Moral enhancement

Moral enhancement is the proposal to improve human moral dispositions — altruism, impartiality, aversion to causing harm — by biomedical intervention rather than by education or institutional design. It is the branch of Human enhancement whose advocates present it not as an optional improvement but as a response to a mismatch: a moral psychology shaped for small groups now controls technologies capable of catastrophic harm. Its critics reply that the psychopharmacology on offer is far too crude for the job, and that the interventions proposed would damage something essential to moral agency even if they worked.

§The argument

Ingmar Persson and Julian Savulescu set out the case in a 2008 paper and expanded it in Unfit for the Future.¹² The argument has four steps.

Human moral psychology evolved under conditions of small-group cooperation. It is biased toward kin and near neighbours, toward the immediate over the distant future, and toward acts over omissions. These biases were adaptive and are now liabilities.

Technology has made the harm a small number of people can cause vastly greater than the good a comparable number can do. Persson and Savulescu call the worst case ultimate harm: an outcome that permanently destroys the prospect of a worthwhile life on Earth. The relevant class of hazards is the one described in Existential risk and, for the biological cases, Dual-use research of concern.

Liberal democracies cannot solve this by surveillance and prohibition alone, because the number of people who must be prevented from acting is small and the means are increasingly distributed.

Therefore, they argue, cognitive enhancement without corresponding moral enhancement is dangerous, and research into biomedical moral enhancement is urgent. The cognitive enhancement they have in mind is the whole range covered by Nootropics, Genetic enhancement of cognition and Intelligence amplification, on the view that all of them increase capability without touching motivation. Thomas Douglas advanced a narrower and more defensible version in the same year: moral enhancement as the attenuation of clearly counter-moral emotions such as aversion to particular racial groups or impulses toward violent aggression, which he argued is permissible on almost any account of moral worth.³

1. 2008
   The proposal appears
   Persson and Savulescu argue that cognitive enhancement without moral enhancement is dangerous; Thomas Douglas independently defends the attenuation of counter-moral emotions.
2. 2010–2012
   Candidate agents identified
   Studies report that serotonin manipulation shifts harm aversion in hypothetical dilemmas, that oxytocin increases in-group favouritism, and that propranolol reduces implicit racial bias.
3. 2011
   The freedom objection
   John Harris argues that moral worth requires the capacity to choose otherwise, and that engineering good behaviour removes what makes it good.
4. 2012
   The God Machine
   Savulescu and Persson answer Harris with a thought experiment: a system that intervenes only to prevent great harms, leaving ordinary choice untouched.
5. 2014–2016
   Sustained criticism
   Robert Sparrow attacks the political assumptions of the programme, and Harris Wiseman's book argues the neuroscience cannot support anything the proposal requires.

§What would count

The literature distinguishes several targets, which are often conflated.

Motivation. Making people more disposed to act on moral judgements they already hold. This is the target Persson and Savulescu emphasise, on the view that akrasia rather than ignorance is the binding constraint.

Emotional disposition. Increasing empathy or fellow-feeling, or reducing aggression and out-group aversion. The candidate pharmacology is concentrated here.

Judgement. Improving the reasoning by which moral conclusions are reached. John Harris argues that this makes cognitive enhancement the only real moral enhancement, since better information and better inference improve moral decisions without touching motivation.

Behavioural constraint. Preventing specific acts. This is not enhancement of the agent at all, which is why the God Machine thought experiment — a system that intervenes only to block great harms — is presented as a limiting case rather than as a proposal.²

The bootstrapping problem

Any programme of moral enhancement must be designed by people whose moral views would then be propagated. If those views are mistaken, the intervention entrenches the error; if they are not, the argument for the intervention presupposes the moral knowledge it is meant to improve.

§The empirical basis

The candidate agents are borrowed from psychiatry and studied on laboratory tasks. None was developed for this purpose and none has been tested against real moral behaviour. Device-based routes are thinner still: reports of personality and impulse-control change after Deep brain stimulation are the closest thing to a demonstration that a neural intervention can shift how a person behaves toward others, and they are side effects of treatment for movement disorders, not enhancements. The transcranial methods surveyed in Non-invasive neuromodulation have produced nothing durable.

Serotonin. A single dose of the SSRI citalopram increased participants' reluctance to endorse harming one person to benefit several in hypothetical dilemmas, an effect interpreted as heightened harm aversion.⁴ Whether this constitutes moral improvement is contested: greater reluctance to act is not obviously better ethics, and the same shift would make some defensible choices harder.

Oxytocin. Often described in popular accounts as a trust or bonding hormone, it behaves in experiments as a parochial one. De Dreu and colleagues found that intranasal oxytocin increased favouritism toward in-group members and, on some measures, defensive hostility toward out-groups.⁵ For a proposal whose central aim is extending concern beyond the near group, this is the wrong direction. The wider oxytocin literature also has severe replication problems, and whether intranasal administration delivers meaningful quantities to the human brain remains disputed.

Propranolol. A beta-blocker reduced implicit racial bias on an association test without changing explicit attitudes.⁶ The result is small, measured on an instrument whose relation to behaviour is itself contested, and has not been shown to alter how anyone treats anyone.

Psychedelics. Supported psilocybin sessions entered the discussion later, on the strength of reported shifts in the personality measure of openness and in self-described concern for others. Those outcomes are questionnaires completed by people who know exactly what they took, which is the unblinding problem that limits the clinical literature on the same drugs; in people without a diagnosis that literature is small and largely uncontrolled.

The honest summary is that no biomedical intervention has been shown to make a person behave better in any consequential setting. Every effect in this literature is a shift on a questionnaire or a laboratory task, in a single session, in university volunteers.

Distance from the claim

The proposal concerns preventing catastrophic harm by a small number of highly motivated actors. The evidence concerns millisecond-scale changes in association-test scores among undergraduates. No argument has been offered that bridges the two, and the gap is not a matter of degree.

§Objections

Freedom to fall. Harris's objection is that moral worth depends on the agent's capacity to choose otherwise. An intervention that makes wrongdoing impossible, or merely much less likely by bypassing deliberation, converts a moral agent into a well-behaved mechanism, and it collides directly with the claim to mental self-determination set out in Morphological freedom and, in its legal form, in Neurorights.⁷ Douglas and others reply that emotions are not chosen either, and that removing a bias one would reflectively disown restores autonomy rather than reducing it. Harris's second point is harder to answer: if the enhancement acts on motivation rather than on judgement, the agent has no way to evaluate whether the new motivation is correct.

Moral pluralism. Deep disagreement about what morality requires is not obviously an error to be corrected. A programme that increases obedience, or conformity, or harm aversion is taking sides in disputes that are live and reasonable. Robert Sparrow argues that the proposal's political naivety is its central flaw: it imagines an intervention deployed by no one in particular, when in practice a state or a firm would decide the content. The objection converges with the bioconservative worry, stated most sharply by Leon Kass, that a technique for improving people presupposes a settled account of what an improved person is.⁸

Who takes it. The people whose dispositions the argument is most concerned about are precisely those least likely to volunteer. Compulsory moral enhancement is rejected by nearly everyone in the debate, including its advocates, which leaves the proposal aimed at a population it does not need to reach. Persson and Savulescu acknowledge this and argue for research now against the possibility of a future in which the calculus changes.

The science does not exist. Harris Wiseman's book-length critique argues that the neuroscience underdetermines everything the proposal requires: there is no moral module, the pathways implicated are involved in most of behaviour, and any agent potent enough to shift moral motivation would have effects across the whole of personality.

§Standing and outlook

Moral enhancement occupies an unusual position: it is one of the most-discussed proposals in academic bioethics and one of the least-supported by evidence. Two decades of argument have produced a substantial philosophical literature and no candidate intervention that a researcher would defend as ready for study in the relevant setting. Within Bioethics of enhancement it is the clearest case of a proposal whose ethical analysis has run far ahead of anything it could analyse.

Two developments would change that. The first is a pharmacological or neuromodulatory agent with a specific and durable effect on some component of moral motivation, demonstrated behaviourally rather than on a self-report instrument; nothing in the current pipeline resembles this. The second is a shift in the underlying risk argument. If Artificial general intelligence or engineered pathogens make the distribution of destructive capability as wide as Persson and Savulescu assume, the question of whether a crude intervention is better than none becomes practical rather than academic. Their critics reply that in that scenario the crude intervention would not work either, and that the effort belongs in institutions, verification, the precautionary regulation of the underlying technologies, and Differential technological development instead.

§See also

• Human enhancement
• Bioethics of enhancement
• Julian Savulescu
• Existential risk
• Nootropics
• Intelligence amplification
• Precautionary principle
• Differential technological development

§References