How should libertarians conceive of the location and role of indeterminism?

Abstract

Libertarianism has, seemingly, always been in disrepute among philosophers. While throughout history philosophers have offered different reasons for their dissatisfaction with libertarianism, one worry is recurring: namely a worry about luck. To many, it seems that if our choices and actions are undetermined, then we cannot control them in a way that allows for freedom and responsibility. My fundamental aim in this paper is to place libertarians on a more promising track for formulating a defensible libertarian theory. I begin by arguing that Robert Kane's influential formulation of libertarianism actually generates an acute worry about luck, showing specifically that Kane's recipe for solving the problem of luck and his attendant conception of the location and role of indeterminism derived from it are deeply problematic. I then offer a reformulation of libertarianism – particularly a new conception of the location and role of indeterminism – that is capable of avoiding the problems that beset Kane's theory and that, I argue, places libertarians on a more promising track for formulating a defensible theory.

Keywords:

• free will
• libertarianism
• indeterminism
• luck
• Robert Kane
• causal theory of action

Acknowledgements

An earlier version of this paper was presented at the 2011 American Philosophical Association, Pacific Division Meeting, in San Diego, CA. I am grateful to the audience for their helpful comments and especially to my commentator, Al Mele, and fellow presenter, Neal Tognazzini. I am also grateful to two anonymous referees for their careful and instructive comments on earlier versions on this paper.

Notes

I use “action” here as an umbrella term: all agential activity falls under the rubric “action”. So choices and efforts of will count as actions. This use of action is consistent with Kane's. See especially Kane (1996, 125) where choices and efforts of will are classified as actions. In other contexts, I will use “action” more narrowly, referring only to agential activity subsequent to choice, such as overt behavior (as Kane himself does (Kane 2011, 385)). The context should make clear how to disambiguate my use of “action”.

I discuss three other important formulations of the luck argument in Franklin (2011a).

The idea that free will requires scenarios in which agents are torn between two alternatives can also be found in Balaguer (2009), Campbell (1967), and Kant (1785/1998).

To clarify, this is not an effort of will to make up one's mind about what to choose. It is an effort of will to make a particular choice, such as the choice to help the victim of the mugging.

I assume here a product view of action, according to which an event is an action in virtue of its causal history. In this account, agents cause their actions. Alternatively, proponents of component views contend that an event is an action in virtue of its internal causal structure. For example, O'Connor contends that agents do not fundamentally cause free choices, but rather free choices (partly) consist in the agent's fundamentally causing “the-coming-to-be-of-an-action-triggering-intention-to-so-act” (2000, 94; cf. 2009, 195–6). I am sympathetic to Clarke's (2003, 25) contention that nothing of substance as regards free will depends on which theory we adopt. But regardless, none of my arguments in this paper turn on these issues. I will continue, for ease, to employ a product view of action.

The qualification of non-deviance is required to exclude cases in which mental states bring about an event, but, intuitively, there is no action. For example, a spy's belief that if he does not blink his eyes at a certain time, then the mission will go poorly and his desire to see the mission succeed might so unnerve him that he blinks his eyes at the right time (a nervous twitch). This event is caused by his belief and desire, but it does not seem to qualify as action, since it was brought about through a deviant causal path (cf. Davidson 1980).

Again, O'Connor would reject this description and insist, instead, that what we would find is a causally complex event of the agent fundamentally causing the coming to be of an action triggering event.

In Franklin (2011b), I argue that this is exactly the role that indeterminism plays (or can play) in event-causal libertarian models of free will.

What moved Kane to double the effort of will was the following kind of worry: it seems that the businesswoman would not be responsible in worlds in which her effort of will failed to culminate in a choice to help the victim. In such worlds, she tried as hard as she could to execute her intention, but through no fault of her own she is frustrated (cf. Mele 1999). In order to accommodate the possibility that agents can be responsible for whichever choice they make, Kane doubled the effort of will. While this move is odd in certain respects, it is completely intelligible once we realize that Kane is, once again, applying his recipe: show that the agent was trying to do what she did (choose to go to the business meeting) and you have shown that she is responsible despite the presence of indeterminism. Kane thought that if the businesswoman was trying to make both choices, then she would be responsible regardless of which choice she made. And thus, we get a second effort of will. I will largely ignore Kane's doubling of the effort of will as it makes no progress toward answering the objections I raise. Below I will argue that Kane's recipe cannot solve the problem of luck and thus we have no reason to follow Kane in doubling the efforts of will. In fact, I will argue that we have no need to introduce any efforts of will in the first place.

Others (Clarke 2003; Mele 2006) have pointed out this problem. What has not, I believe, been appreciated is that Kane, given his recipe for solving the problem of luck, cannot solve it. That is, while others have picked up on some of the problems that beset Kane's view, the problems I raise, and specifically the problems generated from his recipe (as revealed by the dilemma I am about to present), have not been sufficiently appreciated.

In defense of this claim, Mele (2006, 52) offers a case in which the effort of will is implanted in the agent via manipulation. The causal connection between effort of will and choice is indeterministic and so the effort may culminate in one of two mutually exclusive choices. Yet it seems, given that the agent is not responsible for making the effort, that she is not responsible for the outcome of the effort either.

According to volitionism about freedom and responsibility, an agent is free with respect to, or responsible for, a bodily action only if that action was caused by a temporally prior volition or act of will. Ryle (1949, 67) argued that this view is subject to an infinite regress objection: if what makes an agent free with respect to, and responsible for, a bodily action is that it was caused by a temporally prior volition, won't volitions for which agents bear freedom and responsibility also need to be caused by prior volitions? This objection has no force against Kane since he does not ascribe to a volitionist view with respect to either action or freedom. Indeed, he explicitly rejects the former view (Kane 1996, 24).

My regress objection is not aimed at Kane's theory of freedom, but his recipe for solving the problem of luck. While Kane never claims that in order for an agent to be free and responsible for an undetermined event the agent must be trying to bring about the event, he does claim that if the agent is in fact trying to bring about the event, then he is (or at least can be) free and responsible for it. His recipe is meant to offer a strategy for proving that an agent is free with respect to, and responsible for, an undetermined event. My regress objection is part of a more general objection that targets Kane's strategy for dismantling the luck argument.

Let me be clear what the objection is here. I believe that Kane should grasp the first horn. My objection is not that if he grasps the first horn, then he cannot solve the problem of luck. Rather my objection is that if he grasps the first horn, he cannot use his recipe to solve the problem of luck. The significance of this objection, as we will see, is that it undercuts the motivation for following Kane in his conception of the location and role of indeterminism. If the only reason offered for adopting Kane's conception of the location and role of indeterminism is the theoretical advantages of his recipe, and if this recipe turns out not to offer any theoretical advantages, then we have no reason to adopt Kane's conception of the location and role of indeterminism.

There is some evidence that this is the horn that Kane would grasp, see Clarke (2003, 90) and Mele (2006, 53).

Obviously, more must be added to generate a sufficient condition for control transfer. For example, we might think that the agent will have to satisfy some sort of epistemic condition concerning the relationship between his action and its causal results. But the simple version of the idea should suffice for present purposes.

One might argue that both indeterminism and determinism are incompatible with free will. On such a position, it would not follow from the fact that indeterminism is incompatible with free will (and thus in some sense diminishes control) that determinism is necessary for free will (and thus in some sense enhances control), since determinism is also incompatible with free will. Likewise, someone might be suspicious of my claim that if indeterminism diminishes control, then determinism must enhance it. But such an inference does follow, given the sense in which Kane argues that indeterminism diminishes control. Indeterminism constitutes an obstacle to our carrying out our efforts. If indeterminism were absent, we would be guaranteed to succeed in doing what we were trying to do. This would be an enhancement of our control.

In order to explain away the air of paradox, Kane distinguishes two kinds of control (cf. 1996, 144, 186–7). Under “plural voluntary control”, which is indeterministic in nature, an agent cannot guarantee an outcome before it happens. Under “antecedent determining control”, which is deterministic in nature, an agent can determine an outcome before it happens. Kane concedes that indeterminism eliminates antecedent determining control, but argues that indeterminism is necessary for plural voluntary control and that only plural voluntary control is necessary for freedom and responsibility. In other words, according to Kane, what is at stake in the incompatibilist debate is not the quantity of control but the quality.

It is hard to know exactly how to respond to this unorthodox and provocative suggestion. It seems that Kane is committed to the following claims: plural voluntary control affords an agent less control than antecedent determining control (after all Kane admits that indeterminism diminishes control), that plural voluntary control is necessary for freedom and responsibility, and that antecedent determining control is incompatible with freedom and responsibility. These claims jointly entail the following: an agent S ₁ can have greater control than an agent S ₂ and yet, precisely because S ₁ possesses this greater control, S ₁ is not free or responsible. But surely this implication reveals that something has gone wrong. Part of the attraction of my reformulation of libertarianism is that it allows us to avoid these problems with, or at the very least peculiarities of, Kane's account.

There are two points here. The first point is as follows: whatever actions libertarians conceive as basic should be the actions that they require to be undetermined. If libertarians have reasons for thinking that efforts of will antecedent to choice are the basic actions that agents typically perform, then they should hold that these efforts are undetermined. The second point is as follows: libertarians have no theoretical need for efforts of will, and so we should conceive of choice as the basic action that free agents typically perform.

This account bears similarities to Clarke's “unadorned event-causal libertarian view” (2003, 29–56). Although Clarke offers a limited defense of this view, in the end he rejects it (2003, 93–118). In Franklin (n.d.), I argue that this unadorned account requires further refinements in order to handle other worries about luck, such as the problem of the disappearing agent (Pereboom 2005, forthcoming).

Levy (2011, 45) seems to saddle libertarians with this commitment.

Though see Franklin (2011a, 2011b, 2012, n.d.) for attempts to show that it does have the resources to solve both problems.