Saturday, October 8, 2011

Theories of Desert

Why did she deserve this?

Lot’s of people think they deserve to be treated in particular way. And we often think that others deserve to be treated in a particular way. But why do we think this? What is it — what property or state of affairs is it — that makes certain kinds of treatment deserved? Is it possible that this property or state of affairs does not exist? Is it possible that no one ever deserves to be treated in a particular kind of way?

Each of these questions concerns something we can call the desert relation. This is the putative relationship between a person and a set of outcomes in the world that make a certain kind of treatment deserved. Howard Simmons’s book Moral Desert: A Critique is a pretty decent discussion of this relation. As might be guessed from its title, Simmon’s isn’t convinced that the desert relation is real, but I don’t want to share his overall argument here. Instead, I want to consider a small portion from the first chapter of the book. The portion in question deals with the different theories of the desert relation. The purpose of this post is merely to highlight how tricky it is to get a good theory of this relation.

1. Internalist and Externalist Theories of Desert
Suppose I kill my best friend’s cat. Do I deserve to be punished? Before answering that question, consider two different ways in which I could have killed the cat. First, suppose I deliberately and intentionally kill the cat. I always found its miaowing, purring and scratching annoying, so I decided to put an end to it. I went out, bought some poison, and put it in the cat food. Second, suppose I accidentally kill the cat. I was driving in the driveway of my friend’s house when his cat suddenly leapt in front of the car and it got crushed beneath one of the wheels.

Do I deserve punishment in both scenarios or only one? I’m guessing most people would say that I only deserve punishment in the first scenario. In saying so, they are giving voice to the internalist theory of desert. According to this theory, whether I deserve punishment (or, indeed, praise) for my acts depends entirely on the mental states that accompany those acts. In the first scenario, the fact that I deliberately and intentionally directed my actions toward securing the cat’s death singles me out as an appropriate target for blame; in the second scenario, the lack of deliberation and intent has the opposite effect.

Desert Internalism : Whether an individual deserves a particular kind of treatment in response to their acts depends entirely on the mental states that accompany those acts.

When it comes to the cat example, internalism seems obviously correct. But it’s not all plane-sailing for the internalist. As many have pointed out, it seems likely that internal mental states by themselves are insufficient for desert. Consider the following three cases (based on the ones used by Peter French and discussed by Simmons, but with some variations):

  • 1. Imaginative Murder : Charlie has a rich and vivid imagination. He occasionally has day dreams that are so realistic that he confuses them reality, often thinking he has had encounters with people and said things to them that only ever happened in his dreams. One day he has a dream in which he kills his own mother. In the dream, he acts with malicious intent, carefully planning and executing the murder. But in reality, no one is actually harmed and Charlie relationship with his mother remains intact.
  • 2. Virtual-World Murder : Same as above, except this time Charlie is operating in a hyper-realistic virtual world. The virtual world provides realistic feedback for all his actions. So in every respect it feels like he is committing an actual murder. But, once again, no one is actually harmed.
  • 3. Real-World Murder : Same as above, except this time Charlie is operating in the real world. He actually plans and executes the murder of his mother. She really dies, and he and his family suffer irreparable damage.

How do you respond to these cases? If you think Charlie is equally deserving of punishment in all three cases, then you are a steadfast internalist. But if you think his desert varies between the three cases — for example, if you think he deserves more punishment in the third case than he does in the first case — then you are drifting toward an externalist theory of the desert relation. According to this theory, desert depends, at least in part, on what happens in the external world. So the fact that real people are affected in the third case makes an important moral difference when it comes to assessing Charlie’s desert.

Desert Externalism : Whether an individual deserves a particular kind of treatment depends, at least in part, on the external effects of their actions.

2. Which Theory is Best?
Interestingly, Simmons doesn’t think the three cases involving Charlie do much to upset the internalist. He thinks — and perhaps you will agree — that Charlie is equally deserving in all three cases. This judgment might be displaced if we learned that Charlie could appreciate the distinction between reality and imagination in the first case, or reality and computer simulation in the second. After all, we wouldn’t punish someone for a thought crime. But if we take the cases and presented, and we assume he couldn’t appreciate any distinction, then there is something to said for Simmons’s point of view.

Despite this, Simmons is willing to concede that there are flaws in the purely internalist theory. These flaws come to the surface if we imagine another example. This one involves a person who sincerely believes that premarital sex is morally wrong and yet, nevertheless, succumbs to the temptation. Now, assuming premarital sex is not actually morally wrong, what are we to make of this person? Do they deserve to be punished?

Simmons argues that ultimately they do not: Because they did not break any moral rule, it would not be appropriate to punish them for what they did. This suggests that the internalist theory needs to be qualified in one important respect:

Modified Desert Internalism : Whether a person deserves to be treated in a particular way in response to certain acts depends on (a) the mental states accompanying those acts and (b) the moral status of those acts.

Assuming we hold to a cognitivist metaethics, the moral status of an agent’s actions is something that the agent themselves can be mistaken about. Thus, under this modified version of internalism (which is not really internalism at all), an agent’s desert is not entirely fixed by their occurrent mental states.

Can anything else affect our understanding of desert? Simmons discusses a range of issues that might be relevant. For example, he suggests that the amount of effort an agent requires in order to comply with a moral norm might affect our assessment of their desert. The idea being that the harder it is for agent to avoid moral wrongdoing, the less desert they incur. This brings a sensitivity to the fixed psychological characteristics of the agent into our assessment of desert.

Another, related, issue concerns the opportunities for moral wrongdoing afforded to the agent. The idea being that an agent who is exposed to many such opportunities, but nonetheless avoids them all, deserves more praise than an agent who is never exposed to such opportunities and thereby avoids wrongdoing. The flipside of this is that an agent who is exposed to plenty of opportunities to do good, but nevertheless does bad deserves more blame than an agent who is exposed to lots of opportunities to do bad and who avails of some of them.

This raises the issue of moral luck and the role it should play in our understanding of desert. Unfortunately, that’s too big a topic to get into today. Hopefully, this short post has done enough to show how problematic the notion of the desert relation can be.


  1. Suppose a person deserves simply whatever praise or blame would have cause them to improve their behaviour more than the cost of the praise or blame, had they anticipated the praise/blame. It looks to me like all of the issues you mention would show up as a result of this.

    More purposeful acts deserve more attention, because it is easier to deter them. Imaginary sins of any kind don't deserve much attention because we don't care to deter them, unless we take into account more complicated dynamics such as wanting to demonstrate that we will punish that sort of action in reality. How much effort it takes to avoid an action is always important - you want to punish people when punishment could have dissuaded them, and save your effort when it could not. If a person has a lot of opportunities to do wrong, it is worth incentivising them to do good with praise. If there is no chance of them doing wrong that is a waste of effort. etc.

    What's to stop us abstracting all these features to a principle like this?

  2. Hi Katja,

    Thanks for dropping by. I think that's probably right, and it might be worth saying that Simmons actually argues for that kind of utilitarian theory in his book (with certain caveats about the need for a "humane" system of punishment).

    If I could be devil's advocate for a moment though, I'd say that kind of approach faces a potential objection. The objection has a couple of steps to it.

    The first step is to point out that one feature of our current punishment (and reward) practices is technically unnecessary on the utilitarian theory. The feature being that we typically wait until someone has performed an act before we punish or reward them. But if the ultimate concern is with improving behaviour, why do we bother to do so? Surely if we could act preemptively it would be better? Isn't stopping crime before it happens better than waiting for it to happen? If so, why don't we try to preemptively punish people who are likely to commit crimes? This would, after all, help to improve their behaviour. Now, maybe an action carries some significant epistemic weight in deciding whether someone is an appropriate target for behaviour modification, but it still seems unnecessary on the utilitarian theory.

    This brings us to the second step in the objection. This step consists of a counterexample to the utilitarian theory. The counterexample would be some hypothetical case in which a person is arrested and punished even though they have committed no crime (a Kafkaesque scenario). This is justified on the grounds that their behaviour will be dramatically improved by doing so. A defender of a retributive theory -- like Michael Moore -- would then argue that treating a person in this way is deeply immoral. This could be for a number of reasons: e.g. it's intuitively disturbing; it is unfair; and it contravenes the principle of liberty.

    To sum up, the objection looks like this:

    (1) If the utilitarian theory of desert is true, it would be morally acceptable to punish people before they do something wrong.

    (2) It is not morally acceptable to punish people before they do something wrong.

    (3) Therefore, the utilitarian theory of desert is false.

    Obviously, there are a couple of possible responses here: reject the implication spelled out in (1) or argue that punishment prior to action is not morally objectionable. I think I'd probably go for the second kind of response, but I'm not sure yet.

    (Note: the same argument might work for reward as well as for punishment, I haven't thought that through).

  3. Yeah, I was thinking of moral luck issues as I was reading through this, namely, what would Simmons say about the drunk driver who hit a mailbox, while another drunk runs over a family? The Modified version helps him with that I guess, but you'd still have to consider all the possible outcomes to create a flat punishment.

    As to your objection above, I think utilitarians would reject "the ultimate concern is with improving behaviour" and therefore reject (1) on the grounds that it belies a greater concern for improving (or preserving) an overall general standard of living that doesn't allow for pre-crime punishment. Human behavior is only part of the context.