2.1 Acceptance-based consequentialism

Versions of RC deploy different strategies to rebut the demandingness objection. For example, Hooker’s RC, ideal code RC or ICRC, includes two ‘mechanisms’: (i) the acceptance filter and (ii) a limit to the disaster condition.

The first mechanism works as follows: which ideal code is optimific depends on the benefits that arise from its acceptance by most individuals (90%) in a certain population (Hooker Reference Hooker1998, Reference Hooker2000, Reference Hooker and Portmore2020).Footnote ¹⁰ More specifically, the ideal code for a group is the code that would produce the most overall well-being when accepted by most members. Crucially, ICRC stipulates that members of the relevant group are ordinary people, that is, people not equipped with super-human or saintly moral dispositions and capacities (e.g. the capacity to empathise with everybody at the same level). So, Hooker suggests, it is only to be expected that the benefits of the large acceptance of an ideal code for ‘saints’ may not make it up for the costs of having ordinary persons accept it.Footnote ¹¹ The idea is that it might take such an exorbitant effort to ensure that the majority of a group of ordinary human beings acquired certain saintly dispositions (if it were even possible) that a less demanding code would turn out to have a greater net benefit overall.

Of course, were human agents different, a saintly code would perhaps be the optimific one. However, Hooker suggests that, in calculating which code is optimific, we should not consider the expected benefits of implementing such a code among tabulae rasae capable of moral sainthood, but instead among ‘ordinary’ human agents with a variable but limited amount of moral plasticity. In a recent exposition of his view, Hooker uses such terms as ‘natural inclinations’ to refer to aspects of our nature that should be considered when calculating the costs of implementing an ideal code (Hooker Reference Hooker and Portmore2020, pp. 450–451). From his discussion, it seems that these relevant human inclinations include certain dispositions to be partial towards those we care about, as well as instincts of self-preservation and self-interest, at least to a certain degree. So, in an overall calculation of the expected value of a moral code’s widespread acceptance, a code that strongly contrasts with these ‘natural inclinations’ is unlikely to be optimific.

Hooker also claims that any plausible version of consequentialism (and possibly any plausible moral theory) should take into account how much goodness some deviations from the prescribed code may produce. For instance, ICRC may imply that we have to commit a certain amount of time to helping others but, when non-compliance is higher than 10%, we may have to commit more than the rules would normally prescribe, at least when some additional sacrifice would produce a significant amount of overall benefit. The magnitude of these additional obligations is regulated by a requirement to amend the rules when doing so would ‘prevent disaster’ (Hooker Reference Hooker2000, p. 98; cf. Brandt Reference Brandt1992, p. 151); however, Hooker believes that this requirement should be restricted as well. This thought brings us to ICRC’s second mechanism for curbing its alleged demandingness. Hooker claims that internalising a code that includes an unrestricted ‘preventing disaster’ requirement would be extremely costly – so much so that it may outweigh the benefits of including it. Deploying Cullity’s (Reference Cullity1995) distinction between aggregative and iterative cost, Hooker suggests that ICRC would include this rule or principle:

The last part of this rule is what would limit the demandingness of the ‘preventing disaster’ requirement: the costs of beneficence on any specific occasion, properly calculated, include the aggregated costs of what the agent has already done, and thus not solely the costs to the agent of a single action on a single occasion (cf. Dougherty Reference Dougherty2017; Forcehimes and Semrau Reference Forcehimes and Semrau2019; Murphy Reference Murphy1993). The idea is that small costs may add up to an unreasonable one during a lifetime. Thus, in calculating the costs of a particular action – even those aimed at preventing some disaster – we ought to include all the other sacrifices the agent has already made (Hooker Reference Hooker2000, p. 167; cf. Thomson Reference Thomson1971).

2.2 Teaching-based consequentialism

Tim Mulgan’s teaching-based RC identifies the Ideal Moral Outlook (IMO) through a process of mapping ‘moral outlooks’ onto possible futures. First, moral outlooks are ranked according to the expected value of the consequences if we tried to teach them. Among all the outlooks mapped into possible futures, the best is the one that maximises the total expected well-being, after the costs of teaching it to the next generation are considered (Mulgan Reference Mulgan2017, p. 291). A related RC theory (‘Ideal Moral Outlook Consequentialism’, or IMOC) would suggest that ‘the right act is an act that would be performed by someone who had internalised the ideal code’ (Mulgan Reference Mulgan2017, p. 291). According to Mulgan, this formulation ‘sets aside’ the costs of changing existing moral beliefs, which would otherwise make some moral outlooks too demanding.

The main mechanism this version of TB-RC offers to curb the demandingness of the ideal code or outlook is similar to ICRC’s: as there are limits to what the next (or any) generation of normal human beings can learn, the ideal outlook would not comprise rules that are too demanding. In fact, the reasoning goes, attempts to teach excessively demanding moral codes may require so great an effort and so many resources that less morally ambitious codes would bring about more well-being overall.

3.1 Problems for ICRC and IMOC

ICRC and IMOC may turn out to be demanding for different reasons. For example, given the ambitions of Mulgan’s IMOC, this theory may turn out to be demanding because it implies that certain forms of partiality are forbidden, or because it may require agents to maximise the overall good ‘excessively’. In fact, although Mulgan claims that the learnability of certain outlooks imposes limits that curb the view’s demandingness, the ideal outlook is still supposed to be justifiable to people living in ‘broken futures’ – credible scenarios in which basic resources have become scarce (Mulgan Reference Mulgan2015, Reference Mulgan2017, Reference Mulgan2018; cf. Mogensen Reference Mogensen2021). We may understand these scenarios – for example, a possible future in which climate change has made water and food scarce – as states of permanent near-disaster or emergency (Gardiner Reference Gardiner2011). Now, it seems plausible that the practices we presently justify by our strong partial preferences for ‘conspicuous consumption’ are not ones we could imagine justifying to people living in futures of scarcity generated (at least partly) by such practices. In fact, in his recent work, Mulgan seems to accept that, given the risks humanity faces in the near future, significant aspects of our partiality for the present generation may be unjustifiable. It seems plausible that Mulgan would not think that forms of strong generational partiality could be justifiable to proximate future generations living in conditions of scarcity resulting from our actions.

In what follows, I will assume that the imaginative exercise of justifying our practices to people living in broken futures reveals the degree of reasonable partiality which a given theory tolerates, at least at an intergenerational level. The limits to what is teachable to individuals, on the other hand, constitute a mechanism that would limit both of the ways in which demandingness may vary. In light of these considerations, IMOC may face the demandingness objection head on: it would say that if the best moral theory turns out to be demanding in scenarios involving broken futures, then so much the worse for common-sense standards that are faulty and lax enough to allow us to be strongly partial towards the present generation – see the conclusions of this paper for an elaboration on this idea. Still, although Mulgan recognises that his theory may depart from common-sense morality, at least if common sense justifies a certain unjustifiable degree of partiality towards the present generation, he claims that his IMOC would not deviate too much from what common-sense morality seems to imply (Mulgan Reference Mulgan2017, p. 298). The reason, again, is that the IMO would have to be taught, and human beings are not so indefinitely pliable that they will learn any sort of behaviour. As such, the IMO would at least involve teaching moral dispositions such as honesty, generosity, aversion to murder, and so on (Mulgan Reference Mulgan2017, p. 298). However, critics of RC may reiterate their point by claiming that, at the least, ambitious versions of RC (such as Mulgan’s) would indeed require a demanding form of (intergenerational) impartiality.

Another problem for ICRC and IMOC is that, when the alleged limits to teachability and acceptability are considered more carefully, these mechanisms may not deliver the expected results: they might not significantly limit the demandingness of their respective codes. Some examples suggest that human beings really can be made to believe, and do, what would appear to be against their interests – let alone against their partiality and, arguably, in favour of maximising even their own good. Novice Buddhists are taught an impersonal ethics that is based on the metaphysical belief that there is no underlying and enduring self (Goodman Reference Goodman2009; Siderits Reference Siderits2007). According to several Buddhist schools, personal identity is a sort of illusion: it is literally untrue that you or I persist over even a day. This belief may motivate a more impersonal, and thus impartial, approach to ethics (cf. Parfit Reference Parfit1984/87). This is hardly the most extravagant metaphysical or moral belief that is held by a considerable number of people. After all, if atheists are right, the vast majority of humanity is deluded about a central aspect of their lives.Footnote ¹²

People have been made to believe not only counterintuitive metaphysical claims, but enact moral practices that are apparently self-denying or even morally depraved: the long list of atrocities includes the obligation to perform human sacrifices, the legitimacy of killing innocent Jews, the moral legitimacy of mass killings for ideological or utopian reasons, suicide attacks, witch hunting and the Inquisition. In addition, it is at best unclear whether learning these beliefs and the practices based on them required significant amounts of effort, other than perhaps starting at a very early age.

It may be argued that moral transformations – such as one involving a reduction of generational partiality – do not involve merely the transmission of beliefs, but also the cultivation of certain intellectual capacities. In his recent work (Reference Mulgan2025), Mulgan discusses at length, in the context of a thought experiment in which humanity would be facing extinction in two hundred years, the feasibility of multigenerational projects involving moral transformation. More specifically, Mulgan wants to defend the claim that certain multigenerational projects aimed at reorienting certain moral traditions and practices are feasible – even in the context of his thought experiment. Now, while conventional moral progress involves expanding existing moral frameworks or improving compliance, more radical moral transformations require also fundamentally reimagining moral concepts. In describing the intellectual skills required for moral transformation, Mulgan introduces two distinctions, the first between ‘imaginativeness’ (i.e., the ability to recognize or invent new possibilities) and ‘imagination’ (i.e., the ability to visualise or entertain objects or states of affairs) (Mulgan Reference Mulgan2025, p. 204), and the second between the morally imaginative (i.e., someone who invents new ethical possibilities) and the morally creative person (i.e., someone who puts new ethical possibilities into practice) (Mulgan Reference Mulgan2025, p. 205). Mulgan also argues that moral imaginativeness and creativity – which are central to the kind of moral transformation required in certain multigenerational projects, including those reducing generational partiality – are inherently collective (Mulgan Reference Mulgan2025, pp. 206–207). In fact, the application of these skills requires an inherited foundation in existing moral concepts developed by others, interpersonal justification and debate, social validation of innovations, and supportive institutional frameworks.

Assuming that the above account is largely correct, we may argue that moral transformations involving, for example, an increase in intergenerational impartiality would require considerable costs only in those contexts in which the prerequisite moral concepts or institutional framework are completely or significantly absent. If the costs of such changes are relevant in considering which code is optimific, then a code requiring such changes may be discouraged, and thus the resulting moral prescriptions of the code may not be too demanding after all. However, in those contexts in which the moral changes required to assume a more impartial ethics do not necessitate significant costs to be taught, then the relevant code that would turn out to be optimific may turn out to be the one prescribing, say, a higher level of impartiality. In general, it does not seem that, at least in some contexts, certain dispositions towards partiality would be extraordinarily difficult to replace. In these contexts, it therefore seems questionable that teaching a new generation to be more impartial would be particularly costly or difficult to reconcile with human nature. Different contexts, on the other hand, may require the introduction of certain educational practices aimed at improving skills such as moral imaginativeness. Such an introduction may turn out to be considerably costly. Whereas this may be true in some societies, it seems that most liberal societies do have the educational practices and structures required to foster certain skills conducive to more intergenerational impartiality. If this is correct, then after all IMOC or even ICRC may still prescribe moral codes that may seem demanding to common sense (now) – at least in those contexts in which their teaching would not be comparatively too costly to implement.

To summarise the main gist of this second line of objection: it is at best unclear whether it would be that costly to have the next generation accept or learn a more impartial ethics; with some exceptions, vast numbers of human agents seem receptive to many metaphysical and moral beliefs that are seemingly counter-intuitive and even self-denying. If it turns out not to be significantly costly to accept or learn an impartial and thus (apparently) demanding code, then the optimific code ICRC or IMOC prescribes may turn out to be one that is demanding by the present standards of common-sense morality.Footnote ¹³

In general, the previous objections both show that the limiting mechanisms may fail to ensure that the forms of RC under consideration will not prescribe codes that are too demanding – demanding according to some presumed standards of common-sense morality. However, there is another, more fundamental problem with the proposed mechanisms: neither ICRC nor IMOC provides sufficient reasons for believing that these mechanisms align with what can reasonably be considered the proper limits of our impartiality and of our obligations to maximise the good. For instance, we are not given sufficient or at least explicit reasons to believe that our ‘natural inclinations’ against accepting a code, our resistance to learning an outlook, or the costs of implementing a successful moral education are the right sort of mechanisms or filters to help us judge whether prescriptions, obligations and so on are aligned with proper limitations on our impartiality, or on the maximisation of the good. In other words, we lack a good explanation of why reluctance to accept or learn a moral code would align with what we should properly regard as too demanding.

One approach to provide an explanation is to consider how limits to impartiality and maximising in ICRC and IMOC might be explained by the mechanisms that are supposed to align with them. For example, consider again IMOC. Suppose that the resulting IMO would include a certain limit to impartiality – that it is permissible, when the resulting loss of overall well-being is minimal, to dedicate part of one’s spare time to oneself rather than, say, helping one’s local community. Suppose that the agent’s loss of well-being would not be significant either, so their sacrifice would not be of a kind that, if protracted and universalised among the population, would significantly decrease overall well-being. Now, suppose too that IMOC implied that it is permissible just to count the blades of grass in my lawn for 15 minutes rather than helping the local community. Can the mechanism adopted by IMOC be used to explain why this is the case? Applying it would give us the result that it is permissible, in this case, to count the blades of grass because it would be too costly, overall, to teach the contrary. However, the explanation seems neither convincing nor one that tracks the relevant moral facts – why would the fact that it is hard to learn to abstain from that action be what explains its permissibility? That the basis of its limiting mechanism lacks relevant explanatory value does not imply that IMOC has no other resources for explaining why an obligation to maximise overall well-being in this case would be too demanding. It is striking, however, that the basis of the main mechanism devised to weed out what is too demanding does not always seem to track the right or any explanatory facts. So, it is at least mysterious why we should trust these mechanisms to produce the intended results. It may well be true that, due to the difficulty of teaching a moral outlook, the overall costs of teaching it may decrease the overall benefit of doing so. However, it is at best unclear why costs of that kind should be regarded as the right kind of costs to determine what is properly demanding. It may be true that a requirement for an ideal code to be ideal is that it must be one that can be accepted or taught. Still, such a requirement may not be the main factor determining what is too costly in a cost–benefit analysis. After all, difficulty in learning or accepting an outlook may be due to entirely contingent factors that have little or nothing to do with what ought to be permissible or required. It may be the case that some of these difficulties do have explanatory value in determining what is permissible, but not all difficulties seem equally relevant.

3.2 Demandingness and human nature

In the previous section, I argued that a fundamental problem for both ICRC and IMOC is that how difficult it is to accept or learn a code is not always a relevant factor in a proper cost–benefit analysis comparing different codes. In this section, I outline one way of supplementing ICRC and IMOC that may partly ameliorate this problem.

A plausible starting point is the following question: what would need to be added to ICRC and IMOC to ensure that acceptability or teachability were (always) explanatorily relevant? The reply I sketch here develops what Hooker seems to have in mind when he deploys the expression ‘natural dispositions’. In particular, I understand it to refer, at least implicitly, to the notion of human nature, in the sense that the relevant natural dispositions are those dispositions which would be included in a plausible account of our nature. More specifically, I will articulate the idea that acceptability and learnability are (always) relevant factors in a cost–benefit analysis of the kind at issue because, and insofar as, the difficulties in accepting or learning a code would be contrary to human nature, including our psychological capacities. The explanatory burden would thus shift to those features of our nature which make accepting or learning difficult, rather than acceptability and learnability in themselves. Acceptability or learnability would still be theoretically relevant factors because they would be tangible expressions of certain limits imposed by our nature.

One initial problem with this suggestion is that the notion of human nature is notoriously controversial; it is risky to attempt to clarify one central aspect of RC by deploying a notion that may turn out to be on an even less secure theoretical footing. Still, in recent years, plausible attempts to clarify the notion of human nature have been abundant. Among them, Edouard Machery’s nomological account (Machery Reference Machery2008, Reference Machery and Smith2017, Reference Machery, Hannon and Lewens2018) provides a promising expansion of what seem to be the ‘natural dispositions’ to which Hooker alludes.Footnote ¹⁴ Machery’s account is called ‘nomological’ because it encompasses law-like (nomological) generalisations about humans – what humans tend to possess or exhibit because of their evolutionary history. This account regards as relevant those regularities in human traits that are grounded in our evolutionary past and are therefore relatively stable and predictable, similar to how scientific laws describe regular patterns in nature. In fact, on Machery’s view, ‘human nature’ stands for the set of traits that ‘human beings tend to possess as a result of the evolution of their species’ (Machery Reference Machery2008, p. 323). This definition combines two key elements: the universality proposal (traits must be typical of human beings) and the evolution proposal (traits must have evolved). These traits, in turn, are not exclusively instantiated by human beings – that is, for a property to be part of human nature, it need not be one that distinguishes humans from non-humans. In addition, the traits making up our nature have only to be typical; none of them needs to be possessed by all humans (Machery Reference Machery, Hannon and Lewens2018, p. 20). Bimodal traits (e.g. different male and female mating psychologies) would not be included, as they are not widely shared by humans in general (Machery Reference Machery2008, p. 324), but the ability to speak would be included, as would fear reactions to unexpected noise, biparental investment in children and being bipedal, even though not all of these traits are exclusively human and even if not all human beings have each one (Machery Reference Machery2008, p. 323). Arguably, among these traits we can also include an instinct for self-preservation and some tendency towards partiality (Bloom Reference Bloom2013), and other psychological limits to our capacity to concentrate, patience, perseverance and so on (Chappell Reference Chappell2019; p. 255, McElwee Reference McElwee2023, pp. 924–925). According to Tomasello (Reference Tomasello2019), another typical trait of human beings is our capacity for shared intentionality – the ability to create shared goals, attention, knowledge, and attitudes with others. This capacity emerges in two key stages during development: joint intentionality (at around nine months), which enables infants to engage in joint attention, cooperative communication, and collaboration with individual partners; and collective intentionality (at around three years), which enables children to identify with and conform to cultural group norms and conventions. The properties and limits that belong to human nature are those most humans possess and have because of a causal process: that of evolution.Footnote ¹⁵ Human nature is not itself a cause, but it is the result of a causal evolutionary process.

Now, this account can be combined with IMOC or ICRC by claiming that any prescriptions, obligations and so on that clash with traits that belong to human nature are at least always relevant when the comparative costs of a code including them are calculated. In fact, we may plausibly claim that a moral theory should not require actions that the relevant agents to which it applies cannot perform.Footnote ¹⁶ So, the costs of people’s accepting or learning to behave in ways that would defy their typical nature – neglecting their children, complete sexual abstinence, immense psychological efforts, deprivation from participating in shared goals and activities, etc. – should always count when determining whether a code is too demanding. The appeal to human nature would be the theoretical tool against the claim that certain difficulties rather than others are arbitrary: there does not seem anything arbitrary in claiming that the relevant moral agents can hardly be expected to do what they cannot do given the kind of agents they are. This solution would suggest that difficulties in accepting or learning a code are always relevant in comparative cost–benefit analyses if they arise from violations of our human nature. Arguably, such difficulties also have explanatory value.

Unfortunately, this attempt to amend RC still faces certain limits. First, it may be questioned why one specific individual’s actions should be regulated by what is typical for members of their species, rather than, say, their own particular nature.Footnote ¹⁷ In response to this objection, we may suggest that the relevant aspects of human nature are not all those that are typical of human beings, but those of the relevant individual agent(s). On this modified view, what is relevant for a cost–benefit analysis of a code is the particular nature (and limits) that each individual in the relevant group typically displays (and has) separately and in relation to each required part of the code. In fact, we may argue that, within certain limits, rules should be ‘individualised’ to make them responsive to individuals’ natural capacities and limits. This possible solution has several drawbacks, not least a risk of engendering codes of implausible complexity. In fact, codes would have to contain rules addressed to specific classes of individuals. Still, even considering this drawback, it seems to me that having a more complex code that responds to each person’s natural capacities and limits is the better option – at least insofar as one of the purposes of RC is that of maximising overall well-being. Another possible criticism of this solution is that some aspects of our nature (e.g., some violent dispositions of some individuals) should not have to be counted as relevant in assessing whether a theory is demanding in case the theory requires us to curb them. For example, a code that would be optimific other than for the fact that it requires difficult processes for learning how to reduce some individuals’ natural aggressive behaviour should not be preferred to a code the only comparative advantage of which is making more lenient requests to particularly violent individuals. So, although it is plausible that some moral rules may be too demanding because they violate certain aspects of our human nature – and that difficulty in learning or accepting them is a manifestation of this – not all of our typical traits seem to be relevant to, or deserve to count in, a comparative cost–benefit analysis determining which code is optimific.Footnote ¹⁸ If this is correct, then a version of ICRC and IMOC which is based on a nomological account of human nature would at most be in need of further supplementation – in particular, a method for determining which natural dispositions or traits are relevant to the comparative cost–benefit analysis of codes or outlooks. Still, an appeal to human nature would at least make less arbitrary some claims regarding the demandingness of a moral code.