Moral Responsibility in a Vat

Abstract

This paper investigates an ingenious argument by Andrew Khoury which, if valid, could shed new light on some of the most relevant discussions within the field of moral philosophy. The argument is based on the idea that if we deny the phenomenon of resultant moral luck, then the proper objects of moral responsibility must be internal willings. I analyse the argument and find it unsound. The argument does not adequately account for the positions of all relevant moral actors when it comes to the moral evaluation of agents and their actions.

1 Introduction

Lionel Messi is responsible for Argentina’s win in the latest FIFA World Cup. Joe Biden is (perhaps) responsible for the latest recession in the USA. Jimmy Hendrix is responsible for how the electric guitar is played today. These are all cases of responsibility. Anders Breivik is responsible for killing an unspeakable number of human beings. Bill Cosby is responsible for abusing women. Chesley Sullenberger is responsible for saving many lives. These further examples are all cases of responsibility too. Moreover, these latter examples seem to be more relevant, at least from a certain point of view—they are examples of an agent’s having moral responsibility.

It is beyond doubt that all of the abovementioned agents are responsible for something or action that is external to them, whether a crime or a heroic act. We ordinarily take agents to be morally responsible (also) for things outside of themselves. Recently, however, Andrew Khoury (2018) has presented an interesting argument that seems to negate this position (or to put it in a rather different light). He argues that, if we deny the existence of resultant moral luck, an agent is morally responsible only for things that are internal to him or her.¹

This paper investigates the merits of this argument and finds it wanting. The first section presents Khoury’s argument, and the second section focuses on the central conditions of moral practice—conditions that this argument does not accommodate (which is why it ultimately fails). The third section investigates interesting cases and argues that Khoury’s argument cannot account for them adequately. The paper ends with a brief conclusion.

1.1 The Argument

Khoury’s (2018) focus is on the objects of moral responsibility—i.e. what agents are responsible for. His main argument is that once you assume that resultant moral luck does not exist, you should also agree that an agent can be morally responsible for only some of his or her internal mental states (which he goes on to call ‘willings’). He spends some time arguing for the assumptions of the argument: that resultant moral luck does not exist and that we are always responsible for something.² Let’s unpack these assumptions in further detail.

The moral luck debate will be familiar to many and has many facets, but in general, it is understood that (at least in its modern form) it began as a discussion between Thomas Nagel (1979) and Bernard Williams (1981, 1993), who investigated the supposed tension between the following claims: We should not be blamed for things outside our control; what we do is in many ways influenced by factors outside our control; and we seem to morally praise and blame agents partially for the results of their actions. Nagel introduces four kinds of luck: constitutive luck (your happening to have the character you have, e.g. being peaceful rather than volatile); circumstantial luck (your happening to find yourself in certain situations, e.g. being in a position to save a drowning child); causal luck (being causally determined to act a certain way); and resultant luck (concerning the consequences of our actions).³ These kinds of luck and their role in moral evaluation have received much attention in moral philosophy. The debate on the existence of resultant moral luck is based on the question of whether we do in fact form moral judgements about agents (also) on the basis of the results of their actions. This is nicely demonstrated by a standard example. Consider two shooters with exactly the same upbringing, personal history, motivation, skills, modus operandi, etc., who attempt to kill someone by shooting him. The main point of the setup of the example is that there are no morally relevant differences between the two prior to their taking their shot. As it turns out, one of them is successful and hits his target. When the other shoots his gun, however, a bird suddenly flies by and intercepts the bullet, taking the hit. Now if you think that these agents ought to be assessed differently, then you believe in the existence of resultant moral luck. If, like Khoury, you do not think we should make any distinctions when assessing the agents morally, then you are against the existence of resultant moral luck. My point in this paper does not concern Khoury’s arguments for the non-existence of moral luck; therefore, I do not present them here.⁴ Thus, I grant the assumption that there is no such thing as resultant moral luck.

Let us now turn to the second assumption, i.e. that agents cannot be viewed as morally responsible simpliciter but are rather responsible (also) due to the scope of their actions. This assumption is backed by Khoury, who criticises one of the main views within the moral responsibility debate. Khoury counters Zimmerman (see, e.g. his 2002a, b, 2006), who presents a view according to which we should distinguish between the degree and the scope of moral responsibility. The scope covers the actions for which an agent is responsible. The shooters from the example above differ in terms of the scope of their actions (one is a murderer, the other has merely attempted to murder someone). However, their actions are equally morally blameworthy—the degree of their moral responsibility is the same. But this means that the agent can be held responsible without there being an action for which he is responsible or with the action being irrelevant.⁵ Khoury argues against this conclusion. He explicitly states: ‘It is implausible that the scope of one’s moral responsibility is itself irrelevant to one’s moral responsibility’ (Khoury, 2018, 1363). Again, I will not go into the details of his arguments against Zimmerman. I grant him this assumption, simply because the main problem I have is with Khoury’s central argument, not its premises.

Let us now turn to the argument itself. As is usual in this debate, it is motivated by examples. Consider Lee, who is an assassin. In the first scenario, he plans a murder, prepares himself, aims, takes the shot and hits the target, killing his victim. In the other scenario, everything is the same, except that a bird flies into the assassin’s line of shot, and thus, the target survives. Classically, if you deny the existence of resultant moral luck, you should evaluate Lee the same way in both scenarios: he is equally blameworthy in both.⁶ Up to this point, the argumentation is quite standard. In an interesting move, however, Khoury presents another case:

Case 3: Lee is an assassin who reasonably believes that he has placed himself near a window on the fifth story of a building overlooking Elm Street. He reasonably believes that he carefully and skillfully draws a bead on his target. He reasonably believes that he pulls the trigger, but he fails to kill because, unbeknownst to him, he has just recently had his brain removed from his skull, placed in a vat of liquid, and hooked up to a computer which simulates the external environment.

Khoury (2018, 1361)

This is an ingenious move. There is no difference between Lee’s mindset, motives and actions and those of the man who hits his target. Khoury claims that the same considerations that inform our moral assessment of the first two (non-vat) cases are equally applicable to Lee’s case. As Khoury argues, ‘[t]he fact that Lee in Case 3 fails to actually kill is, as far as he is concerned, just as much a matter of luck as it is [in] Case 2. Neither of the factors that inhibit the execution of his plan in these cases serve to excuse his behaviour’ (Khoury, 2018, 1361). This, if correct, is an interesting point when considering the object of responsibility. The reasons are simple. If we agree that there is no morally relevant difference between the agents in the abovementioned scenarios, then whatever each agent is responsible for should be the same in all three cases. This leads Khoury to identify our inner willings as the proper object of moral responsibility. Even more importantly, this argument, if valid, generalises. For any action, we can think of the agent as simply being a brain in a vat, which supposedly cuts off the possibility of there being something non-internal to the agent that serves as the locus of moral responsibility. This is Khoury’s (2018) main argument. If correct, it would be an interesting development within the debate on moral luck, and moral philosophy more generally.

1.2 The Problem

In the previous section, I presented the relevant parts of Khoury’s argument. In this section, I will show why it ultimately fails. As the examples are described, we are asked to imagine three scenarios. The first two are standard examples from the debate on moral luck, and they are meant to show that if we view the first two shooters as morally on par, we are committed to viewing the third shooter as equally blameworthy. I grant Khoury that. The main step in the argument is whether we agree that there is no morally relevant difference between these two shooters and the shooter who is just a brain in a vat. Our intuition is meant to be that the shooters behave in exactly the same way in all three cases; the only difference is that the shooter does not make the kill in two of them and that this is a matter of luck.

Even though the argument seems promising, there is one salient aspect that is under-described in all of the cases: who are the moral evaluators, and what is their relation to the agent? Let us assume that the relevant evaluators are ordinary people living in the ordinary world, which they share with the shooter. This is clearly not equally the case in all three scenarios. We can grant this in the first two (non-vat) cases. We can assume that we, the evaluators, are living in the same community, governed by the same moral rules, as the shooter. And we can evaluate him with respect to these rules. But this is surely not the case in the third scenario. As Strawson famously sees it:

What I want to contrast is the attitude (or range of attitudes) of involvement or participation in a human relationship, on the one hand, and what might be called the objective attitude (or range of attitudes) to another human being, on the other…

The objective attitude may be emotionally toned in many ways, but not in all ways: it may include repulsion or fear, it may include pity or even love, though not all kinds of love. But it cannot include the range of reactive feelings and attitudes which belong to involvement or participation with others in inter-personal human relationships…

Strawson (1962, 52)

These interpersonal relations form the basis of our moral behaviour (including blame, remorse, etc.). When we try to position ourselves objectively with respect to the agent we are assessing, we are missing the point—we are not treating him as a member of the moral community but as something that can be measured, quantified, etc. Thus our position as moral evaluators outside of the vat, assessing his actions inside the vat, is nothing like the position taken up by moral evaluators in actual moral practice, which is marked by an essential equality between evaluator and evaluatee as members of the same moral community.⁷

The central idea and motivation behind the counter-argument is that there is a significant difference among the cases presented by Khoury in his argumentation. This difference was neither covered nor reflected upon by him, and it holds relevance concerning considerations about the moral responsibility of the agent within the presented cases, specifically, the role, and importantly, the position of moral evaluators in relation to the evaluated agent’s actions. When we take these considerations into account—and it appears we should (or, at the very least, one should provide an argument against it)—there exists a notable difference among the three cases. Namely, the evaluators, presumably not in a vat, can observe that the agent is in a ‘real’ world in the first two considered cases but not in the third case. It is plausible that the evaluators would find the notable difference in the agent’s position relevant when making their moral evaluation of his actions.

The scenario of an agent placed in a vat is not significantly different, in terms of relevant features, from a situation where we morally assess the actions of an agent who had severely limited access to many (or some) pertinent resources while committing those actions. A community (or judge) typically considers the distinction when a child steals food from a shop due to hunger, having not eaten for several days, compared to doing it ‘just for fun’. Similarly, we take into account the restricted information flow and limited accessibility to global news in North Korea when evaluating the actions of individuals there. In another case, the actions of suffragettes in the past century would be viewed quite differently if performed by women who already have the right to vote. In all the mentioned cases, including the shooter in a vat, there is a notable limitation on the agent performing the action. We acknowledge this limitation as relevant from a moral standpoint when making moral evaluations. The child has limited access to food. The person in North Korea has limited access to corroborated independent news. The suffragette has limited (or no) access to the possibility of influencing the situation in her community or country. The shooter in a vat has limited access to, well, reality, or at least the same reality as people who are not considered to be in a vat. Even if the shooter in a vat is not in a place, position, or ability to acknowledge the limitations placed on him, moral evaluators are in such a position, possessing the ability and most probably will take the difference into account.

1.3 Vat of One or Vat of Many?

There is an interesting possible reply to my argument: what if the moral evaluators were also put in vats without their knowledge? Would this solve the problem faced by Khoury’s argument? Or what if Lee is the only one evaluating his actions? Agents can and often do morally evaluate themselves. I will consider these possibilities in turn.

Let us consider the scenario in which not only Lee, the shooter, but also those evaluating his actions are put inside a vat. Both Lee and those evaluating him believe that his missing his target was simply a matter of bad luck. I agree that if we consider a scenario in which both moral evaluators and the shooter were in a vat and then emerged, they would still make moral evaluations of the actions taken in the vat. However, it is essential to note that they would likely take the difference into account. I surmise that the evaluators would assess the actions differently, or at least not equivalently, when they were in a vat without their knowledge compared to after they got out.

Let’s delve deeper into the scenario where both moral evaluators and the shooter are placed in a vat, unbeknownst to them. Assume they make moral evaluations and assign some moral responsibility to the shooter. According to the assumption of the case, none of the relevant actors is aware of the vat-placement. Now, what can we conclude about the validity of the moral evaluations made by the evaluators in such a position? We must consider whether we deem them relevant or not. Arguing for the second option, that they are not relevant, is challenging, as the very relevance of the case stems from the assertion that we should also consider their position. So, let’s consider their evaluation-while-in-a-vat as relevant. Is it of the same validity as one provided in cases outside of the vat? Again, the answer is either yes or no. From their perspective, nothing has changed, so there is no apparent need to alter their evaluation. However, one crucial aspect has changed—the basic assumptions upon which they base their evaluation now have different validity. Despite their lack of awareness, they still assume that the shooter attempted to kill somebody external to him. They do not know they are in a simulation. Yet it is precisely because they are envatted that the information upon which they (also) base their evaluation is invalid. Even if they do not perceive any difference between the cases, there is still one—the validity of the information influencing the epistemic conditions placed on a moral evaluator. Not only do we consider the fulfillment of the epistemic conditions on the side of an agent whose actions we evaluate, but we must also fulfill some (epistemic) conditions to be in a position to evaluate. The fulfillment of these conditions clearly differs between the evaluators outside of the vat and those in a vat. Even if we consider this position, there is still a relevant difference among the cases unaccounted for by Khoury’s argument.

Now, let’s consider the scenario where the evaluators somehow exit the vat. Presumably, they would still make evaluations of the actions they now understand were taken in a vat. They can reflect on the situation and still consider it morally wrong to attempt to kill somebody in a simulation. They might even agree, upon reconsideration outside the vat, that the proper placement of responsibility lies with the shooter’s will in such a case. However, these evaluators, now disenvatted, would precisely discern the difference between their real-life scenario and their simulated experiences. Thus, Khoury’s argument against the existence of relevant moral differences among the cases doesn’t hold as soon as we take into consideration the role of moral evaluators in our moral practices.

Let us now focus on the second possibility. In this scenario, there is no one in the world but Lee and his target.⁸ Let us also assume that Lee would view his action as morally permissible, whether he hits its target or whether his bullet is stopped by the bird.⁹ Now imagine again that he is put in a vat. As far as he knows, he missed his target due to (bad) luck. Again, he cannot ever know that he is in a vat—otherwise, his evaluation of himself would change. So again, as in the previous case, even though a scientist might be able to see that there really is no target and that he is merely in a simulation, this does not make the object of responsibility, from the point of view of the only moral evaluator in this case, internal to the agent.

2 Conclusion

I have investigated an enticing argument by Khoury that seems to have the potential to put at least some of the relevant discussions within the field of moral philosophy in a new light. I have analysed the crucial section of Khoury’s argument, and I have shown that his argument does not properly account for the position of all relevant moral actors within the presented cases. The position of the moral evaluator(s) should not be neglected or glossed over; as Strawson suggests, it may be the reactions of these members of the moral community that provide a basis for moral behaviour itself.