What Are Prisons For? Four Answers That Pull in Different Directions

Each answers the question in its own terms and each implies a different policy:

• •Retribution, holding that punishment is deserved because of what was done, which is backward-looking, requires proportionality between the offence and the sentence, and does not depend on any future benefit
• •Deterrence, holding that punishment prevents future offences, split between deterring the individual punished and deterring others, and which is forward-looking and would in principle justify punishing more harshly than deserved if that worked
• •Incapacitation, holding that a person in prison cannot offend against the public during that time, which justifies detention by predicted future dangerousness rather than by past conduct
• •Rehabilitation, holding that the purpose is to change the person so they do not reoffend, which implies release when that is achieved and treatment rather than suffering
• •Restoration, a fifth position gaining ground, which treats crime as a harm to be repaired between the people involved rather than an offence against the state to be punished
• •Denunciation, sometimes listed separately, which holds that punishment expresses society's condemnation and reinforces shared norms

The tensions are practical rather than academic. Retribution caps a sentence at what the offence deserves, while incapacitation would extend it for someone considered dangerous, which is why indeterminate sentences based on risk assessment are philosophically contested and were abolished in England after serious problems. Deterrence would justify making an example of someone, punishing them beyond desert to influence others, which retributivists regard as using a person as a means. Rehabilitation implies release when a person has changed, which conflicts with a fixed sentence proportionate to the offence and with public expectations. Conditions follow the same split: a retributive prison need only be unpleasant enough, a rehabilitative one needs education, training, therapy and contact with family, and those are different buildings with different budgets. The compromise reached in most systems is to state all the purposes in legislation and leave judges to weigh them, which produces inconsistency that is criticised as unfairness and is an accurate reflection of the underlying disagreement.

Some of these claims are empirical and have been tested. Deterrence research consistently finds that the certainty of being caught deters far more than the severity of the punishment, which implies that investing in detection does more than lengthening sentences, and that potential offenders frequently do not know or consider the sentence for what they are doing. Individual deterrence performs poorly, with imprisonment associated with equal or higher reoffending than non-custodial alternatives in many comparisons, once like cases are compared. Incapacitation works while it lasts and is limited by the difficulty of predicting dangerousness, since risk instruments produce many false positives. Rehabilitation programmes show modest but genuine effects when they are well designed, targeted at higher-risk offenders and delivered as intended, which is a substantial qualification because prison conditions frequently prevent that. Meanwhile prison itself causes harm that affects reoffending, including loss of housing, employment and family contact, and exposure to a concentrated criminal environment.

Imprisonment rates vary between countries by more than an order of magnitude, and the variation tracks policy rather than crime rates closely. The United States imprisons a far larger share of its population than any comparable country, the result of sentencing policy changes from the 1970s onward rather than of crime levels, and the consequences have been documented extensively, including concentration in particular communities and effects lasting across generations. Scandinavian systems imprison far fewer people, use shorter sentences and operate prisons on a normalisation principle, arranging conditions to resemble ordinary life as closely as security allows on the reasoning that people who will return to society should not be taught to live in a way that prepares them badly for it, and reoffending rates in those systems are comparatively low, with the caveat that societies differ in many other respects. Alternatives to custody, including community sentences, electronic monitoring, restorative conferencing and treatment orders, are cheaper and perform at least as well for many offences, which is an argument made increasingly on fiscal grounds where the philosophical argument failed.

For most of history, prison was not itself the punishment. Jails held people awaiting trial, debtors who could not pay and prisoners waiting for their sentence to be carried out, and that sentence was usually a fine, whipping, public shaming, exile, transportation to a colony or death. Locking people up for a fixed term as the penalty became common only from the late eighteenth century. The English reformer John Howard toured jails and published The State of the Prisons in 1777, describing filth, disease and corruption that shocked readers and pushed governments towards change.

Reformers hoped a well-designed building could reform character. In the United States two rival models appeared in the early nineteenth century: the Pennsylvania system kept prisoners alone in cells to reflect and repent, while the Auburn system in New York had them work together in enforced silence by day and sleep alone at night. The word penitentiary, meaning a place for penitence, records that religious aim. Jeremy Bentham's panopticon design, with cells visible from a central watchtower, expressed the same belief in surveillance and routine. Many of today's debates about isolation, work and purpose began in these experiments.

A large share of the people in prison at any moment are not being punished for anything, because they have not been convicted. They are on remand, held while they await trial, and worldwide roughly three in ten prisoners are in this position, with more than half in some countries. None of the classic justifications applies to them, since each assumes a proven offence. Detention before trial rests on different grounds: making sure the accused turns up in court, preventing them from interfering with witnesses or evidence, and in some systems protecting the public from a serious risk.

That creates tensions of its own. Remand prisoners are legally innocent, yet they often face some of the worst conditions, in crowded local jails with little work or education, because they are expected to stay only briefly. If they are convicted, the time served is normally deducted from the sentence; if they are acquitted, it is simply lost, often along with a job and a home. Long waits can also pressure people to plead guilty in order to get out. Where bail depends on paying money, poorer defendants stay inside while richer ones go home on the same charge, which is why New Jersey largely abolished cash bail in 2017.

One of the better-studied questions about prison is whether education inside reduces reoffending after release. A large review published in 2013 by the RAND Corporation, combining many studies from the United States, found that prisoners who took part in education programmes were substantially less likely to return to prison than those who did not, and that the programmes cost less than the imprisonment they prevented.

Studies of this kind face a real difficulty, because prisoners who choose to study may differ from those who do not, and better studies try to account for that. Even so, the finding has been influential in debates about what prisons should offer. It connects to a broader point: if one purpose of prison is to protect the public, what happens to people after release matters as much as the sentence itself.

Retribution looks backward and caps punishment at what is deserved, deterrence and incapacitation look forward and would extend it, and rehabilitation implies release once a person has changed, so the four purposes recommend different sentences and different prisons. Evidence finds that certainty of detection deters far more than severity, that imprisonment frequently increases reoffending relative to alternatives, and that rehabilitation works modestly when well delivered. Imprisonment rates differ between countries far more than crime rates do.