23 Responsibility and the Ability to Do Otherwise
We have by now discussed the two most important moves in the free will debate. There is, on the one hand, van Inwagen’s Consequence Argument, which is designed to show that if determinism is true, then no one has control over what they do: powerlessness transfers, via the causal structure of a determinist world, from the past and the laws of nature to powerlessness over future states of the world. One response deploys Lewis’ distinction between two different interpretations of ‘could have done otherwise’. On a strict, but unreasonable interpretation of the phrase, the agent could have produced a given outcome even if the laws of nature remain as they are in the actual world; the agent has a remarkable power to break the laws. On a weaker interpretation, there is a nearby world in which the historical record and the laws of nature are marginally different, and the agent acts differently in that world. The Consequence Argument does not rule out that the agent could have done otherwise on the weaker interpretation of the phrase.
The exchange reaches a stand off when the incompatibilist insists that only the strict interpretation matters for freedom. The fact that there is a nearby world in which the agent does something different does not by itself show that the agent acts freely in this world. On the other hand, the compatibilist may replay that what matters for freedom is whether the agent has the ability to do otherwise, which is best understood in terms of what the agent does in nearby possible worlds.
(Frankfurt 1969) takes a completely different approach. Rather than focus on how to understand ‘could have done otherwise’, Frankfurt argues that the ability to do otherwise is not in fact required for moral responsibility. This is a radical move, since it suggests that the debate over free will has rested on a mistake. For compatibilist and incompatibilist philosophers alike have often presupposed that much of what is in question in the debate over free will is whether or not we are morally responsible for our actions. If we do not act freely, one might have thought, then I should not be held responsible for my actions. But both parties agree that I do not act freely if I could have not done otherwise. They simply disagree on what is for an agent to be able to act otherwise.
To be more precise, philosophers have often presupposed the Principle of Alternate Possibilities:
An agent is morally responsible for what they have done only if they could have done otherwise.
The thrust of Frankfurt’s paper is to argue that the principle is simply false. But before we look at his alleged counterexamples, we should ask why, on the face of it, the principle strikes one as plausible. The basic thought is that it seems unfair to hold someone morally responsible for an action if they genuinely had no choice over it. If the agent is, say, coerced into doing something or if she is impelled to act by a hypnotic suggestion, then not only does it seem impossible for the agent to avoid acting that way but it would seem unfair to hold them morally responsible for the outcome.
23.1 Frankfurt Cases
(Frankfurt 1969) deploys a family of thought experiments designed to undermine the Principle of Aternate Possibilities.
Black is a neuroscientist, who has implanted a device in Jones’ brain that actively monitors his deliberative states. Black wants Jones to perform a given action, e.g., to vote for a specific candidate. The device does not interfere with Jones’ deliberations unless it detects a sign of Jones deciding not to perform the action Black wants Jones to perform. If the device detects such a sign, then it immediately interferes with the deliberations in order to cause Jones to perform the action and vote for the candidate in question.
As it happens, Jones deliberates entirely on his own and without interference and decides for his own reasons to vote for the candidate Black wants him to support. The device plays no role in the outcome.
In Frankfurt’s basic case, we find:
- Jones could have not done otherwise. Had he shown any sign of deciding to act differently, the device would have intervened and cause him to vote for the candidate Black wanted Jones to support.
- Jones had no knowledge of the device,
- Jones deliberated on his own and acted entirely from his own reasons and motivations.
- The device played no causal role in the outcome.
Frankfurt suggests that Jones is indeed morally responsible for his vote even if he could have not done otherwise. The device played no role in Jones’ deliberation and decision. So, it seems difficult to excuse Jones on the grounds that he could have not done otherwise.
But this now seems a clear-cut counterexample to the Principle of Alternate Possibilities. Jones is morally responsible for his vote even if he could have not done otherwise. The ability to do otherwise is not required for moral responsibility.
Notice that the point of Frankfurt’s argument is not to deny that alternative possibilities never matter for moral responsibility or that coercion is never an excuse. The target of the argument is rather the claim that the mere absence of alternative possibilities is sufficient to excuse someone. Instead, we have to look at the sequence of events that led to the action and whether it resulted from the agent’s own deliberation and decision. The fact that an alternative route to the same outcome existed in the background is irrelevant for the attribution of moral responsibility.
23.1.1 The Flicker of Freedom Response
One common line of response to Frankfurt’s cases is this. For Black’s device to work, it must detect some prior sign that Jones is about to decide otherwise, which may take the form of some neural state, some moment of hesitation, or some early deliberative indicator. But that prior sign is itself something Jones does, and it is something he could have avoided. If Jones had not produced the sign, the device would not have activated. So Jones does retain a thin alternative possibility after all: he could have not produced the sign that triggered the device to interfere with Jones’ deliberation.
The objection concludes that Principle of Alternate Possibilities is not refuted. For the principle only requires that the agent have some alternative possibility, however thin and however early in the deliberative process. Frankfurt cases do not eliminate alternative possibilities entirely; they merely push them back to an earlier moment. The flicker is small, but it is enough to satisfy the Principle of Alternate Possibilities.
This line of objection has generated a substantial literature on how to build variants of Frankfurt’s cases that leave no flicker at all. But Frankfurt’s own response to the flicker objection is that even if Jones retained the ability to produce or not produce a sign that triggers the device, that flicker is itself not a morally significant action. It is at most a precursor of a decision, and not the outcome of a process of deliberation itself. And responsibility attaches to what the agent does and decides to do, not to thin counterfactual precursors that may have played no role in the outcome.
23.1.2 What Grounds Moral Responsibility
Frankfurt’s cases suggest that moral responsibility need not be grounded on the existence of alternate possibilities for the agent. Moral responsibility is appropriate even in cases in which no alternate possibilities are open to the agent. That raises the question of what grounds moral responsibility in such cases.
(Frankfurt 1971) outlines an answer according to which moral resposibility for an action is grounded in the fact that the action flowed from the agent’s own will in an appropriate way. To develop the thought, Frankfurt makes a distinction between first- and higher-order desires.
First-order desires are desires to act in one way or another. The objects of first-order desires are actions, e.g., to vote for a candidate, to write a book, to raise one’s hand, etc.
Second-order desires are desires to have one first-order desire or another. The objects of second-order desires are first-order desires, e.g., to want to vote for a candidate, to want to write a book, to want ro raise one’s hand, etc.
According to Frankfurt, what distinguishes free agents and free actions from animals and unfree actions is the capacity to form second-order desires and to act from first-one desires one reflectively endorses. An agent acts freely when their action accords with a first-order desire that they endorse at the second-level. That is, the action expresses a desire to act in that way that they want to have and identify as their own.
By way of illustration, Franfurt compares the case of a wanton with that of an unwilling addict.
The wanton acts on first-order desires without a second-order desires to act in that way. They simply do what they want without ever considering whether their desires are ones they endorse or would endorse on reflection. Frankfurt’s claim is that wantons are not genuine free agents, not because they lack alternative possibilities, but because their actions do not express a reflectively structured will.
On the other hand, an unwilling addict has a first-order desire for the drug but a second-order desire not to have that desire. They do not want to want the drug. When they act on the addiction, they act against their own reflective endorsement. That is, their action does not express who they take themselves to be. Frankfurt’s account yields the verdict that the unwilling addict does not act freely, even if they could have done otherwise.
There is a contrast between this case and that of the willing addict, who has a first-order desire for the drug and endorses that desire at the second-order level. That is, they want to want the drug, or at least have no second-order desire to be rid of it. According to Frankfurt’s account, the willing addict acts freely, even if they could not have done otherwise. What matters is that the action expresses their reflective will, not whether alternative possibilities were available.