Poetic Injustice:

How Narratives Can Lead Us Astray


Lester H. Hunt

University of Wisconsin – Madison

In Poetic Justice Martha Nussbaum undertakes to explain how “story-telling and literary imagining” can supply “essential ingredients in a rational argument” and thereby improve public discourse regarding important ethical, political, and legal issues.[1] The particular sort of ingredient she investigates is supplied by “the realistic novel,” which she claims works on us, in significant part, by appealing to capacity that Adam Smith called “sympathy,” a certain ability to enter into the thoughts and especially the feelings of others. The paradigm case of this effect is the reaction a sensitive white reader will have to the opening scene of Richard Wright’s Native Son, in which Bigger Thomas, the poor black protagonist, gets out of bed one morning and, before he can begin his day, must do battle with a very large rat, a rat that puts up a determined fight until it is crushed by an iron skillet.[2]

No doubt, many readers encountering Wright’s book for the first time have felt the scales fall from their eyes. One feels that one has gained an insight into how it is to live a life radically different from one’s own. Most importantly, this insight seems morally valuable, and it does modify our moral understanding of Bigger Thomas later in the story, when he murders two women. Throughout Poetic Justice Nussbaum contrasts this sort of “empathetic imagining”[3] with another way of attending to the problems of life, with what she calls the “cruder forms of economic utilitarianism.” What makes these forms of economic analysis crude, apparently, is the fact that they are uninformed and unchecked by empathetic imagining, so that they alienate us from the particularities of individuals’ lives and encourages us to view them as interchangeable tokens subsumed under an abstraction.[4] Though it can be very useful politically and economically, such analysis has a powerful tendency to be morally corrupting whenever it leaks out of the realm of the political and economic and into the realm of moral judgment. Since the problems of public policy themselves often raise moral issues, the possibility of this sort of contamination is omnipresent. This makes it all the more important that such analysis be checked and limited by empathetic imagining. On the other hand, there is no need to use this sort of analysis to check and limit the workings of empathy. In this way, the relations between these two ways of thinking are entirely asymmetrical. Sympathetic understanding always brings to our attention features of the world that are morally relevant, while the sort of analysis that is represented by economics is marked by a seductive tendency to blind us to these very same features.[5]

I will do my best in what follows to convince the reader that this position is not true. I do not deny that literature often appeals to our powers of imaginative sympathy, nor that such appeals have potentially enormous value in making our lives better. I do, however, wish to undermine the peculiar position of moral privilege in which Nussbaum places them. Indeed, there are many cases in which empathetic imagination may be morally inferior to a process that seems in relevant respects to resemble economic analysis.

In order to make the point that I wish to make, I will present the reader with an example of another sort of story, a sort that contrasts sharply with the sorts of narratives that Nussbaum has in mind. Indeed, it differs so profoundly from them that it might be argued that it is not type of narrative at all. I will call quasi-narratives of this sort “scenarios.” For convenience, I will use “stories” as a generic term to refer both to these scenarios and to the clearer instances of narration.

The particular “scenario” upon which I will focus my attention is one that I borrow from the economist James Buchanan.[6] It describes the actions of two “players,” named for the time being A and B, who are interacting in situations having the payoff structures represented in a game-theoretic matrix that can be seen in Figure 1. Player B has two options, represented by columns 1 and 2, while player A’s options are represented by rows 1 and 2. Depending on which row and column are chosen, they will get the results in cell I, II, III, or IV. The results that befall each player are represented by the two Arabic numerals in the cells, with the left one going to A and the right one going to B. The results for each player are ranked, by that player, 4>3>2>1. The players make their choices solely on the basis of expected results. The situation is conceived as an “iterated game,” with both players having a series of chances to make choices. At any point, either player may bring about another play of the game simply by choosing another option. For instance, if for some reason they find themselves in cell III, A can, if he[7] notices this opportunity, move both players to I by choosing row 1. He would thus improve his outcome by changing it from 3 to 4, incidentally giving B a result of 2 (which is worse than B’s former result of 4). On the other hand, B will not depart from cell III by choosing column 2, because that will saddle him with a payoff (namely, 3) that is worse for him than his result in cell III.

What are A and B likely to do? At first look, this may seem an unanswerable question but, for those familiar with elementary game-theory, one prominent feature of the situation can serve to simplify the question considerably. Both players have a “dominant” option: that is, each faces a set of possible choices in which the payoff from one option can be expected to be the best one for oneself, regardless of what the other player does. If A chooses row 1 instead of row 2, he gets a result of 4 instead of 3 if B chooses column 1, and he gets a result of 2 instead of 1 in the event that B chooses column 2. Similarly, column 1 “dominates” column 2 for B, giving him 2 rather than 1, or 4 rather than 3. It seems only natural that each will choose his dominant option, with the result that they will find themselves in cell I. This is a happy ending for A: it gives him his best result (4).

This is not such a happy ending for B, of course. Although B’s result (2) is not his worst possible one, it is two steps, so to speak, below his best. There is, however, a way that B can hope to improve his result. He may come to realize that he is in a “game,” that his outcomes are a product, not merely of what he chooses, but of what A chooses as well. If he does, it will then be possible for him to think “strategically”: that is, it will open the possibility that he might search for choices that will cause A to make the choice that he, B, desires. He might try to induce A to choose row 2. How might A do this? He can threaten to choose, or can actually choose, column 2 until or unless A chooses row 2. This, if successful, would put them in cell III, so that B will have gained at A’s expense. This scenario, however, need not end here. A can, assuming again the needed insight, behave strategically. Once again, this strategic behavior would amount to making it clear that A will not be moved by the utility effects of B’s taking (or threatening to take) column 2. In this scenario, A’s strategy would take the form, not of threatening to take a new option himself, but of making it clear that he will not be influenced by B’s behavior or threatened behavior. If only he can make B believe that he will not give in to pressure and take row 2, he will be able to stay were he is. Thus he will avoid sustaining a result of 3, reaping instead a result of 4. B would have no reason to choose column 2 or threaten to, given that he would know that he would only be assuring himself of a result of 1 instead of the 3 he had in cell I. As before, whether A will behave strategically will depend in part on whether he can steel himself against short-term pain for the sake of improved results in the long run.

The scenario I have just set out differs from narratives as we normally understand them in a number of ways, not the least of which is the fact that “A” and “B,” the names of the two “characters” in the scenarios, are not really names at all, but dummy terms that stand in for an indefinite array of names of an indefinite array of individuals. These individuals are characters in an indefinite array of possible concrete narratives, in which the characters are subject to incentives that have the structure that is represented in figure 1. Rather obviously, we will only be able to see how scenarios like this one apply to the world in which we live if we are able to identify some concrete narratives that illustrate these them. Here is one such illustrative narrative.

A is the President of a small country, and B is the head of a terrorist organization that has been active in this country. Though the President’s government is not opposed to the liberationist goals of the terrorists (aiding the cause of an oppressed religious and ethnic minority in a nearby country) she and her government are active opponents of the methods they use, which include deliberately killing innocent people. Now, at last, the President’s government has captured several members of the terrorist organization and convicted them of murder. They now are now safely behind bars (her country does not use capital punishment for this offense). This is cell I: the President has what she wants and the terrorist, though he could be worse off, does not. However, the terrorist chooses column 2, which is to seize several completely innocent hostages. The terrorist lets it be known that if the President releases the prisoners (an option represented by row 2) the terrorist will then have every reason to release the hostages (moving the players to cell III). But the President has an effective, if miserably painful, defense against this strategy: she can make it clear that the government has an ironclad policy of not giving in to the demands of hostage-takers (a policy, that is, of never choosing row 2). If this had been known before, it would have prevented the kidnapping and the threats from occurring at all. At least making it known now would prevent or reduce the likelihood of future incidents of this sort. Still, there is the awful fact of the hostages who have already been taken captive. The President knows who they are. Their pictures have been broadcast on television. She has read reports on each one of them, compiled by the secret service.

One can easily imagine this concrete narrative spelled out in eloquent and gut-wrenching detail. Such a narrative would naturally operate strongly upon our sympathies. Though our sympathies would fall upon more than one character, their greatest weight by far would be on the hostages. What the game-theoretical scenario suggests, however, is that this is precisely the wrong place to experience our strongest motivation, since it favors a row 2 course of action on the part of the President. We can find the same clash between game-theoretical scenario and concrete, realistic narratives in a wide class of the narratives that illustrate the game-theoretic scenario represented in figure 1. These are cases in which one person is strategically using another person’s capacity for sympathy against them, in order to elicit from that other person behavior that interferes with the goals of the victim but advances the goals of the strategic agent. Such situations are faced by officials who deal with plane hijackers, and hunger strikers, and by university administrators facing student demonstrators who are occupying their offices. A strategic response is available to the victim of the sympathy-based strategy: refuse to follow the course of action prompted by those sympathies. Precisely because these sympathies are real, the choice faced by the victim has a dilemma-like poignancy: Buchanan called this sort of situation “the Samaritan’s Dilemma.”[8]

One difference between the two sorts of “stories,” we have looked at -- game-theoretic scenarios on the one hand and narratives as we usually understand them on the other -- stands out as particularly obvious. One of them tells of the actions of characters who are depicted as specific individuals, while the other, as I have already suggested, recounts the actions of characters who are designated by terms that are not really names at all, but variables that are replaceable by any one of an indefinite array of proper names. One, in other words, is concrete, while the other is abstract. A further difference between them, and a particularly important one for my purposes, has to do with a feature that they actually share in common: the fact that readers of both sorts of stories often have a sense of being enlightened by them. The difference is that the enlightenment seems to be brought about in each case in quite different ways. Narratives often produce enlightenment, or at least an impression of insight, by the means that Nussbaum eloquently describes: that is, they draw us into a sympathetic involvement with the characters as individuals, in which we get a stronger sense of what it might be like to be such people as these. Game-theoretic scenarios on the other hand give the impression of enlightening us precisely because of their frigidly abstract character, which seems to produce a sort of alienating effect on the reader, an effect that makes sympathy impossible.

What is most important in the present context, however, is the fact that these two stories conflict in Samaritan’s dilemma situations: they tend to support opposite courses of action. The game-theoretic framework brings before the reader’s consciousness considerations that tend to support a choice on the part of A which conflicts with A’s sympathies. For every such scenario, it will be possible to construct a plausible narrative that focuses attention on considerations that elicit these very sympathies in the reader. In the case of my little narrative of the President and the terrorist, for instance, one can tell a very powerful story about the hostages B has taken, and their terrible fate in the event that A holds to an iron-clad policy of never negotiating in cases like this one. In every case, the tendency of such narratives will tend to lead the reader to thinking that B should “win” the “game,” that A should not take the strategic course of action that would favor A’s cause at B’s expense. The narrative embodies a state of mind that, if A were to indulge in it, would undermine the fortitude A needs to face and overcome the short-term pain that an adequate defense brings with it.

I hope it is obvious that I am not saying that the courses of action that the game-theoretic stories tend to support is always right, or that the sympathy-eliciting powers of concrete narrative always lead us in the wrong direction. That would clearly be false. After all, the party representing the position of B in my scenario could, to take only one example, be the freedom riders in the American south in the early +sixties. At the time those events actually occurred, there were narratives in the press and on television that elicited our sympathy for the marchers and sit-in demonstrators, as they were attacked with fire hoses, police dogs, and electric cattle prods. These stories had a very powerful tendency to inspire attitudes favorable to the demonstrators. By doing so, however, they led us in the right direction. Sheriff “Bull” Connors’ cruel methods were strategically “right” given his preference rankings, but those rankings themselves deserved to be rejected. Insofar as the two sorts of stories would lead us to distinctively different conclusions in this case, game theory would be wrong and narrative would, quite simply, be in the right.

The clash between these two sorts of story raises interesting issues. Clearly, the question of which one is wrong and which is right is not among them. That is not the question at all. But there do seem to be some very real issues regarding the strengths, limits, and relative merits of each sort of story. These do seem to me to be well worth looking into.

It would actually take us too far afield to attempt a full treatment of these issues here, but I think we can learn something that is very much to the point by considering the implications of what we have already learned. The idea of the Samaritan’s dilemma marks out a context in which it is sometimes arguable that the sympathy-eliciting function of narrative leads us in the wrong direction, while game-theoretic stories prompt us in the right one. The nature of the two sorts of story, together with the nature of the Samaritan’s dilemma, ensures that it will always be arguable that, from A’s point of view, the heart- tugging narratives lead us astray. On the face of it, one reason this is so lies in the fact that, given the way I have characterized the dilemma, A’s preferences will be based to a considerable extent on considerations that are quite different from sympathy. In the case I have recounted here, the President is apparently motivated by such considerations as justice, the rule of law, and her legal and moral obligations to the people she represents. To the extent that we view the situation through the lens of a heart-tugging narrative, such considerations as these will be quite invisible to us. Of course, it is also true that A’s preferences, in cases drawn from life, will be to some extent based on sympathy. The President is no doubt motivated by concern for the pain that might be inflicted on the future victims of the imprisoned terrorists, in the event that she gives in to demands that she unleash them once again on the world. But in her decision-making situation, these other victims are mere theoretical possibilities, bloodless abstractions. The present victims -- the hostages -- are real individuals, and it is to them that one’s sympathies attach, or attach most strongly.

What game theoretic scenarios accomplish is to abstract from those considerations that give sympathy its powerful advantage over other considerations. A’s sympathy for the hostages is expressed in the numbers that rank the different cells from A’s point of view, and so are the non-sympathy-based considerations. They are both reflected and, in a sense, expressed in the rankings, but only as rankings. The emotionally affecting details that give sympathy its advantage over other considerations are omitted. Something similar is true of considerations regarding possible future victims. As we regarded the problem of the President and the terrorists game-theoretically, we were in effect conceiving of the game as one that is played (or can be expected to be played) by a series of As and Bs, each dealing with a decision-making situation that has been affected by decisions made by previous players in the series. This is the view of the problem that casts the most favorable light on the argument for the notion that the President should not give in to the demands of the terrorists. One reason it has this effect is that it converts the present hostages into bloodless abstractions, so that they have no more pull on us than the possible future victims do. The present hostages have lost their advantage over future ones. One could argue that this is all for the good, as this advantage seems to be arbitrary and unfair.

[1] Martha C. Nussbaum, Poetic Justice (Boston: Beacon Press, 1995), p. xiii.

[2] Poetic Justice, pp. 9-10 and 93.

[3] Poetic Justice, p. xvi.

[4] She gives, on p. 3 of her book, two examples of economists who are not culpable in this way, one is Adam Smith, the master-philosopher of sympathy, whose The Theory of Moral Sentiments (Oxford, England: Oxford University Press, 1976) provided some of the inspiration for Nussbaum’s own thinking about sympathy. (See Poetic Justice, p. xvi.) The other is her sometime collaborator, Amartya Sen. In this context an association with empathy-based thinking is most obvious thing these two economists have in common.

[5] Though Nussbaum acknowledges that such understanding must be supplemented by rule-based reasoning (see Poetic Justice, p. xvi) this other sort of reasoning generally bears no resemblance to the simplified models of the economist. “We should be on our guard against the ease with which [such] simplified models tend to take over and begin to look like the whole of reality. We should resist that tendency. To that end, we should insist all the more on novel-reading.” (Poetic Justice, p. 47.) There is no contrary tendency that should also be resisted, one that can be checked by simplified, emotionally alienating models of human behavior. There is one context in which she does acknowledge that “the ‘cold’ techniques of economics might give more accurate guidance” than emotion-based deliberation (p. 69), but the context she has in mind consists of situations in which one’s emotions are influenced by a bad idea (namely, that human life is “sacred” or “of infinite value”). That is, the coldness of the techniques, the fact that they dampen the effects of emotion, is not what is virtuous about them.

[6] James M. Buchanan, Freedom in Constitutional Contract: Perspectives of a Political Economist (College Station, Texas: Texas A & M Press, 1977), pp. 167-185. I make several important changes in Buchanan’s account. One such change is a rearrangement of the payoffs in the matrix in figure 1. In addition, the concrete story that serves to illustrate the matrix is my own invention.

[7] For convenience I will refer to A and B for the time being as “he,” though each can actually be either a “he” or a “she,” or even an “it” (eg., clubs, corporations, states, etc.).

[8] “The Samaritan’s Dilemma,” in fact, is the title of the paper from which I have adapted these scenarios. See footnote 6. Perhaps I should add, though, that the game-theoretic scenario suggests is that the appearance of a dilemma – a choice between alternatives that are all unacceptable – is merely apparent. They suggest that, from A’s point of view, one alternative is preferable.