Is there one kind of responsibility or many kinds? Some theorists hold that there are several distinct kinds. Watson famous distinguished between attributably-responsibility and accountability-responsibility (1996). This distinction can be seen to track character theory (e.g. Frankfurt 1969, 1971; N. Arpaly 2014; A. Smith 2005; H. Smith 2014) and control theory (Fischer 2006). An agent is attributability-responsible if she exhibits a defective quality of will, and an agent is accountability-reponsible if she commits an infraction and is reasons-responsive. Some theorists have also advanced an ‘answerability’ model of responsibility, on which responsible agents must be capable of answering moral demands – they must be competent conversational partners, on one compelling version (McKenna 2012). In addition to these models, Scanlon (2014) has defended a view of ‘substantive responsibility,’ on which a person is substantively responsible if she cannot complain about her situation, and specifically, about any burdens she bears as a result of her choices. Some theorists treat these types of responsibility as distinct and incommensurable. Others think that one type of responsibility is the true account. For example, Arpaly, who offers a character account of responsibility, does not think that her account captures only one type of responsibility – responsibility-as-attributibility. She thinks that it explains responsibility, full stop. And she lodges objections agains the control view, which implies that this view also does not capture a distinct type of responsibility – it is just wrong. And similarly, Fischer does not offer his control view as just one variant of responsibility. It is mean to be a comprehensive account of responsibility, and attributionists are just wrong. McKenna, too, seems to think that his view is uniquely correct, or at least that it is a uniquely correct interpretation of responsibility as an interpersonal practice in which we deploy the reactive attitudes.
It is hard to know what to make of these disputes. One pertinent question is, what does holding someone attributability-responsibility versus accountability-responsibility versus answerability-responsibility amount to? What kinds of reactions are these, and what makes them moral reactions? Strawson offered a very nice answer to this question in his account of the reactive attitudes. When we hold someone responsible, we are prone to deploying the reactive attitudes of praise, blame, resentment, disapprobation, etc. This is a rough sketch, but theorists have since fleshed it. Now many theorists hold that the reactive attitudes are conative-affective states, which consist in an emotional state and a judgment that that state is appropriate (Bennett, Watson, Wallace). Others take a cognitivist position (Zimmerman, Haji). I think that the former account is more plausible because there is a difference between holding someone responsible and merely judging someone responsible. If I hold someone responsible, I exhibit a disposition to respond with a cognitive-cognitive state. This does not mean that I must necessarily express that judgment in overt behaviour: the reactive attitude is a standing disposition, with pro tanto force. If expressing my resentment to a gunman would cause him to kill a room full of hostages, I have all-things-considered reason to bite my tongue. I merely have pro tanto reason to blame him, but this is overridden by countervailing considerations of a consequentialist nature. But this does not mean that I do not (sotto voce) blame him. By contrast, when I judge someone responsible, I make a calculated assessment of the person’s character or capacities. This is how I judge a tomato: I deem it to have certain qualities. This is what attributability-responsibility would amount to on a cognitivist picture. But surely blaming someone is significantly different than judging a tomato to be over-ripe.
I’m going to assume that the battleground on which the debate should be waged is over which account of responsibility justifies the reactive attitudes. Should we deploy these attitudes in response to character flaws or reasons-responsive transgressions/omissions? Even if we make this concession to Strawson, there is still substantial room for debate. We could try to resolve this debate by disputing whether a person can be responsible for an unconscious omission, such as forgetting a friend’s birthday or leaving a pet in the backseat of a hot car, or whether a person can only be responsible for a ‘benighted’ omission, stemming from a prior voluntary choice, in which reasons-responsiveness was ‘online’ (e.g. drinking and driving). This is the crux of much discourse on responsibility, with theorists offering (arguably) intuitive examples to support their respective positions. But this seems to me to be a stalemate. The disputants simply find their opponents’ examples to be untenable, reject them, and then offer their own competing examples.
So here is another point of ingress. Arpaly offers an interesting theoretical explanation for why some theorists, like Kant, have historically favoured control theories, which prize reflective ability. It is because acting on a reflective judgment makes it more likely that the person will do the right thing. When a person does the right thing because she believes it to be right, it is “non-accidentally” connected to rightness (Arpaly 2014: 145). Arpaly objects that there is nothing non-accidental about the relationship between believing an action is right and doing the right thing, and she offers the following counterexample. A Nazi believes that it is right to ensure that Germans are well off, so he rescues a German bleeding on the side of the road. Had the victim been an Italian, however, the German would have left him to die in the ditch. It was an accident that he saved the German: the victim just happened to have the right nationality – the nationality that the German cared about.
Arpaly also offers an example of why unconscious dispositions – non-reflective character traits – can be appropriate objects of praise and blame. The example is Huck Finn, who helps his friend Jim escape from slavery, even though he believes slavery to be justified. Finn is an example of ‘inverse akrasia.’ He acts against his better judgment, but does the right thing (unlike Aristotelian examples of akrasia). According to Arpaly, Finn is both virtuous and praiseworthy, even thought he lacks the reflective competence to understand the normative import of slavery. This means that he can be responsible, in spite of lacking reflective control over his benevolent motives (on one very natural interpretation of control). Some theorists would want to say that Finn is attributablity-responsible but not accountability-responsible. But Arpaly does not want to say this. Finn is just praiseworthy.
Arpaly emphasizes that the Kantian ideal is flawed because reflective competency does not non-accidentally guarantee right action. It might not even facilitate right action. Some ‘moral fetishists’ might dogmatically follow their distorted moral principles come what may.
Julia Driver offers a different perspective. In expositing David Hume’s account of responsibility, Driver explains that some people may be naturally (dispositionally) virtuous, like Huck Finn, whereas others have “an extra very valuable skill,” the ability to apply moral norms, understand moral reasons, etc. (2014: 160). This ‘extra’ ability is the “hallmark of moral agency” (161). So for Hume, individuals can be responsible without having moral agency, the capacity for critical reflection. But this is still a valuable moral capacity.
I would resist Hume’s distinction between moral responsibility and moral agency, but take to heart his comment that critical reflection is an important moral capacity. In most people, this capacity will, as a matter of fact, facilitate ethical behaviour. This goes against Arpaly’s view that critical reflection is not inherently morally valuable at all. It is natural to think that, while good dispositions may sometimes secure ethical behaviour, critical reflection is a valuable asset. However, a person can also be good on the force of her intuitions.
Whether reflective competency is an asset really depends on a person’s background psychology. Someone who lacks good instincts may need critical reflection, while someone who lacks critical reflection may need to rely on instinct and non-reflective cognitive processes. Some people may be particularly good practical Kantians, relying on moral reflection as a guide, while others, just because of circumstances, may be terrible Kantians, but good intuitive actors. There are, as Owen Flanagan put it, a “variety of moral personalities” (1991), and we cannot say that one moral personality is necessarily superior to another. If Amy is a good Kantian and Mary is a good virtue ethicists, there is no need to try to convert either to a different moral personality. Indeed, it would be counterproductive, since if either were to try to follow the other’s example, she would probably fail. This is just because it is hard to change one’s psychological profile on command.
So now let us suppose that intuitions and critical reflection both facilitate ethical behaviour. Neither is more ‘non-accidentally’ connected to right action than the other; whether one or the other is likely to produce right action depends on the agent’s moral personality. If this is right, then Arpaly’s defence of character theory fails. But if we accept her reasoning – that praise and blame should track moral traits that predict ethical behaviour – then the reactive attitudes should track different traits in different people. They should track reflective competency in reflective personalities, and character traits in ‘intuitive’ people like Huck Finn. So, on Arpaly’s logic, combined with Franagan-esque considerations about varieties of moral personality, we can infer that both theories get something right. Sometimes the reactive attitudes should target character and sometimes they should target reflective control, depending on the moral features of the targeted agent.
This is the upshot of the current discussion. Moral responsibility should be taken to consist in deployment of the reactive attitudes, and the reactive attitudes should target character in some people and reflective control in others. Their appropriate target depends on the moral personality – the dispositional sensitivities and strengths – of the target agent.
Note that this is a forward-looking account of responsibility, whereas most theories are backward-looking and desert-based. That is, most theorists ascribe responsibility on the basis of whether the agent had a character flaw at the time of action, or whether the agent possessed reasons-responsiveness at the time of action – a time in the past. The present proposal, by contrast, considers whether a particular type of attribution is likely to facilitate a particular outcome – ethical behaviour. It is, more precisely, a functionalist account, which sees responsibility attributions as fulfilling a certain function: fostering right action. This view is compatible with a character basis and a control basis for responsibility ascription: we can blame a person on either ground, just incase the attribution serves the function of responsibility ascription as deployment of the reactive attitudes.
This is my proposal. It is not fully formed and I welcome comments on it. It resembles, I think, McKenna’s view and Vargas’ view in many ways, only it is functionalist, not consequentialist, and not justified by the fittingness of our conversational responses per se.