Theoretical discussion of categorical norms and game theory applied to US sanctions on Syria

Morality Games 303 been silent over his use of conventional weapons to kill over 100,000 Syrian civilians. A Reuters/Ipsos poll at the time found that only 9 % of Americans favored intervention in Syria, but 25 % supported intervention if the Syrian government forces used chemical weapons against civilians (Wroughton, 2013). In the past, the United States has abided by the norm against the use of chemical weapons even at the expense of American lives: In WWII, Franklin D. Roosevelt chose to eschew chemical weapons in Iwo Jima even though, as his advisors argued at the time, their use would have saved thousands of American lives. It might even have been more humane than the flame-throwers that were ultimately used against the Japanese (“History of Chemical Weapons,” 2013). We say that the norm against chemical weapons is a categorical norm because those who abide by it consider whether a transgression was committed (did Assad use chemical weapons?), rather than focus- ing entirely on how much harm was done (how many civilians did Assad kill?). Other norms are similarly categorical. For instance, in the introduction to this chap- ter, we noted that across cultures and throughout history, the norm against murder has always been categorical: We consider whether a life was terminated, not the loss of useful life years. Likewise, discrimination (e.g., during Jim Crow) is typically based on categorical definitions of race (the “one drop rule”) and not, say, the dark- ness of skin tone. Human rights are also categorical. A human rights violation occurs if someone is tortured or imprisoned without trial, regardless of whether it was done once or many times and regardless of whether the violation was helpful in, say, gaining crucial information about a dangerous enemy or an upcoming terror attack. We even assign rights in a categorical way to all Homo sapiens and not based on intelligence, sentience, ability to feel pain, etc. Why is it that we attend to such categorical distinctions instead of paying more attention to the underlying continuous variable? We use game theory to explain this phenomenon as follows: Suppose that two players (say, the United States and France) are playing a Coordination Game in which they decide whether to punish Syria, and each wants to sanction only if the other sanctions. We assume the United States does not want to levy sanctions unless it is confident France will as well, which corresponds to an assumption on the payoffs of the game (if we reverse this assumption, it changes one line in the proof, but not the result). We model the underlying measure of harm as a continuous variable (in our example, it is the number of civilians killed). For simplicity, we assume this variable is uniformly distributed, which means Assad is equally likely to kill any number of people. This assumption is, again, not crucial, and we will point out the line in the proof that it affects. Importantly, we assume that players do not directly observe the continuous variable, but instead receive some imperfect signal (e.g., the United States observes the body count by its surveyors). Imagine a norm that dictates that witnesses punish if their estimate of the harm from a transgression is above some threshold (e.g., levy sanctions against Syria if the number of civilians killed is estimated to be greater than 100,000). As it turns out, this is not a Nash equilibrium. To see why, consider what happens when the United States gets a signal right at the threshold. The United States thinks there is a HOUSE_OVERSIGHT_015515