Causation and Quantum Mechanics

4. Causation Causation and Quantum Mechanics Fredrik Andersen, Rani Lill Anjum, Stephen Mumford (in What Tends to Be. The Philosophy of Dispositional Modality, Routledge 2018) 1. The challenges of quantum mechanics Quantum mechanics introduces us to an unfamiliar and at times puzzling reality. Experimental results suggest that sub-atomic particles can behave in ways that violate classical laws of physics; in quantum entanglement, wave-particle duality, quantum jumps and superposition, to mention only some. A number of interpretations and theories have been offered as attempts to explain such phenomena. Sometimes those interpretations remain openly perplexing, perhaps suggesting that we have to be content with a theory of the world that is deeply counterintuitive. In the extreme case, it might be admitted that the theory makes no sense but is nevertheless true. Philosophers are interested in quantum mechanics but perhaps not primarily because of the challenges it poses for the classical laws of physics. On a more fundamental level, quantum mechanics confronts some of our most basic beliefs about the world. If things can move without passing through intermediate places, or affect one another instantaneously over vast distances, then we must seriously reconsider our assumptions about identity, continuity and causation. If we adopt the naturalistic approach to metaphysics, as more recently defended by Ladyman and Ross (2007), and conclude directly from physical theory to metaphysics, then quantum mechanics seems to challenge some core tenets of our basic ontology. Against such scientific naturalism, we maintain that the physical theory comes already equipped with metaphysical assumptions that we have reasons to reject (see also Andersen and Becker Arenhart 2016). This essay focuses specifically on the notion of causation. There is a line of argument that quantum mechanics is not a causal theory (for one example, see Feynman 1967: 147, though we offer more below). In contrast, we argue that this is not a purely empirical matter but an ontological and conceptual one. To say what causation is would then be a task for philosophy. Certainly, this does not mean that the philosopher can ignore the debates in physics. If a notion of causation has no real application, our ontology might be better off without it. Some do indeed argue that the concept of causation is too confused or ambiguous to be applicable to exact quantum phenomena (Skyrms 1984: 284, Healey 1992: 193). Here, we will first take a closer look at the original debate over causation from classical quantum mechanics. This is because, before accepting quantum mechanics as a counterexample to the central role of causation in science, we should be clear about what it is exactly that is at stake. We then move on to present our preferred theory of causation, based on an ontology of dispositions, and show how this remains unchallenged by arguments from classical quantum mechanics. The dispositional modality plays a crucial part in this argument. 2. Responses to the problem of causation in quantum mechanics We have said that quantum mechanics poses some ontological challenges. Historically, there are two main responses to this. One is the ontological response, which says that we should keep the physics as it is and instead change some of our ontological assumptions, even if these are common sense. To get rid of the notion of causation would be one such ontological revision. The second type of response is to say that we should keep our basic ontological assumptions, including those concerning causation, and instead improve our physical theory. One way to do this is to continue developing the theory until it matches our preferred ontology. This is a theoretical response. Proponents of quantum mechanics, in this case, Bohr (1937) and Heisenberg (1959), argue that the problem of causation is an ontological one; that is, that the nature of the quantum world shows that causation cannot be universally applied. The theory completely describes the relevant phenomena, but not through the standard application of explanatory tools such as causation, spatiotemporal tracking, identity, position, momentum, and so on. The universal application of these tools is thus a mistaken ideal, on their view. Critics of quantum mechanics typically argue for the reverse (see for instance Einstein 1936, de Broglie 1929, Schrödinger 1933 and Bohm 1957). That causation cannot be applied in quantum mechanics is then treated as only one of many symptoms of the physical theory being incomplete. This means that there are relevant aspects of reality that remain unaccounted for within the existing theory. The problem with this is that it would violate the completeness criterion of Einstein, Podolsky and Rosen, which states that ‘… [e]very element of the physical reality must have a counterpart in the physical theory’ (Einstein, Podolsky and Rosen 1935: 777). If causation is part of physical reality but not included in the theory, then the theory would be incomplete in this sense. This second type of response – that the physical theory must be developed – has motivated a number of alternative theoretical approaches to quantum physics (many outlined by Healey 2009). A common aim for these is to preserve a unified ontological foundation for the whole of physics. We will not go into any of these theories here, but only note that at least some of these interpretations are causal: the hidden variable theory (e.g. Bohm 1957) and the permission of backwards causation (e.g. Price 1996, Price and Weslake 2009). The rejection of causation in quantum mechanics has been accepted by many philosophers. Some might even welcome it, for instance if they think that causation plays a purely explanatory role but without any ontological grounding (Price and Corry (eds) 2007, Reiss 2015). Physicists, on the other hand, are generally reluctant to make this move. This may be because of the essential role that causation plays in classical physics and relativity theory. So unless a causal interpretation can be given of quantum mechanics, there is a fundamental discontinuity between the quantum realm and everything else in the world. Such a division is generally regarded as problematic because it violates the thesis of scientific unity, a philosophical idea that goes back to the pre-Socratics and which still guides scientific development. One philosophical strategy to avoid violating scientific unity would be to reject causation universally, which Russell famously did in his 1913 paper, but later withdrew (Russell 1948). This is not our preferred strategy. Our response to the challenge of quantum mechanics is different from both the ontological and the theoretical response. Independently of this divide, there is one thing upon which all the physicists in the classical debate seem to agree, namely what causation is. They all seem to accept the orthodox view of causation, which has largely been influenced by Hume, but also by the mechanistic philosophy of Descartes, and reinforced by some of Hume’s opponents. We will argue that this is a systematic mistake in the debate which ultimately led prominent physicists to deny that there is room for causation in quantum mechanics. We will show why this discussion cannot be settled by scientists alone, but must be informed by the philosophical understanding of causation, which is a matter of controversy. If it turns out that the debate in classical quantum mechanics is not over causation as such, but instead over one specific understanding of causation, then there might be other understandings that could still allow for a causal interpretation of quantum mechanics. While we are not asserting the truth of any such causal interpretation here, we do propose that an alternative understanding of causation in terms of the realisation of tendencies or real dispositions could be such a candidate. But before we do this, it is vital that we look at the concrete features that motivated many of the prominent physicists to exclude causation from the realm of quantum mechanics. 3. Challenges to causation in quantum mechanics On Hume’s analysis, the causal link is itself not directly observable, but must be derived from something that is itself observable and he famously presented three such observables: constant conjunction, temporal priority and contiguity (Hume 1739: 73ff). These are not the only types of observable features that causation could be thought to have. Others have focused on different features, such as counterfactual dependence (Lewis 1973, 1973a), energy transference (Fair 1979, Salmon 1984, Dowe 2000 and Kistler 2006) or manipulability (Woodward 2003). We will not treat any of these theories here, but instead look at those features of causation that were explicitly discussed and dismissed by physicists in the context of classical quantum mechanics: necessitation, determinism, predictability and separability. 3.1 Non-necessitation We saw that Heisenberg opted for an ontological response to the challenges of quantum mechanics: to keep the physical theory as it is and instead change one or more of the ontological assumptions. In Physics and Reality, he argued that quantum mechanics proves that Kant was wrong in assuming causation as a fundamental principle for any metaphysics of science. The a priori law of causality, Heisenberg said, does not fit atomic physics and ‘is no longer applied in quantum theory’ (Heisenberg 1959: 82). But what exactly is this law that cannot be applied to atomic physics? Heisenberg discusses causation in different places in the book, often with explicit reference to Kant. As example we take the law of causality. Kant says that whenever we observe an event we assume that there is a foregoing event from which the other event must follow according to some rule. This is, as Kant states, the basis of all scientific work. (Heisenberg 1959: 81) Heisenberg here mentions two assumptions about causation. The first, to which we will return later, is that all events have a cause. The second is causal necessitation: that the effect must follow the cause. The latter is a feature that Hume famously denied to be part of causation, though he also denied the first assumption (Hume 1739: 78-9). To him, causation does not include a necessary connection between cause and effect because none such is known in experience. But despite his denial of causal necessitation, Hume’s formulation of this principle has remained influential and is often what anti-Humeans want to add to their accounts. But as Hume said of necessity: ‘Such a connexion wou’d amount to a demonstration, and wou’d imply the absolute impossibility for the one object not to follow, or to be conceiv’d not to follow upon the other’ (Hume 1739: 161-2). We agree with Hume that it does seem to follow that necessitarians are committed to this being true of causation, or at least the first part of it. Heisenberg could not find any causal necessity in quantum mechanics. We will come back to his reasons for saying this in the next section. What is worth noting now, however, is that he concluded from this that there is no causation to be found in quantum mechanics, simpliciter. Another physicist who discussed causation in detail was Bohm, especially in Causality and Chance in Modern Physics (Bohm 1957). He took a more radical stance than Heisenberg and argued that not only is there no causal necessity to be found in quantum physics; there is no causal necessity anywhere in nature. We have thus been wrong to assume such necessity, also in classical physics. According to Bohm, any causal law might be interfered with by external factors and must therefore be taken as conditional upon a number of unforeseen events. … the necessity of a causal law is never absolute. For example, let us consider the law that an object released in mid-air will fall. This in fact is usually what happens. But if the object is a piece of paper, and if ‘by chance’ there is a strong breeze blowing, it may rise. … Hence, we conceive of the necessity of a law of nature as conditional, since it applies only to the extent that these contingencies may be neglected. (Bohm 1957: 2) Similar concerns have been posed by philosophers (e.g. Peirce 1892: 304ff, Cartwright 1983 and 1999, Dupré 1993, Mumford and Anjum 2011: ch. 3). But a common response is that if one takes all these contingencies into account, then the cause would necessitate its effect after all (essay 2). All one has to do is to include in the cause all the positive and negative factors that might affect the outcome, and hold these fixed. The idea is then that the effect will be guaranteed by this total cause (Mill 1843: 332). Now there might not be a finite list of such factors, so instead of listing them all, one could instead say that a cause necessitates its effect under some ideal conditions, or ceteris paribus: all else being equal. A worry concerning this move is that it can be used to make any causal claim trivially true. If the effect does not follow the cause, then the conditions simply weren’t ideal, which seems an unfalsifiable claim. The total cause response might work for classical mechanics, which is generally thought to be a deterministic system. At least Bohr (1928: 584) seems to accept that the strict necessities, although they are abstractions, are still justifiable in classical mechanics because quantum effects can be neglected there. In quantum mechanics, however, such a response would not hold, since even a theoretical abstraction could not make an outcome necessitated, neither strictly nor conditionally. 3.2 Non-determinism That causation involves necessitation is not a new idea, but one that has philosophical roots going back to Aristotle, Spinoza and Kant, among many others. One reason why necessitation is thought to be an essential feature of causation is the central role it plays for determinism. Both in philosophy and physics, there is a frequent assumption that causation is the provider or vehicle of determinism, via causal necessitation. If the cause guarantees its effect, and every event is caused, then from any given state of the universe, there could be only one possible future (Laplace 1814, Popper 1958). In principle, this would mean that everything that happens in the history of the universe was already fixed by its initial state immediately after the Big Bang. According to Heisenberg, this type of determinism is not part of quantum mechanics, and he took this to mean that there is thus no causation either. In the following quote, we can see how he explicitly identifies causation with determinism: Even there [in the science of psychology] one tried to apply the concepts of classical physics, primarily that of causality. In the same way life was to be explained as a physical and chemical process, governed by natural laws, completely determined by causality. (Heisenberg 1959: 169, our emphasis) This assumption is not unique to Heisenberg and should not be read as a mere slip of the tongue. As shown by Suárez and San Pedro, it has been an almost universal idea that causation and determinism amounted to the same thing. And, as they say, ‘even those who regretted the demise of a causal picture attempted to restore a causal understanding of quantum mechanics precisely by restoring determinism’ (Suárez and San Pedro 2011: 173). A failure to save determinism is thus a main reason for why the theory of quantum mechanics has been understood as non-causal, for instance by Heisenberg (1959) and von Neumann (1955). One philosopher who explicitly denied the link between causation and determinism, however, was Anscombe (1971), arguing that causation is consistent with indeterminism. For instance, it allows genuinely probabilistic causation. So if determinism were to be true, then this would have to be some extra fact over and above the facts of causation. We agree with Anscombe, but will return to the reasons why later. Some philosophers have offered probabilistic theories of causation (Reichenbach 1956, Suppes 1970), but a more common view in philosophy seems to be that probabilistic causation occurs only when the probability of the effect collapses or resolves to 1 (Armstrong 1983: 132-3, Popper 1990: 13, 20, Mellor 1971: 65-9). Consider, for instance, what Bohm (1957: 21) thought to be a chancy case: a traffic incident. The chance of a pedestrian being hit by a car on an empty road might not be high but the closer the car gets to the pedestrian, the closer the chance of being hit gets to 1. In the exact moment that the probability reaches 1, causation happens. But if this is how causation actually works, then any apparent stochastic element is effectively reduced to a deterministic one. This shows how strongly determinism is linked to causation, even by those who are inclined to treat them as separate. We saw how Bohm (1957) rejected necessitation both in quantum mechanics and in classical physics. In the same context, he rejected determinism, and argued that any deterministic assumption is a result of theoretical abstraction and causal isolation: Very often we may for practical purposes isolate the process in which we are interested from contingencies with the aid of suitable experimental apparatus and thus verify that such an abstract concept of the necessity of the causal relationship is a correct one. Now, here it may be objected that if one took into account everything in the universe, then the category of contingency would disappear, and all that happens would be seen to follow necessarily and inevitably. On the other hand, there is no known causal law that really does this. … In other words, every real causal relationship, which necessarily operates in a finite context, has been found to be subject to contingencies arising outside the context in question. (Bohm 1957: 2-3) Bohm refers to the strategy, mentioned earlier, of fixing all the factors that might influence the outcome in order to guarantee causal necessitation. But here the strategy is taken a step further to include the whole universe. The idea seems to be that determinism can be secured by making an artificially closed system, thus shielding off all possible contingencies. Bohm argues that this strategy is based on a mistaken mechanistic perspective on the world: … the inadequacy in the microscopic domain of the mechanistic form of determinism into which causality was restricted by classical physicists helped to provoke a very strong reaction in the opposite direction, and thus helped to encourage modern physicists to go to the opposite extreme of denying causality altogether at the atomic level. (Bohm 1957: 34) So by assuming determinism to be provided by causation, physicists had no other choice than to deny the existence of causation in quantum mechanics since no such determinism can be found there. This means that they must also deny the principle from Kant that we mentioned earlier: that all events must have a cause. Since there is no causation in quantum mechanics, there are events that don’t have a cause. An example of such an event could be radioactive decay, which seems to happen without prior causes. We will return to this example in section 4 when we present an alternative understanding of causation. 3.3 Non-predictability In classical physics, determinism is a central notion and philosophers have gone to great lengths to defend causal necessitation and determinism at least here. The main reason why determinism is regarded as so important, besides its supposed link to causation, is that it provides an ontological basis for our predictions. So even though perfect predictions remain a practical challenge, because we cannot take into account everything in the universe, determinism would make reliable predictions possible at least in principle. Laplace explains this idea by imagining a creature, or a super-scientist, that has enough brainpower and knows all the initial conditions and laws. Given for one instant an intelligence which could comprehend all the forces by which nature is animated and the respective situation of the beings who compose it – an intelligence sufficiently vast to submit these data to analysis – it would embrace in the same formula the movements of the greatest bodies of the universe and those of the lightest atom; for it, nothing would be uncertain and the future, as the past, would be present to its eyes. (Laplace 1814: 4) This creature, commonly referred to as Laplace’s demon, would be able to predict with perfect accuracy any future event. But it could do so only under the assumption of determinism, which cannot plausibly be defended for quantum mechanics. Without determinism and causal necessitation, there is no guarantee that the same event would follow from two identical sets of initial conditions or states. Perfect predictions are thus not even in principle possible in quantum mechanics, and Bohr took this to mean that there could thus be no causation either. The reason was that he thought causation was somehow dependent on predictability: In physics, causal description, originally adapted to the problems of mechanics, rests on the assumption that the knowledge of the state of a material system at a given time permits the prediction of its state at any subsequent time. (Bohr 1948: 141) In classical physics, our predictions can be improved by performing more experiments. But this is not the case in quantum mechanics because of the extreme sensitivity of the system in quantum physical experimentation, where even the instruments used can interfere with the result. Again, Bohr: It must not be forgotten, however, that in the classical theories any succeeding observation permits a prediction of future events with ever increasing accuracy, because it improves our knowledge of the initial state of the system. According to the quantum theory, just the impossibility of neglecting the interaction with the agency of measurement means that every observation introduces a new uncontrollable element. (Bohr 1928: 584) Predictions are not entirely impossible in quantum mechanics, but they must be spelled out in terms of probabilities. According to Bohr, this is because ‘… the specification of the state of a physical system evidently cannot determine the choice between two different individual processes of transition to other states’ (Bohr 1948: 142). Heisenberg raised similar concerns to Bohr when dealing with the apparent spontaneity in radioactive decay for an atomic particle: … the law of causality is reduced to the method of scientific research; it is the condition which makes science possible. Since we actually apply this method, the law of causality is ‘a priori’ and is not derived from experience. Is this true in atomic physics? Let us consider a radium atom, which can emit an a-particle. The time for the emission of the a-particle cannot be predicted. We can only say that in the average the emission will take place in about two thousand years. (Heisenberg 1959: 81-2) We here seem justified in concluding that one of the reasons physicists had for rejecting a causal interpretation of quantum mechanics is that predictions in the quantum realm cannot be more precise than probabilistic ones. 3.4 Non-separability The fourth disputed issue is not whether quantum mechanics deals with genuinely non-necessitated, non-determined and non-predictable phenomena. Rather, the question is what this implies for the understanding of causation. Given the classical notion of causation, these are alone very good reasons to exclude causation from the quantum realm. In the next section, we argue that this move is not justified given any notion of causation. But Bohr also discusses a fourth problem for causation, related to the separability of cause and effect. This alleged feature of causation comes directly from Hume’s analysis, unlike necessity, determinism and predictability, although these have become parts of the standard notion (what Norton 2007 and Kutach 2007 call the ‘folk theory of causation’). On his theory, causation is a relation between two events or objects. But this requires that the two relata can be clearly separated. For Hume, the cause and effect were what he called distinct existences. It had to be possible, in principle at least, for one to exist without the other. This was an essential aspect of his theory, as it ruled out for instance that logical relations could qualify as causal ones. After all, if A logically entails B, there will be a regularity of As followed by Bs, but we do not think of it as a causal regularity. Hume thought that we gain the notion of causation where, although A and B could exist apart, as a matter of fact, A is always followed by B. This is what gives us an expectation of B whenever A. Thus, causation for Hume has to be a relation between distinct existences, where logically connected existences do not count as distinct. If the cause and effect are to be distinct existences, however, they must also have some form of individuality, which Bohr (1937) says cannot be found in quantum mechanics. The renunciation of the ideal of causality in atomic physics which has been forced upon us is founded logically only on our not being any longer in a position to speak of the autonomous behaviour of a physical object, due to the unavoidable interaction between the object and the measuring instruments which in principle cannot be taken into account, if these instruments according to their purpose shall allow the unambiguous use of the concepts necessary for the description of experience. (Bohr 1937: 87) The general problem that he refers to is this. In the quantum mechanical set up one cannot talk about an atomic system in isolation. Instead, the system also includes the whole experimental context, including the measuring apparatus. However, this makes the experimental context a part of the definition of the target system, which means that a strict separation between the experimental set up and the target system cannot be drawn. In a quantum mechanical experiment, it is therefore impossible to determine what exactly it is that caused the effect, since this is produced partly by the measuring itself. Since full separability is impossible in quantum mechanics, Bohr concluded that we must also reject universal causation. He repeated this point several times over a 20-year period and with slightly different formulations (see for instance Bohr 1928: 580, 1935: 75, 1938: 95-6, 1948: 141-2), so it was clear that he saw non-separability as a crucial argument against causation. 3.5 A change of framework: from Newtonian to Aristotelian For Bohr (1948: 142-3) a description of physical objects must be given within the classical conceptual framework of Newtonian physics. This would include a Cartesian metaphysics and all the usual mechanistic concepts. Within this framework objects are understood in the classical sense, having a clear individuality. That a classical object has such individuality may be taken to imply that we can give distinct values for its momentum, mass, position, velocity, shape, spin, angular momentum, and so on. In quantum mechanics, on the other hand, objects do not carry such values for all elements of a description at an instant. Either we can measure the position, or we can measure the momentum, but we cannot measure both. If we ascribe momentum to an atomic system, this excludes the possibility of ascribing a position to it, and vice versa. According to Bohr’s complementarity thesis, position/momentum is an example of a conjugate pair of variables. These are variables that complement each other. There are many such pairs and the central one for our discussion is the spacetime/causation pair. An atomic system has clear values for only one of these variables at a time. We cannot, therefore, talk of a universal application of spatiotemporal location and causal interaction. We can only talk of partial or complementary descriptions of the system. We can interact with the quantum object either in such a way that the object gets a clear spatiotemporal position, or in such a way that it has clear causal interactions. We cannot do both simultaneously. The Newtonian ideal of a full-blown causal and spatiotemporal description of a single object is therefore no longer tenable, and should no longer be treated as an explanatory ideal. In contrast to Bohr, Heisenberg (1959) suggested that a way to understand the nature of a quantum object in isolation is to understand it as energy, where energy is a substance. This substance in isolation is best understood as pure potentiality. Heisenberg thus proposed to move away from the classical Newtonian concepts and instead adopt an Aristotelian framework with unformed energy, which seems similar to Aristotle’s materia prima. All the elementary particles are made of the same substance, which we may call energy or universal matter; they are just different forms in which matter can appear. If we compare this situation with the Aristotelian concepts of matter and form, we can say that the matter of Aristotle, which is mere ‘potentia’, should be compared to our concept of energy, which gets into ‘actuality’ by means of the form, when the elementary particle is created. (Heisenberg 1959: 139) Potentialities, or potencies, would according to Heisenberg give a better description of the quantum phenomena. Rather than having a specific nature that is independent from the experimental context, what exists before the experiment is a potential that is actualised through the experimentation. The experiment thus creates a novel phenomenon. This means that one cannot infer from the outcome of the experiment what was there before the experiment, which one could have done had the experiment involved classical objects. Instead, one can say something of the potencies that were there. This also gives us a different understanding of the seemingly dual nature of quantum objects. A clear distinction between matter and force can no longer be made in this part of physics, since each elementary particle not only is producing some forces and is acted upon by forces, but it is at the same time representing a certain field of force. The quantumtheoretical dualism of waves and particles makes the same entity appear both as matter and as force. (Heisenberg 1959: 139-40) Heisenberg’s suggestion to replace the Newtonian framework with an Aristotelian one clears the way for an alternative understanding of causation in quantum mechanics that is neo-Aristotelian in essence. We will now offer an alternative, dispositional model of causation that accords with the Aristotelian notion of potentiality. The aim is to show that this model differs substantially from the notion of causation that has been rejected in classical quantum mechanics. What this means is that there might still be some room left for a notion of causation in quantum mechanics, if this does not include the problematic features discussed above. 4. An alternative notion of causation In the previous section we saw that quantum mechanics leaves us with some phenomena that cannot be fitted neatly into the classical Newtonian world-view. While the systems of classical physics are deterministic and predictable with universal laws, quantum mechanics involves genuinely chancy events that are not predictable even in principle. A further complication is added by the fact that there is no way to treat atomic objects as classical objects. To pick out the cause and the effect as two spatiotemporally separated entities is therefore not possible. What we are left with, then, is a part of reality that behaves in a very different way from the rest. But does it also lack causation? Once we have taken out strict laws, determinism, predictability and separability, we seem to have lost everything that characterises causation, as it was assumed to be by Bohr, Heisenberg and others. As a result, they rejected the presence of causation in quantum mechanics. But is this rejection justified? That quantum mechanics lacks certain features is a problem for causation only if these features really are essential parts of causation. And one might ask what could possibly be left of causation without them. We now move on to argue that there is still ample room left for a notion of causation, even with the restrictions discussed above. We will show that there is at least one plausible theory of causation that includes none of the features that those physicists thought of as problematic for a causal understanding of quantum mechanics. But this does not mean that we are left with a causal notion that has no application outside quantum mechanics. Importantly, the same theory we outline allows us to account for classical causal phenomena just as well as more messy and indeterministic ones. A further strength of this theory is that it can include and explain features that are often treated as problematic by other theories, including causal complexity, context-sensitivity, nonlinear interaction and the possibility of prevention and interference. The move we recommend has a degree of novelty. It is not among the options directly considered by Healey (2009), for instance. We will now offer a brief outline of the most relevant features of this theory, which we refer to as causal dispositionalism. The aim is not to argue for the truth or details of the theory, but to show that it is possible to have a useful notion of causation without the features of necessity, determinism, predictability and separability, which is thus apt for an understanding of causation in quantum mechanics. 4.1 A neo-Aristotelian framework Aristotle thought that there were real potencies in the world and that these can be actualised under certain conditions (see Metaphysics Θ, in Makin 2006, and Physics II). For every new actualisation, however, new potencies emerge. We agree with Heisenberg that the Aristotelian terminology of potencies is better suited for describing quantum phenomena than the Cartesian mechanistic language. But we take the Aristotelian framework to have a more universal application than quantum mechanics. In the last decades a neo-Aristotelian ontology of dispositions, or causal powers, has been developed by a number of philosophers (see for instance Harré and Madden 1975, Cartwright 1989, Mumford 1998, Molnar 2003, Martin 2008 and Marmodoro 2017). One motivation for taking dispositions as fundamental parts of reality is that they are thought to provide an ontological basis for causation, laws, natural kinds or even modality. Causal dispositionalism, as developed by Mumford and Anjum (2011) is a recent attempt to provide a comprehensive theory of causation based on an ontology of dispositions. Causal dispositionalism is a neo-Aristotelian theory in a very concrete sense. Every time a potency gets actualised, causation happens. In other words, causation is the manifestation of dispositions. The ontology of dispositions offers a different perspective on reality than the classical Newtonian or Cartesian one. Instead of explaining physical events in terms of governing laws that determine their behaviour, a dispositionalist looks to the thing’s properties (Mumford 2004). The idea is that objects behave the way they do, not because of some external laws that push them around, but because of their own intrinsic dispositions and their interactions. Dispositions can be ascribed to particular objects, but also to events, fields or processes. Take for instance an autumn leaf falling from a tree. Within an idealised context, the leaf will fall straight to the ground in accordance with classical gravitation. But in a natural context, this fall will be influenced by a number of things: the weight and shape of the leaf itself, the direction and intensity of the wind and rain, the heat from the ground, the draft from an industry fan, and so on. The gravitational attraction on the leaf is thus only one of the causally relevant factors for its behaviour. The actual movement of the leaf is impossible to predict from the laws of physics, as has been duly noted by Cartwright (1999: 27). But if we understand the situation in terms of dispositions, instead of external and strict laws, we can still make perfect sense of the fall as caused. Causation should not be thought of as a perfect fall that is interfered with by the wind, the fan or the rain. All these aspects of the causal situation contribute to the outcome with their own causal powers. Rather than deriving causation in concrete cases from exceptionless laws of closed systems and idealised conditions, causal dispositionalism motivates us to look at how things are naturally disposed to behave in virtue of their own properties. This behaviour will typically be a complex matter including a number of factors, some of which will interact in nonlinear ways. In any open system, the potential interactions that can affect the outcome will be infinitely many. We will not offer a full argument for this dispositional theory of causation here and we concede that it is far from universally accepted. The reality of causal powers is rejected by Humeanism in particular. What we can do, however, is show how this neo-Aristotelian conceptualisation gives us an understanding of causation very different from the classical notion that we got from Descartes, Newton and Hume. Furthermore, because it does not contain the features that were problematic for quantum mechanics, it is quite possible for it to be applied there. 4.2 Non-necessitation We saw that Heisenberg and Bohr rejected causation partly because they could not find any necessitation in quantum mechanics. On the classical notion, this would be a good reason to reject causation, since causal production is there collapsed into causal necessitation. A cause then produces its effect by necessitating it. On causal dispositionalism, however, there is no such necessitation. Instead, there are irreducible tendencies. A cause is on this view something that disposes or tends towards its effect. Often the cause will produce its effect, but there is no guarantee that it will do so. And even when the effect was produced, it was not through necessity. It is easy to see that the external principle of tendency (from essay 1) can be applied in physics. An effect can be interfered with in two ways; either by subtracting or adding further dispositions. Subtractive interference is to remove something from the causal set-up that tends towards the effect. One might remove friction from a fall by creating a vacuum, for instance. Additive interference means that one adds something to the causal situation that tends away from the effect. One might add friction to interfere with a fall by introducing air, wind or water. This latter form of interference shows that causes don’t necessitate their effects. Something could always have been added that prevented the effect from occurring. A cause is then never in itself sufficient for its effect, even when it succeeds in producing it. This is why we adopt the dispositional modality view specifically, rather than conditional necessity, for instance. The possibility of causal prevention is a problem for the classical understanding of causation, and we have already mentioned some of the strategies to avoid any interference: adding ceteris paribus clauses or ideal conditions to the causal claim. On causal dispositionalism, however, the possibility of causal interference is not something that we need to treat as problematic. Instead, it is an essential feature of causation that it can be counteracted in this way. The possibility of interference is thus a symptom of the irreducibly tendential nature of causation, and thus a more reliable symptom of causation than necessitation. Indeed, our account of causation does not require even that constant conjunction accompanies causation, let alone constant conjunctions that are also necessary. 4.3 Non-determinism A second reason for rejecting causation in quantum mechanics was the indeterministic behaviour of quantum objects. Instead of strict laws, there seem to be some irreducibly probabilistic phenomena. Since causation is often linked to determinism, this behaviour was taken to challenge a causal interpretation of quantum mechanics. We saw that even those who allow some probabilistic causation in their ontology, usually end up analysing it into a non-stochastic matter, where causation happens once the effect reaches probability =1. Causal dispositionalism is an alternative to both classical determinism and classical indeterminism. A tendency is less than necessity, so the effect is not guaranteed by its cause. This gives us less than causal determinism. But it is also offers more than a pure indeterminism, if by that we were to mean that the effect is entirely random, chancy or spontaneous. On causal dispositionalism, an outcome can be caused even if it is probabilistic because it allows genuinely probabilistic causation in addition to the more common non-probabilistic instances. In both cases, causation is grounded in dispositions, but in different types of dispositions. Probabilistic causation happens when a probabilistic disposition manifests itself (see essay 3). A typical example of a probabilistic disposition is the 50:50 propensity of a fair coin to land heads or tails if tossed, while a non-probabilistic disposition could be the propensity of a vase to break if dropped onto a hard surface. It’s a philosophical controversy whether setups involving coin tosses, roulette tables or decks of cards should count as genuinely probabilistic or deterministic cases. For a determinist, these cases will only appear to be probabilistic, but the initial conditions plus laws will still determine the outcome. Regardless of this debate, quantum mechanics presents us with phenomena that seem irreducibly chancy in nature. Radioactive decay is one such example that has been widely discussed in philosophy. In quantum mechanics, this is usually offered as an event that is stochastic, in the sense random, but that can be given a probabilistic prediction of half-life. On causal dispositionalism, radioactive decay would be the effect of a propensity or disposition manifesting itself. It would have to count as causation because it is the manifestation of the atom’s propensity to decay. We will return to this example shortly. But what is important to notice in this context is that on causal dispositionalism, determinism is not provided by causation. Instead, the claim of determinism would have to be an extra claim that must be justified separately from the presence of causation. A worry about stating a non-causal determinism, however, is that it could in principle be applied to any system. Even a Humean world of pure contingency could thus be a deterministic one in this sense, for instance, if all the contingencies were fixed or pre-determined. Popper (1958: 6-8) refers to this as metaphysical determinism, and it is not the type of determinism about which philosophers and scientists usually worry. It is also not what is at stake in quantum mechanics. We have seen, then, that on causal dispositionalism there is no determinism that comes for free with causation. A dispositionalist can be a determinist or an indeterminist, depending on considerations other than causation. Such independence from determinism and indeterminism should be taken as a marker of the credibility of a theory of causation rather than a weakness. We will now look further at the type of indeterminism that quantum mechanics involves. 4.4 Non-predictability The indeterminism of quantum physics leaves us with a more practical concern about causation. If the effect is not determined by its cause, we seem to have no ontological basis for our causal predictions either. We will now look at some different ways in which we might lack predictability in quantum mechanics. For this purpose, we will take the case of radioactive decay, which seems to be unpredictable because of its indeterministic behaviour. According to quantum mechanics, such decay is stochastic, random and spontaneous. All these features are commonly linked to indeterminism, but they seem to relate to causation in slightly different ways. It will therefore be useful to treat them separately. This is important for our purpose, which is to show that there could still be some place for a notion of causation in quantum mechanics even if it includes indeterministic features. We begin with the stochastic element. One way to understand an event as stochastic is that there is some genuinely probabilistic element involved. In other words, the event is chancy. This is an ontological interpretation of probability, which contrasts with the purely epistemic notion of credence. We have already seen how dispositionalism is able to allow chanciness as a case of genuinely probabilistic propensity (essay 3) and therefore also as a matter of causation. We therefore move on to the next indeterministic element: randomness. That an event is random could mean that it is unpredictable on the individual level, while still displaying a clear tendency over a sequence of trials. Bohm (1957: 21) illustrates this phenomenon with the example of a traffic accident where two cars collide. On the individual level, he argues, this outcome was not predictable, since there could be infinitely many contingencies that could causally affect the outcome. If one of the drivers had received a text message before starting the car, or stopped to buy a newspaper on the way, then the cars might not have collided. But despite the randomness on the level of the individual, on the statistical level we often find clear tendencies. For instance, there might be a high statistical tendency towards traffic accidents at a particular crossroads, among a specific age group or linked to certain types of behaviour. On causal dispositionalism, such tendencies could be emergent, higher-level dispositions of traffic systems and their users. This means that there could be dispositions on the individual level that increase one’s propensity to be involved in a traffic accident: lack of observation skills, alcohol consumption, speeding, and so on. But there could also be some dispositions that belong to the infrastructure itself, and that makes it more likely that risky behaviour results in an accident. An individual event could therefore still be caused, even if it is random to some degree and not predictable. The third feature of spontaneity might represent a greater challenge for causation. On the classical notion of causation, there is a set of (ideal) background conditions, a stimulus (efficient cause) and a response (effect). Typically, causation will then happen when the effect is triggered by the right stimulus and under the right conditions. When no such trigger is found, we call the effect spontaneous. Since the trigger is thought of as the cause, the most obvious conclusion is that the event was uncaused. On causal dispositionalism, however, this does not follow. There could be dispositions that manifest spontaneously. Something that is very explosive might for instance explode without any extra stimulus, and a poisonous gas won’t need any trigger to poison those who go near it. To prevent such dispositions from manifesting, one will often add something to contain or counteract them. But for a dispositionalist, causation happens once the disposition manifests itself, whether or not it was triggered. So while the indeterministic nature of quantum phenomena threatens predictability, it does not necessarily threaten the presence of causation. On the contrary, the fact that effects cannot be predicted with certainty is something that a dispositionalist sees as an essential feature of causation, not as a problem. A feature of causation that makes our predictions fallible is that causal processes are extremely sensitive to a change in context. Any difference in the causal setup might also change the outcome. And even a tiny change in the cause might result in a vast change in the effect because dispositions often interact in nonlinear ways. All causal truths are also vulnerable to the so-called problem of induction, as famously pointed out by Hume. In our context, this is because we will never know if we have taken into account all relevant factors, unless we make our prediction within an artificially closed and deterministic model. Instead of attempting to save predictability by stipulating closed systems and strict laws, which is anyway not possible in quantum mechanics, we should take it as a symptom of causation that our causal predictions are fallible and vulnerable to the problem of induction (essay 8). Still, this doesn’t mean that we cannot predict anything. We can predict how something will tend to behave given the factors that are taken into account. A good prediction will therefore tend to be right. And, if there is a stochastic disposition involved, the prediction will be irreducibly probabilistic. 4.5 Non-separability We saw that Bohr had an extra worry about causation in quantum mechanics, concerning the separability of quantum objects from the experimental setup. In the classical model of causation, the cause and effect have to be clearly separated, both in time and space. Hume’s perfect instance of causation was two billiard balls colliding, which has inspired the neuron diagram as the preferred model for causal representation (as, for example, in Lewis 1973a). This model is perfect for the classical understanding of causation, according to which causation is a relation between two distinct objects or events. But it does not help us much in quantum mechanics, where the typical autonomous behaviour of physical objects is not found. Instead, there are contextual interferers even on the level of the instruments used in the experiments. In contrast to the classical model, a causal dispositionalist should think of causation not as a relation between two separate events or objects, but as a continuous, unified process that typically takes time to unfold (Anjum and Mumford 2018). A causal process will begin once a disposition meets its appropriate partner(s) and starts interacting. During this process, some properties will be lost and new properties and new interactions might be introduced. Imagine for instance salt dissolving in hot water. This is surely a typical example of causation. On the classical understanding, placing the salt in the water counts as the cause, and the dissolved salt is the effect. But on a dispositional account, we will represent the situation differently. Causation starts once the salt gets into contact with the hot water. The salt has the dispositions of being soluble and salty and the hot water is an appropriate solvent. They are what Martin (2008: 48) calls mutual manifestation partners, meaning that they can produce an outcome together that neither of them could have produced alone. In this case, that outcome is a saline solution, which we think of as the effect. But the saline solution is only the end product of the causal process. The causal interaction between the salt and water is a continuous process in which the cause and the effect seem to happen simultaneously (Mumford and Anjum 2011: ch. 5). The causal process is thus gradual, starting with the two dispositions in separation, but eventually fusing into something with different dispositions, the salty liquid. Other examples can be used. Firewood is flammable but the process leading up to it burning requires a number of manifestation partners: a suitable fireplace, a proper ignition, enough oxygen, and so on. Once all these dispositions meet, causation happens. The firewood might then burn until it has run its course or it might be interrupted, for instance if a damp log is added, or the ventilation is closed, preventing sufficient access to oxygen. On the classical view, we might think of the firewood as the cause, the ignition as the stimulus, and the oxygen and the stove as background conditions. In contrast, a dispositionalist will treat them all as more-or-less equal partners and causes merging into and becoming the effect. Another point is worth noting in this connection. Within the dispositionalist theory, change and flux is the default and any stability is a result of counteracting dispositions producing a state of equilibrium. Sometimes dispositions or powers cancel each other out, and the result is that nothing happens. Think of the thermostat in a refrigerator that is set to 4 degrees. Whenever we open the door and warm air comes in, the temperature rises. To counteract this, the fridge will start cooling down, thus keeping the temperature stable. Similar systems of equilibrium are found in any object or system that is kept relatively stable. On the surface, nothing seems to happen, and we might think that there is no causation going on. But on the micro-level we often see the complexity of powers involved to keep it in a stable condition. The stability is thus the effect of multiple causal powers balancing each other out, which means that it is as good an example of causation as one involving change (Mumford and Anjum 2011: 37). From this perspective we see that, rather than thinking of cause and effect as spatiotemporally distinct, causation often happens within one particular object, system or process and simultaneously. A system is kept stable only as long as the counteracting dispositions are doing their causal work. What we pick out as the cause and the effect might just be different stages of a continuous causal process. On dispositionalism, no strongly ontological distinction is drawn between properties belonging to the object undergoing change and the contextual properties in this process. They are all causes of the outcome. So whenever we pick out something as a cause in isolation from its context, this will usually be a result of an abstraction and not a description of what actually happens in real causal processes. The genuine interaction of multiple factors should therefore be seen as a symptom of causation rather than as something that needs to be analysed away by theoretical abstraction. 5. The way forward Heisenberg suggested that we adopt an Aristotelian framework in order to describe what happens on the quantum level. We have here offered one such framework involving an ontology of dispositions. This ontology does not require necessitation, determinism, infallible predictions or strict separability, which we saw were the main concerns about including causation in the quantum realm. There are of course challenges to other basic ontological categories that come from quantum physics, such as to notions of continuity and identity (see French and Krause 2006). Without these, we will struggle to understand dispositions and causation, but we will also struggle to understand fundamental ontological notions including substance, process, change or even time. In this paper we have restricted our discussion only to the notion of causation and what were believed – mistakenly so, we have argued – to be its essential features. In place of those alleged essential features we have offered an alternative characterisation of causation. On this view, causation i) involves irreducible tendencies, not necessity, ii) is independent from, not the driver of, determinism, iii) supports predictions of what tends to happen, not what will happen with certainty, and, finally iv) causation is a continuous, unified process, not a relation between two distinct events. What we have tried to show is that there is a defensible notion of causation that could in principle be applicable in quantum mechanics. Significantly, this is a causal notion that could be applied generally, to any scientific realm. We need not treat causation in quantum mechanics as a special case, therefore. At the very least, we hope to have shown that there should be no rush to judgement on whether there is causation in quantum physics, because that depends on what we take causation to be. View publication stats