Political Theory Is Not a Game

Human Agency and Interrogational Torture

The application of a formal game theoretic model to interrogational torture is dangerous in Howes’s sense only if game theory is even applicable to the question of interrogational torture and its effectiveness, something Howes rejects in the course of the first section of his reply and later in his discussion of the will. ¹ Following Scarry (1985) and others, he asserts that the pain and domination in torture are so agency destroying, reason destroying, information destroying, strategy destroying, even world destroying that they preclude any reasonable conception of choice, preference, calculation, or strategy. As a result, the torture victim does not even possess the “ability to know what game is being played” (Howes 2012, 21) and so “rational, strategic interaction is, by definition, made impossible” (Howes 2012, 24).

This model of “utter domination” (Howes 2012, 24) by the torturer is almost precisely the caricature of interrogational torture presumed by its strongest proponents. ² Torturers know how to inflict pain so that all room for agency, calculation, strategizing, and choice is removed. The torturer controls the environment so completely, breaks the victim down so thoroughly, that he or she does exactly what the torturer desires. The will of the torturer becomes the will of the tortured. This is exactly what the designers of the Bush interrogational torture program believed: they would “break” detainees who thought they were strong and could hold out into weak detainees who would talk. Such a model, whether in the guise of the television program 24, in the mind of Dick Cheney or SERE psychologists, or in a work of political philosophy such as Howes’s here, is a fantasy. As a result, both perspectives “obscure the fundamental character” (i.e., what actually happens empirically) of interrogational torture (Howes 2012, 20). ³

In addition to beatings, burning, and waterboarding, Henri Alleg was electrocuted during his interrogation by French paratroopers in Algeria in 1957. Variously attaching the clips to his finger, chest, ear, groin, and the palette of his mouth, they caused in the last instance his jaw to lock for a time and his eyes to feel as if they were being pushed out of his head from the inside (Alleg [1958] 2006, 56-57). And yet, he recalls, “stupefied as I was by the blows and the tortures I had undergone, one single idea was still clear in my mind: ‘Tell them nothing, don’t help them in any way.’ I didn’t open my mouth” (Alleg [1958] 2006, 63).

Now although Howes does not say what is left to a human being after the comprehensive destructive effects of torture, nor can it be inferred from his alternative model of human action, since he offers none, he characterizes resistance to interrogational torture as “the sheer force of will” (Howes 2012, 22). I return to this point in a moment. I first want to point out that it is clear from torture victims’ own accounts that they value things—avoiding pain, protecting information, convincing the interrogator that they are innocent—and order them in particular ways. The fact that “torturers may want to elicit some words or behaviors from their victims” (Howes 2012, 28), as Howes is grudgingly willing to acknowledge, indicates they have goals too. Victims by their own accounts think of themselves as having at least two actions, opening or not opening the mouth, to paraphrase Henri Alleg. The same is true of interrogators in their selection of whether or not to torture. Finally, both sides understand, even if imperfectly and under uncertainty, how those actions lead to outcomes.

All of this is starkly clear from testimonials from torture victims across the world and through time. Consider again Henri Alleg. After electrocuting him with a magneto and failing to elicit any information, his torturers threatened to bring his wife from France and torture her. Alleg reports that “in this nightmare, it was only with the greatest difficulty that I was able to separate the menace I had to fear from the blackmailer’s bluff” (Alleg [1958] 2006, 58). He later rejected an opportunity to commit suicide (though he suspects his captors may have deliberately created that opportunity) because the idea that his captors would think he did so out of fear of torture “revolted” him (Alleg [1958] 2006, 79). It is precisely then, “in ordinary language, the phrase ‘prefers to suffer torture’ rings” true, and not “false” (Howes 2012, 22). To deny that these actions were the result of a conscious choice and agency by Alleg but rather of a free floating will beyond his control is far more “detached” and “inhuman” (Howes 2012, 23) than representing them as choices based on an ordered set of goals, values, or preferences.

For Howes, the locus of resistance to torture is the “the specifically human quality of the free will,” not what people esteem or hold dear or consider important enough to guide their actions. In an extended discussion of Arendt’s conception of the will, Howes argues that it is a “unifying force” that “inheres in the internal experience of accepting or rejecting what is” and is capable of directing the senses, desires, acceptance, or rejection of pain and objects presented to it by the intellect and reason. It imagines ends for itself and “creates objects or activities to prefer” though it “cannot be reduced to preference orderings” (Howes 2012, 22).

But the model does not reduce the will to preference orderings. Reduction entails a complete explanation of the complex whole of one thing in terms of something (or several somethings) simpler or more elemental constituting it. By providing such a complete explanation, the entity at the higher level becomes epiphenomenal: it is completely explained by (the interaction of) its components. But this is not what a model—formal or otherwise—does. The model makes no claim that the preference ordering –k > –i is an elemental component of the will; it does not reduce the will to –k > –i.

What the model does do is represent (a “strong”) will with the ordering –k > –i. Models, like other representations of things in the world, simplify, but do not reduce, reality to re-present it in a way that is more intelligible, meaningful, or revelatory for particular purposes, in my case to assess the effectiveness of interrogational torture. The will is relevant for an examination of interrogational torture insofar as detainees possess it or are moved by it to different degrees, because it affects their behavior. The model recognizes both the will and how it varies over people using the ordering of –i and –k without reducing the will to that ordering.

Promoting Bureaucratic Violence

What makes the application in my article dangerous, according to Howes, is that it “invites policy makers and social scientists to use formal models to improve or perfect torture” (Howes 2012, 20) and so is “conducive to bureaucratic violence” (Howes 2012, 23). Rather than issue an invitation, however, my article responds to one: the implicit model of interrogational torture defended by the Bush administration and others. It is a continuance—not a misuse—of the liberal tradition to deploy reason to challenge the government’s model of interrogational torture, to expose its assumptions and force its defenders into the open. Could a government mathematician design a game in which interrogational torture works like they want it to? Undoubtedly, but the assumptions would then be out in the open, subject to challenge and perhaps refutation.

Moreover, what is the alternative approach to critically assessing the claims of interrogational torture proponents? I see three possibilities: a nonformal critique of the effectiveness claim based on some notion of rationality, a critique of the effectiveness claim that makes no reference to rationality, and finally offering no critique at all. A nonmathematical, nonformal, but still rational analysis of interrogational torture’s effectiveness would look at the interests of each side, assume goal-seeking behavior, and examine what happens. Indeed, we see this in the passages in Augustine and Beccaria referenced by Howes. Augustine, for example, says “now if this accused has followed the wisdom of our philosopher friends in choosing to escape from this life rather than endure those tortures any longer, he confesses to a crime he has not committed” (Augustine [1467] 1984, 860). Both argue from an analysis of motivations, the information available to the relevant parties, and their rational behavior, rather than from empirical evidence. Indeed, Beccaria goes so far as to say, “[I]t would be superfluous to confirm these reflections by examples of innocent persons who, from the agony of torture, have confessed themselves guilty: innumerable instances may be found in all nations” (Beccaria [1764] 2011, chap. 16, para. 9). For Beccaria, pure “reflection” is sufficient, though he thinks it would be confirmed empirically. In short, this approach is an acceptable method of critique: “[T]here is nothing wrong with these arguments in and of themselves” (Howes 2012, 23). “The problem” for my argument, according to Howes, “is that, unlike the works of Augustine and Beccaria where strategic dynamics take a back seat to nuanced and rich accounts of torture, [my] formal model is conducive to bureaucratic violence” (Howes 2012, 23). First, strategic dynamics do not at all take a back seat; they are just implicit rather than explicit in Augustine’s and Beccaria’s descriptions. Second, bracketing the question of just how nuanced and rich their discussions are, how is it that natural language, however rich and nuanced, somehow avoids being conducive to bureaucratic violence, while more compact, formal language somehow is? ⁴ Game theory simply formalizes the same type of reasoning in Augustine and Beccaria but is more systematic and covers more ground. The problem for Howes is thus the step from nonformal to formal rational choice theory, not with the rational approach per se. Indeed, he admits as much himself, when he claims in the same context that I “simply formalize what others have already said”. What exactly it is about a formal rational choice argument that leads to the bureaucratic violence that Augustine and Beccaria manage to avoid in their nonformal—but still rational choice—arguments is left unargued. ⁵

Another alternative might be a nonrational choice critique of effectiveness that simply looks at outcomes to determine whether or not interrogational torture provides valuable information. Indeed, drawing on Rejali (2007) and Rumney (2006), this is Howe’s very first criticism; there is plenty of empirical evidence that interrogational torture is unreliable, rendering my article an unnecessary—though “dangerous”—exercise (Howes 2012, 20). As I argue in the article, however, this alternative is not as straightforward as Howes would make it seem. The empirical work in Rejali and Rumney (especially the former) does indeed make a strong empirical case that interrogational torture is ineffective. A strong case is not, however, coextensive with a decisive case.

First, as they both admit in the passages cited by Howes, interrogational torture sometimes works. Second, adding “too rich” to “too fragmentary” supports rather than undermines the claim that the available evidence falls short of conclusive or decisive, at least for many (and not just the Cheneys, Thiessens, and Yoos of the world) (Howes 2012, 20). The complexity of cases adds to the difficulty of assessing effectiveness. Third, this is demonstrated by the many who argue that it does indeed work, including those who oppose it on other grounds, whom I discuss in the appendix but whom Howes ignores. On balance, I happen to think the empirical evidence, such as it is, probably favors the “doesn’t work” camp, but the one-sided references to supporting evidence and the complete dismissal without argument of countervailing evidence—what one anonymous reviewer aptly labeled “the current fog of assertion and counterassertion”—hardly counts as a successful refutation of the effectiveness claim. Nor does any of this gainsay an attempt to make the argument against interrogational torture stronger by providing the underlying logic that accounts for the varied outcomes we see empirically.

The last alternative is to mount no critique of the claims to effectiveness at all. We might do this from a Kantian or a virtue ethics perspective, arguing that interrogational torture’s effectiveness is beside the point. Perhaps, but if so, we must then take responsibility for the decision not to try and persuade the nearly two in three Americans who support interrogational torture only if it works (44% of Americans overall; PublicMind 2011). To the degree that this large group is pivotal for the future of interrogation policy in America, the potential for promoting bureaucratic violence would appear greater by deliberately choosing to remain silent rather than attempting to change those minds via reasoned argument, as we see, for example, in Wisnewski and Emerick (2009).

In short, unless Howes wants to assert that there should be no challenges to the claims of effectiveness, lest they result in invitations to bureaucratic violence, he must demonstrate why a formal model is susceptible to such invitations but other approaches are not. “If creating a formal model of interrogational torture is a legitimate way to argue against it, then social scientists could legitimately use the same methods to argue for it” (Howes 2012, 23). Well, yes. But the following is also true: if creating a nonformal model of interrogational torture is a legitimate way to argue against it, then social scientists and philosophers could legitimately use the same methods to argue for it (for an example from philosophy, see Bagaric and Clarke 2007). And indeed, the fact is that other nonformal critiques of interrogational torture have already “invited” responses from proponents. Former Bush speechwriter Marc Thiessen, for example, makes it very clear that he was prompted to write his detailed (and best-selling) apologia of the Bush interrogational torture program by its many critics (Thiessen 2010, 18). Foremost among them is Mayer (2008), whom Howes repeatedly cites favorably, rather than criticizing for promoting bureaucratic violence (Howes 2012, 20, 21, 24). The upshot is that if the government is going to engage in interrogational torture, if it is going to make an argument that interrogational torture works in a particular way that alone could ever justify for some people that torture, and it is possible to demonstrate that, in fact, it does not work in that way, then bureaucratic violence is in fact facilitated by avoiding the question rather than taking it on.

Formal Models and Normative Political Theory

Howes acknowledges that “certain uses of . . . esoteric works in political theory and philosophy . . . encourage a detached, inhuman, . . . view of politics” not unlike some corrosive forms of rational choice theory. I agree with his assessment of both rational choice and political theory—there are more and less esoteric forms of each enterprise, and I am concerned that Howes’s response is an exemplar of the former camp. Even if we restrict our attention to Howes’s narrower focus (the effects of torture on the victim) rather than remain with mine (interrogational torture), the representation he offers is beset by three problems: logical inconsistency, empirical aridity of concepts, and a kind of arbitrary domain restriction.

First, although Howes challenges the model of the rational actor, he offers no alternative model of human action and indeed seems to reject the idea of any such model in his discussion of the will. As an entity that can pursue and imagine ends, which can take a position, which can create objects and activities, control desire, direct the attention of the senses, the will would appear to have the status of a subject. Yet to the degree that it is a “faculty” that can be invoked, exercised, or drawn on, it seems also to be something that is an object of choice or manipulation (Arendt 1978b). In still other places, as a “unifying force,” as something inhering internal experience,” or, as Arendt says in other places, as something that “fashions” the individual, “an organ of free spontaneity that interrupts all causal chains of motivation that would bind it” (Arendt 1978b, 195; 1978a, 213), it appears to be neither subject nor object, but something beyond both. It may be one or even some of these things, but it cannot be all of them. ⁶ Nor does it help to simply state baldly that the will is “paradoxical and self-contradictory,” as Arendt (1978a, 214; also see 1978b, 68) does. In short, it is difficult to see how such a representation is more “commonsense” and less “detached” and “inhuman” than a model built on the actual desires, beliefs, and actions of human beings (Howes 2012, 23). What makes a normative theory less esoteric is not whether it is expressed in natural or formal language but whether it is at once logically coherent and firmly tethered to the actual experiences of human beings.

Second, normative political theory is not empirical political science, but it benefits from a more thorough and systematic grounding in empirical work. Howes and those whom he cites argue that torture destroys victims so completely along a number of dimensions (perception, agency, reason, etc.) that they are left with nothing but the free-floating will, a “sheer force” that is (may be?) somehow outside the control or consciousness of a person. But as the few examples of victims’ own accounts cited above make clear (and there are unfortunately many more), human beings still manage to deliberate, to make choices, to act even under such horrific circumstances. Ignoring these accounts goes much further than formal modeling to bury human suffering by refusing to credit victims of interrogational torture with the particularly human quality of reasoned action even under enormous pain and trauma. Excessively esoteric political theory of the kind sketched by Howes is “blind to this potent human potential” (Howes 2012, 22) because of a commitment to philosophical concepts divorced from human experience.

Finally, not only does an exclusive focus on will as he conceptualizes it ignore the elements of choice and strategy clearly in view in interrogational torture victims’ own accounts, but it is incomplete on its own terms as well, resulting in a form of unrecognized, if ironic, “arbitrary domain restriction.” ⁷ Howes introduces the discussion of the will to account for successful resistance to interrogational torture, but there is no discussion of the role of the will for those victims of interrogational torture who eventually falsely confess or divulge information. Consider Sheila Cassidy (1977, 188), an English doctor tortured by the Pinochet regime in 1974. After lying to her interrogators, she was subjected to additional torture and eventually provided information to her interrogators, although she “told them as little as she could, always hoping that a minimum of people would be involved.” Denying Cassidy agency adds insult to the suffering she experienced at the hands of her torturers. And since Howes does not suggest any alternative model of human action for detainees under interrogation generally or victims of interrogational torture in particular, we are left with no account of these cases, cases that are just as much a part of interrogational torture, which involve the will, as the cases of successful resistance.

Formal argumentation does not, of course, necessarily prevent this sort of domain restriction. But a formal model’s clarity can help to define that domain clearly, and to the extent that a single, logically coherent framework does manage to capture an important range of cases in the world without arbitrarily excluding others, then that model is useful. If it also generates novel insights, then it is analytically powerful in addition to being useful. To this extent, Howes misses the point when he says at various points that the model “simply formalize[s] what others have already said” (Howes 2012, 23). Yes, the model does generate the outcomes discussed by Augustine and Beccaria (the torture of innocents resulting in false confessions and strong detainees refusing to confess). But it also captures outcomes they do not discuss, such as ambiguous information and surprise torture. The goal of the model is to cover coextensively the cases we see in the real world, neither missing any nor generating implausible outcomes we do not see. To this degree, the article aims to complement, not substitute for, empirical work such as Rejali’s.

Not all arguments in normative political theory can or should be formalized. Still fewer should be rational choice models. But game theory does have a place in normative theory. Lean descriptions and a narrower focus on key features not only help preserve logical consistency and improve analytical tractability of complex problems but also help identify the appropriate empirical referents and scope of the model, helping to keep it honest and rooted in real human experience. Finally, there is a way in which formal theory connects to liberalism that is worth highlighting. The explicitness and openness of formal argumentation are reflected in both our models of reason and liberalism themselves, as well as in the elective affinity between reason and liberalism. If liberalism’s origins are to be found in challenges to arbitrary and irrational authority via the open light of reason, then formal theorizing about interrogational torture continues that tradition.