Printer Friendly

Are coerced acts free?

After a busy day of Christmas shopping, Anne walks back to her apartment through the hustle and bustle of the city streets. As she moves through the crowd, a ringing sound grows increasingly louder until she spies a volunteer, clad in the customary black uniform, bell in hand, standing beside the Salvation Army kettle. As she passes, she dutifully deposits some change. She proceeds to round the corner onto the street where she lives, only to be confronted by a man carrying a gun. This man, obviously not in the Christmas spirit, threatens her, saying "Your money or your life!" Immediately she hands over what's left of her money to the crook who grabs it and flees.

What we have in this case is a simple, if extreme, instance of coercion. Anne, threatened by the mugger, gives up her money for fear of losing her life. It is traditional philosophical fare to hold that actions such as these, viz., coerced actions, are the antithesis of acts freely performed. In fact, most compatibilist accounts of freedom cite instances of coercion as among the paradigm cases of unfree action. But there are some prima facie considerations that seem to indicate that such an action was not at all unfree. It certainly seems that Anne could have simply refused to comply with the robber's demand. As a result, it appears that she could have done something other than what she did, and, on many accounts of freedom, this is a sufficient condition for an acts being free.

It is puzzling issues of this sort that are brought to the fore by the phenomenon of coercion. What does it mean to say that the victim's action was coerced? Should we categorize the victim's action as unfree even though it seems she could have done otherwise? And if we count the coerced action as free, how is it different from other free actions from which we distinguish it, but which seem to share a number of relevant features with it (e.g., even in the Salvation Army case the "victim" could feel coerced - not by a gun, but by a guilty conscience)? In this essay we will propose an answer to these questions by arguing for a certain account of coercion and its relation to freedom and moral responsibility. Of course, we could not hope to provide a complete analysis of the relationship between these three formidable philosophical topics in one essay. Nevertheless, the account of coercion we develop here requires some comments concerning the way in which it affects these closely allied concepts.

We will begin our investigation by looking at some accounts of coercion which are representative of the recent literature. Our purpose however is not to provide a comprehensive look at these alternative accounts, since many of them approach the problem with concerns orthogonal to our own, concerns arising primarily in contexts of political or legal philosophy. While we do not argue that discussion of coercion in these contexts is misguided, our intention is to show that approaching these issues from another flank yields some unique insights into the problems arising in the literature. As a result, our goal in our first section is simply to do some stage-setting to show how other accounts have approached the problem. In the second section we will use features of these other accounts to set forth our own proposed set of necessary and sufficient conditions for coercion. In the third section we will examine the implications of this account for the related notion of freedom. Our comments here are primarily aimed at the implications of our account for the view of freedom we favor: libertarianism. However, the account of coercion that we develop in the second section uses language which is indifferent between libertarian and compatibilist accounts of freedom. In this third section we argue that a strong case can be made from principles likely to be plausible to the libertarian for interpreting coerced acts as free acts. Finally, we will investigate the issue of whether or not the coerced individual can be held culpable in performing the coerced act.


It is not overstating the matter to say that many quite different accounts of coercion have been proposed in the literature. For our purposes we will focus on three representative positions set forth by Harry Frankfurt, Bernard Gert, and J. P. Day. Again, we do not pretend that these positions by any means exhaust the spectrum of accounts offered by those who have attempted to characterize coercion. Instead, each of these accounts has exercised some measure of influence over the literature and raises issues that will be important for the development of our own account.

Briefly, Frankfurt argues that coercion occurs only when a person is overwhelmed by a threat to the extent that any contrary action is rendered impossible. In Frankfurt's words, coercion "requires that the victim of a threat should have no alternative to submission."(1) However, there are a number of ways one might dissect this brief characterization. What exactly is the force of the claim that the coerced individual is overwhelmed or has "no alternative to submission?" According to Frankfurt, the coerced individual acts in an unfree manner because coercive settings render the coerced unable to choose. When we are coerced, we are so overtaken by the circumstances, especially the threat, that we act in an almost mechanical way. When coerced, it is as if we are moved by an outside source. My hand is not moved by me, but by force of the coercer's will or by the threat itself. As a result, coerced activity is better rendered as a coerced behavior rather than a coerced act (since we normally reserve the latter title for free and deliberate activity). This means, according to Frankfurt, that coerced behavior cannot be cause for praise or blame since it is not behavior that is brought about by the person but merely, as it were, through the person.(2)

While Frankfurt's position has a number of things that recommend it, it ultimately fails in important respects. While it may be true that in many cases threatened individuals are overwhelmed by the threat to such an extent that it is as if the threat moves them directly, bypassing their deliberative and willful faculties, this is not always, or even usually, the case. Consider, for example, cases in which one submits to extortion or blackmail. Both of these cases involve giving something to a threatener of one sort or another. But we can easily imagine that the strength of the threat is as significant as it is in the case of Anne and the mugger described above. J.P. Day and H.J. McCloskey have argued, convincingly, that we must distinguish between the sort of case discussed by Frankfurt and cases in which the threatened makes a deliberate, albeit coerced, choice.(3)

To this end, McCloskey introduces a distinction between force and coercion. According to this distinction, individuals who are thrown into a blind panic or driven by an irresistible impulse are not coerced but forced to do what they do. These are not legitimate cases of action since they do not involve any deliberation or consideration on the part of the one performing the relevant behavior. In cases of coercion proper, McCloskey argues, there is no such irresistible impulse or blind panic but a choice, which arises out of deliberation about alternatives, i.e., an act. Those who are "Frankfurt-coerced" do not act, properly speaking; they are simply one link in a chain of unfree events, events as determined in their nature as the falling of a stone. In such cases, a person does not act, she is acted upon. We believe some such distinction between forced behavior and deliberate action is surely needed and will be mindful of it in the remainder of our discussion. Consequently, we will restrict our use of "coercion" to those cases in which the threatened individuals are not forced, in the sense described above, to do what they do.

According to the second account, proposed by Bernard Gert, coercion involves both a choice and an act, but the choice involved is "internally" different from choices in non-coercive contexts. When one is legitimately coerced, it appears as though all of the options save one are, in some sense, unchoosable or ineligible for that person. Gert, like Frankfurt, agrees with this condition. Unlike Frankfurt, however, Gert holds that the victim's alternatives are not narrowed to one by force but by the fact that the threat introduces "incentives" for the threatened to choose in a certain way. Gert argues that coercion is "the result of a threat of evil which provides an unreasonable incentive"(4) and that "an incentive is unreasonable if it would be unreasonable to expect any rational man in that situation not to act on it."(5) On this view, coercion is an instance of normal human action, but with a twist. The twist is that all of the alternatives, save one, are made "unreasonable" by means of the incentives introduced by the coercer's threat. Thus, the coerced individual acts but the action is unfree since, according to Gert, an action can be free only when there are no unreasonable incentives leading a person to perform or not to perform that action.(6)

While this explanation of coercion more accurately depicts the nature of the choice involved, we find it to be unsatisfying for the following reason. What "the reasonable man" would or would not do in the context of a threat just does not matter. Consider the case of one who has a great fear of traveling over 20 miles per hour in an automobile. Could such a person be coerced if he was threatened with a trip on the Interstate? It seems so, even though he would be succumbing to a threat which any "reasonable man" would not find coercive in the least. Thus, our account of coercion must take into consideration the fact that even threats that the "reasonable man" would not find compelling can still introduce coercion-producing unreasonable incentives. Thus, whether or not alternatives are unreasonable is, in a sense we will describe shortly, more a function of what is reasonable for that individual or reasonable as that individual sees it.

Yet, there are certain pressures that lead one to attempt to index coercion to the reasonable man, as Gert does. The main one results from the fact that we have a tendency to make coercion a sufficient condition for morally excusing the coerced act. For example, in the case of our "Interstataphobic," it might be fair to say that he was coerced when confronted with the threat "Kill the guy in the back seat or we are going up the entrance ramp!" What makes one hesitate to call such an individual coerced is that we think that the threat of driving fast constitutes insufficient grounds for morally excusing murder. The person, one is inclined to say, should not have been coerced by such a threat. And so, in an effort to satisfy our sense of justice, we become inclined to admit that the individual was not coerced. But to do this is just to confuse the relationship between coercion and moral responsibility.(7) As we will show below, the issues of coercion and moral responsibility need to be more carefully teased apart, allowing us to dispense with the indexing of coercion to what the "rational man" would be inclined to do.

A third approach to coercion, proposed by J.P. Day, contends that coercion occurs when someone is placed, in virtue of a threat by an intentional agent, in a situation where she cannot fulfill a conjunctive desire, specifically, a desire which, in the absence of the threat, she would have been able to fulfill.(8) As a result, on Day's view, a coerced individual has her "liberty" restricted since she is unable to satisfy as many conjuncts of her desire as she was able to before being threatened.

Yet while Day's approach seems to capture an important feature of coercion, it does not go far enough. Coercive situations are unique, not simply in that the coerced individual is suddenly unable to fulfill a conjunctive desire, but in the fact that the very act of choosing is special. Consider our standard case again. Here the gunman is presenting the victim with a choice: she can keep either her money or her life, but not both. As such, Day's criterion is satisfied, and so, he would say, this is a legitimate case of coercion. Yet there is more to be said about what is taking place here. The gunman does not wonder which option his victim will choose; he is quite sure he will be leaving the scene a richer man. The certainty of the victim's selection, the fact that no other choice seems to be "available," is the heart of coercion, and it is this issue that Day does not fully consider. While the initial loss of liberty is correctly assessed by his view, there is more at stake in a coercive situation.

It should be clear to the reader that the approaches taken in discussing this issue are widely varied. In the following section, we will borrow some elements of the above accounts in order to establish what we take to be a more satisfying account of coercion.


To begin, we reiterate our disagreement with Frankfurt. The type of case he describes will not be in view here, since we are concerned only with cases of coerced actions, and not the cases of forced behaviors he describes. Having said that, we admit that we do find something valuable in the latter two accounts, which we have attributed to Gert and Day.

To begin, we distinguish various key features of the coercive situation. Specifically, we distinguish between the act that the threatened person is being asked to perform by the threatener and the consequence the threatener threatens to bring upon the threatened if the threatened action is not performed. We will refer to the former as "AR" for the "act required" and the latter as "TC" for the "threatened consequence." Having made this distinction, we propose the following definition of coercion:

A person P is coerced if and only if P is placed in a situation in which:

1) P believes that P is unable to have both [similar to]AR and [similar to]TC.

2) P's situation is such that choosing {TC and [similar to]AR} is ineligible for P.

3) The threat is the reason for P's choosing AR.

4) P does not want TC to be actualized.

5) TC is intentionally imposed on P by another rational agent and these intentions are understood by P.

The first condition is self-explanatory and is included to incorporate Day's observation concerning the way in which coercive situations essentially curtail "liberty" by preventing the coerced individual from fulfilling her conjunctive desire.

The second condition is the most difficult to characterize. As noted above, our project is not just to describe the external circumstances that surround an instance of coercion. The goal is to understand just what it is about a coercive situation that makes the coerced individual's decision to comply with a sufficiently strong threat a virtually foregone conclusion. In other words, what is it that makes the threatened say, when later reflecting on her action, "There was nothing else I could do?"

It is our contention that in cases of coercion, while the victim is faced with a choice, each of the alternatives, save one, is ineligible for the coerced individual. In other words, in a coercive situation, the threatened individual could choose to do nothing other than comply. This claim is initially surprising since it seems to entail that no other choice was possible for that individual. And this seems to be clearly false. Couldn't the individual who complies with the threat have chosen to resist the threat instead? The short answer to this question is no. Detailed discussion of just what we take to be impossible for the coerced individual comes below.

In the remainder of the discussion of this condition we will lay out the somewhat complex parameters that determine when one is in a situation in which a threat has limited one's options to a single alternative. On our view, an individual who is coerced cannot choose to do anything other than comply, because no other alternative is perceived as choosable or electable or eligible for that individual at that time under those circumstances. As a result, all such alternatives are ineligible for P; or, more specifically, when P is being coerced to do AR, any alternative, A, to doing AR is such that A is ineligible for P.

But what brings it about that in some cases all alternatives to AR simply become ineligible? What is needed here is a way of characterizing choice-situations in which every alternative to AR is ineligible. This characterization can be found in a function with two variables, one corresponding to the AR and the other to the TC. Assume that our coerced individual assigns a subjective utility to all potential AR's for that individual in specified circumstances, from the most disagreeable to the least. Thus, it may be for me, as it appears to be for Aristotle, that the AR with the lowest utility assignment would be matricide, whereas the one with the highest assignment could be any number of trivial acts. Likewise, one could also assign all of the potential TC's a subjective utility, ranking them from the most to the least objectionable. We can then use these utility assignments to assign certain numerical values, from 1 to 100, which we will use to define a certain function. Let us call these numerical values [AR.sup.*] and [TC.sup.*]. For the AR, [AR.sup.*] is directly proportional to the subjective utility assigned to that AR, i.e., the AR with the lowest utility assignment is given [AR.sup.*] value 1, the one with the highest assignment, value 100. For the TC's we assign a number from 1 to 100 which is inversely proportional to the assigned subjective utility. Thus, the most disagreeable TC has a [TC.sup.*] value of 100, and the least, a value of 1. From [AR.sup.*] and [TC.sup.*] we can define a function, call it the coercion function, or C-function, which defines a value, the C-value. We propose that for any threat, the C-value can be calculated as the product of [AR.sup.*] and [TC.sup.*]. When this C-value is greater than some number, n, the threatened will be such that any alternative to AR will be ineligible.

Thus, the two extreme cases here would be the ones in which [AR.sup.*] and [TC.sup.*] are minimal, i.e., where the C-value approaches 1, and where [AR.sup.*] and [TC.sup.*] are maximal, i.e., where the C-value approaches 10,000. In an example of the former case, if one were to propose to me that either I kill my mother or be called a dirty name, the degree of coercion would be minimal, i.e., the C-value would approach 1. This is so because the AR, killing my mother, has a very low utility assignment and thus is such that I am very unlikely to do it. As a result, [AR.sup.*] approaches 1. Likewise, the TC, calling me a dirty name, is not very threatening and thus unlikely to compel me to do anything, especially something so grave. As a result, its subjective utility assignment is not significantly low and thus [TC.sup.*] also approaches 1. The product of the two, then, also approaches 1, yielding a low C-value. Since this minimal value would surely not surpass the value required to coerce, earlier called n, coercion does not occur here. On the other hand, as an example of the latter case, if one were to threaten to kill me unless I scraped my fingernails on the chalkboard, the values of the [AR.sup.*] and [TC.sup.*] are close to maximal. This is so since, the AR, scraping my fingernails, is something about which I am indifferent, thus, I am as likely to do it as not. Consequently, any threat could easily move me to do it, and so [AR.sup.*] approaches 100. But the TC in this case, killing me, is extremely unpalatable to me, and thus such that it compel me to do a great number of things. Consequently, [TC.sup.*] is also high. In this case, the C-value will be close to maximal, thus surely surpassing n and making every other alternative ineligible.

One might wonder why a different function would not serve our purposes better here. Instead of making the C-value the product of [AR.sup.*] and [TC.sup.*], it might be simpler to make it a function of the difference between AR and TC so that C-value = TC - AR. On this latter function, cases where significant coercive pressure is present will yield values as high as 99, and cases of minimal coercive pressure will yield values of -99. Such a function is not only simpler but, one might argue, philosophically preferable because it allows a more natural understanding of cases such as the following noted by Feinberg:

Suppose that the powerful villain A makes the following coercive proposal to B, a woman in his power: "Sleep with me or I'll break both your legs." Suppose further that B is utterly repelled by the demand . . . but even more terrified by the threat . . . since the cost of the threat is even greater than that of the demand, she very reluctantly succumbs to the coercive pressure. Now consider a second example involving the same persons. A says to B: "Get me a cup of coffee or I'll break both your legs." Fetching a cup of coffee is a mere minor inconvenience to B, but getting her legs broken is as terrifying as in the first example(9)

But contrary to Feinberg, our account seems to yield the result that the second example involves a situation of not only greater coercive pressure, but much greater coercive pressure than the first. And this result is exaggerated by the fact that our function requires that the two values we use in calculating the C-value, [AR.sup.*] and [TC.sup.*], be multiplied.

However, it seems that Feinberg is just mistaken in his analysis. While it may be true that the woman in the first case is in far worse circumstances than the woman in the second case, this does not mean that the woman in the former case experiences greater coercive pressure. In fact, we might see why this is so by considering that the woman in the former case might well be led to consider which alternative is worse, and to feel that opting for broken legs might really not be as bad as succumbing to the threat. Because this is so, it is imaginable that someone in these circumstances might truly be in a position where other alternatives look eligible to her. But surely this is not even an entertainable position for the woman in the latter case. As a result, we think that our account yields the right answer in such cases.

Such cases also indicate why our original function is philosophically preferable to the alternative function proposed. In the first case, coercive pressure is mitigated to the point that the woman could be in a position where alternatives still look eligible to her. But in the second case, the AR is so trivial in comparison with the extreme TC, that no reasonable person would refuse to comply with the threat. We believe that the multiplicative function best captures this difference between these two cases since coercive pressure really does seem to differ by orders of magnitude here.(10)

However, even this characterization of the first two conditions is not complete. To see why, consider a city in which the police have concluded that all of the muggings in the city are being committed by two criminals, A and B. Criminal A never has a weapon when he commits his crimes and, in fact, always flees the scene if the victim shows any resistance. Criminal B, on the other hand, is armed and invariably shoots victims who show any resistance. Finally, the police have determined that A is responsible for three-quarters of the muggings. Now consider the case of one who is confronted on the street and told to hand over their money or be shot. Does this person "believe that P is unable to have both [similar to]AR and [similar to]TC," as condition 1) requires? Is choosing {TC and [similar to]AR} ineligible in this situation, as condition 2) requires? These questions force one to recognize that coercion is not a one-dimensional affair, as the C-factor analysis might lead one to believe. Thus, ineligibility in these contexts is determined not only by TC relative to the AR, as reflected in the C-factor, but also by a second dimension which has been called elsewhere "imminence factors."(11)

As noted, the C-factor measures the strength of the threat relative to the gravity of the act required. Imminence factors measure certain qualities attending one's belief concerning the fact that the TC will follow, given that AR is not performed. There are at least three of these factors: probabilistic imminence, temporal imminence, and epistemic imminence. Probabilistic imminence is the degree to which the threatened believes that the threatener can or will in fact successfully execute the threat. Thus, if I am threatened by a mugger who I think is blind and would likely miss me if I ducked, or if I believe that the mugger has a carrot rather than a gun stuck in my back, the C-factor might be quite high, while the probabilistic imminence is quite low. As a result, I still might believe that I am able to have both [similar to]AR and [similar to]TC, against condition 1), and thus not be coerced.

Temporal imminence reflects the extent to which the execution of the threat is temporally removed from one's failure to act in accordance with the threat's conditions. The more temporally removed the execution of the threat, the lower the temporal imminence. Thus, consider two cases. In both, I am told that if I fail to hand over my money I will be poked with a cattle prod that will deliver a strong, painful electrical shock. However, in one case I will feel the shock immediately whereas in the other case I will not feel the shock for fifty years. It seems that while I might be coerced in the first case, I might not feel coerced in the second. The reasons for this are not quite clear, nor particularly relevant for our purposes. It will suffice to note here that temporal imminence does seem to play an important role in determining whether or not one is coerced. Thus, again, even though the C-factor might be high, low temporal imminence, like low probabilistic imminence, may suffice to undercut the coercive force of the high C-factor.

Finally, epistemic imminence is the degree to which the threat and its consequences are epistemically forceful to the threatened. The role of epistemic imminence can be shown indirectly through the example of advertising campaigns intended to dissuade people from certain harmful behaviors. So, for example, we believe that showing people images of accident scenes will reduce the incidence of drinking and driving. And in some measure such campaigns are successful. Why? Certainly not because the provide the viewer with information he or she lacked. The answer is that such images make the dangers of drinking and driving more epistemically forceful. We "realize" in a more powerful way just how dangerous such activities are for us. Something similar can be said concerning cases of coercion. Threats presented to us in a less epistemically forceful way are simply not as coercive as ones presented in a more forceful way.

Thus, the factors which play a role in satisfying conditions 1) and 2) are somewhat complicated. Both threat strength, as represented in the C-factor, and threat imminence play a role in determining how one sizes up the situation. Thus, a lower C-factor might suffice to coerce if probabilistic, temporal, and epistemic imminence are high. To the extent that these factors are mitigated, the C-factor must be correspondingly higher if the threat is going to be coercive. In most cases, these factors will be high enough that coercion ultimately depends on the C-factor value alone, and thus they can usually be ignored. However, there are cases where they play a role, and we must be aware of just what they are and what that role is. For simplicity's sake, imminence factors will be ignored for the remainder of our analysis, with the understanding that the appropriate qualifications need to be made when decreasing imminence becomes relevant to the coerciveness of the threat.

From all this we take the first two conditions to be necessary, although not jointly sufficient, conditions of coercion. The reason they are not sufficient becomes evident in the following example. A philanthropist, whom we will call T, has in the past made generous donations to numerous charities. Yet, he has found such endeavors to be unfulfilling because he is never allowed to experience the joy of giving directly to one who is in need. As he embarks on his daily walk, he decides that instead of donating the $10,000 he is carrying to a faceless organization, he will give it to the first person he meets, regardless of who it is. Before he has traveled a block from his home, a man (call him Q) jumps out from some bushes. Although he is initially startled, T remembers his intention and gets ready to give Q the $10,000. Before he can, however, Q pulls out a gun and, in a scene reminiscent of our standard case, issues the threat, "your money or your life." T is surprised by Q's threat, but he proceeds to hand over the $10,000 willingly and walks away.

What should be said about this case? First, Q brought it about that T was unable to have both his money and his life. Second, Q's threat brought it about that choosing to keep the money was, under the circumstances, ineligible for T. However, the philanthropist was not coerced into handing over the money, because while no alternative was eligible, he was acting in accord with his original intent. Thus, the third condition is necessary in order to rule out threats that do not alter the threatened's ability to carry out his original intention.(12)

Condition four states that the threatened must not want the threatened consequence to be actualized. This condition is necessary in order to rule out cases in which the TC is something that is desired by the threatened. Surely I could not be coerced by someone who threatened to give me an ice-cream cone if I failed to give him all my money. While this claim seems self-evident, it is necessary in order to address the troubling status of offers. The question as to whether or not offers can be coercive has received much attention in the literature. Our approach to offers, however, will be virtually the same as that taken towards threats. As in the case of threats, whether or not offers put a person in a situation in which every other alternative to accepting the offer is ineligible depends on the C-value in that case. Thus, on our view, there may be some offers you really can't refuse. The fourth condition allows for these "God-father-cases" by stating only that the threatened not want the TC. Therefore, on our view, a TC can consist either of the threat of imposing a negative consequence or, as occurs in the case of offers, the "threat" of withholding a positive benefit. Offers, then, can be coercive in cases in which the five conditions are met.

The fifth condition requires that the threat be intentionally imposed on the threatened by another rational agent and that these intentions be recognized by the threatened. This condition is important not because the individual's freedom is impaired in some unique way when the threat originates in some rational agent, but simply because it is necessary to distinguish the use of the word "coercion" from other related terms. Frankfurt describes a pair of cases in each of which a driver comes to a fork in the road he is traveling. In the first case the traveler notes that if he chooses the easier path, let us say it is the northern one, he will surely set off an avalanche that will crush him. His situation, then, can be summarized as follows: 1) he is unable both to travel the northern path and not to be crushed; 2) choosing to take the northern path and be crushed is ineligible for the traveler; 3) the threat of the avalanche constitutes the travelers reason for not choosing that path; and 4) the traveler wants not to be crushed. Although this case meets all the criteria for coercion, it simply does not conform to standard usage of the term. The same can be said for a case in which the avalanche would actually be caused by an intentional agent, if the agent's intentions were unknown to the traveler. Such a situation could arise in a least two ways. First, a potential threatener may be trying to coerce you by threatening to crush you with an avalanche if you take the northern path. He stands on the cliff rocking a stone which is so large you cannot even see him or, it turns out, hear him shouting his threats. All you perceive is a large stone, looking as if it is about to come tumbling down the hill. As a result you choose the southern path. Here the intention is present in the threatener but you know nothing of it and as a result that you are not coerced by it. In the second case you do see the threatener rocking the stone and hear him shouting at you. However, he is shouting to you in Chinese, and rather than feeling threatened you surmise he is a Chinese construction worker trying to clear away roadway hazards. Consequently, while his action prevents you from going that route, and you know it is the intentional agent's action that is so preventing you, you are not coerced.

Yet while such cases do not count as "coercive," it is clear that the situation vis-a-vis the traveler's freedom would not be altered if, instead of a naturally imminent avalanche, the traveler saw a troll standing on top of the cliff threatening to initiate an avalanche if the traveler chose the northern path. Thus, while these scenarios are indifferent as far as the metaphysics is concerned, our need to introduce these distinctions for other purposes, legal and political, necessitates this last condition.(13)

In addition, the fifth condition requires that the threat be imposed by an agent distinct from the one being threatened. This serves to rule out cases where one imposes threats on oneself such as, "If I eat one more piece of chocolate, I will starve myself for a week." While this may be a legitimate (self-imposed) threat, it does not count as a legitimate instance of coercion.


Earlier, in the discussion of our second condition, we stated that although the victim of coercion is faced with a choice, all of the alternatives save one are ineligible for this individual, i.e., all the alternatives are such that, in some significant way, the victim could not choose to act otherwise than the threatener demands. In this section we will discuss the "significant way" in which the coerced individual "must" comply with the threat. In addition, the modality of the claim that alternative actions were impossible, and the way in which this is to be understood in terms of the agent's deliberation will also be discussed. What we propose is two-fold. First, we will argue that the person who is coerced could not have done otherwise in any meaningful sense. Second, we will argue that in spite of the former claim, even a libertarian can admit that the coerced person acts freely, as long as certain conditions are met.

Once again, consider our standard case. Following the libertarian convention, we can ask whether or not our threatened individual is "able to do otherwise" here by using the device of possible worlds. That is, given the history of the world, the laws of nature true in the world, along with all the metaphysically necessary truths, is the choice of the individual indeterminate? If we conceive of alternate possibilities for choice in this way our problem can be framed as follows. Keeping in mind the world segment containing our coerced individual and all the attending circumstances of the threat up until the time of choice, what possible futures are open to her? To answer this within the picture we have described, we must first answer the following question: what possible futures can be conceived that are coherent with this initial world segment (i.e., up until the time of the choice) in which she chooses another alternative, say, keeping her money? Is there anything, given her history, her character, the threat, and its attending circumstances, that would lead her, under these very circumstances to choose otherwise? The answer appears to be a clear no. Let us attempt to imagine such a possible world - one that shares an initial world segment with the actual world, and one in which she chooses to keep her money. Just how would this person construct a coherent explanation for what led her to such an action (as she did in the actual world by saying "There was nothing else I could do")? She would have to say something like, "I believed that choosing my life was the only option. I believed it before the choice, I believed it as I decided to keep the money, and I believe it now. I never saw any reason for choosing the money, yet that is what I did."(14) This scenario seems quite incoherent; in fact we would argue that it is simply impossible for someone, speaking truthfully, to respond in this way.

At this point a critic might question an apparent discrepancy in the above account. Earlier in this essay we argued that Frankfurt's account of coercion was defective on just the point we seem now to be endorsing, namely that on his view the coerced action is forced. What began as an attempt to differentiate between our view and that of Frankfurt may seem to be accomplishing just the opposite. Wouldn't Frankfurt agree with our assertion that the coerced individual cannot do otherwise? Maybe so, but the significant difference between the two views lies in the explanation for this lack of alternatives. For a compatibilist such as Frankfurt, this explanation consists in either an appeal to a panic-stricken frenzy or to the laws of nature that determine all actions, free or not. Given the situation, the coerced person is physically determined to perform the alternative she does; the laws of nature necessitate her actions in this case, much as they do in any situation. What differentiates coerced from non-coerced acts for Frankfurt is not so much the "ability to do otherwise" as libertarians render it, but the type of causal antecedents that give rise to the action.

On our account, however, the ineligibility of all but one alternative is explained not in terms of natural necessities but by (libertarian) freely cultivated dispositions, desires, rankings - that person's character. One point neglected by contemporary libertarians is that many, if not most, of the choices we make in life are dictated by the various habits and dispositions we have cultivated. For example, a person who has cultivated a virtuous character may feel as though she simply cannot harm an innocent; such an action is, for her, as ineligible as is choosing the money for the victim in our standard case of coercion. This ineligibility is a result of a certain character, cultivated by a history of free choices of one sort or another. By making repeated choices of certain sorts, we come to have habits of certain sorts, habits both of perceiving certain types of things as goods and habits of desiring certain sorts of things over others. And it is these freely cultivated habits or dispositions that lead us to rank various objects of choice or courses of action the way we do. Thus, the subjective utility an individual assigns to AR's and TC's depends on these freely cultivated dispositions. Furthermore, the choices responsible for cultivating these dispositions were morally significant, free, and such that the person in question could, in those circumstances, still do otherwise. Thus, for these earlier choices, coherent depictions of possible worlds in which she chooses to act morally could be described, as could worlds in which she chooses to act immorally. These choices led to the cultivation of a character that now governs her choices, making certain alternatives ineligible in certain situations.

Thus we are confronted with a set of actions for which there are no (accessible) alternatives. How should we characterize these actions? Some libertarians have argued that a person's action can be considered free only if it is such that that person can do otherwise, where this is understood as having more than one eligible alternative.(15) As a result, they would hold, actions of the sort we describe are not free.

In the end, it seems that how we classify such acts is an arbitrary matter. While characterizing actions in which only one alternative is eligible (due to freely cultivated dispositions) as unfree accords well with the traditional libertarian view, there are important considerations that lead us to argue that the opposite convention should be adopted. First, classifying such acts as free accords better with common usage. Since the majority of our actions are such that the alternatives are ineligible, and these actions are considered free in non-philosophical discussion, it seems reasonable to side with common usage here, since we would otherwise be forced to admit that many, if not most, everyday actions that we had hitherto called free are actually not so. Second, regarding such acts as free is simpler. If we were to hold such acts to be unfree, we would have to introduce additional theoretical apparatus that would distinguish between non-free acts for which one is morally responsible and those for which one is not morally responsible. The introduction of such apparatus would dissolve the strict isomorphism that will be preserved between freedom and moral responsibility on our account. Thus, in the interest of preserving the relationship between freedom and moral responsibility and in order to satisfy common usage, the traditional libertarian account would have to be amended as follows:

S freely performs action A only if either

i) S could do other than A, or

ii) S acts as a result of a disposition formed by actions that S freely performed.

But the most significant reason for taking coerced actions to be free is that there is an important sense in which coerced acts are of a piece with non-coerced free actions. All actions take place in contexts where we must analyze the risks attached to making one choice over another. And, in cases where there is no coercion, free individuals will choose that act which best accords with the goods they value and/or desire at that moment. In most cases, the risks attached to certain alternatives are not as stark as they are in cases of coercion, nor are they (usually) imposed by other agents in the same way that they are in cases of coercion. Nevertheless, choices made in contexts of coercion do not differ from choices made in ordinary circumstances in ways that justify classifying the former as unfree. Bernard Williams summarizes these sentiments as follows:

If we are not to count as exercising freewill in cases of [coercion], then we never exercise it, since all choices operate in a space of alternatives constrained by the contingent cost of various possibilities, and these exceptional [coercive] cases are simply dramatic cases of that, where the space has been unexpectedly restricted. . . . What is peculiar about [coerced acts] is that they witness to the agent's vital interests or deepest needs only negatively, as things to be protected: the actions required are the expression of someone else's intentions. . . . Such considerations can indeed help to explain why [coercion] is perceived as specially opposed to freedom, and more so than the nasty choices, or lack of choices, laid in our path by nature. But the very fact that decisions taken under constraint are decisions, and can take the form of practical necessity, a form that belongs to some of the most serious and responsible decisions we take, itself shows how constraint has nothing to do with the question of freewill.(16)

Thus, while there may be some reasons for considering the coerced act to be unfree, there appear to be considerations significant enough for regarding the act as free when it meets the conditions described above.(17)


On the account we have sketched so far, the coerced individual can be said 1) to be in a situation in which every alternative to the action demanded by the threatener is ineligible and 2) to be acting freely in performing the act required by the threat. Yet something remains troubling here. First, there is the counterintuitiveness of the claim that the coerced individual acts freely. This we have addressed above. Second, there is the fact that free acts are usually taken to be acts for which the agent bears moral responsibility. What is troubling about this claim is that in the case of a coerced action, coerced individuals are not held morally responsible, or at least not morally culpable, for the act they perform. To illustrate this, consider the following cases. In the standard case, if the victim was carrying her family's weekly grocery money, we would not deem her morally culpable for surrendering it to the coercer. But if this same individual were simply to hand the money over to a passerby, we would be inclined to hold that she had acted irresponsibly, and thus be worthy of some sort of moral condemnation. All of this may be held to constitute an objection to our view. Since, one might argue, we contend that coerced acts are free, we are stuck with endorsing the counter-intuitive position that these same acts are ones for which the coerced bears moral responsibility.

This sort of objection is especially likely to be voiced by certain coercion theorists, viz., those who endorse what are known as "moralized" accounts of coercion. According to these "moralized" accounts, one can be said to be coerced only if the action in question was "morally permissible" for the agent in those circumstances. Thus, if someone performs an act under a threat, where that threat does not suffice to excuse the agent's performing that act, that person, regardless of the circumstances, cannot be properly said to be coerced. As a result, on such accounts, being coerced entails that one is excused from moral responsibility, or at culpability, for coerced acts.

We think the right response is to admit that what the objector claims follows from our position does in fact follow from it, but that rather than constituting a reductio for our view, it instead shows that our prima facie inclination to regard the coerced individual as lacking moral responsibility is in error. The victim of coercion, as noted earlier, cannot fulfill her conjunctive desires. And, ipso facto, she is placed in the position of having to choose between the conjuncts. Inevitably, the victim chooses that which is, to her mind, the "lesser of two evils," or, alternatively, what she perceives to be best under those circumstances. But how one judges or "perceives" what counts as best or better in a given situation is, we have argued, a result of certain dispositions cultivated through prior free, morally significant choices. As a result, even though she cannot do otherwise, the choice is free.(18) Thus, the way one "judges" or "perceives" the situation is itself something that the agent is responsible for, and thus the same can be said for the actions that follow from such judgments or perceptions.

But while this is the case, it does not entail that our moral assessment of all actions committed "under duress" will be the same. In the most widely cited treatment of the relation between coercion and moral responsibility, namely Aristotle's in the third book of the Nicomachean Ethics, it is urged that the captain who jettisons the ship's cargo in order to prevent the ship from sinking deserves different treatment than one who kills his mother because his own life was threatened, and this seems right. The crucial distinction between these cases, however, is not so obvious. We take it to be this: in the former case, the captain had rightly ordered dispositions that led him to see his situation in a certain light and to do what was, under those circumstances, the right thing. In the latter case, however, the son had disordered dispositions that led him to assess his situation in a certain light and, in light of that assessment, to make the wrong choice. Dispositions can either be cultivated in such a way that they lead us towards acting rightly or not. If we cultivate morally appropriate dispositions then, even in cases where we are coerced, we will choose the morally appropriate alternative, i.e., the true "lesser of two evils." If we have not cultivated such morally appropriate dispositions, however, the choices that arise from our character will be morally defective.

In our standard case, then, the common reaction is to hold that the victim was not morally responsible for giving her money away. But this is not quite right. We ought to say that the person was morally responsible and that, under the circumstances, she did the right thing. Given her dispositions, she assigned morally appropriate rankings to the AR and TC as they were presented to her. As a result, such an individual is coerced but morally praiseworthy for having dispositions which led to her doing the right thing. On the other hand, consider a case in which an avid car collector, Fred, becomes deeply attached to his rare and expensive 1948 Tucker. On one of his journeys, Fred is stopped on a back road by a masked gunman who forces him from the car and presents him with the following threat: "Go into that farmhouse and kill my ailing mother or I'll bash this car to smithereens." Given his strong attraction to the car and his lack of concern for the old lady, Fred takes the gun, walks inside, and finishes her off. On the analysis we have given it could well be the case that Fred was legitimately coerced in this situation. The reason we do not find Fred free from blame, however, is that we think that he ought not to have been coerced under these circumstances. The cultivation of dispositions that led him to rank the AR and TC as he did in this case, that is, to see murder as a preferable alternative to having his car damaged, indicates a defective, morally culpable, character on his part. He is both coerced and morally blameworthy.

Thus, while we cannot fully defend the claim here, we think that this account has certain advantages over its "moralized" competitors. The reason for this is that moralized accounts seem to neglect that fact that one who is presented with a threat can be in the same metaphysical boat with respect to their eligible alternatives whether they are entitled to be in that boat or not. Since coercion is, we take it, a description of one's metaphysical status with respect to the available alternatives in light of circumstances, what one is entitled to do seems to become important only when we are led to consider one's culpability for the action performed in those circumstances. But this sort of assessment appears to be a wholly separate matter. Still, there is some room for the moralizer to maneuver here. The moralizer might say, and we think should say, that our reading of coercion is wrong-headed from the start. What the moralizers have in mind when they think of coercion is not simply a description of one's metaphysical circumstances, as we would have it, but a description of the relationship between one's metaphysical state and one's legal (or, sometimes, moral) obligations. But if this is what the moralizers have in mind, then there really is no debate, because the advocates of moralized and non-moralized accounts are using the term equivocally. The concept they have in mind is "moralized" from the start since it is a term intended to help us to solve matters of legal responsibility and culpability in legal contexts. Thus, coercion in the moralized sense is narrower than it is in the sense described in non-moralized accounts, the former being a proper subset of the latter.(19)


In this paper we have provided a set of conditions for coercion that resolves a number of the difficulties of earlier views. While our account of coercion is indifferent between various accounts of freedom, libertarian and compatibilist, we have shown that it leads to certain philosophically illuminating reflections when considered in libertarian terms. In particular, it leads to a certain accommodation on the libertarian's part by leading him to admit that there are cases where we are free despite the fact that we have no eligible alternatives and thus are not able to do otherwise. In light of this, we have argued that it is preferable for the libertarian to regard coerced acts as free. While this view is admittedly unconventional, it creates no special problem since coerced acts do appear to be similar in nature to other (free) acts where we are constrained in our choosing, not by threateners, but by natural circumstances. Furthermore, this account of coercion and freedom is coherent with the view that coercion is at least sometimes an excusing condition. On our view it can be an excusing condition, but only in those cases in which the coerced individual does that which, under those circumstances, was the morally appropriate or at least morally defensible act. We thus have an account of coercion that adequately characterizes the phenomenon and resolves the puzzling status of the freedom of coerced acts.(20)

Franklin and Marshall College


1. Harry Frankfurt, "Coercion and Moral Responsibility," in T. Honderich (ed.), Essays on Freedom of Action (London: Routledge and Kegan Paul, 1973), p. 77.

2. Ibid., p. 65.

3. J. P. Day, "Threats, Offers, Law, Opinion, and Liberty," American Philosophical Quarterly, vol. 14 (1977), p. 262; H. J. McCloskey, "Coercion: Its Nature and Significance," Southern Journal of Philosophy, vol. 18 (1980), pp. 335-52.

4. Bernard Gert, "Coercion and Freedom," in J. Roland Pennock (ed.), Coercion (Chicago: Atherton Inc., 1972), p. 32.

5. Ibid., p. 34.

6. Ibid., p. 37.

7. There is, in fact, a class of accounts of coercion that define coercion in just this way. On these accounts, known as "moralized accounts," coercion can be said to occur only when one accedes to a threat that one is morally permitted to accede to. For more on such views and our reaction to them see the section entitled "Moral Responsibility" below.

8. Day, Op. cit.

9. Joel Feinberg, Harm to Others (Oxford: Oxford University Press, 1984), p. 204. We thank John Martin Fischer for providing us with this example.

10. Of course, this account may represent an attempt to introduce more precision into this matter than the subject allows. For that reason it may be silly to argue whether or not the function ([AR.sup.*] x [TC.sup.*])=n is the only right way to characterize the relationship between threats and acts required. What is important is that threats and acts required have some such relation and that the strength of a threat increases as or decreases as a function of these two factors, and in a way which is much the way our formulation characterizes it.

11. See Michael J. Murray, "Coercion and the Hiddenness of God," American Philosophical Quarterly, vol. 30 (1993), pp. 27-38.

12. One might think that an easy way to avoid "mugged-philanthropist" examples is to make condition 3) a counterfactual such as 3[prime]) If P were unthreatened, P would not have chosen AR. But notice that counterfactual characterizations such as these fall prey to numerous troubling counter-examples. Consider Johnny, a young boy anxious to prove his machismo to his friends, who decides that he will follow them in jumping from the bed of a speeding pickup truck. Further, suppose that Johnny, if he were confronted with a relatively weak threat, being punched in the stomach, would not jump from the truck, since by jumping he would appear to be acting out of fear of this weak threat. Further, while Johnny is anxious to prove his machismo, he is not a complete fool, and were he to be threatened with the prospect of a stronger threat - say, a vial of acid being thrown in his face unless he jumped - would jump from the truck. In other words, in the case where there is no threat, he jumps; in the case where there is a weak threat, he does not jump; and in the case of strong threats, he jumps. Now in this last case, the case of the strong threat, it seems that were Johnny to be so threatened, he would be coerced into jumping. But while he meets the other conditions, 3[prime]) fails to yield the right answer here because Johnny would still have jumped if unthreatened (although for a different reason).

13. Robert Nozick, for one, denies this claim by arguing that the two cases differ in one important respect (see, for example, his Philosophical Explanations, Cambridge: Harvard University Press, 1981, p. 49). In the case where the troll is threatening to crush me, Nozick would say that my refraining from taking that path is an act that is more "closely connected" to the troll's will than to mine. Things are different in the standard avalanche case, however, since in that case my will is the only one involved and thus it is my own will that is most closely connected with the act. On this reading, what distinguishes cases of coercion from ordinary free actions is that when I am coerced I am compelled to do an act which springs from another's will. We will argue in the next section that this view is mistaken and that, in fact, actions done in the context of coercion evidence reflect one's value commitments in an important way.

14. This expression is essentially reproduced from a paper by Peter van Inwagen, "When Is The Will Free?" in James E. Tomberlin (ed.), Philosophical Perspectives; 3: Philosophy of Mind and Action Theory (Atasadero, CA: Ridgeview Publishing Company, 1989), pp. 407-8.

15. See the article cited in note 14. Van Inwagen uses the phrase "indefensible" to mean the same as our "ineligible." In this essay he argues that, when choices are made in a context where only one alternative is "defensible," the choice is not free.

16. Bernard Williams, "How Free Does the Will Need to Be?" The Lindley Lecture, University of Kansas, March 27, 1985, Department of Philosophy, University of Kansas, pp. 2-3.

17. It is quite possible that some will not be swayed by our efforts to justify calling these actions free. But we find another compelling reason for regarding such acts as free, namely, that, with slight modification, it provides for a solution to the puzzle found in orthodox Christian theism between the claims that God is free and also that it is not possible for Him to perform any evil action.

This same tradition holds that God's actions are, in a sense, necessitated by His character. God can do no evil - there is no possible world in which God chooses to perform an evil action. Consequently, on a traditional libertarian account, many of God's actions could not be considered free since He must choose to do the morally appropriate thing. Yet it looks as if on our account things will fare no better. Even if we argue that God's character or His "habitual dispositions" are such that evil courses of action are never "eligible" for Him, He is still not free because condition ii) is not satisfied. But notice that the role condition ii) plays is to insure that the dispositions from which we act are ones we are responsible for, ones that respect the autonomy of the agent. If these habitual dispositions are cultivated by us rather than simply in us, then the actions that follow from them can legitimately be said to be our actions. But if this is so, then the orthodox Christian picture has the resources for resolving the difficulty, since God and free creatures are simply not on univocal ontological footing here. What is unique about God's character is that it, along with the rest of the divine nature, exists ase; that is, from its own resources. God's character yields free actions in virtue of the fact that He creates His own character, and consequently only He is responsible for it. Creatures, however, are not so fortunate. As finite creatures we do not exist ase but ab alio, or from another. Thus, if we are going to be responsible for our own characters, we must be created with some mechanism that permits their autonomous cultivation, and libertarian freedom seems to suit just such a purpose. As a result, we need to revise our definition of freedom as follows:

S freely performs action A only if either

i) S could do other than A, or

ii) S could not do other than A but the doing of A follows from the agent's character which

a) was created by that agent a se, or

b) was formed by actions that S freely performed.

18. One potential difficulty here is sorting out the relationship between higher and lower order desires and their relation to freedom. For example, one could argue that the drug addict is in a situation such that he has cultivated a certain set of habitual dispositions in virtue of past free choices and that as a result the drug addict takes drugs freely. But consider the case of the drug addict who wants for all the world to rid himself of his destructive habit. Each time he rolls up his sleeve he agonizes over the state of self-induced slavery in which he has placed himself. Many are inclined to say that, although this person acts from freely cultivated habits, he is not free. While these are substantive questions which have an important bearing on the account we offer here, we will take refuge in the fact that one can't do it all in one paper.

19. It may be this sort of confusion that leads advocates of moralized accounts, such as the one offered by Alan Wertheimer (Coercion, Princeton: Princeton University Press, 1987), to argue that moralized accounts are superior since, in their legal application, non-moralized accounts must ultimately make appeals to the sort of moralizing considerations built into the moralized accounts: "nonmoralized theories also have their price. For nonmoral theories of coercion seem to either collapse into moralized theories or to be unable to distinguish the coercion that bars the ascription of responsibility from the coercion that does not" (p. 288).

20. An earlier version of this paper was presented at the Central Division meeting of the American Philosophical Association, April 1992. Special thanks are due our commentator, John Martin Fischer. We would further like to thank Glenn Ross, Leon Galis, and an anonymous referee for this journal for numerous helpful comments on earlier drafts. Research for this paper was funded in part by the Hackman Scholarship Program through Franklin and Marshall College.
COPYRIGHT 1995 North American Philosophical Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1995 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Murray, Michael J.; Dudrick, David F.
Publication:American Philosophical Quarterly
Date:Apr 1, 1995
Previous Article:Rejection letters to famous philosophers.
Next Article:Indeterminism and control.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters