Wrongs and Crimes is about the relationship between wrongs and crimes! It is about the nature and sources of wrongdoing, why wrongdoing can make a person liable to punishment. In the light of that it is about the scope of the criminal law. The book covers far too many issues – everything from the nature of wrongdoing, to debates about free will, to the nature of harm and the harm principle, to consent, inchoate wrongdoing, and firearms possession. As taster, I focus only on one issue that I address in chapter 5: can punishment be justified in the face of challenges from free-will sceptics?
Consider this familiar manipulation argument against desert (a rough version Derk Pereboom’s manipulation argument):
- A person deserves to suffer punishment only if she is responsible for her conduct.
- If a person’s wrongful conduct is fully determined by evil scientists, who have manipulated her brain to ensure that she acts wrongly is not responsible for her conduct.
- There is no responsibility-relevant difference between a person whose conduct is manipulated by evil scientists in this way and a person whose conduct is fully determined by an otherwise identical causal process, but where evil scientists do not bring about this process.
- All of our acts are fully determined by a causal process relevantly like that in 2).
- Therefore no one is responsible for her conduct.
- Therefore no one deserves to suffer punishment for acting wrongly.
If this argument succeeds, it shows that one popular theory of punishment – retributivism – fails.
There is an equivalent argument that is a challenge to a wider range of theories of punishment – those that aim to justify punishment by showing that by acting wrongly a person makes herself liable to punishment. My own view – the duty view – is one such theory. The duty view is that punishment is justified because of the goals that it instrumentally serves – most importantly protective goals. Wrongdoers are harmed to serve these goals, and the challenge is to show why harming them for the sake of these goals does not wrong them. But wrongdoers incur enforceable duties to serve the aims of punishment by acting wrongly. It does not wrong a person to harmfully use her against her will to serve an end that she has an enforceable duty to serve. Therefore, it is permissible to punish wrongdoers to serve the ends of punishment.
Now consider this challenge: like retributivism, the duty view depends on the idea that wrongdoers are responsible for acting wrongly. But if so, the duty view, like retributivism, is vulnerable to challenges from free-will sceptics. Defenders of the duty view can meet this challenge by showing that even if manipulated wrongdoers do not deserve to suffer, they do incur duties as a result of acting wrongly. I offer a much fuller argument that this is so in Wrongs and Crimes, but here is a brief case based argument.
First consider:
Bridge: Dorabella is on a bridge with Fiordiligi. A trolley is heading on a track under the bridge towards five people who will be killed if Dorabella does nothing. Dorabella can save the five only by throwing Fiordiligi from the bridge onto the tracks. Fiordiligi’s body will stop the trolley, saving the five, but Fiordiligi will be killed.
It is intuitively wrong for Dorabella to kill Fiordiligi.
But whilst harmful using is often intuitively wrong, it is not always intuitively wrong. Consider:
Wrongdoer on the Bridge: As Bridge except Fiordiligi has wrongly started the trolley in order to kill the five, simply because she will enjoy seeing them die.
It is intuitively permissible to use Fiordiligi to save the five. Here is a rough argument.
- Because Fiordiligi wrongly started the trolley she is required to save the five, even at the cost of her life.
- This duty is enforceable.
- It is often permissible to use a person to serve an end that she has an enforceable duty to serve, as long as the costs that she bears do not exceed those she is required to bear to serve the relevant goal.
- Therefore, Dorabella is permitted throw her off to save the five.
But free will sceptics may claim that premise 1) is unsound because it depends on the unreliable intuition that Fiordiligi is responsible for her wrongdoing. To meet this challenge, consider:
Manipulated Wrongdoer on the Bridge: As Wrongdoer on the Bridge, except that scientists have manipulated Fiordiligi’s brain to ensure that he acts wrongly. However, Fiordiligi fulfils all plausible compatibilist conditions of responsibility – his effective first-order desire to kill the five conforms to his second-order desires; his process of deliberation from which the decision results is reason-responsive, in that it would have resulted in him refraining from posing this threat were his reasons different; his reasoning is consistent with his character, because he is egoistic; but he sometimes regulates his behaviour by moral reasons; he is not constrained to act as he does and he does not act out of an irresistible desire.
It is plausible that Fiordiligi does not deserve to suffer because she is manipulated in this way. But is it plausible that he does not incur any duty to save the five in virtue of his wrongful attempt to kill them? To consider whether it is, imagine that you are Fiordiligi. You have started the trolley to kill the five, but now recognize that what you did was wrong. However, you also discover that scientists manipulated your brain to get you to act wrongly. Even if the manipulation weakens Fiordiligi’s duty to save the five, it does not plausibly negate it altogether. Surely she has a more stringent duty to save the five than an ordinary bystander. That is so simply because of the deep connection that there will be between his wrongful deliberations and acts and their deaths if he does nothing about it.
This suggests that even if sceptics are right that the manipulation argument should make us lose confidence in the significance of desert, it should not make us lose confidence in the idea that wrongdoers have duties to respond to their wrongdoing. Thus the duty view of punishment can respond to an important challenge from free-will sceptics.
Those who employ manipulation arguments against desert would likely push back on the claim that Fiordiligi can acquire new (or strengthened) duties in light of actions or decisions that flow from manipulated mental states. In a non-trivial way, Fiordiligi is a bystander to his own deliberations and actions. If he is a globally manipulated agent, he has no control over what he does (and has no control over how he reasons about what he does) since these features of his “agency” flow from covert neuroscientific interventions. Even if Fiordiligi comes to be aware of the fact that he was manipulated into setting up the trolley, this does not implicate his agency and this seems to absolve him from acquiring special duties as a consequence of “his” past deliberations and actions. The intuition in Tadros’s case seems, to me at any rate, to depend upon the idea that Fiordiligi has meaningful control over what ultimately happens to the five people that he was manipulated into targeting with the trolley. But if determinism is true, or if Fiordiligi is globally manipulated, that claim is dubious at best. Moreover, duties to perform (or refrain from performing) particular actions are plausibly acquired only in contexts where the agent is free to do otherwise. On one popular gloss on the “ought-implies-can principle,” it would make no sense to say that Fiordiligi ought to save the five if it is metaphysically (or nomologically) impossible for him to save the five. If that option is closed to Fiordiligi, he cannot have a duty to choose it — especially if his choice ultimately flows from factors over which he has no control.
Thanks for your interesting post Brandon.
As I understand the manipulation argument, the presence of the neuroscientists in the first case plays the role of brightly illuminating the fact that even though the person satisfies all compatibilist conditions of responsibility, his conduct is fully determined. Once this is illuminated in this way, the case directly elicits the intuition that the person is not responsible (in the sense that he does not deserve blame). If that is right, our question about manipulated wrongdoer on the bridge is whether the intuition that Fiordiligi lacks any more stringent duty than a bystander is directly elicited, and it seems to me (and I think to others that have reflected on the case) that it is not.
So one way of reading your post is that even if this argument can be met, there are other free-will sceptical challenges to the idea that we can acquire duties as a result of wrongdoing. For example, you might find this principle independently attractive:
FREEDOM FOR DUTIES (FRD): a person can acquire a duty only by the performance of some free action (in some robust sense of freedom that we do not have)
For my part, I don’t find FRD a plausible principle. There are many duties that we find ourselves with that do not result from free actions, even in the non-robust sense: duties to assist our parents, for example. If we can acquire duties by the connection between us and our parents, why can’t we also acquire duties by the connection that we establish with others by threatening them with death as a result of our decisions, regardless of whether those are free in some robust sense?
The last part of your post is even more difficult to address. I am sceptical about the very demanding version of ‘ought implies can’ that the argument relies on, but I don’t know how to show that it is false. (Of course, if we don’t have duties, we don’t have duties not to punish, so perhaps there is no interesting question about the justification of punishment in that case – let’s leave that aside though).
But suppose that it turns out that we don’t have duties in the strict sense. Pereboom, who believes this, still believes that we have something in the ballpark, and that something in the ballpark might be enough for me. Suppose, for example, that Fiordiligi has good reason to value his not being involved in the deaths of the five. In the absence of duties, that might be enough to make a difference to justify involving him in saving them at some cost to him.
Thank you for your thoughtful reply Victor. I was a little pressed for time earlier, so I’d like to clarify and elaborate upon my initial comments in light of your response. The dialectic involving free-will skepticism, determinism, and justifications for punishment is primarily centered on issue of moral desert. What I appreciate most about your approach is that it offers a justification for punishment that is independent of desert-based considerations, yet avoids the pitfalls of many consequentialist justifications. However, I worry that the very considerations hard incompatibilists take to undermine desert may also undermine your duty-based justification.
In my mind, the most salient (and controversial) argument against the compatibility of duty and determinism comes from Ish Haji, who argues that while moral responsibility is not threatened by determinism, deontic categories such as obligation, duty, rightness, and wrongness are threatened because deontic moral judgments require alternative possibilities. Leaving the issue of responsibility aside, there seems to be a great deal of plausibility to the claim that deontic assessment of agents (and their actions) depends on their possession of a non-compatibilist-friendly ability-to-do-otherwise. A central motivation for this view is the Ought Implies Can principle, which I referenced to in my first post. Though there is debate about how OIC is to be interpreted, Haji’s proposal does not seem to be overly demanding; it merely states that an agent has a moral obligation to perform (or not to perform) an action only if it is within that agent’s power to perform (or not perform) that action. Perhaps this principle, if true, does not ultimately undermine the obligations of determined agents (or strip their actions of rightness and wrongness) but it appears to pose a prima facie problem for duty-based justifications for punishment that strive to be compatible with determinism.
Perhaps, as you suggest, skepticism about obligation and deontic categories wouldn’t be so devastating after all; maybe, following Pereboom, we can get “in the ballpark” by using alternative normative schema that don’t appeal to deontic considerations, such as consideration of an agent’s reasons for action. But it might be the case that determinism threatens not only deontic categories, but also objective reasons for action. The move away from an agent’s obligations or duties to a consideration of their reasons may not yet admit us to the ballpark.
Apart from general skepticism about the deontic, my other main worry was that manipulated agents might not acquire new duties because of their apprently wrongful deliberations and actions. This worry is not so much a consequence of a principle like FRD, which I reject as implausible for precisely the sorts of reasons you mention. Rather, it stems from doubts about the “deepness” of the connection a manipulated agent’s actions and deliberations and the harms they ultimately cause. To motivate this concern, I’d like to consider the final case you present.
You invite the reader to imagine that they find themselves in Fiordiligi’s position of having set up the trolley to kill five people, but coming to realize that this is wrong. Where my intuitions become murky is the next step, where the reader gains knowledge that her decision was actually the result of covert manipulation and had been causally determined by neuroscientific hijinks. Upon discovering that fact, I would likely view myself as estranged from my earlier actions and the deliberations and psychological states from which they flowed. Given that these states were implanted in me, I would regard them as altogether alien and inauthentic – even if they seemed well-integrated with my psychology at the time. Where first there appeared to be a deep connection between my thoughts, intentions, and actions and a significant harm, the revelation of manipulation dissolves this bond by showing it to be a consequence of nefarious deeper factors beyond my control.
Faced with this revelation, I imagine that I would view myself as a sort of strange, first-personal bystander to my own experienced agency and behavior. Would I have an obligation to try to stop the trolley and save the five people in peril? Of course. But would this obligation be different in kind or degree from that of Dorabella, standing beside me on the bridge? There I’m not sure. But I certainly do not think Dorabella would be justified in killing me to stop the trolley on the basis that I had somehow acquired an extremely demanding enforceable obligation to kill myself in order to stop the trolley (if this was, indeed, the only way it could be stopped).
To put a point on it, I’m not sure that I agree with the first premise of the rough argument you’ve presented (or that, if I do agree with it, it does not capture the morally salient elements of Knowingly Manipulated Wrongdoer On Bridge). If Fiordiligi wrongly started the trolley, she may be required to save five at the cost of her own life. So far, so good. But if Fiordiligi was extensively manipulated into acting badly, she is arguably more akin to an innocent bystander (wronged by her manipulators) than a wrongdoer. Perhaps, owing to the aberrant causal history of her deliberations and actions, her decision and behavior should be viewed as amoral or simply bad rather than wrong. Or perhaps these actions and decisions, even if wrongful, cannot be fairly attributed to her given their deviant origin.
At any rate, I appreciate and welcome your approach and look forward to spending time with your book once it becomes available.
Thanks for coming back Brandon. These are great challenges. I have read Haji’s work on this, but I haven’t thought about it as deeply as I should. One general question is how to understand ‘can’ and ‘within that agent’s power’ when interpreting OIC. I am optimistic that there are compatibilist-friendly versions of these concepts, and that OIC is plausible only on these versions. But I admit that I don’t have an argument for that.
Let’s suppose that this is not so, and that we lack duties. The duty view of punishment responds to a challenge to justifying general deterrence – the challenge that it is normally wrong harmfully to use a person, even if that averts more harm. If we lack duties, it is not clear that this challenge even arises, because the challenge seems to rest on a non-consequentialist principle that might itself depend on our having duties.
A similar thing might be true if we should generally be alienated from our actions. The truth of that idea might undermine non-consequentialist principles, such as the principle against using others, because these principles rely on the importance of personal autonomy.
All of this is to say that if non-consequentialist ethics generally fails because of deterministic challenges, there is no challenge for the duty view to respond to.
In your reaction to the Knowingly Manipulated Wrongdoer on the Bridge case, you seem to be reacting especially to manipulation, and not simply because this brings to light the fact that your actions are fully causally determined. In that case, though, an alternative response to the manipulation argument against the duty view might be available – even if we do not incur duties in manipulation cases, we may incur them in relevantly similar non-manipulation cases. In Wrongs and Crimes I consider the idea that manipulation might itself make a difference, and so we should not rely on manipulation cases to brightly illuminate the fact that are actions are causally determined when assessing the warranted response to wrongdoing – doing so in this way introduces a confounding factor. I think that this response is also promising.
Finally, in order to defend the duty view, I don’t need it to be true that Fiordiligi has a duty to kill herself to save the five in cases like Wrongdoer on the Bridge. I would need that only if I was defending the death penalty (something which I’m not inclined to do anyway). I only need it to be true that her obligation to save the five is significantly more stringent than a bystander. You don’t seem hostile to there being this difference.
Sorry to be late to the party, Victor. I like the way Brandon has been pushing you, but I’m interested in another thought, named, that the stringency of the duty ostensibly is a function of the deep connection between F’s deliberations and her actions (causing the deaths). So she has a more stringent duty than a mere bystander. This seems right. I’m wondering, though, the extent to which this has any connection to responsibility. Suppose that after deliberating and acting, F dies but is immediately replaced with a replica (F*), someone who didn’t cause the situation but who is nevertheless exactly like F (and so seems to remember — mistakenly — having caused the situation). It seems to me that F* also has a more stringent duty than a bystander to save the five. But now, given the absence of identity between them, would you think there’s also no connection of responsibility (or at least no connection between F*’s deliberations and the harm-causing actions)? If so, then perhaps worries about responsibility here are red herrings?
Thanks for writing Dave. This is a tricky case, and I’m not sure exactly what to say about it, partly because it’s difficult to know what to say about the relevant replica cases more generally. For example, suppose that I am cloned. To what extent does the clone have reason to respond to the things that I have done? I am tempted to say that these reasons are at least weaker than my reasons, and perhaps the clone has no more reason than anyone else. But I can’t say that I’m sure about this. Suppose it is right. Now suppose that I am destroyed straight after I am cloned. I doubt this makes any difference to the reasons that the clone has.
So I am tempted to say that even if F* has stronger reasons to respond to F’s actions than a bystander, she has weaker reasons to respond than F, and perhaps no stronger reasons than a bystander. But, of course, as F* reasonably believes that he has caused the threat, F* has good reason to believe that he has as stringent a duty as F would have had.
Suppose that this is wrong, and F* has as stringent a duty as F. I doubt that this casts doubt on the significance of responsibility to our duties. It’s just that something else – being exactly like someone who would have been responsible – can also ground a duty. And that idea depends on responsibility. F* has a duty because F would have had a duty had she survived, and F* is exactly the same as F. If this is right, we are still entitled to conclude that responsibility grounds duties in more standard non-clone cases – the fact that being exactly the same as a person who is responsible can ground a duty does not show that responsibility does not ground a duty.
Thanks, Victor. For my part, I think you’re right to start thinking that “something else” may ground the relevant duty here. Plenty of responsibility theorists, for instance, are going to be grumpy if you allow for moral responsibility without causal responsibility (as would be the case for F* here). I’m happy, as a result, to think in terms of *attributability*, as in, “F’s action is attributable to F* in virtue of F*’s being a closest continuer, or in terms of F*’s continuity of beliefs, desires, values, character, etc. But my point was that if you make the switch to talk of attributability — which doesn’t require causal responsibility or identity with the doer of the deed — you will have perhaps moved sufficiently beyond the domain of responsibility PROPER that standard worries about manipulation and such no longer clearly apply.
Hello Dave,
It’s a really interesting case. I certainly agree that our duties have more than one source. And the claim that one person has a duty that another lacks does not always depend on standard conditions of responsibility. Your case may be one such case – as I suggested in my earlier reply I am not sure whether it is – and your explanation may be right (I am still troubled, though, by the idea that whether F* has a duty depends on whether F is destroyed, and so whether F* is the closest continuer. That makes F*’s duties depend on the difference between F existing and being unable to help and F not existing. But it is not clear why this difference should matter morally).
The more general question that I am trying to address is whether the manipulation argument counts decisively against the idea that standard conditions of responsibility ground duties. The fact that duties can be grounded without such conditions can contribute to the general case that this is not so. Roughly, the manipulation argument depends on the idea that our common-sense conception of responsibility is unreliable because we perceive that we have options that we lack in a deterministic world. Manipulation cases are supposed brightly to illuminate this fact, and when it is brightly illuminated we lose confidence in desert based judgements. The question is whether we also lose confidence in our judgements that we have duties to respond to our wrongdoing because we are ‘responsible’ (in a compatibilist sense) for it.
In one way your case, if it is successful, might be thought friendly to my cause, because it suggests that duties arise through what you call attribution without the spectre of the manipulation worry. This is grist to the mill of the more general argument that the manipulation argument will be unsuccessful in showing that we lack stringent duties to respond to our wrongdoing – it is part of the general case that we can have more stringent duties without having made choices at all. If this is so, we should be less concerned about the idea that our choices in more standard cases are determined: their importance in grounding duties need not depend on our judgement that our choices are not determined. Manipulated Wrongdoer on the Bridge suggests that this is so – even when causal determinism is brightly illuminated, we have more stringent duties to respond to our wrongdoing. And it seems natural to conclude that this is because of our decisions, judgements and so on.
But in another way, your case might be thought to pose a challenge. That is because it might be thought to support the idea that ‘compatibilist’ responsibility for wrongdoing is not a source of our duties at all. That explains my earlier reply – that the case, even if successful, would not show that this is true.
Yes, you’ve put your finger on the various options well, Victor. I was thinking what I said was meant to be a way to buttress your general view, but to show that worries about the manipulation argument aren’t directly relevant to its defense, in the end. (In general, I find myself unmoved by the manipulation arguments, for many reasons, including reasons of identity, attributability, applicability, and humanity.)
But you’ve also noted what is definitely a puzzling implication, namely, that if one goes with a “closest continuer”-type theory here, then *whether or not* F* has a duty depends on the existence or non-existence of F. And that IS puzzling. However, it might be that if F survives, then F has the *overriding* duty, one that, if she fails to discharge it, F* must then take up (i.e., F* has a kind of “back-up” duty). I’m not sure. These are strange waters.
Strange waters indeed! Thanks again for engaging with me on this.
I do think that manipulation arguments are relevant, though, in that our intuitions about the stringency of our duties are affected in manipulation cases. And even if our duties in non-manipulation cases are more stringent than our duties in manipulation cases, reflecting on these cases together makes me less confident that the duties that we incur through wrongdoing are as stringent as our pre-reflective intuitions suggest. This does not lead me to the conclusion that the duty view of punishment is wrong, but it does lead me to think that less punishment may be inflicted on wrongdoers for the sake of deterrence than I thought before reflecting on these cases.