Consequentialist theories justify punishment by its effects. The English philosopher Jeremy Bentham put the starting point bluntly in An Introduction to the Principles of Morals and Legislation (1789): “all punishment is mischief: all punishment in itself is evil.” On his view, punishment is justified only if it prevents a greater harm, and it should be no more severe than needed to do so. There are three main ways punishment might prevent harm.

Deterrence

General deterrence means discouraging other people from committing crimes by the example of punishment. Specific (or individual) deterrence means discouraging the punished person from offending again. Both assume that people know about penalties and weigh them before acting, at least sometimes.

What does the evidence say? Stated carefully: reviews of the research, including a US National Research Council report in 2014, have generally concluded that the certainty of being caught deters more reliably than the severity of the punishment. Increasing already long sentences appears to produce small deterrent effects at best. This does not mean severity never matters, and results vary by crime, place and method. Beccaria made a similar argument about certainty as early as 1764 (Level 4).

Incapacitation

Incapacitation means preventing crime by removing a person's ability to commit it, most often by imprisonment. While someone is in prison, they cannot commit crimes in the community. This is a strong argument for holding people who are clearly dangerous, though it depends on how well anyone can predict who will offend in the future.

Researchers note limits. Offending generally peaks in the late teens and early twenties and declines with age, a pattern called the age-crime curve. So very long sentences often hold people well past the years when they would most likely have offended. And for some crimes, such as street-level drug selling, an imprisoned offender may simply be replaced by another.

Rehabilitation

Rehabilitation aims to change the offender so they no longer wish or need to offend, through education, work training, treatment for addiction or mental illness, and programs that address patterns of thinking. In 1974 the sociologist Robert Martinson published a review of rehabilitation programs that was widely summarized as showing that “nothing works,” a conclusion he later qualified. Later research, pooling many studies, found that some programs, especially cognitive-behavioral ones matched to a person's risk and needs, reduce reoffending modestly, while poorly designed programs often do not.

Strengths

Consequentialism asks a hard-headed question: does this punishment actually make people safer, and at what cost? It treats punishment as a public policy to be tested by evidence, and it refuses to inflict suffering that achieves nothing.

Criticisms

• The scapegoat problem. If punishing an innocent person would calm a riot and prevent many deaths, pure consequentialism seems to permit it. Critics say this shows the theory misses something essential about justice.
• Disproportion. If a harsh penalty for a minor crime deterred well, the theory seems to allow it.
• Indefinite treatment. A purely rehabilitative system could hold someone until they are judged “cured,” however minor the offense.

Consequentialists reply that framing innocent people or imposing wildly disproportionate penalties would destroy public trust and so produce worse results in the long run, and many adopt rules or desert-based limits for that reason. This is the path toward the mixed theories described in Level 1.

Worked example: a city council and a car theft wave

An invented city sees car thefts double in a year. The city council considers three proposals and asks how each would work, using the tools of this level.

Proposal A: double the maximum sentence for car theft. This aims at general deterrence through severity. The research suggests caution: many car thieves are young, act on impulse and do not expect to be caught, so longer sentences may change little. It would also cost a great deal in prison places.

Proposal B: more patrols in hot spots, better street lighting and a public campaign to get owners to lock cars and use steering locks. This aims at certainty of being caught and at reducing opportunities. Research on certainty suggests this is more promising, and it may cost less per crime prevented, though results should be measured.

Proposal C: for young first offenders, a program combining a job placement, mentoring and a cognitive-behavioral course on decision-making, with prison for persistent offenders. This combines rehabilitation for those most likely to change with incapacitation for a small group responsible for many thefts. Its success depends on program quality.

A purely consequentialist council would choose whichever mix prevents the most theft at the least cost. A council with retributive concerns would also ask whether each sentence fits the seriousness of the crime. Both would want to evaluate the results after a year rather than assume the policy worked. This habit of testing punishments against their actual effects is the central contribution of the consequentialist tradition.