Punishment Notes for April 12

Main points

Feinberg lays out three justifications for punishment: retributivism, utilitarianism, and vengeance. He explained how these views work to justify punishment and some of the major problems that each one faces.

At the end, I proposed a way of combining the retributivist and utilitarian views.


The major objection to retributivism is that its sufficient condition for punishment would recommend punishment too often, when punishment doesn’t do any good (beyond ensuring that the guilty get what they deserve, of course).

Max argued effectively that neither Feinberg or I had made a persuasive case for this objection. I gave an example of someone who had committed a small crime in the distant past and led an exemplary life since. The idea was that there is little point to punishing this person now.

But Max pointed out that quite a lot turns on the crime and punishment in question. If it’s a minor crime, then the penalty will be minor too. That makes imposing the penalty even late in life seem less objectionable. If it’s a major crime, like murder, then it’s harder to see that there’s no point in punishing the criminal, even if he has led an exemplary life for years, and so on.

It remains true that if we take seriously the retributivist’s sufficient condition for punishment, then we will ignore the social costs of punishment. So it must be at least possible to come up with a case where the cost of punishment vastly exceeds the benefit, even taking into account the value of ensuring that the guilty get what they deserve. But Max is right that I didn’t present one. Score one for Max!

In addition to the hypothetical example, I presented some facts about the social cost of punishment in the US. These were meant to suggest that if we did the hard empirical work of tallying up the cost and benefits of punishing all the guilty as we do, that we would find the costs outweigh the benefits.

Of course, a retributivist could say that this is due to the fact that the US employs punishments in excess of what criminals deserve. However, the connection between giving people only the punishments they deserve and keeping the costs of punishment reasonable would be entirely fortuitous.


Utilitarianism pays greater attention to the costs and benefits of punishment than retributivism does. But it pays much less attention to what people deserve. It’s at least possible that a utilitarian would punish the innocent or punish people more (or less) than they deserve. If the consequences of doing these things for the society as a whole are positive, that’s the correct thing to do, by the standards of utilitarianism.

Utilitarians, of course, can correctly note that there’s a gap between the logically possible and the plausible. There are very strong utilitarian reasons for restricting punishment to the guilty and apportioning the amount of punishment to the gravity of the crime. Still, it’s not obvious that this is enough to blunt the objection. If you’re willing to punish the innocent, something has gone wrong.

A hybrid

The problems with both views come from their sufficient conditions for punishment. So I proposed merging their necessary conditions and, implicitly, abandoning each one’s sufficient condition. That is, I proposed that punishment is justified only when it is deserved and when the benefits of punishment exceed the costs.

The hybrid lacks a sufficient condition for punishment. It is also open to a complaint about fairness. You and I might commit the same crime but receive different punishments. That is because the circumstances might change, such that the benefits of punishing you outweighed the costs of doing so but the opposite is true for me. If so, you would be liable for punishment while I would not, despite the fact that we did the same thing.

This page was written by Michael Green for Philosophy of Law, Philosophy 34, Spring 2010. It was posted April 16, 2010.
Name of website