Lucy's and Brian's throwing. The audience member, I think, is treating it as two events: (1) Lucy's throwing and the window's reacting to it and (2) Brian's throwing and the window's reacting to that. (Or maybe she considers it to be 4 events: Lucy's throwing, the window's reacting to Lucy's throwing, Brian's throwing, and the window's reacting to Brian's throwing.) The problem isn't from extraneous factors but from the unclear boundaries of events or perhaps their infinite analyzability. This is why the audience member goes on to say "it works out if you view it from _the window breaking_ as one event..." She realizes that the example shows what it's supposed to when the window's reacting to Lucy's and Brian's throwing is treated as its own, third event (assuming I understand her correctly).

Really unfair, I really enjoy her way of teaching difficult concepts to grasp.

Because she does not know - in the olden days we used to call it "tap dancing"