The Interventionist Theory of Causation: A Challenge for Incompatibilism

Introduction

Debates about moral responsibility depend crucially on accounts of causal relations and causal sourcehood. Manipulation arguments such as Pereboom’s (2007) four-case example have served to draw attention to these issues. A challenge facing manipulation arguments is a recent approach Deery & Nahmias (2016) call The Interventionist Theory of Causation. The purpose of this essay is to explain and defend this attempt to block the incompatibilist conclusion of Pereboom’s (2007) manipulation argument. Once key terms have been defined I will sketch Pereboom’s four-case account. From there I will outline the interventionist theory of causation and argue that it offers a plausible compatibilist account of moral responsibility. Once I have considered an anticipated line of defence from incompatibilists, the following shall become clear: The interventionist theory of causation provides an approach towards causal sourcehood that buttresses compatibilist accounts of moral responsibility. The burden now rests with incompatibilists to critically examine and respond to this account. 

Compatibilism & Incompatibilism

I shall start with a brief overview of the theories and positions I shall be discussing. Given the scope of this paper they shall be somewhat generalised, although I will provide enough context for subsequent sections. Causal determinism is, roughly, the thesis that every event or action in the physical universe is the inevitable result of prior states and actions playing out in accordance with the laws of nature (Caruso, 2018; O'Connor & Franklin, 2018). As for free will, there are differing conceptions. The two most common are (1) that an agent has the ability to do otherwise, and (2) that her choices are up to her in such a way that she is the source of her actions (Mele, 2006; Caruso, 2018; O'Connor & Franklin 2018).

The traditional problem of free will and determinism has centred on reconciling our phenomenological sense of freedom with the idea that forces over which we exercise no control are causally determining our choices and actions. A number of approaches to this problem have emerged, but for the purposes of this essay I shall focus on two: compatibilism and incompatibilism. Compatibilism is the view that free will can co-exist with determinism: therefore, even if determinism is true, we can still have free will. In contrast, incompatibilists believe that if determinism is true, free will cannot exist, because we can’t exercise ultimate control over a causal chain of events that we, as physical beings, are governed by. While these are the positions I shall focus on, it’s worth noting that Pereboom (2007) is a hard incompatibilist, which is to say that regardless of the truth of determinism he doesn’t believe we can have free will. His manipulation argument, however, deals specifically with determinism, and other positions fall outside the scope of this essay. For that reason, I will focus on standard incompatibilism. With a grasp on these theories we can now turn to Pereboom’s (2007) influential manipulation argument.

Pereboom’s Manipulation Argument

Pereboom’s (2007) four-case manipulation argument was intended to highlight the difficulty in pinpointing a relevant difference between cases where an agent’s actions are caused by an external manipulator and where they are caused by determinism. We are introduced to Professor Plum, who in four cases murders Ms. White for selfish reasons. Pereboom (2007) ensures that Plum meets the typical, uncontroversial compatibilist conditions for responsibility. He’s a sane and rational human being. He conforms to Frankfurt’s (1971) requirement of acting in accordance with second order desires; he has reflected on his desires and they are consistent with his deeply held values. Plum is also responsive to reasons in the sense that Fischer and Ravizza (1998) outline as a condition for responsibility; if he were presented with good reasons not to murder Ms. White, he would capably consider those reasons and would commit the act regardless. Pereboom then outlines the following four cases to prompt our intuitions about when Plum is morally responsible and when he is not.

The Four Cases

In the first case presented by Pereboom (2007, p.94), neuroscientists manipulate Plum’s brain just before the act and cause him to murder Ms. White. Pereboom argues that intuitively, we would not hold him accountable in this case, as his action was obviously outside of his control. In the second case, Plum was programmed by the neuroscientists at birth to give significantly more weight to his desire to murder Ms. White. Again, Pereboom argues that intuitively he is not responsible. In the third case, Plum was conditioned as a young child in such a way that he was unavoidably socialised to commit such an act. Here again Pereboom argues that, as Plum was unable to resist this conditioning, he is not responsible. And lastly, Pereboom gives us Case 4, where causal determinism is true, and in the absence of any other mitigating factor, Plum was simply causally determined to kill Ms. White. Pereboom (2007, p.97) argues that if we exempt Plum in Case 1 then it would be inconsistent not to exempt him in Case 4 – his central claim being that there is no relevant difference between the manipulation and deterministic cases. The challenge laid out by Pereboom, then, is for compatibilists to locate at which point between the cases – and for what reason – Plum becomes morally responsible for murdering Ms. White.    

Pereboom’s argument was influential. As a compatibilist, Watson (2000, p.67) notes the difficulty in locating a difference between natural, deterministic causes and purposeful causes relevant to issues of control and freedom. Fischer (2000) was slightly more forceful in summing up the issue, stating: “I think that manipulation cases are compatibilism’s dirty little secret. Compatibilists don’t like to admit that this is a problem. It is to Bob Kane’s and other incompatibilists’ credit that they have pushed us to confront cases of covert non-constraining control...We compatibilists have to deal with this” (Fischer 2000, p.390). Now I have outlined manipulation arguments and the challenge they have traditionally posed for compatibilism, we can examine a recent and salient objection: Deery & Nahmias’ (2016) Interventionist Theory of Causation.

The Interventionist Theory of Causation 

In developing this theory, Deery & Nahmias (2016) aimed to demonstrate that there is a relevant difference between manipulation and determinism in a direct rebuttal to Pereboom’s (2007) aforementioned claim that there is no relevant difference. The central tenet of the theory is that regardless of the truth of determinism, different causal factors can bear varying levels of strength when it comes to determining which was more significant in bringing about an event. The interventionist causal modelling for this theory is directly below, followed by an explanation of how it works.

“X = x is the causal source of Y = y iff X bears the strongest causal invariance relation to Y among all the prior causal variables (including X) that bear such relationships to Y.” (Deery & Nahmias, 2016, p.1263).

The idea being expressed here is that you could have causal source (X) and resulting event (Y). There may be multiple causal relationships from sources to event (Y). For instance, there may be causal relationships from both (X) to (Y) and (W) to (Y). The point is, that we can ascertain which relationship: (X) to (Y), or (W) to (Y), bears a stronger causal invariance relation. Specifically, if a prior variable bears a stronger causal invariance relation to an action than variables describable at the time or immediately prior to an action then the causal source of the action is identifiable with the prior variable with the stronger causal invariance relation. (Deery & Nahmias, 2016, p.1263). The below example serves to clarify this further.

Danny and Manny

In Deery & Nahmias’ (2016, p.1257) paper, they use the example of Danny and Manny. Danny is a rational agent in a deterministic setting who has not been manipulated by an external controller. At the age of 30 Danny finds a wallet containing money on the street, thinks rationally about what to do with it and decides to keep it. Manny (because he is manipulated), is identical to Danny in all other ways except that he was programmed by an all-powerful Goddess, Diana, at the time of his conception. Diana programmed Manny and all related external factors to ensure Manny would find a wallet at the age of 30 and choose to keep it. Deery & Nahmias (2016) argue that there is a relevant, principled difference between Danny (who is subject to causal determinism), and Manny (who is also subject to causal determinism but in addition has been manipulated). 

The Principled Difference

This principled difference, they claim, in line with the theory above, is that the causal invariance relationship between Diana’s programming and Manny’s action is stronger than the causal invariance relationship between determinism and Danny’s action. That is to say that regardless of interference from background conditions, Manny will always keep the wallet because Diana will always intervene to ensure that he does. The authors use the example of Danny receiving an important and surprising text message just before making the decision to keep the wallet. The message is unexpected and engrossing enough that he leaves the wallet and continues on his way. The point here is that even if Manny received the same message, because he had been programmed to keep the wallet, he still would have kept it, regardless of the same interference from background conditions, because Diana’s prior programming has a stronger causal invariance relationship to the event. 

I argue the authors are on the right track here: If we want to more justifiably point out the source of an action, the difference in the strength of the causal invariance relationship allows us to locate that source more accurately. The most convincing factor seems to be that, unlike Danny’s situation, in Manny’s case the interference of Diana will override any future changes in background conditions after the fact of the programming that could change Danny’s outcome. If Deery and Nahmias (2016) are on the right track with this, it seems plausible there is a relevant principled difference between manipulated and ordinary agents in deterministic settings. Whether this attempt to defend a compatibilist account of moral responsibility can survive critical scrutiny from incompatibilists remains to be seen. For now, I will consider one anticipated line of response – the question of ultimate causation. 

The Source Incompatibilist Objection

A serious challenge to the interventionist theory approach can be expected from source incompatibilists, who typically require an agent be the ultimate source of her actions in order to be morally responsible (Deery and Nahmias 2016, p.1269). Proponents of this position are likely to claim that the proposed account of responsibility is too weak because the truth of determinism would entail that all decisions made by non manipulated agents like Danny are still outside of his control; he is still not the ultimate source. Deery and Nahmias (2016, p.1270) have a ready response. They note that when this ultimate source defence is used, it is typically only applied to free will and moral responsibility; its application in other scientific disciplines that measure causation is usually not addressed. The implication here is that this defence loses its force when it comes to describing, in a useful and practical way, the causal sources of real life, day-to-day events. 

The Mechanic’s Purpose

I agree with Deery and Nahmias (2016) here. In claiming that all causal relations can be tracked to prior causes, source incompatibilists risk implying that no causal relation carries more weight than another. This would appear to hinder progress if one is trying to develop a practical process of determining degrees of moral responsibility. Therefore, I consider Deery and Nahmias (2016, p.1270) correct in claiming that their account provides a useful way to discern which variables are more causally significant in a given situation. 

They employ a relevant example to demonstrate the force of this argument. Imagine a mechanic who is to fix an oil clog in your car. In order to most effectively locate and fix the problem she would causally model the engine and try to find the relevant causal variable to focus on to bring about the desired change and fix your car. She would not just be looking for the relevant variable, but for the strongest causal invariance relation “obtaining between the variable representing the clog’s occurrence and any of its prior causal variables, in order to identify the causal source of the problem.” (Deery and Nahmias, 2016, p.1270). In other words, she would try to find the specific problem with your car that is having the strongest causal impact on any other variables that may be contributing to the oil clog. Prior factors such as what terrain you were driving on recently or which mechanic fixed your engine last are simply not relevant to the mechanic unless they relate to her purpose in locating the causal source of the specific problem she is tasked with fixing. In this way, perhaps we can ask ourselves what our purpose is in investigating an agent’s level of moral responsibility and use that answer to determine what the strongest causal factor is without entering into a regress of infinite prior causes. 

Conclusion

Deery and Nahmias (2016) have argued that interventionist causal modelling is a practical and useful way to examine claims about causal relationships as they relate to debates about moral responsibility. In this essay I firstly outlined the key positions in the debate, followed by Pereboom’s (2007) four-case manipulation argument and it’s challenge for compatibilism: to demonstrate a relevant difference between cases of manipulation and cases of determinism. Danny and Manny were introduced to emphasise the point that there may be a principled relevant difference between agents who are manipulated to act in accordance with another’s intentions and agents who are subject to determinism but not interventionist manipulation. The crux of the argument was presented: if a prior variable bears a stronger causal invariance relation to an action than variables describable at the time or immediately prior to an action, then the causal source of the action is identifiable with the variable with the stronger causal invariance relation. Deery and Nahmias (2016) have produced the most recent argument against the intended incompatibilist conclusion of Pereboom’s (2007) manipulation argument. As I have demonstrated, it seems technically promising and practically sound. The compatibilist approach towards moral responsibility has therefore been fortified by this argument. The burden now rests with incompatibilists to find a vulnerability with this account if they wish to reject it. 

References

Caruso, G. (2018) Skepticism About Moral Responsibility The Stanford Encyclopedia of Philosophy (Spring 2018 Edition), Edward N. Zalta (ed.), URL = <https://plato.stanford.edu/archives/spr2018/entries/skepticism-moral-responsibility/>.

Deery, O., & Nahmias, E. (2017) Defeating Manipulation Arguments: Interventionist causation and compatibilist sourcehood. Philos Stud 174:1255–1276 DOI 10.1007/s11098-016-0754-8, 2017

Fischer, J. M. (2000). Responsibility, history and manipulation. The Journal of Ethics, 4(4), 385–391

Fischer, J., & Ravizza, M. (1998) Responsibility and control: A theory of moral responsibility. New York: Cambridge University Press. 

Frankfurt, Harry G. (1971), 'Freedom of the will and the concept of a person', Journal of Philosophy, vol. 68, no. 1, 5-20.

McKenna, M. (2008). A hard-line reply to Pereboom’s four-case manipulation argument. Philosophy and Phenomenological Research, 77, 142–159

McKenna, M., & Coates, J. D. (2016) "Compatibilism", The Stanford Encyclopedia of Philosophy Winter 2016 Edition, Edward N. Zalta (ed.), URL = <https://plato.stanford.edu/archives/win2016/entries/compatibilism/>.

Mele, A.R. (2006), Free Will and Luck, New York: Oxford University Press. doi:10.1093/0195305043.001.0001

O'Connor, T., & Franklin, C. (2018) "Free Will", The Stanford Encyclopedia of Philosophy (Fall 2018 Edition), Edward N. Zalta (ed.), URL = <https://plato.stanford.edu/archives/fall2018/entries/freewill/>.

Pereboom, D. (2007) Hard Incompatibilism, in Fischer, J.M, Four views on free will. Blackwell, Malden, MA, 85-125 .

Watson, G. (2000). Soft libertarianism and hard compatibilism. In M. Betzler & B. Guckes (Eds.), Autonomes Handeln: Beitrage zur Philosophie von Harry G. Frankfurt, Berlin: Akademie Verlag., 59–70