Early advocates in the moral bioenhancement (MBE) debate claimed that our atavistic moral hardware is ill-equipped for modern challenges, therefore humankind has an imperative to morally enhance itself [1]. Meanwhile, sceptics maintain that morality is too complex, too tangled in vast agential and socio-cultural processes, to be amenable to crude meddling at the level of biology [2, 3]. Are these latter theorists correct in claiming that MBE is patently unfeasible, or are there blind spots in the debate where the idea of MBE might have some purchase? This paper advances pragmatic considerations about the bio-targets of moral behaviour in regard to a concept that has been roundly neglected in the discourse: aggression. If we are to preclude (or endorse) proposals for MBE then we ought to establish sound reasons for doing so. However, paying due attention to contemporary aggression studies shows that efforts to rule out aggression as one such bid are unpersuasive or altogether unviable. Once standing assumptions are elucidated it becomes evident that attempts to dismiss aggression as a focus of enquiry are premised on faulty grounds and that, consequently, there is currently no basis for disqualifying aggression as a prospective target. I therefore posit that a scientifically-informed analysis of aggression is necessary before we can advance or discount this suggested enhancement. The first section relates what I term the ‘Collateral Damage Argument’ against aggression-reduction as a mode of MBE. I identify two strands of this argument: 1) the ‘baby and the bathwater’ objection; and 2) the ‘blunt instrument’ objection. The next section unpacks the assumptions of these objections to reveal two incompatible views. Then, I introduce a bimodal classification of aggression which is applied to demonstrate why extant non-feasibility arguments misfire. The final section advances a preliminary argument for establishing a normative differential in subtypes with a focus on reactive aggression. Ultimately, I argue that existing arguments against aggression as a target of MBE are untenable, and that recognising a specialist account opens potential space for MBE in targeting malignant subtypes.

The Collateral Damage Argument Against Moral Bioenhancement

While recent focus in the MBE debate has shifted to practical considerations [4], many bioethicists remain convinced that MBE, in any form, is not possible. These theorists make a common argument which has held sway in the literature and has gone largely unchallenged, at least in so far as it relates to aggression [2, 3, 5, 6]. I’ll call this the ‘Collateral Damage Argument’ against MBE [7].Footnote 1 Essentially, the argument claims that any biomedical attempt to modify human traits to improve moral functioning will have unintended harmful consequences that will outweigh the benefits of intervention. The standard method of this approach is to identify a potential target of MBE (such as empathy, altruism, or certain cognitive capacities) and show that, because the trait is entangled in behaviours both good and bad, attempts to modify it will have global consequences for moral functioning. Critics subsequently argue that because we cannot untether beneficial trait expressions from unsavoury ones, efforts to eliminate the latter will also destroy the former, whilst jeopardising other valuable capacities. Fundamentally, the Collateral Damage Argument asserts that any attempt at MBE will do more harm than good.

Although it is here misapplied, I do not mean to dismiss all applications of the Collateral Damage Argument – it may, in some cases, provide valid reasons for rejecting specific motions. Take recent research suggesting that particular traits are squarely dual-use in their moral orientation, and thus inappropriate candidate-targets for MBE. For instance, much has been written on the double-edged sword effect of empathy (and empathogens), unveiling empathy-inducing treatments as morally ambiguous, if not counterproductive [8,9,10,11]. While the case against empathy may hold up, issues arise when the Collateral Damage Argument is discharged more broadly. A number of bioethicists have applied its rationale to other proposed bio-targets – problematically, they have done so without convincing support [2, 3, 5, 6]. One of the capacities that has been treated as such, and therefore dismissed as a possible target of MBE, is aggression.

A General Problem for the Collateral Damage Argument

The Collateral Damage Argument faces a fundamental problem: it depends on folk psychological assumptions from which proceed false descriptive conclusions. Based on a priori conceptual views of aggression, claims issue in the absence of substantive support as to what is (or is not) possible with respect to biomedical interventions that might target moral capacities. The theorists proffering these claims thus make general assertions about aggression on the presumption that it is self-evident what aggression is, how it functions, and why it cannot be an appropriate target of MBE. Such approaches entail several operating conditions that conjointly serve to obscure a more veridical understanding of aggression and its bearing on MBE debates. First, they invoke and embed lay beliefs that fail to capture the complexity of certain subjective states and associated moral actions. Second, they neglect an established body of interdisciplinary research on aggression, which integrates rich epistemic perspectives from the fields of psychiatry, psychology, and the biosciences. As a result, the simplistic notion of aggression at play in the MBE debate is significantly out of step with a scientifically-current understanding.

This mischaracterisation gives rise to several complications. One corollary is the conflation of discrete psychological states into homogeneous types. Above reifying bogus conceptions of aggression, this obscures crucial differences between distinct subjective states, the understanding of which has consequence for present accounts. When discrete states are clumsily subsumed under broad affective categories, the result is a mismatch between certain normative actions and the states of which they are claimed to be a product. This tethering of disparate moral actions to loosely formed affective categories impacts considerations of treatments that aim to modulate particular traits. If given practices are wrongly characterised as motivated by aggression, then attempting to moderate such practices by targeting aggression will be a non-starter. Similarly, falsely attributing certain virtuous conduct the same psychological basis as vicious conduct leads to the unsafe conclusion that it is impossible to modify the latter without modifying the former. A further problem concerns the mechanistic processes of cognitive states linked to particular behaviours. Misrepresentation of aggression fails to discern a potentiality of mechanisms for different types of deleterious conduct and associated moral functions. Without an informed analysis of the processes that contribute to aggressive states and related actions there is no probative basis for determining whether, or to what extent, biomedical processes might be effective in influencing moral functioning.

To be certain, I am not endorsing wholesale MBE as assuredly viable. Rather, my aim is to demonstrate that extant arguments against one proposal, which have been taken largely on faith, are misguided. My critique is therefore directed at the Collateral Damage Argument against the feasibility of aggression-reduction as a MBE. I argue that it is premature to make this non-feasibility claim in advance of critical enquiry, and that by extension, aggression remains a legitimate research focus that deserves due regard prior to determination. Allowing for normative appraisals requires attention to the concept of aggression, as well as to the varied psychological states implicated in associated moral actions. Therefore, what is urgently called for in the discussion is a more granular representation of aggression. This compels an analysis that acknowledges prevailing epistemic structures, including recognising a diversity of aggression subtypes and the particular behaviours they might realise. Before exploring these features, it is worth reviewing the ways aggression has been discounted.

Two Objections

The Collateral Damage Argument with respect to aggression comes in two varieties. The first objection claims that biomedical efforts to moderate pernicious manifestations of aggression will unwittingly eliminate any associated moral goods. For example, theorists argue that while reducing aggression might lower the incidence of antisocial behaviours like violent assault, it will also destroy the capacity of persons to utilise aggression in the prevention of such behaviours [2, 3, 5, 6]. They therefore claim that attenuating aggression would be to discard the good qualities of aggression with the bad. I will refer to this as the ‘baby and the bathwater’ objection (henceforth, BBWO).Footnote 2 The second form of the Collateral Damage Argument I call the ‘blunt instrument’ objection (henceforth, BIO). This claims that biomedical interventions are too imprecise to moderate aggression in isolation without damaging other traits of distinct moral and/or subjective value [3]. Even if we can eliminate harmful aspects of aggression, this will come at too great a cost to the person and their moral life to be properly deemed an enhancement. Ultimately, both claims converge on the view that any attempt to realise MBE via aggression cannot be done without spoiling something of value. However, these objections face a compatibility problem with respect to each other: they present contradictory views of aggression that cannot mutually obtain. While this is an issue for proponents, I will argue that both objections fail on independent grounds, and that, as they stand, Collateral Damage Arguments should be rejected. To see this, it is necessary to detail each objection in turn.

The Baby and The Bathwater Objection

The primary claim against aggression comes in the form of the BBWO [2, 3, 5, 6, 12]. This type of argument is developed by John Harris in How to be Good, which offers an incisive critique of the MBE debate. However, one line of objection Harris raises against MBE warrants serious scrutiny. He states:

A very fundamental problem…is that the sorts of traits or dispositions that seem to lead to wickedness or immorality are also the very same ones required not only for virtue but for any sort of moral life at all ([2], 61).

Harris believes that substrata of moral actions are multifunction and therefore that the biological basis of certain pernicious behaviours cannot be modified without also afflicting those responsible for certain virtuous behaviours. In context, the argument holds that aggression is one such case as it is implicated not just in harmful conduct but in prosocial action too. This gives rise to what I call the ‘harm-prevention claim’: the belief that aggression is necessary to prevent harm to oneself and others in cases of assault, threats, and similar violations.

This apparent truism is advanced using a common example. Consider these comments from Harris, who fears that aggression-targeting interventions would inhibit “the ability to do violence where morally appropriate” and “reduce the sort of aggressiveness that leads people to defend themselves and others” ([2], 82–83). Harris is not alone in this claim. Lavazza & Reichlin [6] state:

[W]e may agree that it would be good to decrease aggression, generally speaking, but a morally wholehearted individual also ought to be able to react firmly to defend vulnerable subjects and not be overly submissive ([6], 3).

Similarly, Wiseman argues [3]:

[A]n impulse to do a good thing may well actually benefit by being enriched with a certain amount of aggression, again, such as intervening to prevent a hostile individual doing a tremendous harm ([3], 49).

To summarise, the harm-prevention claim holds that while aggression is responsible for vicious actions like assault, it is also vital in stifling such acts, or otherwise necessary for virtuous conduct. The conclusion is that diminishing violent tendencies will impair the capacity to employ meritorious uses of aggression. Critics thereby assert that aggression-reduction will not evince an overall moral improvement to the agent, therefore this proposed form of MBE is a dead-end.

The Blunt Instrument Objection

The second form of the Collateral Damage Argument I describe as the BIO. This is formulated by Harris Wiseman, whose emphasis on a “practical reality-first” ([3], 13) approach makes a gainful contribution to progress in the MBE discussion. That stated, his account contains some key inconsistencies that must be recognised if we are to conduct an honest appraisal of aggression-reduction as a prospective enhancement. The BIO recognises the reality that aggression can already be tempered by biomedical means ([3], 11). However, it claims that where procedures are able to treat problematic forms, an inevitable side-effect is that valuable independent traits will be blighted, and that in all but extreme cases this trade-off will be too costly:

[T]here can be a tremendous cost to using such agents…the costs to the personality, well-being, and agency of the individual in question are tremendous ([3], 11).

And similarly:

Whatever moral enhancement might be possible, it must be broad-scope and very limited in nature. With that, we have returned to the sledgehammer approach that is already available, so-called because such interventions can only be effective, at best, by making tremendously broad changes to the organism’s biology as a whole, creating numerous side-effects that might just as well diminish other moral powers ([13], 47).

Wiseman does acknowledge that some forms of aggression are plainly damaging and therefore rightly annulled at the price of certain moral goods. But according to Wiseman, these distinctly pernicious forms are limited to severe cases of psychiatric disorder. This amounts to a tacit denial of their existence within the general population – a condition that if true, would preclude their candidacy for moral enhancement. I will return to this point later. Even in cases where unacceptable aggression-types can be attenuated, Wiseman argues that this process will affect a necessary loss of capacities that contribute to normative goals:

In cases where we are dealing with an aggression unleavened by any orientation to a greater good, such persons could well benefit from anti-aggression medication, even if the cost is their ability to do the good things that aggression might facilitate. This being so, apart from very extreme cases, the point should be clear, good and bad impulses cannot be readily severed, since they are ambiguous. Our embodied nature defies this sort of atomic analysis, and knock-on effects from trying to remove “bad” impulses…may yet result in the loss of good too ([3], 49-50).

Unlike the BBWO claim (which Wiseman also makes), the BIO does not claim the moral goods that would be lost through intervention are specific products of aggression. Rather, it claims that procedures targeting specifically virulent forms can operate merely as blunt instruments by diminishing not just the unsought aggressiveness but, along with it, independent traits of distinct social and agential value. Essentially, Wiseman accepts that aggression types are amenable to biomedical treatment, but maintains this can only work in a sweeping, unwieldy fashion by simultaneously removing either:

  1. i.

    valuable manifestations of aggression (BBWO); or

  2. ii.

    other morally and/or subjectively valuable traits (BIO).

The fundamental claim behind both objections is that any attempt to modify aggression will inevitably result in damage to traits of substantive value.

Two Conflicting Views of Aggression

Having surveyed both versions, we can now discern a foundational issue of the Collateral Damage Argument: it tacitly alternates between varying conceptions of aggression depending on which objection is deployed. Specifically, the BBWO presents aggression as homogeneous, by assuming it is a singular phenomenon uniformly responsible for moral, nonmoral and immoral actions. The BIO, on the other hand, recognises aggression as a complex, heterogeneous phenomenon made up of various subtypes. To foreground this, we must examine these accounts in turn.

Review of the first objection reveals aggression construed as an endogenous subjective property – such as a trait, an emotion, a biochemical, or a psychological state. In the absence of explicit definition, we must rely on close reading to unwrap this latent interpretation. For instance, it emerges in Harris’s assertion that aggression “leads people to defend themselves and others” ([2], 83), and similarly in Wiseman’s supposition that moral tendencies require “a certain amount of aggression” ([3], 49). The suggestion is that aggression is a quality or capacity that exists internally in some biophysical form, and which provides the impetus for various sorts of moral action. However, this is not the only difficult presupposition the BBWO has about aggression.

Further, it assumes that aggression is a homogeneous construct. This reification of aggression as a uniform entity is observed in Wiseman’s claim that aggression can only be “clubbed down wholesale” ([3], 28), thereby dismantling the moral and neutral goods that it (allegedly) produces:

The fact is that aggression informs a huge number of various activities we perform on a daily basis, many of which are morally neutral (e.g., sports), and indeed in certain circumstances aggression can be a necessary evil, and in extremis a righteous moral good ([3], 49).


We fail to recognize the ambiguity of certain qualities like aggression and the immense plurality of ways in which something like aggressive impulses can be mediated, shaped, and transformed into such a wide variety of possible manifestations, some of which may be morally bad (attacking someone for no reason), some morally good (defending one’s family), and some completely morally neutral (becoming a feisty sportsman) ([3], 104).

Wiseman’s treatment of aggression here implies that the varied actions – whether moral, immoral, or morally neutral – all derive from the same singular source which is consistent in kind and only variable post factum. As I will show, this presents a serious complication for the BBWO. But first, let us notice the switch in the alternative view.

Shifting to the BIO, we witness a far more erudite conception. Paradoxically, it is here recognised that aggression is “not an easily defined, singular phenomenon” and that “[t]here are many, many kinds of aggression ([3], 105)”. Wiseman goes even as far as to distinguish the established subtypes of reactive and premeditated aggression, among others. Further, he acknowledges that these subtypes can be affected differentially by existing treatments. Yet despite evidence that different types of aggression can be altered in distinct ways, Wiseman maintains that interventions are incapable of isolating subtypes to make discrete changes that will not sabotage independent moral goods. Here surfaces a problem: observed in these remarks is a view of aggression that contradicts the earlier one. This more discerning account is applied in positing a new argument which is, effectively, an argument to the same end.Footnote 3 The claim is that even once we recognise aggression subtypes, and that some of these may be more perilous than others, and that it is possible to target these subtypes, it is nonetheless impossible to target them without destroying something of considerable value. The incongruity between this account and the earlier one has strong implications for non-feasibility arguments.

I have now shown that the Collateral Damage Arguments applies different conceptions of aggression. The problem is that these two conceptions are incompatible with one another: either aggression is of a singular type that varies only in degree across wildly divergent phenomena, or it is a multifaceted concept entailing various types that manifest diverse realisations – proponents cannot have it both ways. If they maintain the former (simple) view of aggression then they commit to the BBWO. Alternatively, if they hold the latter (complex) view of aggression then they must commit to the BIO.Footnote 4 From these two competing accounts emerges the obvious question: which one is correct? An answer requires something which has been notably absent from the MBE debate – turning to the scientific literature on aggression.

What is Aggression?

Although a cursory profile is all that can be afforded here, some basic distinctions are necessary if we are to move towards a resolution. The consensus view in contemporary psychology is that aggression refers to any human behaviour that is motivated by the intention to cause harm (to a person who does not wish to be harmed). This includes direct physical or psychological harm (e.g., striking or insulting someone), or indirect harm (e.g., spreading hurtful rumours). Consensual harm is not included because it is instrumental in the target achieving a desired end (e.g., surgery, masochism, body modification). Accidental harm is also excluded from the category of aggression as it is non-intentional. This definition marks a vital distinction from non-specialist usage, where ‘aggressiveness’ is often used synonymously with ‘proactivity’ or ‘assertiveness’. For example, a poker player might be described as ‘aggressive’, but this does not convey an intention to hurt other players. Similarly, although parents might be assertive in disciplining their toddler, they do not wish harm upon the child [14,15,16].

Aggression is further distinguished into subtypes which aim to reflect its heterogeneous nature. These are based on ‘form’ and ‘function’, where form refers to the kind of aggressive act, and function refers to the purpose of the aggressive act. Functional subtypes are divided into two broadly recognised categories: reactive aggression (also: hostile, affective, or impulsive aggression) and instrumental aggression (also: proactive or premeditated) aggression. Reactive aggression is ‘hot-blooded’ in that it is always accompanied by angry, hostile feelings and occurs in immediate response to a (real or perceived) threat or provocation. An example is a child punching one of their peers in response to being teased. Reactive aggression is associated with a lack of impulse control, and aggressors often express regret over their actions. Instrumental aggression is described as ‘cold-blooded’ as it refers to extrinsically motivated acts of harm that involve controlled, calculated actions associated with low emotional arousal. It is employed to achieve an ultimate goal, such as obtaining someone else’s property or boosting one’s social status. Forms of aggression can be physical (e.g., assault, property damage), verbal (e.g., insults, abuse), or relational (e.g., social exclusion). These can be either direct (where the victim is present), or indirect (where the victim is absent) [15,16,17].

For the purpose of this paper, I take the primary distinctions of reactive and instrumental aggression to be established in line with current consensus. However, it should be recognised that not all theorists accept the above framework. Recently, some have argued for the designation of another functional subtype termed ‘appetitive’ aggression [18,19,20]. Others argue that all aggression is to some degree instrumental [21, 22]. Although I think my view can readily accommodate these accounts, they are not strictly necessary for the purpose of my argument and so will not be the focus here. Whilst conceptual debates continue, the bimodal classification I have just outlined is enough for present purposes. What should now be apparent is that this invites crucial questions about the moral dimensions of aggression, suggesting that subtypes may exhibit differential normative qualities. Before exploring this aspect, we must review earlier arguments in light of the adjustment.

Problems with the Baby and the Bathwater Objection

Evidently, the view of aggression assumed by the BBWO is at loggerheads with the consensus account. We have seen that the objection employs a crude notion of aggression that fails to recognise its bimodal structure, instead rendering the concept as a singular endogenous state inherent in a variety of divergent behaviours across the moral spectrum. This misconception raises several issues. First, it results in an obvious contradiction for adherents. We might recall the thrust of the argument, that reducing aggression will not enhance moral functioning because it is necessary for moral behaviour, i.e., preventing harm (the harm-prevention claim). However, appreciating the technical view presents us with a stark incompatibility: if aggression is a behaviour enacted with the intent to harm, then preventing that initial behaviour (if otherwise realised) will ipso facto reduce harm itself. So, contrary to the harm prevention claim, an aggression-reducing intervention would – by its very nature – be a harm-prevention measure.Clearly though, the claim that aggression is necessary to prevent harm rests on a conceptual anomaly; as seen, claimants misinterpret aggression as an internal quality synonymous with anger, hostility, or assertiveness. A more charitable take might interpret this divergent usage as intending to convey the underlying qualities that lead to aggression, as opposed to the received meaning. However, I would like to bracket this difference for now to focus on deeper concerns.

A more substantial flaw in BBWOs is the assumption that aggression is a singular, homogeneous construct, rather than a diverse phenomenon consisting of multiple modes and mechanisms. While it appears to align (somewhat) with the vague notion of trait aggression, no clear definition of this usage is ever specified. But even if proponents were to embrace a trait view of aggression, they would first need to stipulate what exactly this means, and secondly, address the patent challenges faced by this account. For example, criticisms of trait aggression include that it is poorly defined, that it does not capture a coherent trait, and that it does not map onto accepted personality frameworks within psychology [23, 24]. In any case, BBWOs present a problematic rendering of aggression not as a heterogeneous cognitive-behavioural concept, but as a uniform, embodied quality which applies consistently across a spectrum of moral action.

This has two adverse consequences. First, it involves the dubious ascription of diverse subjective states and behaviours to the umbrella term of ‘aggression’. Second, it leads theorists to conclude that it is only possible for biomedical therapies to target aggression indiscriminately by reducing or removing ‘it’ as a singular entity, thereby making universal changes to supervening actions regardless of moral bearing. But once we identify the misconception this conclusion cannot be granted. If aggression is pluralistic this not only admits of distinct types, but suggests the operation of distinct substrata for these various types. Furthermore, there is no a priori reason to assume that discrete varieties share equal moral standing; there may be grounds to believe that certain kinds of aggression are more malignant than others. This being the case, it may be possible – in principle, minimally – for biomedical interventions to act on acutely hazardous modes of aggression (or rather, their biological correlates) whilst leaving more benign forms unaffected. If support can be established for this line of argument, it would prove detrimental for the claim that severe cases of aggression cannot be discretely targeted without coincident costs.

A further challenge confronts this objection once we adopt a specialist account. The conceptual shift calls into question folk psychological assumptions about the inherent presence of ‘aggression’ in prosocial behaviour. If the baby and bathwater objection is mistaken about the general nature of aggression, then related assertions about its necessity in salutary actions are likely mistaken as well. When we recognise the non-uniformity of aggression, this invites serious scrutiny over the validity of the overall claim. We might then ask whether aggression of any type is a necessary component of the diverse operations it purportedly facilitates, or whether some of these normatively gainful behaviours are the product of other neurocognitive mechanisms. I contend this is indeed possible, and that it is likely that inordinate value has been placed on the function of aggression in prosocial behaviour. Against these views, I here suggest that aggressive tendencies (at least of a certain kind) are less constructive than assumed, and furthermore, may in fact be counterproductive to social outcomes. Further questions arise over the epistemic basis for the unsubstantiated claim that aggression is essential for beneficent conduct like protective actions, and the assuredness of this posit (more on this later). Clarifying the assumptions behind these claims reveals a disregard for contemporary accounts, as well as for alternative mechanistic explanations for the aforementioned types of moral action. From a sheerer perspective, the BBWO crystallizes as an expedient bid to dismiss further inquiry into the prospect of aggression-reduction as a MBE. In any case, its purchase on the non-feasibility claim of aggression as a target of MBE is shown to be precarious.

It should now be evident that the BBWO is undermined by its failure to appreciate the multiform nature of aggression and its diversity of encompassing behaviours. This oversight raises open questions about the essentialism of aggression in normative conduct, and about the kinds of behaviours that certain subtypes may or may not realise. Whatever the case, it has been recognised that such arguments are premised on a view of aggression that is at odds with the conventional understanding. Of course, one might argue that it is unfair to simply declare the received psychological definition of aggression as the final word and hold proponents of the BBWO to that standard.Footnote 5 There are several responses to this. As to the privileging of a psychological account (as opposed to one from, say, evolutionary biology); simply, as the primary field concerned with individual human behaviour, psychology so far provides the most comprehensive treatment. That is not to assume it offers the final word – rather, aggression is an ‘open concept’; definitional debates persist, and conceptual revision should be expected with further empirical and theoretical progress. Nevertheless, as the most developed account to date, the psychological standard ought to be taken as the starting point from which MBE debates proceed.

Second, BBWO proponents could respond by rejecting the received view in favour of another conception. To do this, however, they would first need to specify an alternative, and next, justify why it ought to take precedence over the one I have outlined. Given that proponents have not distinguished any particular account, nor made attempts at conceptual clarification, it is only reasonable that one adopt the standard view until or unless a good alternative is provided. Note that this does not mean that the BBWO fails absolutely – there may yet be alternatives open that yield more success for proponents. As it stands, however, the BBWO should be resisted unless advocates can substantiate their claims with some form of tenable support, or show how they might accommodate the bimodal classification (or a feasible alternative). Nevertheless, this still leaves a second appeal in the BIO. I next examine some of its claims, and whether they are sturdy enough to rule out aggression as a potential target of MBE.

Problems with the Blunt Instrument Objection

The BIO marks progress in recognising a more precise view of aggression. However, it goes awry by drawing baseless conclusions about empirical matters related to the impact of biomedical tools targeting distinct subtypes. Positively, there are no grounds for accepting that aggression-reducing therapies will have the ineliminable side-effect of marring vital peripheral capacities – at any rate, adherents would first need to specify what these threatened species are, and explain how exactly they are surrendered due to aggression-reducing interventions. Even accepting the claim that MBE can only have broad-brush application, it does not follow that this would spoil valuable appurtenances. One might readily draw the opposite conclusion, by positing that coarse-grained MBE would improve peripheral capacities; reducing hostility could very well unencumber one’s cognitive faculties to facilitate better judgement and lead to overarching agential benefits, for instance. In any case, once subtypes and subtype-targeting processes are recognised, the question becomes one of a descriptive kind which cannot be satisfied with recourse to convenient presuppositions.

However, let us take one example that is of concern to proponents. Wiseman points out that SSRIs have been shown to decrease the tendency for reactive aggression:

[W]hile citalopram may well reduce reactive aggression, it does not reduce premeditated aggression. Indeed, in some cases various SSRIs, including citalopram, have been found to increase premeditated aggression ([3], 106).

Wiseman stresses that SSRIs do not lower premeditated aggression, and may actually increase it. This point might be an issue if it were conclusively shown to be true. But while some preliminary studies suggest that reducing tendencies towards reactive aggression has in some cases resulted in an increase in premeditated aggression, there is no evidence to indicate that this is a necessary relation, nor any prima facie reason why it must be the case. As it stands, the matter is undecided on whether decreasing reactive aggression using SSRIs promotes premeditated aggression, with recent meta-studies inconclusive as to whether or not SSRIs cause an increase in aggression (of any type) [25, 26]. An objection on these grounds is therefore premature; it presents a speculative claim as an empirical truth when the jury is still out.

For the sake of argument, though, suppose we grant that decreasing the proclivity towards reactive aggression using SSRIs does increase premeditated aggression. Even if this were the case, it need not be detrimental to current prospects. First, we can ask if this might not be an acceptable trade-off in terms of moral outcomes – perhaps it is better on balance to reduce the incidence of reactive aggression in the population at the cost of a periodic increase in premeditated aggression. If this is the case, SSRIs may actually support a move to MBE, at least from a utilitarian standpoint. Second, we might consider the possibility of subsequent procedures that could mitigate the spike in premeditated aggression caused by the initial intervention. It is not unreasonable to envisage a two-step process consisting of a primary intervention to target the correlates of reactive aggression, and a secondary one to target those of premeditated aggression, whatever they might entail. Furthermore, there may be other means besides SSRIs of decreasing reactive aggression that would not have the unwanted side-effect of increasing premeditated aggression (e.g., genetic editing, neuromodulation, psychedelic-assisted therapy). Additionally, discounting MBE prospects based only on extant therapies fails to consider the likelihood that advances in genetics, neuropharmacology, and so on, will greatly improve current methods to make targeted interventions far more precise. Without venturing any such possibilities, the BIO urges us to abandon further enquiry and dismiss aggression as a potential target of MBE. With that, we have returned to pretentions that discount aggression without a proper assessment of its intricacies.

While I have pointed to some issues with the BIO, an additional word is necessary before concluding this part of the discussion. It must be stated that Wiseman’s criticism of MBE does not hinge on the Collateral Damage Arguments thus far outlined. In fact, Wiseman does not rule out MBE definitively. Rather, he offers a multi-level case that raises concerns around bio-reduction and bio-primacy which conclude that the possibilities for MBE are, at best, trivial (the ‘ambiguity/social context argument’; the ‘systems biology argument’; and the ‘biopsychosocial argument’) [3, 13]. I can neither detail nor address these in sufficient detail here. However, I will offer a brief comment, and take it that the remainder of the paper forms a partial response.

A broad concern about MBE proposals is that they neglect the moral ambiguity of actions and action-producing capacities. The moral standing of an aggressive act and its respective biophysical causes is context dependent; what is good in one environment may be bad in another, and vice versa. Similarly, no matter how finely we slice the biological machinery, we cannot point to any component part and attribute it a normative value, so the argument goes.Footnote 6 These points are true, strictly speaking. I agree with such views in so far as biological influences are not the only things that matter to moral functioning, and that improving morality is optimally achieved with a well-scaffolded biopsychosocial approach. But these details do not preclude the point that some human tendencies, in greater or lesser degree, are generally considered better or worse than others from a moral perspective. If that were incorrect, pharmacotherapy for antisocial behaviours would not exist. Positively, there are some biologically-mediated capacities and dispositions that are almost wholly unidirectional in their moral bearing, and which may be amenable to biomedical intervention.Footnote 7 In the final sections, I aim to illuminate one such capacity.

Consequences for the MBE Debate

What happens when we give up lay assumptions in favour of a technical understanding? Acknowledging a multidimensional view of aggression raises unexplored issues for the MBE debate. First, there is the question of whether different kinds of aggression share equal moral standing: perhaps particular types are more hazardous than others, or certain aggressive functions do serve some altruistic aims. We might then ask: how does aggression partition along moral lines? It may be that, say, reactive aggression has less normative worth than assumed, and that instrumental aggression is more propitious in terms of social outcomes. I next suggest this is a very real possibility. There are provisional reasons for thinking that reactive aggression is typically vicious and rarely (if ever) morally advantageous.Footnote 8 Certainly, this is already recognised in pathological cases, but we ought not make the reckless assumption that diagnostic criteria mark firm moral boundaries. In other words, this raises questions about latent and undiagnosed dispositions to reactive aggression in the general population. If my concerns are accurate, such cases plainly qualify as candidates for MBE. Next, we should consider whether purported ‘aggressive’ acts of rescue, protection, and self-defence of the socially constructive kind are not better understood (some of the time, at least) as rational, goal-directed actions that involve justifiable, teleological uses of force, rather than aggression per se. Neutralising aggression may demand acts of physicality, but this need not entail intentions of harm-infliction. This therefore opens a space for investigating the separation between harm-directed actions and acts of harm-prevention. Lastly, a refined account of aggression invites examination of the biological mechanisms that underpin functional subtypes. Given their radical differences, it is likely that each displays independent biomarkers [27].Footnote 9 This being the case, it raises the possibility for biomedical processes to target aggression subtypes differentially. If there are plainly deleterious subtypes that can be discretely attenuated via biomedical means, then given these present in non-clinical cases, such procedures would offer one potential route to MBE. For this to be feasible, a first step would require identifying normative differences between aggression subtypes.

The Moral Problem of Reactive Aggression

One form of aggression already clinically recognised as problematic is chronic impulsive aggression. Characterised by sporadic violent or hostile behaviour that is excessively disproportionate to the provocation, the DSM-5 describes Intermittent Explosive Disorder (IED) as “recurrent behavioral outbursts representing a failure to control aggressive impulses” [28] that are manifested by verbal or physical aggression which is neither instrumental nor premeditated. Positive IED diagnosis requires the patient exhibit a consistent pattern of explosive aggression over a specific time period (twice weekly for at least 3 months for Type-A1, and at least three times over a 12-month period for Type A2) [28]. As medically recognised by this condition, such behaviours are clearly damaging. However, framing problem-cases of reactive aggression in these limited terms raises a concern that is of significance to current MBE debates.

The behaviours implicated in IED are surely troublesome, but this is true regardless of whether they are chronic or not. Frequency does not bear upon the force of a moral act itself, and a single episode of impulsive violence has the potential to cause devastating harm, including death. This is an issue when we consider that one could be (genetically or otherwise) prone to explosive aggression yet fail to meet the requisite frequency of the diagnostic standard. Despite a disposition towards runaway violence, such an individual would not necessarily be identified as disordered, simply because of a relatively low incidence of aggressive acts. Thus, it may be the case that those with an erratic or inconsistent history of explosive aggression are classified as non-clinical (i.e., general population). Of course, it is recognised that the DSM-5 does make provisions for those who do not meet the full range of symptoms for particular disorders, and such cases may warrant a diagnosis of ‘unspecified’ or ‘other specified’ (previously ‘NOS’). Certainly, grey spots and borderline cases can be expected in the diagnostic testing of mental health, but there will also exist subsyndromal and prodromal cases that fall beneath the clinical standard, despite a presence of one or more respective symptoms. Furthermore, there may be individuals who have a genetic liability to explosive aggression but without any history of such behaviour, simply because of insufficient epigenetic or environmental triggers.

It is reasonable to ask why we should be concerned with such cases – one might argue that if a person is without incident then there is no problem to either warrant or justify intervention. However, simply because cases evade clinical recognition does not assure they are not morally problematic. We should be concerned about persons with an unrealised disposition towards reactive aggression for the same reason we regulate the treatment of dangerous goods like explosives, toxic chemicals and other volatile substances: an individual that, under the right conditions, would be given to bouts of frenzied violence poses a potentially fatal risk. Many bioethicists already endorse gene therapies for the elimination of serious illness (e.g., cancer, cystic fibrosis, Huntington’s disease) which have the potential to kill one person (i.e., the carrier). Fatal risk is clear justification for biomedical intervention in these cases. Now imagine for a moment there is discovered a biopredictor that increases one’s disposal to impulsive aggressive behaviour – let’s call it the ‘IED profile’. Whereas terminal diseases will only result in one death if realised, the IED profile could result in the deaths of many (through acts of violence perpetrated by the carrier). We should therefore be concerned about its presence in the population. It is not difficult to imagine the naive IED candidate – an otherwise ordinary person, who, on a particularly bad day, is beset by a series of escalating troubles which serve to compound their frustration to the point of triggering an unprecedented episode of violence. This might culminate in any form of grievous act, like lashing out at a spouse, punching a co-worker, or smashing a nearby object. Worse, it could result in the death or serious injury of one or many innocent persons. One might object that such instances are irregular, but even if this is true it does not detract from their seriousness.

Obviously, it would be misguided to think that the etiology of aggression can be traced solely to biological causes. But there is no dispute that biology plays a contributing role in aggressive behaviour. For instance, the first genome-wide DNA methylation study on IED recently identified novel pathways that may be associated with impulsive aggression, with the authors suggesting this could lead to preventative treatment for individuals in whom impulsive aggression might not be clinically recognized [29]. That stated, behavioural phenotypes are highly polygenic, meaning that traits are determined by combined fractional effects from a multitude of genes, rather than by a few easily-identifiable targets [30]. In fact, a recent Genome-Wide Association Study (GWAS) identified over 500 genetic loci for externalizing behaviours, with researchers determining a polygenic score that accounts for just 10% of variance in externalizing phenotypes [31]. Additionally, phenomena such as epistasis, pleiotropy, and gene-environment interactions further complicate the process of identifying risk profiles for reactive aggression.Footnote 10

With that stated, given that there already exist a range of treatments for moderating harmful impulses in aggressive individuals, further reducing such conduct may be more a matter of fine-tuning risk prediction and early intervention methods than a need to seek novel therapies. In this case, advances with respect to biodata and bioinformatics may benefit predictive modelling to help identify risk profiles and facilitate preventative measures before aggression manifests. One possible avenue here is further genetic research on impulse control, with some suggesting that reactive aggression shares a strong association with deficits in self-control, self-regulation, and emotional stability [32]. Another suggestion is the use of neurotechnologies for affective priming in preclinical subjects at risk of developing conduct and/or antisocial behavioural disorders [33]. Risk prediction and treatment need not take the shape of invasive medicine – social and psychological mechanisms will likely prove supportive – but, in any case, we should remain open to potential biological levers of aggression, however minimal they might initially seem.

Does Reactive Aggression Serve Any Good?

It might be accepted that reactive aggression exists outside a clinical scope but argued that it holds some moral worth that warrants its preservation. Of course, it is possible to conjure hypotheticals where reactive aggression is able to serve some moral benefit. But I argue that in almost any such instance, we can substitute reactive aggression for its instrumental counterpart without any loss of the good. At this, one might be inclined to question the purpose of doing so, if we are simply able to equate the two subtypes as interchangeable. The difference concerns the risk inherent in reactive aggression. Consider three different bystander interventions to the scenario of a random attacker assaulting an innocent person on the street:

  1. 1.

    Physically restraining the assailant with minimal force until police arrive.

  2. 2.

    Striking the assailant repeatedly, knocking them unconscious, resulting in serious injury.

  3. 3.

    Striking the assailant repeatedly, knocking them unconscious, then kicking them whilst unresponsive, resulting in life-threatening injuries.

The bystander response in scenario 3 is a very clear case of unrestrained reactive aggression, and an act of excessive force constituting a gross moral infraction. Even if we diminish the extent of violence (scenario 2), the action is still disproportionate to the threat, and well beyond the necessary preventative measures. Nonetheless, this response is likewise indicative of rash, spontaneous behaviour that carries with it significant risk of harm. Scenarios 2 and 3 are thus illustrative of the immanent liability bound up with reactive aggression. In contrast, whether the moral imperative or an act of supererogation, we would consider bystander response in scenario 1 to be the ideal. It is not a hostile, affective reaction but a rational response directed at preventing harm. It is arguably a case of instrumental aggression. I say ‘arguably’ not because it is debatable whether this qualifies as an instance of reactive or instrumental aggression – the act is clearly teleological – but because it is debatable whether such an action even qualifies as aggression.

The Non-Necessity of Aggression in Beneficent Action

It is not given that aggression, of any type, is an essential component in the realisation of moral goods like acts of rescue, protection, and self-defence. As earlier observed, we must be careful of folk psychological views that conflate formal definitions with independent concepts like anger, proactivity, and assertiveness. Similarly, we should not confuse actions commonly perceived as aggression with rational, goal-directed actions that involve justifiable uses of force. To illustrate the difference, take the act of saving a drowning child: despite employing a serious degree of physicality to avert an imminent moral wrong, it is not ‘aggressive’. The same can be true of intervening to prevent violent assault; while achieving a morally desirable end might require the use of force, it need not involve aggression. A police officer subduing an attacker (in minimal terms) is an instance of rational force – a teleological action necessary for the officer to perform their duty in upholding the law. If, however, the apprehending officer proceeds to taser the perpetrator once restrained, that is an instance of aggression (specifically, violence).

Neutralising aggression may demand physical acts that involve resisting, restricting, inhibiting, suppressing, and so on, but it does not demand harm-infliction. Just as a fire is not extinguished with fire, harm is not countered by harming. Recall that aggression concerns intentions to harm. This considered, there is nothing to indicate that motives towards harm are an essential element in protecting oneself or others. The intention in such cases is very plainly not to cause harm to the instigator, but to prevent harm from being done to the victim. Physical force against someone’s person is therefore not tantamount to violence; only when force is enacted with the purpose of damaging does it constitute aggression, properly understood. This does not mean that aggression is never present in protective acts; as we saw in the earlier scenarios, one may go ‘too far’ by overstepping the boundaries of what can be considered reasonable. Similarly, reactive aggression might serve in strictly limited circumstances, but in any case, its concomitant risk far outweighs any goods. We have already seen that reactive manifestations are hazardous in such contexts, and this detail simply reinforces the previous point: if aggression is necessary to realise certain moral goods, this is optimally achieved via comparatively tenable instrumental forms.

MBE and Reactive Aggression

The latter sections have aimed to demonstrate the moral asymmetry of aggression subtypes, with attention to the inherent risk of reactive aggression as compared to morally ambiguous instrumental forms. I have suggested that reactive aggression is plainly damaging and that reducing it would, all things considered, evince a moral enhancement. Further, it has been shown that aggression of any kind is non-essential in the realisation of moral goods like protective actions and acts of rescue. The burden therefore rests on those who object to aggression as focus of MBE to demonstrate the worth of reactive aggression and present good reasons for why it ought to be retained. I am doubtful that a convincing case for this position can be made, and suggest that this is because reactive aggression offers little value worth keeping. Until and unless such a case is presented, attenuating reactive aggression thus offers an in-principle mode of MBE.

To be clear, we must be critical of the assumption that injurious manifestations of aggression are limited to psychiatric cases and thereby within the purview of existing therapy. The issue is not simply that IED-qualifying behaviours warrant legitimate mental health treatments (as is already the case), rather that there are common iterations of reactive aggression which are not captured by any current psychiatric diagnosis. Whether addressing these cases should be regarded as ‘treating mental health problems as targets for MBE’ as some have suggested is up for debate [3]. On one hand, we could argue for an expansion of diagnostic criteria, in which case such practices might preselect for clinical status and eventually qualify for treatment. On the other hand, such phenomena may in fact signal pervasive yet unrecognised action-tendencies in a sizable portion of the general population. In addition to the everyday examples of reactive aggression earlier mentioned, there are studies suggesting this latter proposal may have credibility, and that these types of harmful conduct are hardly uncommon [34,35,36]. In this instance, unless medically indicated, biomedically addressing ‘ordinary’ cases of aggression would fall outside the bounds of conventional treatment and positively merit enhancement status. While there are thorny questions about medicalization, ‘concept creep’ and ‘diagnostic inflation’, there is a similar need to balance these concerns by scrutinizing the entrenched norms and attitudes that can enable harmful everyday behaviours to proceed unchecked [37,38,39]. These are considerations that require far more discussion than can be had here. At any rate, the general cases I refer to are at present not clinically recognised. Although the wider point is concerned with harm reduction (rather than category concerns), by a customary understanding of the treatment/enhancement distinction it follows that targeting these cases, as it stands, would meet the classification of enhancement [40,41,42].

It might be argued that this is to commit to a thin notion of MBE. Whether one accepts that (further) reducing reactive aggression with early or pre-emptive treatment rightly counts as MBE will depend on a number of factors, such as their view of the treatment/enhancement distinction, and the definition of MBE to which they subscribe. This is all well-trodden ground, and the minutiae of such debates need not be rehearsed here. Some advocates of MBE might find this conclusion unsatisfying. I suspect that certain critics will view it as further confirmation of the limits and import of the MBE proposal. However, accepting the broad argument, its weight is contingent upon the cost and frequency of reactive aggression in the wider community. As I and others have already suggested, this type of behaviour is more pervasive than we might assume. Determination of relevant constraints and possibilities therefore requires that the preliminary efforts advanced here be followed with further enquiry.


This paper has argued that present objections to aggression as a candidate-target for MBE are untenable, and resultantly, that it may be possible to etch out a space for MBE in attenuating reactive aggression. I began by introducing the general problem of non-feasibility arguments, and then reviewed two strands of the Collateral Damage Argument in the form of the BBWO and the BIO. The next section applied the received classification to reveal discrepancies within the MBE debate in showing objections to be premised on non-viable views of aggression as either 1) a singular endogenous quality; or 2) a capacity that cannot be discretely surrendered without countervailing damage. With this, I made normative comparisons of subtypes to highlight the moral ineptitude of reactive aggression. In concluding, I have endeavoured to show that extant arguments in the MBE debate are premature in dismissing aggression as a research focus, and therefore that aggression ought to be reconsidered as a legitimate target of interest. While the feasibility of reducing harm with biotechnology is ultimately an empirical question, once we update the concept of aggression to be scientifically current then it becomes possible, in-principle at least, to identify one viable target for MBE. If we can mitigate harm with the aid of biomedical interventions, we should. A suggestion is that proceeding work aim at mapping the normative standing of subtypes in the context of functional settings to further assess the value differential of reactive and instrumental forms (among others). Feasibly, forward-looking efforts will be better placed to navigate the moral significance of aggression subtypes and further trace out their biopsychosocial roots.