Can Sociopaths Be Fixed?

Two Cochrane reviews published in 2020 are the most rigorous summaries available. The review of psychological treatments identified 19 randomised trials covering 18 different interventions, with usable outcome data from 10 trials and 605 adults. Its conclusion was that there is not enough good quality evidence to recommend or reject any psychological treatment for people diagnosed with ASPD. The parallel review of drug treatments covered 11 trials and 416 participants and concluded that the evidence is insufficient to draw any conclusion at all.

Those two sentences are the state of the field, and they are not the same as "nothing works." They mean the studies that would answer the question have not been done to an adequate standard. Sample sizes are small, often between 14 and 83 per comparison. Most comparisons rest on a single unreplicated trial. Attrition was high in 7 of the 11 drug trials. Almost no drug trial recruited people because they had ASPD; most enrolled people for substance use problems and identified ASPD afterwards.

The outcomes measured are also narrower than the question. Trials measured aggression scores, substance use, social functioning scales, and occasionally reconviction. Change in the diagnosis itself was almost never the endpoint. A trial can report an improvement without anything having changed about the underlying personality pattern.

The strongest single trial is recent. The MOAM trial, published in The Lancet Psychiatry in 2025, randomised men on probation in England and Wales who met DSM-5 criteria for ASPD and had recent aggression to either mentalisation-based treatment added to probation or probation alone. At 12 months, mean aggression scores were 90 in the treatment group against 186 in the probation-only group, an adjusted difference of -73.5 with a confidence interval from -113.7 to -33.2 and an effect size of 0.74. That is a large effect from a well-conducted multicentre trial, and it is one trial. Longer follow-up on offending was planned and should be read when it appears.

Everything else is weaker. Cognitive behavioural therapy added to usual care was tested against usual care alone in 52 men with ASPD and recent violence, and found no difference in physical aggression, with an odds ratio of 0.92 and a confidence interval from 0.28 to 3.07. Both groups reported less aggression a year later.

Contingency management, which rewards the absence of a target behaviour, produced the only statistically significant primary-outcome result in the entire Cochrane psychological review: a small improvement in a social functioning subscale in one trial of 83 people, rated low certainty. The measure was drawn from an addiction severity index rather than from antisocial behaviour.

Schema therapy showed no interpretable difference in reconviction. Dialectical behaviour therapy produced a suggestion of fewer self-harm days from a comparison involving 14 people, rated very uncertain. Impulsive lifestyle counselling, a drink-driving programme, and a post-release risk management programme all showed no difference from usual care. Democratic therapeutic communities, often cited in this context, had no randomised trial meeting the review's criteria at all.

No medication is approved anywhere for antisocial personality disorder. The UK's clinical guideline advises against routinely using drugs for ASPD or for the aggression, anger, and impulsivity associated with it, and recommends treating comorbid conditions on their own terms.

The Cochrane drug review rated the certainty of evidence as very low for every outcome it examined. The single positive signal most often quoted is phenytoin, an antiepileptic, which in one six-week trial of 60 aggressive male prisoners was associated with a mean of 0.33 aggressive acts per week against 0.51 on placebo. One unreplicated trial with very low certainty is not a basis for treatment.

The negative results are worth stating alongside it. Desipramine showed no difference in social functioning in 29 participants. Nortriptyline showed no difference in global functioning in 20. Bromocriptine showed no difference in 18 participants, and five of them dropped out within two days from severe nausea. Even a drug with a marginal benefit has to be tolerable to a group that disengages from treatment readily.

This is where the evidence is strongest, and it is not about adults. Parent training programmes reduce conduct problems in children with small to large effects depending on the programme and the measure. A meta-analysis of one widely used programme across 50 studies found an effect of 0.27 on disruptive behaviour across informants. A meta-analysis of parent-child interaction therapy across 23 studies and 1,144 participants found a standardised mean difference of -0.87 on externalising behaviour.

Callous-unemotional traits, the childhood features closest to the affective side of psychopathy, mark a harder subgroup. A meta-analysis covering 60 studies and 9,405 children found that treatment reduced disruptive behaviour about as much in children with these traits as in those without, but that the children with them started with more severe symptoms and finished with more severe symptoms. They improve without catching up. The direct effect of treatment on the traits themselves was null overall, and became detectable only in studies with parenting-focused components, where it was small.

The claim that early intervention prevents adult ASPD is plausible and not established. These studies measured conduct problems and callous-unemotional traits in children. Almost none followed participants far enough to count adult ASPD diagnoses. The reasoning that childhood is more malleable is sound, and the long-horizon trial that would confirm it has not been run.

Two problems undercut every result above, and they pull in opposite directions.

The first is engagement. People with ASPD leave treatment at elevated rates. Across psychotherapy trials for personality disorders generally, median non-completion sits near 37%. One study of 236 people in residential substance treatment found that those with ASPD who were there voluntarily dropped out more than any other group, while those mandated by a court stayed. Completer analyses therefore describe an unrepresentative slice, and mandated retention is not the same as engagement.

The second is age. Antisocial behaviour declines over the life course without any treatment, one of the most replicated findings in criminology. Long follow-ups of people diagnosed with ASPD find substantial minorities improving decades later: one cohort of 82 found 12% fully remitted and 27% improved, another of 45 men assessed an average of 29 years after discharge found 27% remitted and 31% improved. Any uncontrolled study that reports improvement over time is partly measuring this.

Together these mean a trial can show a drop in aggression that reflects the participants who stayed, the passage of time, and legal supervision, in any combination, before any therapeutic effect is counted.