Effective Altruists aim to use a significant share of their resources to do the most good, which is largely a theory-neutral goal. However, in practice, many effective altruists are sympathetic to consequentialism. It has been persuasively argued before that the core ideas of EA, like doing the most good via charity and career choice, are not dependent on consequentialism. Arguments about which cause to prioritize, however, often do rely explicitly on consequentialism, especially arguments about animal welfare and the long-term future.. In this post, I will be evaluating how non-consequentialist ethical positions affect cause prioritization.
Table of Contents
1. My Approach to the Question
1.1 Why We Care About Non-Consequentialist Ethical Theories
1.2 Sub-Questions I Answer
1.3 How to Read This Article
2. The Value of Animal Welfare
2.1 Refined Intuitions
2.2 Broad Deontological Considerations
2.2.1 Kantianism
2.2.2 Neo-Kantianism
2.2.3 Regan
2.3 Virtue Ethics
2.4 Contractualism
2.5 Summary and Takeaway
3. Extinction Risk and Value of the Long-Term Future
3.1 Refined Intuitions
3.2 Contractualism
3.3 Kantianism
3.4 Virtue Ethics
3.5 Summary and Takeaway
1. My Approach to the Question
1.1 Why We Care About Non-Consequentialist Ethical Theories
Many EAs are interested in taking account of moral uncertainty. This is the idea that just as we take into account empirical uncertainty in our decisions, we should take into account the possibility that our moral frameworks could be incorrect. There are many different views about how this should be done, such as maximizing expected choice-worthiness, the parliamentary model, or the portfolio approach.[1] I will often be discussing things in the context of the portfolio approach, which holds that you should have buckets of resources for each worldview, whose size is tied to your credences in the validity of the worldview, because this is the method Open Philanthropy uses (See here and here for Open Phil’s explanation and defense of this method).
If moral uncertainty is worth correcting for, then even EAs sympathetic to consequentialism should be interested in what non-consequentialist theories recommend because this will affect their decision-making. Given that nearly 2/3s of philosophers are non-consequentialists, we should give these positions significant weight in our decision-making process (Philpapers Survey).
1.2 Sub-Questions I Answer
I split my thoughts into two separate articles. This is part 1, which analyzes how good improving animal welfare and shaping the long-term future look under non-consequentialist theories. I do not consider global poverty because there is wide agreement that this matters morally from non-consequentialist perspectives. [2] Part 2 considers if humane animal agriculture and intervening in wild-animal suffering are permissible under non-consequentialist theories. I will be publishing part 2 in a few days.
1.3 How to Read This Article
I first want to note that I am not a professional philosopher or expert in ethics. I am a philosophy undergraduate at the University of Chicago, and I conducted this research for a 6-week summer fellowship with CEA.
Half of these posts are a review of the non-consequentialist literature on EA topics. The non-consequentialist positions I reviewed were virtue ethics, contractualism, many forms of deontology and what I am calling refined intuitions, which is the view you come to after careful reflection upon specific intuitions.[3] I won’t be going into much detail about the theories themselves, so if you are not familiar with these theories I recommend checking out these articles.[4]
The other half of these posts are my commentary on the non-consequentialist literature. This includes if I think the applications of the non-consequentialist theories are correct, how seriously we should take the considerations, and a final take away section where I summarize the main claims and what I think was most important. Many of these views bottom out in my own intuitions, so I recommend taking these sections as a proxy of what one person thinks, rather than the definitive view.
Finally, these posts are long. If you are only interested in the big-picture takeaways, you can just skip to the takeaway section for each major part e.g. section 2.5 and 3.5 of this article.
2 The Value of Animal Welfare
Almost every ethical theory gives animals at least some moral weight, condemning, for example, torturing animals for fun. However, theories diverge from each other on how much moral weight to give animals. Some are dismissive, giving animals weak or indirect importance, while others are serious, giving animals full moral status, which is roughly the same normative importance as humans.[5] [6]
2.1 Refined Intuitions
Many philosophical arguments do not appeal to any specific theory, but rather use thought experiments and refined intuitions. Peter Singer’s argument from marginal cases is a famous example of this type of argument (Singer 2009). Singer holds that in order for all and only human beings to have full and equal moral status, all and only humans must have some property that grounds this status. However, for any reasonable property, it is not the case that all and only humans have this property.[7] For example, one might guess that rationality is what grounds moral status, but this excludes severely mentally disabled people and babies. Most people find it obvious that babies or mentally disabled people do have moral rights, so the starting assumption that all and only humans have moral status (at least on the grounds of having rationality) must be incorrect. In order to include all marginal cases of humans, we must choose a sufficiently broad category, like sentience, but this will include almost all animals inside our moral circle.
My view: The fully fleshed out version of this argument in the literature is very strong. I have yet to see a persuasive morally relevant distinction after 40 years of attempts.[8] I am fairly confident that it is impossible. I also have a strong intuition that biting the bullet by accepting that certain humans do not have moral worth is completely unacceptable.
2.2 Broad Deontological Considerations
2.2.1 Kantianism
Kantian ethics holds that there are three tests to determine if something is immoral: if it can’t be universally willed, meaning that everyone couldn’t all do it; if it treats someone as a means to an end; and if it cannot be legislated in the Kingdom of Ends, which is an idealized world where all agents treat each other as ends in themselves and create moral laws together.
Kantian ethics is famous for being dismissive of animals, as Kant himself argued that because animals had no rational nature, they were not morally relevant in themselves; however, he argued that animals mattered indirectly because if we hurt animals, we damage our own moral character. Philosopher Matt Altman has pointed out that this view is less dismissive of animal welfare than it may initially appear. In fact, “Kant would condemn the deplorable conditions on factory farms” and held that “human interests do not automatically trump animal interests” (Altman, 2011). However, it is clear that on Kant’s account, the duty to help animals is much weaker than to help other humans, so I characterize the duty as weak overall.
My view: This position is weak for three reasons. Firstly, it cannot accommodate the marginal cases argument well. Secondly, it cannot accommodate the wider intuition that animal suffering is wrong in itself, not just indirectly. Thirdly, I find the theoretical arguments of Neo-Kantians for animals’ moral status to be at least somewhat convincing.
2.2.2 Neo-Kantians
Many contemporary Neo-Kantians find Kant’s view on animals to be a misapplication of his own theory. Most famously Christine Korsgaard has argued that animals have a strong moral status under Kantian ethics for two main reasons. Firstly, Korsgaard argues that humans are ends-in-themselves only because things can be good or bad for them, but of course, the same is true for animals (Korsgaard, 2018, part 8). Secondly, because humans have an animal nature, when we legislate moral laws in the kingdom of ends, these moral laws naturally extend to animals as passive citizens. [9] Overall, this view takes animal welfare seriously, perhaps even more seriously than consequentialists do, because it argues that animals have the same inviolable status as humans do.[10]
My view: I find Korsgaard’s view much more persuasive than Kant’s own take. Korsgaard’s view gets us a more intuitive account of the value of animals but does so with some theoretical motivation. I have some uncertainty about theoretical arguments, but they are at least plausible.
2.2.3 Regan
Another famous deontological stance on animal ethics is posited by Tom Regan. Regan used the marginal cases argument to motivate a conception of rights that included animals. He argued that as long as a being was a subject-of-a-life then it was morally relevant (Regan, 2010). This ends up with a practical view similar to Korsgaard’s.
My view: Regan’s arguments are less persuasive than Korsgaard’s because there is less theoretical motivation in the actual grounding of moral status. Regan’s position gets almost all of its strength from the marginal cases argument. As I mentioned, I think this is a strong argument, but I am more persuaded by a combination of theoretical and intuitive insights, rather than relying on just intuitive considerations.
2.3 Virtue Ethics
Virtue ethics is an ethical theory that is agent-centered rather than act-centered. It asks what a virtuous person would do, and then recommends that action. The practical upshot of this is that instead of asking what acts are moral, virtue ethicists often ask what kind of person would perform an act. Philosopher Rosalind Hursthouse has persuasively pointed out that you cannot be truly compassionate “while I am party to such cruelty” in factory farms (Hursthouse, 2006, p. 142). Therefore, a virtuous and compassionate agent must morally care about the suffering of animals. Hursthouse hints that this might make vegetarianism obligatory.
My view: I find that virtue ethics is always a bit unclear and vague when it comes to action guidance; however, my intuitions certainly track with Hursthouse’s point. Although, It is still a relatively weak obligation all things considered.
2.4 Contractualism
Contractualism is an ethical theory pioneered by T.M. Scanlon. Scanlon argues that something is moral if it is something that no one could reasonably reject. Someone can only reasonably reject a principle if it inflicts a greater harm upon them than any other option would inflict upon another individual. This is distinctive from consequentialism because 1) it does not allow for interpersonal aggregation, but rather is a comparison between what individuals can reject 2) it allows for non-welfare based harms and 3) it is fundamentally based on the idea of justifying yourself to others, which makes it less friendly to animals.
Scanlon himself has outlined two possible ways the moral status of animals could be accommodated under contractualism. Firstly, he concedes that contractualism only covers a subset of morality, namely what we owe to others. “This leaves open the possibility that our obligations to animals fall outside this part of morality” (Ashford, 2007). Secondly, contractualism could accommodate animals “via the notion of trustees, to whom justifications of proposed principles can be offered, on behalf of the animals they represent” (Ashford, 2007). So either contractualism tells us little to nothing, or gives animals full moral status and takes their welfare as seriously as consequentialists do.
My view: Neither option is very attractive. The first view conflicts with the marginal cases argument, while the second view seems unmotivated by the theory because contractualism itself cannot justify why we need to give certain things trustees. Ultimately I lean a bit further towards the second because I think the marginal cases argument is overwhelmingly strong and Scanlon admits that non-rational humans have trustees.
2.5 Summary and Takeaway
Many different non-consequentialist ethical theories converge on animal welfare being very important, e.g. Neo-Kantians, refined intuitions, the trustee model of contractualism, and virtue ethics to some extent. Other non-consequentialist ethical theories give it at least some importance, e.g. Kantian ethics and the broad model of contractualism. Moreover, I find the views that give animal welfare importance to be much more plausible than those that give it little. Also, my findings are consistent with polls of philosophers, as about 80% of philosophers think that factory farming is morally wrong and 27% consider it a serious “moral abomination” (Leiter, 2017). I expect this number to be even higher among moral philosophers. [11]
If you prefer a robustly good cause, I think animal welfare is normatively about as robust a cause as global poverty, although there are many other considerations, such as knock-on effects and other empirical questions, that are involved in comparisons between animal welfare and global poverty.
3 Extinction Risk and Value of the Long-Term Future
Almost all ethical theories agree that human extinction would be a bad thing. However, they disagree about how bad it would be. Some theories take a dismissive view and think it is only bad because of the pain and suffering that people would go through when the extinction event occurred. Other theories take a more serious view and think that it would be a disaster not only because of the direct deaths of people but because of the loss of potential future lives.
3.1 Refined Intuitions
For the sake of categorization, I am grouping population ethics into the category of refined intuitions, as many considerations in it are a combinations of moral intuitions and rational judgement. The most pertinent debate for long-termism and existential risk is on person-affecting views.[12] Person-affecting views argue that it is neutral to add happy people into the world, while non-person-affecting views argue that it is good to add happy people. This debate has been hashed out extensively in many other places, so I won’t say more on it here.[13]
Even if you accept a person-affecting view, there are also many non-welfarist reasons to care about the long-term future. To name a few: first, we lose potential massive advancements in science, art and other ideal goods if we go extinct (Parfit, 1984, p. 453). If you believe these are intrinsic goods, losing them would be massive waste even under person-affecting views. Second, there is the argument from the final value of humanity. This argument holds that certain things have final or intrinsic value, and that the appropriate response to this intrinsic value is to prevent the extinction of this thing. For example, we seem to assign great works of art, cultures and languages final value, and this is why we mourn the extinction of these things.“The analogous claim that, given humanity’s final value, we have moral reason to help it survive therefore strikes me as plausible” (Frick, 2014, p. 172). Third, the loss of intelligent life could mean the loss of an important part of the universe, something akin to a complete loss of meaning. Finally, from a self-interested standpoint, many of the things we do, like building long-lasting infrastructure or fighting for long-term justice would become meaningless if we were the last generation. Therefore, we must prevent extinction to preserve our own sense of meaning (Scheffler, 2013, p. 171).
My view: The literature on person-affecting views is mixed but I think leans against the person-affecting view because no one has really found an account of person-affecting views that has acceptable implications. [13] Moreover, I think roughly all the non-welfarist reasons to care about extinction hold as well, although it is unclear to me how strong a reason it gives us to stop extinction. It’s clear to me that these reasons are non-trivial, but they are certainly less powerful than the totalist reasons to prevent extinction. I’m not sure where in-between these extremes these considerations fall. Overall, because person-affecting views looks implausible, combined with the moderate non-welfarist reasons to prevent extinction, I think intuitive concerns take existential risks seriously and count for prioritizing it as a plausible top cause.
3.2 Contractualism
Contractualism is probably the most robust case against prioritizing existential risks, for two reasons. First, contractualism cannot consider the interests of people who will never be born because “we can only wrong someone who did, does or will actually exist because wronging involves failing to take a person’s interests into account” (Finneron-Burns, 2017, p. 331). This is roughly a person-affecting view that is just baked into the theory.[15] Second and more importantly, as philosopher Tim Mulgan points out, because “contractualism regards sacrificing an identified person as worse than imposing a small risk on everyone” reducing identified suffering from say global poverty would be more important than reducing the risk of extinction for everyone (Mulgan, 2017, p. 3).[16] Therefore, even if you accepted a totalist view of population ethics, contractualism would prefer donating to high probability global health interventions over x-risk, because an identified individual with a disease has their entire life at stake, while potential people have relatively little to gain as individuals in reducing x-risk because it will only slightly increase their chance of living [17]. Contractualism would certainly condemn extinction, but it would take the dismissive view and would prioritize other causes over it.
My view: This is the most coherent and best argument for prioritizing global poverty over existential risk. I think it clearly follows from the theory itself. If you take a portfolio approach to moral uncertainty, I think that this justifies a cut going to interventions that have a high probability of impact such as certain global poverty interventions.[18]
3.3 Kantianism
The Kantian position on the long-term future is a bit hazy. I could only find one Kantian talk about our obligations to future generations, Korsgaard. Korsgaard accepts a person-affecting view because she believes that anything that is good is good for a person. However this runs into the non-identity problem, the fact that helping make future generations better off also changes the identities of the future people and thus makes no specific individual better off (see here for more explanation of the non-identity problem). Because this has absurd conclusions like that it is meaningless to prevent the long run impacts of climate change, Korsgaard attempts to avoid this conclusion by arguing that we have relational duties to people as occupants of a role. So we have duties to future generations, whoever they might be, because they are the continuation of our species. (Korsgaard, 2018, Part 5 p.21). This gives us reason to care about future generations. Although Korsgaard does not mention this herself, I do not think this view can care about future generations who won’t exist because then we will have no relation to them.
My view: Korsgaard’s position is very unlikely to be correct. Not only does it skirt over many important issues, but it also seems unmotivated by Kantian ethics itself. I think this appears because Kantians do not have much to say about potential people; not much comes out of their theory. Therefore, I think we should treat it as a wash and largely pass over Kantian considerations for this issue.
3.4 Virtue Ethics
Similar to Kantians, most virtue ethicists have not discussed population ethics or the badness of extinction. However, I think virtue ethics takes human extinction seriously for two reasons. First, Toby Ord has made a few arguments about how existential risk shows individual and civilization-wide vices, such as carelessness, selfishness, lack of generosity, etc. (Ord, 2017). Second, Rosalina Hursthouse has argued that virtues arise from consistency with human function, one function being the continuation of the species (Hursthouse, 1997).
My view: These arguments are strong enough to say that virtue ethics takes extinction somewhat seriously, although not necessarily as the top cause. I have a lot of uncertainty because virtue ethicists themselves write about the topic so rarely.[19]
3.5 Summary and Takeaway
The overwhelming value of the long-term future is a bit less robust than the case for animal welfare, although overall it does look pretty good. Refined intuitions and virtue ethics to some extent take extinction seriously, while contractualists treat extinction dismissively. Kantians also do not take extinction seriously, although I am skeptical of approaches to population ethics offered by Kantians.
Another consideration is that contractualism robustly justifies a focus on global poverty over x-risk mitigation even if you accept a totalist view of population ethics. If you take a portfolio approach to moral uncertainty, this gives good reason to donate a certain amount to global poverty reduction.[20]
Endnotes
[1] Click on these for a further explanation of Maximizing expected choiceworthiness, parliamentary model, and portfolio diversification.
[2] The only question non-consequentialists typically have about global poverty is how strong of an obligation individuals have to stop it. Everyone agrees that it is good to help, but it is a open question if it is obligatory. These questions about obligation are largely irrelevant to cause-prioritization because they apply roughly equally to every cause.
[3] Here I have in mind the type of philosophy that Jeff McMahan or Frances Kamm engages in.
[4] Virtue ethics, Kantian Ethics, Contractualism
[5] I’m using serious and dismissive as value-neutral terms, as used here. I don’t mean that there is something inherently wrong with being dismissive or good with being serious, just that theories that take the issue seriously think it is an important issue, while dismissive theories take it to be less important.
[6] By normative equality, I mean that an equal amount of suffering from an animal and human would be equally morally relevant. This still allows for descriptive inequality, meaning that humans could have a greater capacity for suffering, and therefore matter more.
[7] By reasonable, I mean some category that is plausibly morally relevant. For example, you might argue that having human DNA separates all humans from all animals; however, it is unlikely that human DNA is the basis of moral status.
[8] The best attempt to answer Singer’s argument that I have seen is Shelly Kagan’s modal person argument, which can be found here, but I think it fails because of the reasons highlighted here.
[9] Korsgaard’s entire argument is a bit confusing, so if my short summary leaves you wanting, I recommend reading part 8 of her book Our Fellow Creatures.
[10] I am assuming here and elsewhere that consequentialists care about non-human animals in the standard Singer-esque way. This conclusion can be avoided with certain theories of philosophy of mind, personhood or of sentience, but most consequentialists seem to have the Singer-esque view so I will be assuming that view in my description of the article.
[11] This article explains that ethics philosophy professors are more likely than non-ethics philosophers to give animals moral status.
[12] Alternatively, you might think that the future has diminishing marginal utility, but this is a view that I find implausible. For why see chapter 5 here.
[13] Places that discuss this well: Derek Parfit’s Reasons and Persons Part IV, Nick Beckstead’s thesis, Johann Frick’s thesis
[14] Some seemingly unacceptable conclusions of person-affecting views include 1) Biting the bullet on the non-identity problem. Personally, I think this is the worst counter-example to person-affecting views. I am unwilling to accept an an ethical theory that says, for example, preventing the long-run impacts of climate change is meaningless because it will change the identities of who is alive. 2) Either biting the bullet on the morally neutrality of creating miserable lives or justifying the asymmetry between creating miserable and creating happy lives. 3) That there is nothing wrong with painless extinction. 4) Cyclical preferences once you try to incorporate the person-affecting view into your axiology. For a paper on a person-affecting view that tries to avoid these problems see here. I think it is the best person-affecting account.
[15] You could try to modify contractualism to accept a more totalist view; although, I think this stretches contractualism a bit far. However, I think it is not completely implausible. Derek Parfit tries to modify contractualism like this in On What Matters.
[16] This assumes that you are assessing the situation ex ante, before we know who the risks will kill. I find this view very plausible within contractualism, because otherwise things like driving are immoral. See here for more discussion.
[17] Not all global poverty charities save identified lives. In fact, most probably save statistical lives. If contractualism is right to prioritize identified lives, then this would change what charities are best in this area. I haven’t personally given this much thought, other than I am relatively confident that global poverty saves more identified lives than x-risk mitigation.
[18] A concrete example of this would be the Open Philanthropy Project’s cause prioritization, which currently donates a cut to global poverty for signalling and virtue ethical reasons. See here for their reasoning. My view would switch out the virtue ethical justification for a contractualist one. Another change would be favoring high probability interventions of identified lives rather than low-probability interventions of statistical lives within the cause of global poverty alleviation.
[19] Toby Ord is a consequentialist, and I’m extrapolating Hursthouse’s views to the question of extinction, rather than using her own words.
[20] Some of this depends on if you think contractualism accepts animals as morally relevant. However, even if it does, plausibly humans have more to lose by dying.
Works Cited
Altman, Matthew C. Kant and Applied Ethics: the Uses and Limits of Kant's Practical Philosophy. Wiley-Blackwell, 2014.
Ashford, Elizabeth. “Contractualism.” Stanford Encyclopedia of Philosophy, Stanford University, 20 Apr. 2007, plato.stanford.edu/entries/contractualism/.
Donaldson, Sue, and Will Kymlicka. Zoopolis: a Political Theory of Animal Rights. Oxford University Press, 2011.
Frick, Johann. “'Making People Happy, Not Making Happy People': A Defense of the Asymmetry Intuition in Population Ethics.” 'Making People Happy, Not Making Happy People': A Defense of the Asymmetry Intuition in Population Ethics, Harvard, 2014, dash.harvard.edu/handle/1/13064981.
Hursthouse, Rosalind. “4. Virtue Ethics and the Treatment of Animals.” The Oxford Handbook of Animal Ethics, 2006, pp. 119–144., doi:10.1093/oxfordhb/9780195371963.003.0005.
Hursthouse, Rosalind. On Virtue Ethics. Oxford University Press, 1997.
Korsgaard, Christine M. Fellow Creatures: Our Obligations to the Other Animals. Oxford University Press, 2018.
Leiter, Brian. “What Is Your Opinion of ‘Factory Farming’?” Leiter Reports: A Philosophy Blog, 2017, leiterreports.typepad.com/blog/2017/12/what-is-your-opinion-of-factory-farming.html.
Mcmahan, Jeff. “The Comparative Badness for Animals of Suffering and Death.” The Ethics of Killing Animals, 2012, pp. 65–85., doi:10.1093/acprof:oso/9780199396078.003.0005.
Mcmahan, Jeff. “Eating Animals the Nice Way.” Daedalus, vol. 137, no. 1, 2008, pp. 66–76., doi:10.1162/daed.2008.137.1.66.
Muglan, Tim. What Exactly Is Wrong with Human Extinction? 2017.
Ord, Toby. “Why the Long-Term Future of Humanity Matters More than Anything Else.” 80,000 Hours, 2017, 80000hours.org/podcast/episodes/why-the-long-run-future-matters-more-than-anything-else-and-what-we-should-do-about-it/.
Parfit, Derek. Reasons and Persons. Clarendon Press, 1984.
“Preliminary Survey Results | PhilPapers Surveys.” Philpapers, philpapers.org/surveys/results.pl.
Regan, Tom. The Case for Animal Rights. University of California Press, 2010.
Sandberg, Anders. Practical Ethics, 2015, blog.practicalethics.ox.ac.uk/2015/09/the-moral-limitations-of-in-vitro-meat/.
Scheffler, Samuel. Death and the Afterlife. Oxford University Press, 2013.
Singer, Peter. Animal Liberation: the Definitive Classic of the Animal Movement. Harper Perennial, 2009.
Williams, Nancy M. “The Ethics of Care and Humane Meat: Why Care Is Not Ambiguous About ‘Humane’ Meat.” Journal of Social Philosophy, vol. 46, no. 2, 2015, pp. 264–279., doi:10.1111/josp.12094.
So I've not written extensively on this (and in some cases not at all), but I'm a virtue ethicist and I care about animal welfare and x-risks (or just future-folks more generally) as an expression of compassion. That is, satisfying the virtue of compassion that I aspire to (now drawn from Buddhist notions of compassion, but originally from a more folky notion of it that I learned to adopt as a virtue from my upbringing in secular Protestant America) encourages me to give consideration to the welfare of animals and future folks, and this results in my choice to eat almost-only plants and to work on addressing AI-related x-risks.
This is not exactly something you can cite nor a full-fledge argument, but figured you might find it worthwhile to hear more from someone in EA who has one of these non-consequentialist moral views. I expect there are a number of crypto-Kantians around (crypto only in the sense though that they just don't bring it up much because it's not part of normal EA conversation to reason via deontology) and a decent number of contractualist given that position's affinity with libertarian ethics and the number of libertarian EAs drawn from the rationalist community.