Robust Local Alternatives and Responsibility*
A Frankfurt case is supposed to provide an example of an agent who is responsible for some action to which there was no "local" alternative: a way to avoid performing the action that existed within the situation ‘at hand’. That is, it is intended to show that an agent may be praiseworthy or blameworthy- an appropriate object of one of the Strawsonian "reactive attitudes"- for a certain action despite currently being unable to do otherwise. (A local alternative is to be contrasted with a "historical" one: something that could have been done prior to the situation ‘at hand’ so that it would have had a different outcome.) The inevitability of a Frankfurt case’s outcome is a matter of a process unfolding unbeknownst to the agent, set in motion by someone desiring that outcome, that will force her to perform a particular action should she not decide to do it ‘on her own’.
Here is a case in point. Debbie, a renowned neuroscientist, wanted to sleep with her married colleague, Phil. To that end, she slipped a nanobot into his tea that would have compelled him to make an advance by midnight on Friday, had he not decided by then to do so ‘on his own’. Assuming that this device would have worked as intended, it seems that Phil had no choice but to sleep with Debbie. Moreover, despite his inability to avoid becoming involved with her, it is something for which he appears blameworthy. Certainly his wife, having discovered his unfaithfulness, would not absolve him, even were she apprised of Debbie’s device. However, defenders of PA, or the principle of alternatives according to which an agent would not be responsible for her action were it to lack alternatives, have challenged both of these claims.
Some maintain that a "Frankfurt agent" (FA) can be seen as having more than one option, depending on how we specify his action.1 Timothy O’ Connor, e.g., holds that an FA, although he cannot avoid performing a certain action, can avoid its voluntary performance. The claim here is that an FA has more than a "flicker of freedom" in the form of a local alternative, viz., acting in a certain way involuntarily, rather than voluntarily. Thus, if we hold an FA responsible, not simply for what he did, but for doing it ‘on his own’, we will not be holding him responsible for something to which there was no alternative, so that his case satisfies PA. It would fail to satisfy PA only if we held him responsible for voluntarily or involuntarily performing his action, which was unavoidable. In the above Frankfurt case, since Debbie has determined that she will have Phil by midnight on Friday, he has no choice but to sleep with her ‘one way or the other’. But when he approaches her on Thursday afternoon, he is still able to put off initiating the affair ‘on his own’. If his exercise of that option is that for which he is blameworthy, then it is not true that he is responsible for doing something to which there was no alternative: he could have involuntarily entered into the relationship instead. Thus, this case does not seem to count as a counterexample to PA.
A "friend of Frankfurt cases" can respond here by pointing out that such a case would fail to be a counterexample to PA only if its agent not only had an alternative to what he did, but the possibility of doing something else entailing less blame or praise than his actual course of action, which would constitute a "robust" alternative on his part.2 It will not do for PA’s defenders to assert that an FA has alternatives unless they are options he can be faulted or praised for not exercising. A Frankfurt case is supposed to show that an agent, without having had any way of avoiding performing an action having a certain moral or practical value, can still be praiseworthy or blameworthy for what he did, having voluntarily committed his action. If an FA cannot avoid doing something that is right/wrong or prudent/imprudent, he may yet be an appropriate candidate for one of the reactive attitudes provided that he acted ‘on his own’. Frankfurt cases establish as much unless PA’s proponents point up morally or practically unequal options on the part of an FA. That is to say, the version of PA that is under attack by the friends of Frankfurt cases is "robust" PA:
(RPA) An agent is responsible (praiseworthy/blameworthy) for performing a given act only if he could have performed an alternative with a different moral/practical value
Were alternatives of this sort available to Phil? The only existing one entailed acting involuntarily. Still, that may have been the morally preferable option, depending on why it would have been chosen. The involuntariness of the act involved does not by itself make it insufficiently robust to satisfy RPA, since there may be circumstances in which acting ‘on one’s own’ would not be morally/practically equivalent to acting involuntarily, given the "internal" reasons for doing one rather than the other. E.g., if at the onset of a war I enlist in the service rather than be conscripted, so as to gain the satisfaction of having volunteered, then, depending on whether or not the cause for which I will be fighting is just, I will be more or less praiseworthy than I could have been. Thus, the answer to the question of whether or not Phil should be faulted for not acting involuntarily depends, I would maintain, on what he believed and intended.
There are several possibilities to consider. First, supposing that he was aware of the presence of the nanobot and wanted to initiate the affair voluntarily, his unfaithfulness seems more reproachable than it would have been had he simply allowed it to occur, revealing as it does a lack of compunction. On the other hand, there is something to be said, as Martin Luther maintained, for sinning wholeheartedly. Fortunately, we do not have to settle this issue, since, as noted, it is of the essence of an FA that he not be aware that he is about to do something involuntarily should he not decide to do it ‘on his own’. What should we say, then, of Phil who is ignorant of Debbie’s device? Should we fault him for not doing that which would have caused him to act involuntarily, viz., delaying taking his decision until sometime after midnight Friday, an option, for all he knew, that existed? Not unless we have reason to believe that he thought that exercising it would have led to forbearance sans her interference. If he had merely been struggling with qualms that he expected to eventually set aside, his conduct would have been no less contemptible than that in which he actually engaged.
A friend of Frankfurt cases, thus, emerges from the debate over robust alternatives having to make a small concession to her opponent. To wit, she must stipulate that an FA justifiably rejects the option of exercising restraint until intervener’s deadline has passed, otherwise there would be a robust alternative to the action he can perform voluntarily: viz., allowing his desire to perform that action to dissipate, so that he would be performing involuntarily an action from which he thought that he would have probably refrained had he been ‘left alone’. Obviously, our Frankfurt case can be revised so as to meet this requirement. We just need to imagine that Phil is under the perhaps mistaken but justifiable impression that his passion for Debbie will not cool any time in the foreseeable future, so that any delay on his part would be caused simply by apprehensiveness. He would, thus, be left with no robust alternative to having an affair: either he will give in to his desire and initiate the affair himself or the nanobot will put an end to his dilly-dallying. For him, the option of doing something less reprehensible than deliberating violating his marriage vows, viz., involuntarily being "unfaithful" for having resisted the temptation to have an affair, will not be one that he should be faulted for not exercising, since it will not be amongst his epistemic possibilities. An FA meeting this requirement we will refer to as an ‘FA*’.
Carl Ginet counters that even an FA* has a robust alternative. It would have been morally preferable, he maintains, for Phil to have been involuntarily unfaithful for having been unable to overcome his compunctions before Debbie’s deadline- even if he was just about to do it ‘on his own’.3 It makes no difference whether or not he was on the verge of starting the affair, he would still have been less reproachable than he actually is, having not only contemplated being unfaithful, but also decided to set aside whatever qualms he had had about breaking his marriage vows. For the purpose of judging morally, the only relevant fact is whether or not wrongdoing occurred; the reasons why something was or was not done at a particular time should not matter here.
This move, however, entails that a person’s moral luck should not be disregarded in evaluating his conduct. The only difference between actual philandering Phil and counterfactual philandering Phil is temporal: the latter is vacillating when the nanobot effects him- an action over which he has no control. Had it not effected him, he would have done the very same thing as actual Phil, only later. Let us suppose that counterfactual Phil would have ‘made his move’ at 12:01 p.m. that night. Thus, by meeting her deadline, actual Phil has, by Ginet’s lights, made himself more blameworthy than he would have been had he hesitated for another minute. With this view in mind, we can imagine him saying to himself, upon discovering what Debbie had been up to, ‘too bad, if I had waited just a little longer I wouldn’t have become as bad a person as I am now’. Or to look at things from counterfactual Phil’s perspective, it was his good fortune that the nanobot stimulated him when it did, otherwise he would have been actual Phil’s moral equal. As things stand, in virtue of nothing he did, he is less blameworthy. The nanobot’s intervention accounts for that moral fact.
It appears, then, that treating an FA* as having a robust alternative leads to the conundrum that has vexed philosophers such as Thomas Nagel and Bernard Williams: it seems unfair that a person’s moral standing in relation to others should depend upon factors over which she lacks control and, yet, that is precisely what is entailed by many of our moral judgments.4 By treating an FA* as lacking a robust alternative, we sidestep this issue here: Phil could not have made himself less blameworthy by delaying his decision unless it was going to issue in the resolution to remain faithful. The nanobot’s intervention would not alter his moral status.5
A supporter of RPA may object at this point that an FA* could not be responsible for his action, given his belief that it is inevitable. In the present context, however, Phil’s belief that he cannot restrain himself is not mitigating. If his inability is the result of free choices that he made prior to the situation at hand, then he is blameworthy for being unfaithful, as even some libertarians have conceded. Remember, for present purposes, we’ve set aside the question of historical alternatives. I am only trying to convince libertarians such as O’ Connor and Ginet that an agent who is responsible for her action need not have had alternatives in the temporal vicinity of its commission.
O’ Connor’s specification move, thus, fails to clearly demonstrate the existence of robust local alternatives for an FA*. Ginet maintains, however, that this move does not exhaust the conceptual resources available to the incompatibilist. There are other ways of qualifying judgments of responsibility, which, when combined with an incompatibilist friendly revamping of the parameters of a Frankfurt case, entail robust alternatives on the part of an FA*. By Ginet’s lights, an FA* (and any agent) should be held responsible for having acted precisely when he did, and not for having acted at any less specific time.6 Moreover, according to him, to avoid begging the question against the incompatibilist, the friend of Frankfurt cases must place the deadline for the voluntary performance of the desired act beyond the time at which it is actually performed. If this "time-shift move" is not made, if the actual and counterfactual actions are taken to have the same time of occurrence, then the process that would have led to the involuntary performance of the act would have failed to achieve that result only because the monitoring of the FA*’s neural processes revealed a sufficient condition for its voluntary performance, in which case the incompatibilist will claim that it was not a free act, being determined.7
In response, I will adjust the parameters of my Frankfurt case in the way Ginet prescribes and see whether or not it entails robust alternatives on the part of an FA*. Thus, let us stipulate that Phil made his overture at 11:59 p.m. on Friday and that it is that for which he is blameworthy.
The problem with Ginet’s proposal is that it makes responsibility temporally relative. It allows for the possibility of someone being responsible for doing something at a certain time without deserving to be held accountable for its being done simpliciter.8 In our example, although Phil would be responsible for being unfaithful at 11:59 p.m. on Friday, which was not inevitable, it would wrong to hold him accountable for being unfaithful, since that is not something he could have avoided doing, thus preserving RPA. Thus, if we equate being an appropriate candidate for one of the reactive attitudes with being responsible, Phil’s wife would be justified in being indignant with him for being unfaithful at 11:59 p.m. Friday, but not in feeling resentful towards him for being unfaithful. If his conduct came to light, however, one imagines that such hair-splitting on the part of Phil, suggesting that the inevitability of his unfaithfulness mitigates his guilt, that he is not as blameworthy as he would have been had Debbie not implanted the nanobot, would not change the way his wife feels. (‘You betrayed me period!’ she would respond.) Thus, Ginet’s proposal fails to track the way in which the reactive attitudes are actually held.
Ginet responds to this criticism with a putative counterexample to the principle in question: if an agent is responsible for an act being committed/consequence occurring at t, then he is responsible for that act being committed/consequence occurring at any time including t. Booth, he says, was responsible for Lincoln dying on 18 April 1865, but not for Lincoln dying sometime before 1920 (when he would have been 111 years old).9 It must be asked, however, why shouldn’t we hold Booth accountable for the fact that Lincoln died sometime before 1920? Admittedly, it is not a fact that would interest an historian, whose concern is with contingencies. That it was going to obtain regardless of what Booth did may tempt us to say that it is something for which no one is responsible. We must resist this temptation, however, as giving in to it would, in the present context, amount to question begging, since the issue at hand is whether or not inevitability entails lack of moral responsibility: the Booth case is part of a defense of RPA. Thus, the question becomes: are we dealing here with just another Frankfurt case? If we are, then we may adjudicate it in the same manner as we treated the case of Phil, which shows that inevitability does not preclude moral responsibility. If it is not, then it is irrelevant to our present concern: determining whether or not an FA* should be held accountable for what is inevitable. That is to say, if Booth turns out not to be responsible for Lincoln’s death simpliciter, but is not an FA*, then his case cannot serve the purpose of defending RPA against the use of a Frankfurt counterexample. The friends of Frankfurt cases will concede Ginet’s point about the required specificity of moral judgement vis-à-vis the Booth case, but insist that an FA* should be held accountable for that which is inevitable, viz., the less specific action entailed by the temporally indexed action he commits.
It is obvious that Booth was not an FA*, having been certain of Lincoln’s mortality. The reason he is not responsible for the (inevitable) fact entailed by the more specific (contingent) fact for which Ginet does hold him accountable is that he willed that the latter obtain rather than the former, not because its obtaining was inevitable.10 An FA*, however, is not cognitively in a position to make such a choice: Phil, in our example, does not reject making the less specific (and inevitable) fact of coming-on to Debbie at some time or other obtain in order to have an illicit relationship with her at a definite time. He simply decides to attempt to seduce her. Thus, the reason that Booth, who is knowledgeable in the way that Phil would have been had he been aware of Debbie’s device, is not responsible for what is inevitable does not apply to an FA*.
There does not, then, appear to be a non-question begging way of demonstrating that responsibility is temporally relative that would help in explaining away the intuition that an FA* should be held accountable for performing an unavoidable act. It will not do to attribute this judgment to the failure to temporally specify that for which an FA* is responsible. Even having shifted an FA*’s action backwards in time, so that its counterfactual involuntary performance and actual voluntary performance are temporally separated, the friend of Frankfurt cases can still maintain that an FA* is responsible for what is inevitable- his performance of that action at some time or other- given his responsibility for what is not unavoidable- his performance of that action before it would have been done involuntarily. The Booth case is a counterexample to the principle ‘Any agent who is responsible for the obtaining of F @ t is also responsible for the obtaining of F at some time or other’ but not the narrower principle ‘Any agent, who does not specifically intend that F obtain @ t rather than at some time or other, but who (nonetheless) is responsible for F obtaining @ t, is also responsible for F obtaining @ some time or other’. The obtaining of the fact that an FA* performed a certain action @ some time or other is covered by the latter principle. Thus, that for which he is responsible is inevitable.
Notes
* I am indebted to Carl Ginet and Michael McKenna for their helpful comments on earlier drafts of this paper.