Abstract

The hierarchical analysis of autonomy, and in particular the version developed by Harry Frankfurt, has been considered inadequate due to its inability to account for the apparently heteronomous nature of action taken under threat, or more precisely, the autonomy-undermining effects of what Frankfurt calls “duress.” Both reforms to hierarchical analysis and competing non-hierarchical paradigms have attempted to repair this oversight. I assess three competing theories of action taken under duress according to their plausibility when applied to this problem: James Stacey Taylor’s situation-averse analysis, Pascal Brixel’s intrinsic-extrinsic desire analysis, and Thomas Nys’ volitional necessity analysis. The former two fail, I argue, because they permit second-order volitions to remain in causal control of the will, effectively retaining autonomy of action under duress, though they insist such autonomy is absent. Volitional necessity analysis, by contrast, locates the potency of duress in the most reasonable location, that being the threatened individual’s own deepest cares. Though it possesses the greatest explanatory power, this theory is bound to the position that duress does *not* infringe upon autonomy, but rather freedom, where freedom is understood as the range of possible actions open to the individual in a given situation. Since volitional necessity analysis is nonetheless the most plausible interpretation of duress, this feature suggests a potentially oppositional relationship between autonomy and freedom—one which, I conclude, is especially consequential to our understanding of social and political arrangements, and ought to be examined in future literature.

Frankfurt's Hierarchical Analysis of Autonomy

     To begin, one should have a firm grasp of Frankfurt’s theory of free will and free action. As previously mentioned, Frankfurt’s analysis is referred to as “hierarchical,” and this is due to its positing of so-called “hierarchies of desires” as central to the concept of free will (6). Frankfurt conceives of persons as possessing two relevant “levels” of desire: first-order desires, which are “desires to do or not to do one thing or another, (7)" and second-order desires, which are desires to possess certain first-order desires or for certain first-order desires to become “effective (8)," meaning for them to move the person to action (9). He refers to the latter sort of second-order desire as a “second-order volition (10)," identifying, furthermore, effective first-order desires (which we might also call first-order volitions, following his terminology) with the “will (11)." *Freedom* of the will, on this view, is constituted by a particular relationship between one’s will and one’s second-order volitions: “It is in securing the conformity of his will to his second-order volitions, then, that a person exercises freedom of the will (12).” In other words, free will is the causal determination of one’s will by one’s second-order volitions. Importantly, however, Frankfurt is careful to distinguish freedom of the will from freedom of *action*, which merely consists in the individual possessing the ability to act on their first-order volitions (13). Such freedom would be restricted if, for instance, one was physically restrained by being placed in a prison cell, thereby preventing them from acting on their desire to leave (or any other desire contingent on being outside of the cell).

      In Frankfurt’s own work and related literature, the possession of a “decisive” second-order volition has been described as “identifying with” or being “autonomous with regard to” the relevant first-order volition (14). Identifying with or being autonomous with regard to one’s actions would be, by extension, freedom of action as described by Frankfurt—causal determination of actions by one’s first-order volitions (15). These, then, are the concepts of autonomy we begin with when analyzing the metaphysics of action under threat.

     According to the hierarchical framework, actions taken under threat—i.e., actions taken *only* to avoid a threatened penalty—emerge as rather peculiar cases. As Brixel notes, where threats do not involve the direct application of physical force, they must use “volitional coercion,” meaning they fundamentally *require* the use of a victim’s own first-order volitions to be effective (16). It is not the direct force of the threatener, it seems, that ultimately pushes their victim to act as they do. Rather, the victim’s motivational source appears *internal*—it must be that they form a first-order desire to act as the threatener wishes, which they may or may not then endorse with a second-order volition (17). If autonomy is a matter of causal alignment between second-order volitions, the will, and action, and such alignment seems present (even necessary) in at least some instances of volitional coercion, then how can hierarchical analyses hold that threatened victims experience diminished autonomy, as we might intuitively believe they do?  

      This is precisely the objection posed by Irving Thalberg, who observes that Frankfurt’s own solution is to posit that threatened victims experience disharmony between their second-order volitions and their first-order desire to comply, indicating a loss of causal control by the former over the latter. Thalberg argues that this model is erroneous because it counterintuitively assumes that threatened victims form a second-order volition *not* to comply with the threat (hence causing the disharmony of the will), yet it seems empirically true that most such people do precisely the opposite; they later reaffirm their choice to comply (18). If there is in reality no disharmony of the will in these cases, then hierarchical analysts would be forced to say that a threatened victim acts fully autonomously, which seems absurd. Furthermore, according to Thalberg, hierarchical analyses of autonomy have a second problem: whereas Frankfurt posits that the threatened victim’s main object of disdain must be their own first-order desire to comply (having such disdain being equivalent to forming a second-order volition *not* to comply), it seems more likely that victims instead take issue with the *consequences* of complying, such as the loss of money to a robber (19). Thalberg asserts that these objections together warrant rejecting hierarchical analyses altogether (20). According to Taylor, however, both of them rely on a misunderstanding of Frankfurt’s position, and by clarifying where these misunderstandings lie, one may arrive at a more plausible (yet still flawed) account of autonomy under threat (21). 

     Recall Frankfurt’s distinction between freedom of action and freedom of *will*. To Taylor, Thalberg is mistaken to assume that Frankfurt takes autonomy to be “transitive” across action and the will, or in other words, that possession of autonomy with regard to one’s will is the sole guarantor of autonomy with regard to one’s actions (22). If this principle is true, then the only way one could be deprived of autonomy of action would be through a loss of autonomy of will. If the principle does not hold, however, then Frankfurt could conceivably describe a threatened victim as having lost autonomy of action while retaining autonomy of will. This would answer to Thalberg’s first objection, since hierarchical analysis would no longer rely on incongruity between one’s first and second-order volitions to account for the victim’s lost autonomy. And indeed, in Frankfurt’s wider corpus, he does describe certain instances of action under threat this way, calling these situations ones of “duress (23)." In cases of what Frankfurt calls “coercion,” by contrast, the victim genuinely does repudiate their first-order volition to comply at the second-order level, but due to the former’s irresistibility, the latter simply has no say in the matter (24). Under duress, however, the victim is *not* moved to act by an irresistible desire, but rather one which they voluntarily endorse at the second-order level (25). Despite this, Frankfurt still maintains that victims of duress are motivated by desires they would rather not possess, which appears to contradict their supposed harmony of will (26).

     Taylor resolves this apparent paradox by observing that to “prefer not” to be motivated by a particular desire does *not* necessarily imply the presence of an oppositional second-order volition (27). Rather, such a preference might be of a “relative” nature—it might express a preference for a *different* desire which, due to the conditions of duress, cannot be fulfilled (28). According to this view, a victim of a robbery (if they are under duress, rather than coerced) would suffer from a lack of autonomy with respect to their action (*not* their will, as their first and second-order volitions are in agreement) because they experience an aversion to the *circumstances* of duress, since such circumstances force them to act on desires they would prefer not to, in the aforementioned relative sense (29). Such an account of duress provides a response to Thalberg’s second objection, as it is no longer a disdain for one’s own will that defines a loss of autonomy, but instead disdain for the situation. Due to this feature, Taylor refers to *this* interpretation of Frankfurt’s theory as a “Situation-Averse” analysis (30). 

Autonomy as Intrinsically Motivated Action

     Brixel distinguishes between two categories of desire: intrinsic and extrinsic. Intrinsic desires are those aimed at activities the agent desires to do because of perceived value in the activity itself, i.e., because the activity has intrinsic value. Extrinsic desires, by contrast, are aimed at activities which, while not valued by the agent intrinsically, are expected to yield intrinsically valuable *outcomes* (40). Of extrinsic desires, some are partially intrinsic (e.g., the desire to pursue a fulfilling career) while others are entirely extrinsic, or “merely instrumental,” (e.g., the desire to clean purely for the objective of creating a cleanly environment) (41). Utilizing these concepts, Brixel argues that one necessary condition of autonomy is that the agent acts on partially or fully intrinsic desires, rather than merely extrinsic ones (42).This he justifies on the basis that intrinsic desires appear to constitute an agent’s “practical identity,” or who the agent believes themselves to be, as expressed by their actions (this being the same kind of thing Frankfurt and Taylor theorize as second-order volitions) (43). Thus, an individual who acts on merely extrinsic desires never truly identifies with their actions, and therefore does not act autonomously (44).

     Applied to situations of duress, this criterion appears to explain the victim’s deprivation of autonomy in the following way. Since the victim acts *only* to avoid the threatened penalty when they acquiesce to a threat, we can say that the desire to acquiesce is a purely extrinsic one. There is nothing in the action itself that the victim finds intrinsically valuable, as evidenced by their requiring a threat to motivate the action (assuming the threat is a necessary one—the victim could have already been intent on acting as the threatener desired prior to the threat, in which case the desire would remain intrinsic, and so autonomy would be retained (45). The reason an individual acting under duress experiences diminished autonomy, then, is because the circumstances of duress are such that *if* they desire to act autonomously in the future (i.e., to act on intrinsic desires in the future), then they *must* act non-autonomously now. This is where the potency of threats—or what Brixel calls their “necessity”—arises from on this model: threats invoke “mere instrumental necessity (46).”

     There are two advantages to this theory over Taylor’s. To begin with, it seemingly requires no appeal to second-order volitions, since the necessity involved with duress is already present in the victim’s first-order desires by way of the logical structure of means-ends reasoning. If someone possesses the desire to continue living, for instance, then *given this desire* it would be logically necessary to acquiesce to a robber issuing a threat to their life. Because we need not assume that the victim endorses the first-order desire to hand over their possessions at the second-order level, we also need not contend with the objection that they *could* conceivably choose to identify with some other first-order desire. Such identification, according to Brixel, is immanent in intrinsic first-order desires simply by virtue of their intrinsic nature.

     Additionally, Brixel offers a more plausible explanation for perceived differences in the degree of autonomy lost under differing circumstances of duress. To illustrate this function of the theory, he uses a comparison: imagine a call center worker who labors purely for the end of survival, and a wealthy banker who invests in order to fulfill her desire to “fund her dream of going on a commercial space flight or indulge some other ‘expensive taste (47).'" In both cases, the activity undertaken by the individual is purely extrinsic—it is done for the sake of some further intrinsically valuable activity. However, relative to the general *ultimate* good both individuals aim at (“having a good life”), the satisfaction of the call center worker’s extrinsic desire is clearly more integral than that of the wealthy banker’s, since survival is a requisite condition for any kind of life at all, as well as a wide range of other intrinsic goods (48). Therefore, while both the banker and the call center worker might suffer from impaired autonomy, the latter suffers to a greater degree.

     Analyzing duress in terms of the intrinsic-extrinsic desire distinction, however, does not avoid problems associated with first-order desire analyses of duress in general. As Brixel himself acknowledges, the extrinsic nature of certain desires can only explain the “external dimension of autonomy,” which he asserts is the domain affected by threats: “In this dimension, the question is not whether an agent has the right desires or whether they were formed in the right way, but whether her activity is an unqualified expression of her intrinsic desires (49).” It appears, however, that instrumental necessity depends upon some deeper level of necessity, though explicating the former has brought us closer to the latter. As in Taylor’s solution, Brixel requires that we assume a victim of duress already possesses certain psychological states (these being certain first-order desires, rather than second-order volitions) which ensure they will act in a certain way. But if threatened individuals *only* suffer from unfreedom of action (as Brixel seems to believe), then what prevents them from utilizing their freedom of will to modify their desires in such a way as to not be affected by the threat? There must be, in other words, something more fundamental about the individual’s psychology with which duress interacts.

Conclusion and Implications

     Duress, then, can be defined in the following way: the intentional use of one individual’s volitional necessities by another for the purpose of artificially restricting the number of possible actions the individual under duress may take in response to a given situation, where the degree of restriction is dependent on the amount of harm threatened towards their object(s) of care. Such constraint on action is similar to that involved with physical restraint (e.g., when in chains), but importantly different in that it involves no disharmony between second-order volitions and action. This difference is not one that Taylor’s theory can account for, since, according to Taylor, the potency of duress is located in the *content* of the relevant first-order desire, rather than a care-informed interpretation of second-order volitions. Although Brixel’s emphasis on mere instrumental necessity brings us closer to volitional necessity, it still falls short of explaining precisely why the action of a person under duress is *necessitated* in some way, since it, too, does not incorporate the most plausible interpretation of second-order volitions. Nys’s model provides precisely this.

     An individual under duress may be said to be “less free” insofar as they literally cannot take actions they otherwise would have been able to take (due to volitional necessities), but this lesser freedom is self-imposed, and therefore, their autonomy remains intact. It seems, then, that the discussion of duress has revealed a peculiar relationship between autonomy and freedom. On the one hand, duress wields the most distilled form of autonomy, as it forces individuals to confront and assert that which fundamentally motivates them. On the other, duress is clearly a restriction of “freedom” understood as the amount of possible actions one may take in a given situation. These two concepts—autonomy and freedom—are thus not always in alignment with one another, yet both seem morally relevant. Any activity necessarily involving the use of duress—including social or structural activities, such as those of an economic or political nature—must thus engage in a delicate balancing act between these two goods. How exactly this process might look, however, is a consideration for another time.