I. PARTIAL CONSTRAINTS RULE-CONSEQUENTIALISM

According to Galvão, Partial Constraints Rule-Consequentialism is superior to alternative views about what we owe to animals. Consider, first, what he calls

Typical Contractualism

1. (a) The moral considerability of moral patients is entirely indirect, since it is grounded on the interests of moral agents.

2. (b) On the principles that would be chosen in ideal conditions, all humans are accorded moral considerability and are protected by rights, whereas non-humans are not.Footnote ²

Like Galvão's rule-consequentialist view, (b) is a version of

Nevertheless, Galvão claims, Typical Contractualism must be rejected since (a) implies that it is permissible to harm animals for trivial purposes, which he believes to be at odds with very common moral intuitions.Footnote ³ Consider now

Typical Act-Consequentialism

1. (c) The well-being of all sentient individuals is directly morally considerable.

2. (d) Whenever frustrating an interest is optimific, there are decisive reasons to do so.Footnote ⁴

This view would be partly true, since it accepts the direct moral considerability of non-human animals. According to Galvão, however, (d) is incompatible with the also strongly held intuition that it is not justified to inflict some severe harms on human beings without their consent, such as during ‘painful and lethal experiments’, even when doing so would be optimific.Footnote ⁵

Moreover, Galvão claims, rule-consequentialism offers an independent argument in favour of that intuition. On rule-consequentialism, each agent has the supreme moral aim of following the set of principles which, if generally accepted, would bring about the best possible consequences. Central to the ideal rule-consequentialist code would be a requirement to promote the good (a ‘duty of beneficence’). But, Galvão says, a code merely containing such an overly demanding duty would not be optimific.

Following Harsanyi and Hooker,Footnote ⁶ he claims that, if such a rule were generally accepted, societal trust would collapse and there would be widespread fear among moral agents at the expectation of being harmed when other agents considered that to be impartially best. In addition, they would be afraid of being harmed when others believed they were permitted to do so under their prerogative not to maximize the good, either for their own sake or for the sake of those to whom the code allows them to be partial. Finally, social stability would also be seriously threatened if it was generally accepted that human moral patients to whom agents are meaningfully related (such as infants and aged relatives) might be similarly harmed. Therefore, everyone would accept that, at least in most circumstances, there are decisive reasons against harming any human being, even when that would be optimific.

This argument can be generalized. Typical Act-Consequentialism is a version of

Galvão could have claimed that, on its best version, rule-consequentialism is preferable to any No Constraints view.

Finally, Galvão argues that Partial Constraints Rule-Consequentialism is preferable to what he calls

Reganian Deontologism

1. (c) The well-being of all sentient individuals is directly morally considerable.

2. (e) The interests of all sentient individuals are protected by agent-relative constraints against promoting the good.Footnote ⁷

This view considers that animals are directly morally considerable and recognizes that human interests must be protected against optimization. But it entails, as well, that biomedical research involving the use of non-human animals is morally unjustified, no matter how large its expectable benefits. Galvão disputes this. There is, he claims, a strongly held intuition that it is justified to use animals in biomedical research and for xenotransplantation, provided that the consequences of doing so are sufficiently good.Footnote ⁸

Rule-consequentialism can again explain why non-human interests need not be protected against optimization. Non-human animals cannot experience widespread fear, or severely threaten social stability, if moral agents generally accept that it is justified to harm them in order to promote the good, since they cannot be aware of that general acceptance.Footnote ⁹ Non-human animals are sentient beings, so they also fall under the protection of the code. But, Galvão claims, that merely implies a prima facie duty not to harm them (e.g. not consuming factory-farmed products or cosmetics tested on animals) and a weaker prima facie duty to help them.Footnote ¹⁰

As before, Galvão's argument can be generalized. Reganian Deontologism is a version of

Some other deontological accounts are versions of this view.Footnote ¹¹ Full Constraints is also compatible with rule-consequentialism. Galvão's argument can be extended to claim that Partial Constraints Rule-Consequentialism is preferable to any of these theories.

Thus, according to Galvão, of the two Partial Constraints theories assessed only rule-consequentialism accommodates the intuition that non-human interests are directly morally considerable. Moreover, of the three views about constraints, only Partial Constraints can accommodate strong intuitions about which interests are protected against optimization. Therefore, we would have decisive reasons to prefer Partial Constraints Rule-Consequentialism over all other views.

II. FULL CONSTRAINTS RULE-CONSEQUENTIALISM

The possibility of miscalculating the expectable value of an outcome provides us with further reasons, which Galvão overlooks, for recognizing agent-relative constraints.Footnote ¹² Consider a community of impartial moral agents with epistemic limitations similar to ours. Because they are impartial, they always and effortlessly act on a disposition to promote the overall good. If only rules allowing for optimific harms were generally accepted, these agents would perform their moral calculations without all the relevant information. Similarly like us, they would be prone to making mistakes.Footnote ¹³

One possible mistake consists in erroneously believing that certain harms are optimific when in fact they are not. Plausibly, an optimific set of principles should contain rules designed to offset the loss of value which would follow from causing such harms. The fact that some harm is expectably net positive would be necessary, but not sufficient, to justify causing it. In addition, the resulting expected net good must be sufficiently high to compensate for the possibility of mistakes in the moral calculus. Now Galvão states that constraints may be absolute – decisive in all possible circumstances – or moderate – admitting circumstances in which they can be outweighed by other reasons, like the reason to promote the overall good. On Galvão's definitions, these rules would constitute moderate constraints.

Another possible miscalculation consists in erroneously believing that some harms which in fact would be optimific would not be so. In these cases, a net loss of value would result from not causing those harms. Thus, the ideal code may also contain moderate constraints designed to prevent false negatives. These may require inflicting some harm, believed not to be optimific, when its expected value is sufficiently close to net-positivity.

This entails that the existing epistemic limitations of human agents give us enough grounds for accepting that, on the ideal code, there are agent-relative constraints of one kind or another. Given that actual human agents are not impartial, our reasons for recognizing those constraints are even stronger, since their partiality makes them prone to cognitive biases from which impartial agents would be free. These epistemic limitations bias human agents in a way that renders them susceptible to causing non-optimific harms to other agents. Even when striving to be as impartial as possible, they may overstate the benefits which causing some harm may have for themselves or for those with whom they are meaningfully related.Footnote ¹⁴ In addition, these epistemic limitations make human agents susceptible to causing non-optimific harms to moral patients, both human and non-human. Thus, the interests of these individuals should also be protected by agent-relative constraints.

Moreover, human beings seem to be partial to their species co-members. For instance, they are willing to inflict severe harms on animals, but generally not on other humans, for fairly trivial purposes. Suppose that, on the ideal code, that disposition was justified to some extent. As Galvão admits, the code should disincentivize harmful actions believed to be optimific, but regarding which the calculus is biased by the agent's partiality. That would give us additional reasons to recognize constraints against harming non-humans.

III. ASYMMETRICAL CONSTRAINTS RULE-CONSEQUENTIALISM

Galvão, however, claims that there are compelling reasons against Full Constraints. He takes this view to imply that it is not justified to use animals in biomedical research or xenotransplantation, even if the benefits are sufficiently large. Yet Full Constraints can avoid these implications or yield them in a way which suggests that our intuitions should be revised.

There are two possible versions of Full Constraints. According to

Alternatively, according to

A constraint is weaker than another protecting a similar interest when the former is moderate, but the latter is absolute; or when, being both moderate, the first provides decisive reasons against frustrating that interest in fewer circumstances than the other. Galvão argues that Symmetrical Constraints is to be rejected. Yet he says nothing about Asymmetrical Constraints. We have reasonable grounds to believe that, on the ideal rule-consequentialist code, non-human interests must also be protected by agent-relative constraints, even if these are weaker.

First, the facts about human agents discussed above provide us with some reasons for Asymmetrical Constraints. Galvão was right in pointing out how, on rule-consequentialism, some facts about the partiality of human agents give us additional reasons to protect human interests. Constraints for agents must be designed not only to prevent the loss of value resulting from epistemic errors, but also to prevent widespread anxiety at the possibility of being harmed. The resulting value of causing serious harm must be large enough that human agents do not ordinarily entertain the possibility that others may consider violating their rights to promote the good.

Thus Asymmetrical Constraints can also accommodate Galvão's intuition that it is justified to cause some severe harms to animals, but not to humans, if the benefits are sufficiently large. Like Partial Constraints, this view is compatible with the claim that, if necessary to promote the good, it would be permissible to harm an animal, but not to similarly harm a human being. Unlike Partial Constraints, this is not achieved by granting rights only to human beings, but by arguing that those protecting non-human animals are weaker.

Nevertheless, Galvão could still claim that the intuition we must preserve is that animals may be severely harmed whenever doing so is optimific. I believe that this intuition should be revised. Let us compare these implications about non-human animals to what Asymmetrical Constraints tells us about human moral patients. We may ask whether some human beings could be singled out, on rule consequentialism, for weaker constraints. If that were the case, then it could be argued that our insistence that all humans have similarly strong rights, but non-humans have none, should be regarded with suspicion.

There is a clear case of human moral patients to whom moral agents are not meaningfully related: abandoned, unwanted newborns. There is a strong disposition against harming newborns. But historical evidence shows that societies can have different attitudes, postponing their recognition as fully considerable beings.Footnote ¹⁵ We may wonder whether an optimific rule is conceivable which would justify using these infants in biomedical research, or for transplantation, provided that the benefits are large enough and no better alternatives are available.

Consider first these two possible practices:

Suppose that we could internalize these rules so that, even though we retain our affection for some human infants, we are not disturbed at others being used in these procedures. On this assumption, of the two possible practices, the second seems to bring about better consequences. This is because, first, neither arrangement causes widespread anxiety or threatens societal collapse. Second, similar non-lethal harms are caused in both situations. Third, however, on Compensated Newborns each human infant will have a life of net positive well-being, whereas that is uncertain in Uncompensated Newborns. Moreover, a disincentive (the compensation) is established to offset partiality towards the well-being of human agents and possible epistemic mistakes. In the long run, by minimizing false positives, a greater total sum of well-being is obtained.Footnote ¹⁶

Therefore, even assuming a scenario of no affection towards abandoned human infants, our epistemic limitations give us, again, decisive reasons to design rules which allow us to avoid overestimating the resulting good. This is done by committing ourselves to incur heavy costs in case we choose to harm. Yet this operation is equivalent to requiring more than merely a net positive outcome in order to justify the harm – it must be sufficiently net positive. This implies that, on the ideal code, these newborns must also be protected by agent-relative constraints, even if weaker than those protecting other human beings. If Galvão wants to vindicate his intuition that all human beings must be protected by rights, including abandoned unwanted human infants, then he must consider these epistemic facts significant.

Consider now what practices of seriously harming non-human animals would be justifiable, when no better alternatives are available:

Given our current attitudes towards non-human animals, we have compelling evidence that neither arrangement would be socially destabilizing. Just as with human infants, it seems that on that assumption, Compensated Non-humans should be chosen because it would result in a greater total sum of well-being.

Additionally, it must be noted that, if we chose Compensated Newborns before, consistency requires us to choose now Compensated Non-humans. Thus, even assuming a scenario of no affection towards abandoned human newborns, on Full Constraints these infants would still be protected by constraints against receiving harm. If generally accepted, such a rule would not have the negative effects of widespread anxiety and collapse of societal trust, for moral agents and their loved ones would still be protected by the stronger constraints. On this view, non-human animals would also be protected by these agent-relative constraints. Because, however, abandoned infants and non-human animals would be protected by weaker constraints, given the right circumstances it would be preferable to experiment on them rather than on human agents or other human beings.Footnote ¹⁷

Galvão may still insist that the implication that some human infants are protected by weaker constraints than human agents is counterintuitive. So would be the implication that non-humans are protected with such constraints, however weak. Nevertheless, the fact that a theory has counterintuitive implications does not always provide us with decisive reasons to reject it or revise it. Some suggest that our intuitions should be given no credence.Footnote ¹⁸ Others claim they have some methodological value, even if it is first necessary to identify those intuitions whose origin makes them suspect.Footnote ¹⁹

Arguably, our intuitions about our moral obligations to human infants, on the one hand, and to non-human animals, on the other, are suspect in this way. It is convenient to regard viable offspring as fully morally considerable, requiring as much protection from harm as the human adults they will become. Animals can be processed into resources which were necessary for the survival of ancestral human communities, so it is adaptive to consider that their interests are less morally important or not important at all. Thus, we have an evolutionary explanation about why we were caused to have these beliefs, even if they were false. Provided that we also have a credible moral theory implying that they are wrong, it is justified to dispense with them and act on our reflected moral principles.

Asymmetrical Constraints is also compatible with non-consequentialist positions.Footnote ²⁰ Proponents of those views could claim that sentient individuals endowed with autonomy, rationality or moral agency must be protected by stronger constraints. Whether any version of rule-consequentialism is preferable to these other theories will depend on how plausible we believe it to be, all things considered, when compared with them.

IV. THE DUTY TO HELP WILD ANIMALS SUFFERING FROM NATURAL HARMS

According to Galvão, the ideal code includes a prima facie duty to help wild animals. He distinguishes between two kinds of situations in which humans might be in a position to help them.Footnote ²¹ The first are those cases in which an animal can be helped without thereby harming another. These include helping animals suffering from anthropogenic harms as well as some natural harms, like natural disasters or serious diseases.Footnote ²² In these cases, he correctly claims we may have a duty to help. The second are those cases, like predation, in which it is necessary to harm some animal in order to help another one suffering from some natural harm. In these cases, however, he claims that helping is wrong in virtue of the stronger prima facie duty not to harm. This is so even if the harm that could be averted is worse than the harm that would thereby be caused.

Now Galvão argues that ‘since helping is generally harder than simply not harming, a due consideration of the internalization costs will presumably lead to the conclusion that the duty to help animals is much weaker (i.e. more easily overridable by other factors) than the duty not to harm them’.Footnote ²³ Yet it does not directly follow from this that, in those cases in which our duty to help conflicts with the duty not to harm, the latter must prevail. Here Galvão is conflating two senses in which some duties are stronger than others. The first is the one Parfit calls the cost-requiring sense. In that sense one duty is stronger than another if we are morally required to bear greater burdens, when that is necessary to fulfil it. The second is the conflict-of-duty sense. In this second sense one duty is stronger than another if the former would outweigh the latter when these duties conflict.Footnote ²⁴

Importantly, neither of these different notions of the relative strength of a duty implies the other. Thus, even if the duty to help is more easily overridable by other factors than the duty not to harm, that doesn't imply that, necessarily, whenever these two duties conflict, the latter prevails. Whether that is so must be established separately – an argument which Galvão does not provide.

Let us suppose that an agent can, at no cost to herself, prevent a prey from being devoured and killed by a predator. Let us further suppose, as Galvão does, that the predator's harm in being prevented to eat is less severe than the one her prey would otherwise suffer. If no further consequences result from the intervention, then the agent has decisive reasons to help the prey – doing so is optimific and not worse for herself or others to whom she is meaningfully related.

But suppose, alternatively, that though intervening were optimific, doing so was sufficiently costly for the agent. Given this construction of the prima facie duty to help, it would be overridden. This implies that it would not be wrong for the agent to refuse to assist the prey. Yet it does not necessarily imply that it would be wrong for her to help. It seems that a rule that, in those circumstances, gave agents permission to help, even if they do not have a duty to do so, has better consequences than a rule forbidding them to help, all else being the same. If generally accepted, on the first rule there may be more optimific actions. By internalizing such a rule, agents would be disposed against blaming those who do not help at great cost to themselves, and yet would be inclined to praise those who do so. It seems that, in the long run, the first rule generates a greater sum of total well-being.

It could still be objected that, since we don't know the long-term consequences of even these individual interventions in nature to help wild animals, it is best if we avoided them entirely. Thus, in order to forestall the catastrophic destabilization of ecosystems, on the ideal code everyone would accept that humans should avoid interfering with nature.

Such a rule would not result in the best consequences, since it presupposes a false idyllic view of nature. Natural processes cause most wild animals to have lives of net suffering.Footnote ²⁵ Moreover, it must be noted that the virtual totality of sentient beings consists of wild animals.Footnote ²⁶ Thus, the majority of sentient individuals plausibly have lives not worth living. This will continue to be so unless human beings obtain the knowledge and means necessary to ameliorate this situation.

If we are to take seriously the requirement of impartiality, then on the ideal code everyone must generally accept that there is a duty to modify nature in the ways that will allow wild animals to have net positive lives. Just as the ideal code must contain a motivation to prevent disaster which outweighs all other motivations,Footnote ²⁷ so it must incorporate a similar motivation to mitigate or end ongoing disasters. Furthermore, it is necessary to maximize the chances that future generations engage in optimific large-scale interventions in nature. Therefore, whilst that is not possible, the ideal set of dispositions plausibly includes one to help individual animals in distress, even from predators, associated with an inclination to praise those who do so. The long-term consequences of agents being so generally disposed appear to be best in terms of total well-being.Footnote ²⁸