Retributivism and consequentialism

The most prominent political philosophers of the seventeenth century, such as Hugo Grotius, Thomas Hobbes, and John Locke, all thought that punishment was justified only if two conditions were met: punishment would do some good in the future and the person being punished was guilty of a crime.

Bentham and Kant split those two propositions apart. Bentham articulated the utilitarian or consequentialist view that punishment is justified if and only if it brings about the best consequences. Kant took the retributivist position that punishment is justified if and only if someone deserves it.

Feinberg laid out the central propositions for both views and some of the major problems with each.

Problems and impure solutions

The chief problem with consequentialism is that it is willing to punish the innocent. The chief problem with retributivism is that it is committed to punishing even when doing so has prohibitive costs or does no good.

The obvious solution is to recreate the seventeenth-century mixture: require both guilt and desirable consequences. That is what Hart will try to do in the reading for Thursday and what Goldman will criticize in the article we will discuss next week.

One thing to keep in mind is that mixed views do not solve all of the problems with the utilitarian approach. In particular, the requirement that punishment bring about desirable consequences could mean that people receive different punishments for the same crime. That will strike many people as unfair.

What about a sufficient condition?

I proposed adopting two necessary conditions that have to be met for punishment to be justified: guilt and desirable consequences.

But, as Leo pointed out, I did not say anything about sufficient conditions for the justification of punishment. Shouldn’t we have those as well?

It’s a good point. I have not thought this through, but I think that you could make the two conditions jointly sufficient without any obvious problems. That is, you could say that guilt and beneficial effects are a sufficient justification for punishment.

I think a partial explanation for my neglect of sufficient conditions is a point that the seventeenth century authors made explicit but that I had forgotten, namely, the state has the power to pardon criminals. If we think that the state should have that power, we may not actually think that guilt and the desirable consequences of punishment even together are sufficient reasons for punishing someone. We want to leave room for pardons.

That said, I am not sure that this is a good reason for neglecting sufficient conditions.

First, as Mollie noted, a lot depends on what you mean in talking about “sufficient conditions” on the justification of punishment. It’s one thing if you mean that you have good enough reasons to show that punishment is permitted, but not necessarily mandatory. But it’s something else if your sufficient conditions are supposed to explain why the state must punish. The power to pardon is compatible with the first but, arguably, not the second. I suspect that when Leo was asking for a sufficient condition, he had in mind conditions that would be sufficient to show that punishment is permissible. Kant, by contrast, thought that punishment was mandatory.

Second, you might ask if the power to pardon isn’t just a way of ensuring that punishment has good consequences. We generally think that the state should pardon criminals when there is no purpose to punishing them or some great benefit from not doing so. If that’s all the power to pardon involves, then it doesn’t really require anything beyond the two conditions: guilt and good consequences.

Finally, there is a good case for thinking that punishment really is mandatory. The state would be falling down on its job if it failed to punish criminals. And pardons or a refusal to punish mean that justice is not done. This is a significant issue in Ferguson, Missouri right now, for instance. Goldman will have some interesting things to say about this next week.

Key concepts

  1. The utilitarian (or “consequentialist”) approach to punishment
  2. The retributivist approach
  3. How a mixed view seems to address the main problems with both


In discussing Feinberg, I said that one problem with retributivism is that it seems to be impossible to fulfill its necessary condition: only the guilty can be punished. The reason is that we know that punishment hurts innocent members of the criminal’s family and community.

The very next day, I read a magazine article about research on exactly that point. Here’s the summary.

Today, about one-in-31 adults are in prison. This is a human rights crisis for the people that are incarcerated, but its impact also echoes through the job sector, communities, families, and the hearts of children. One-in-28 school-age children—2.7 million—have a parent in prison. In a new book, Children of the Prison Boom, sociologists Christopher Wildeman and Sara Wakefield describe the impact of parental imprisonment on children: an increase in poverty, homelessness, depression, anxiety, learning disorders, behavioral problems, and interpersonal aggression. Some argue that taking parents who have committed a crime out of the family might be good for children, but the data is in. It’s not.