- In 1992, as part of a larger charitable campaign, the Prince of Wales (Prince Charles, Queen Elizabeth's older son and heir) launched a line of organic food products called 'Prince's Duchy Originals'. The first product that went on sale was an oat cookie: 'the oaten biscuit.' Since then the oaten biscuit has been joined by hundreds of other products and Duchy Originals has become one of the leading organic food brands in the United Kingdom. Presumably, the Prince of Wales is very proud of his Duchy Originals products, and of the oaten biscuits in particular. Let's imagine that he is so proud of the biscuits that he eats them regularly. Also, let's imagine that one day Queen Elizabeth asks the prince to water her plant. As she explains to him, she'll be gone for the day and the plant needs to be watered every afternoon. But the prince decides not to water the plant. Instead of watering it, he spends his afternoon savouring some oaten biscuits, and the plant dies.
- What caused the plant’s death? If you were to ask the queen, she would presumably say: the prince, plus some 'natural causes' (including the fact that the plant was particularly delicate and needed intensive watering). Now, in virtue of what could the prince be a cause of the plant's death? When we say that an agent caused some event in the world, we typically mean to say that there is something that the agent did, or something that the agent failed to do, which caused the outcome. There are several things that the prince did and failed to do that afternoon: he ate some oaten biscuits, he read the newspaper, he scratched his nose, he didn't phone a friend, he didn’t watch TV, he didn't water the queen’s plant, and so on. Among these, we clearly want to say that his not watering the plant is relevant to the plant's death: the plant died because he didn't water it. Under slightly different circumstances, some of the things he did would also be relevant. Imagine, for instance, that the oaten biscuits are so amazingly good that they induce some kind of psychological trance that makes you forget any obligations that you might have. So maybe the prince was determined to water the plant until he ate the biscuits, at which time he forgot all about it. In that scenario his eating the biscuits would also be a cause of the plant's death. But note that, even in that case, the prince's contribution to the plant's death is ultimately 'negative' in nature. For his eating the biscuits causes the plant's death by means of causing the prince's subsequent failure to water the plant. At the end of the day, the plant still dies because of something that the Prince doesn't do: it dies because he doesn't water it.
- Scenarios of this kind suggest that omissions, and absences in general, can be causes, and that our reconstructions of the causal histories of the outcomes are somehow flawed if they don't include the omissions of agents or the absences of certain events but instead include only 'j five’ causes. For, again, in these cases, the outcomes seem to happen, at least partly, because of something that someone doesn't do, or because of something that doesn't happen, not (just) because of something that someone does, or because of something that actually happens. I will call the apparent failure of positive causes to adequately account for the outcome's occurrence in these cases 'the inadequacy fact about positive causes.' The inadequacy fact about positive causes is an important motivation for accepting negative causes.
- Now, assuming that we want to make room for negative causes, how could we make sense of omissions and absences being causes? A natural thought is to appeal to the notion of counterfactual dependence. We can say that the prince's not watering the plant is a cause of the plant's death because the plant's death counterfactually depends on the prince's failure to water it: had his failure to water the plant not occurred (i.e., had he watered the plant), the plant wouldn't have died. In other words: in the closest possible world(s) where the prince waters the plant, the plant doesn't die. Counterfactual theories of causation claim that the causal facts are grounded in facts about counterfactual dependence. On these views, causes are ‘difference-makers’ with respect to their effects in that effects (at least typically) counterfactually depend on their causes (Lewis 1986a).
- Now, this idea has to be refined in two kinds of ways.1 First, as cases of 'pre-emption' suggest, sometimes effects don't counterfactually depend on their causes. For example, an assassin can cause his victim's death even if the death would still have happened if he hadn't shot him, given that a backup assassin would then have shot the victim himself. This suggests that counterfactual dependence is not necessary for causation. At least originally, Lewis thought that we can sidestep this problem by taking causation to be, not simple counterfactual dependence, but the ancestral of counterfactual dependence. Second, counterfactual dependence also doesn't seem to be sufficient for causation: some counterfactual dependencies track 'tighter,' non-causal connections, such as logical and mereological relations. For example, my writing the word 'cat' counterfactually depends on my writing the letter 'c' but my writing 'c' isn't a cause of my writing 'cat' (Kim 1973). So the relevant concept of counterfactual dependence would have to be circumscribed accordingly. Lewis does this by setting constraints on potential causes and effects. On Lewis's view, a necessary condition for C to cause E is that C and E be fully 'distinct,' where C and E are not fully distinct if, for example, one is part of the other. Also, some counterfactual conditionals express counterfactual relations that aren't causal because they arc backtracking - as when I say 'If my friend had invited me to his birthday party today, then we wouldn't have had a fight yesterday.' Lewis's suggestion is that we should restrict our focus to ordinary or standard contexts, in which backtracking counterfactuals aren't true.
- In spite of these problems, the claim that counterfactual views have at least identified a sufficient condition for causation once counterfactual dependence has been restricted in these ways has seemed quite plausible to people. In particular, at least in recent times, it has seemed much more plausible than the converse claim that counterfactual dependence, or something close to it, is necessary' for causation (the consensus seems to be that counterfactual theories have a really hard time addressing the pre-emption problem). From now on I will focus on the sufficiency claim only. I'll call it 'the counterfactual criterion':
(CC) If there is counterfactual dependence of the ordinary (nonbacktracking) kind between C and E, and if C and E are fully 'distinct' (e.g., they are not logically or mereologically related), then C is a cause of E. CC seems to be initially plausible: if E counterfactually depends on C, then C is a difference-maker with respect to E - it makes the difference between E's occurring and E's not occurring - and so (if the counterfactual dependence is of the ordinary kind and if C and E are fully distinct) it is plausible to think that C is one of the things (among potentially multiple things) that causally contributed to E's occurrence.
- Given CC's initial plausibility, an advantage of counterfactual theories seems to be that they have the basic resources to accommodate causation by omission, which many other theories lack. For example, theories according to which a causal relation requires the transfer of some physical quantity, like energy or momentum (Salmon 1994), or any other kind of physical interaction, don’t have the resources to do this. For there is no physical interaction between the Prince and the plant in virtue of which he caused the plant's death. Counterfactual dependence, by contrast, doesn't require the existence of physical interaction: on the basis of CC, we can say that the Prince caused the plant's death even if he never physically interacted with it. So the ability to accommodate causation by omission appears to be at least a prima facie advantage of counterfactual theories over theories that don't allow- for this type of causation.
- I will argue that this is a misconception. I will argue that, despite appearances to the contrary, the ability to accommodate causation by omission is not a prima facie advantage of counterfactual views, at least to the extent that we take the main motivation for believing in causation by omission to be the inadequacy fact about positive causes (as I am assuming we do). For I will argue that, even if omissions are causes, and even if counterfactual views can accommodate causation by omission, those views still fail to respect the inadequacy fact about positive causes. Although my main focus will be counterfactual theories of causation, in the final section I will suggest that the arguments of this chapter apply, more generally, to theories that attempt to account for the contribution of agents' omissions in counterfactual terms, regardless of whether this is a causal contribution or not.
- I have argued that counterfactual views of causation cannot accommodate causation by omission while remaining faithful to the motivation for accepting that kind of causation. In this final section I attempt to generalize this result.
- Some philosophers reject the possibility of causation by omission. Still, some of these philosophers feel the need to provide an account of the relation that omissions can bear to outcomes, in virtue of which agents can be morally responsible for those outcomes. Notably, Dowe has offered an account of a relation of this kind, which he called 'quasi-causation' (Dowe 2000, 2001, 2004). According to Dowe, omissions can quasi-cause outcomes, and it is in virtue of this relation of quasi-causation that agents can be morally responsible by omission. Quasi-causation is a counterfactual relation. Basically, whereas for a counterfactual theorist of causation who believes in causation by omission counterfactual dependence between an omission and an outcome is the mark of a causal relation between the omission and the outcome, for Dowe it is the mark of a quasi-causal relation. As far as I can see, everything I’ve said about the prospects of a counterfactual theory of causation can be said, mutatis mutandis, about a counterfactual theory of quasi-causation like Dowe’s.
- Briefly, for someone like Dowe, the problem arises as follows. We think that agents can be responsible for outcomes in the world in different ways. Sometimes they are responsible in virtue of having caused those outcomes. Other times they are responsible without causing those outcomes. So we need to find a new way to account for the agents' responsibility in these scenarios. Let’s call this fact 'the inadequacy fact about causes.' The inadequacy fact about causes motivates the search for a new theory, a theory of 'quasi-causation.' It is natural to try to give such a theory in counterfactual terms. But a counterfactual theory of quasi-causation would face the Prince of Wales problem. For similar arguments to those offered here would show that there are 'fake' counterfactual dependencies between genuine causes and the upshots of quasi-causes as well as between quasi-causes and the upshots of genuine causes. As a result, a counterfactual theory of quasi-causation would fail to respect the initial motivation for giving a theory of that kind, i.e. the inadequacy fact about causes.
- To conclude: the Prince of Wales problem is not just a problem for counterfactual theories of causation. It is a more general problem that arises for any theory that attempts to understand the contribution of omissions in counterfactual terms.
Text Colour Conventions (see disclaimer)
- Blue: Text by me; © Theo Todman, 2020
- Mauve: Text by correspondent(s) or other author(s); © the author(s)