One of the most profound accomplishments of the Enlightenment is the idea that a ethics from the ground up can succeed in establishing robust moral principles without the necessary intervention of any kind of objective moral lawgiver giving us ethics from the top down. In a recent essay for philosophical academia, I argued that even if a moral theory that successfully provides a sound, internally coherent, rigorous map of how humans ought to behave could be derived from unchallengeable premises, such a theory could not in principle be successful in modifying human behavior.

I felt this for a number of reasons, some scientific and some abstract. Because I know that Edger’s audience consists mostly in freakin’ brilliant navigators of the human condition of all stripes, I have decided to here briefly summarize and expand my argument into three primary points for your consideration. I can think of no finer peer review process than letting Edger’s readers be the first to examine, discuss, and hopefully dissect my argument.

It seems implicit in the act of moral theory-making that the moral philosopher wants to present some system for advising rational moral actors in how to respond to problems. Even if only a slim minority of people know what “utilitarianism” means, the utilitarian moral philosopher still has at least some interest in having other real people maximize goods and minimize harms. It is in this respect that I think no moral theory can succeed.

First, let me define a what I mean by a “moral theory:” a moral theory is any behavioral heuristic that compels one to respond to moral dilemmas by evaluating the morally salient features of those dilemmas. For example: “I want to maximize goodness while minimizing harmfulness” is probably the most intuitive moral theory ever devised. It meets my definition of a moral theory because, given a moral problem, the theory asks you to look at morally salient features of the problem (the “goodness” and the “harmfulness” of your choices) and to make a decision. By contrast, “flip a coin” is not a moral theory, even if it is a heuristic for solving moral dilemmas. Flipping a coin is completely irrelevant to the moral right or wrong of a particular choice, and so even if it could guide your behavior, it would not meet my definition of a moral theory. My definition of a moral theory is not, I think, controversial.

That really is the only definition you need (as a philosophy student, I know that 95% of philosophy is a morbid obsession with definitions, so I’m glad that my argument only needs one!) for my argument to proceed. Here, then, are two good reasons for why no moral theory can succeed in modifying human behavior:

1. The morally salient features of moral dilemmas are often less important than morally inert factors in deciding how people respond to those dilemmas.

Ok, so what do I mean by this? Basically I mean that, even if people think that they are responding to a moral dilemma by evaluating right and wrong, this is often an illusion. Instead, people are often mislead by completely morally irrelevant devices such as framing effects. A framing effect is any effect on your responses to a problem caused by something like how a problem is phrased or presented. Walter Sinnott-Armstrong’s fantastic Framing Moral Intuitions from his 2008 compendium Moral Psychology vol.2 (there are three volumes) discusses a number of these effects.

Sinnott-Armstrong cites a 1981 study by Tversky and Kahneman in which subjects were asked to choose between two risky treatment plans for an imaginary impending disease outbreak. For half of the subjects, one of the imaginary treatment plans will definitely save exactly one third of those infected. The other treatment plan has a 1 in 3 chance of saving everyone and a 2 in 3 chance of saving no one.

For the other half of the subjects, one of the imaginary treatment plans will definitely kill exactly two-thirds of those infected and the other treatment plan has a 1 in 3 chance that nobody will die and a 2 in 3 chance that everyone will die.

It should be obvious that, objectively speaking, both groups had the same exact plans. However, the unconscious influence of the “save” vs. “kill” in this study produced a dramatic effect: 72% percent of people chose the safer treatment in the first instance, but only 22% chose the safer treatment in the second instance. This is an obvious example of a morally irrelevant feature of a problem actually influencing peoples’ decisions: a rational moral theory would target the morally salient features of these dilemmas (who lives and who dies), but as we see here, the rightness or wrongness of an answer was completely overwhelmed by the morally irrelevant question of how the experimenters worded the problem.

Another study by Petrinovich and O’Neill found that, not only can framing effects take place within problems, they can even take place between problems. In a 1996 study, they presented four different groups of subjects with the same three dilemmas, only they offered them in a different order for each group of subjects. Some groups started off with a dilemma whose most beneficial choice involved action, others a dilemma whose most beneficial choice involved inaction. In short, the researchers found that the order in which certain problems were presented had a statistically significant impact on peoples’ answers to those questions. Again, this is an example where a trivial fact of presentation actually overrode a neutral sample’s moral judgments.

The final example I’ll offer (but by no means the final example in the literature; further data can be provided on request) comes from a 2007 study performed by four behavioral psychologists that tried to look at factors completely outside the scope of moral dilemmas. This study, by Schnall, Haidt, Clore, and Jordan, had two groups of subjects sit at a desk and give answers to written moral dilemmas. For one group, the desks were neat and tidy. For another group, the desks were filthy, with a trashcan full of old wrappers and food within sight nearby. Those seated at the filthy desks delivered far harsher moral judgments than those seated at the clean desk. The point here is clear: completely irrelevant trivialities have the power to override moral judgments in truly profound ways.

Why this is a problem for a successful moral theory is obvious: if the morally salient features of dilemmas are less relevant than irrelevancies to moral decision-making, then a moral theory is completely barking up the wrong tree in terms of guiding behavior.

2. Fast and frugal amoral heuristics can override strong moral judgments.

One possible objection to my first area of argumentation would be that “well, maybe these guys got a bit confused by the wording of some problems, but there are obviously some real-life moral problems that you just can’t mess up with a framing effect.”

Suppose I told you that a completely amoral wrinkle in a situation can lead 500 good, liberal men to murder over a thousand Jews.

Gerd Gigerenzer’s 2008 essay Moral Intuition = Fast and Frugal Heuristics?: quotes Christopher Browning’s 1993 Ordinary Men in telling a story about 500 German men, from the liberal middle-class of generally Nazi-hostile Hamburg, who carried out an order to round up and massacre more than a thousand innocent, unarmed civilians from an undefended civilian area.

The commanding officer received the order, and he assembled his men. He told them the orders, but then said that anyone who wanted to opt out of the mission could do so without punishment. Fewer than 3% of them opted out. The rest carried out their orders, despite having graphic, physical reactions of anguish and horror all the while. These men were not evil. Most of them probably knew what they were doing was wrong. So why did they do it?

They did it because of a fast, frugal, completely amoral behavioral heuristic: Don’t break ranks. Gigerenzner goes on to provide a wealth of data defending the existence of this heuristic. Another heuristic that is massively substantiated by the data: if there is a default, do nothing about it. The example here is organ donation: America’s organ donation apparatus has the default position being that one is not an organ donor, France’s organ donation apparatus has the default position being that one is an organ donor. Even though high numbers of Americans report endorsing organ donation, only 28% opt in. In France, where a roughly equal number of people endorse organ donation, only 1% opts out. Where the default is, so goes the majority.

Notice that neither of these heuristics has anything to do with morality. Whether or not all your friends are doing it (”don’t break ranks”) has literally no bearing whatsoever on the goodness of that action, and yet as we have seen, it is a far keener influence on human behavior than on any moral theory (nearly all of which would probably describe murdering innocent civilians just for being Jewish as being morally prohibited). The “go with the default” heuristic is the same story. Any utilitarian moral story will tell you that, on the balance, you should probably be an organ donor. And yet only 28% of people opt into the system. But, all of the best evidence demonstrates that “it is the default rule rather than alleged preferences that explains most people’s behavior” (Gigerenzer 2008).

This is another clear problem for moral theory-making: if even morally unthinkable scenarios can be occluded by completely amoral confounding factors, such as whether or not all of your friends are doing it too, then is anything sacred?

Formalization:

(1) If heuristic x is a moral theory, then x should be expected to provide guidance for moral behavior.
(1a) If heuristic x is a moral theory that should be expected to provide guidance for moral behavior, then x does so based on the morally salient features of problems.
(2) If x should be expected to provide guidance for moral behavior based on the morally salient features of problems, then x will not be able to provide responses to moral dilemmas because morally inert factors of these problems (like framing effects) have a greater on moral decision-making than the morally salient features of those problems.
(3) If x does not affect responses to moral problems because of morally inert factors of these problems, then it is not the case that x can be expected to provide guidance for moral behavior.
Therefore: If heuristic x is a moral theory, then x cannot be expected to provide guidance for moral behavior (HS 1-3).

Premises (1), (1a), and (3) all derive from the definition of a moral theory I have provided. Only definition (2) actually needs defending. I hope that the evidence I have provided from the literature (a bounty of additional evidence is available on request) sufficiently justifies this step of the argument.

It should be noted that this argument is not precisely moral nihilism (the philosophical position that no moral truths actually exist). In fact, my argument works completely independent of whether or not moral truths exist or can be tracked. All that matters is that any moral theory is irrelevant to behavior in the sense that moral theories target moral features of problems, but these features are less important than amoral heuristics and framing effects, even in dramatic cases.

Once doubt has been cast on a few moral judgments, doubt is cast on them all. Walter Sinnott-Armstrong provides the wonderful example of a bathtub with some thermometers. Suppose you are trying to determine the temperature of the water in a bathtub, which for some reason you can’t touch. All you have is a box full of thermometers. Unfortunately, all you know is that some number of the thermometers is off, and you don’t know by how much. What is the temperature of the water? You can’t know.

Moral decision-making by appeal to a moral theory works the same way. If the bathtub is a moral judgment, and the thermometers are your moral heuristics, you will find that you could never know whether or not your judgment is being occluded by a framing effect, or some other completely morally inert triviality that has nothing to do with right or wrong but which still has a strong effect on your moral decision-making. Not only do you not know which judgments are being occluded, you don’t know how many are altered by amoral trivialities.

And that is the crux of the problem: you don’t even know when your moral theory is broken. You have no way of knowing if your deployment of a moral theory is broken, or when it is working, because you cannot possibly know whether or not you are making a genuine moral judgment or whether you are simply giving a predictable unconscious response to a morally irrelevant framing effect or other morally inert confounding factor.