- The Moral Weight Project assumes hedonism. It also assumes that in the absence of good direct measures of the intensities of valenced experiences, the best way to assess differences in the potential intensities of animals’ valenced experiences is to look at differences in other capacities that might serve as proxies for differences in hedonic potential.
- Suppose that these assumptions lead to the conclusion that chickens and humans can realize roughly the same amount of welfare at any given time. Call this “the Equality Result.” The key question: Would the Equality Result alone be a good reason to think that one or both of these assumptions is mistaken?
- We don’t think so. To explain why not, we consider three bases for skepticism about the Equality Result. Then, we consider whether the Equality Result should be surprising given hedonism.
- Three Bases for Skepticism
- First, someone might balk at the implications of the Equality Result given certain independent theses. For instance, given utilitarianism, the Equality Result probably implies that there should be a massive shift in neartermist resources toward animals, and someone might find this unbelievable. But the Equality Result isn’t to blame: utilitarianism is.
- Second, someone might be inclined to accept some theory of welfare that does not support the Equality Result. Fair enough, but as we’ve argued elsewhere, that only gets you so far.
- Third, someone might balk at the Equality Result even given hedonism. The basic problem with this is that the anti-Equality-Result intuition is uncalibrated, uncalibrated intuitions are vulnerable to various biases, and there are some highly relevant biases in the present context.
- If Hedonism, Then the Equality Result Shouldn’t Be Surprising
- We quickly consider three popular theories valenced states and argue that there are plausible assumptions on which each one leads to the Equality Result.
- This isn’t an argument for the Equality Result. It is, however, a check against knee-jerk skepticism.
This is the seventh post in the Moral Weight Project Sequence. The aim of the sequence is to provide an overview of the research that Rethink Priorities conducted between May 2021 and October 2022 on interspecific cause prioritization—i.e., making resource allocation decisions across species.
As EAs, we want to compare the cost-effectiveness of all interventions, including ones that benefit (sentient) animals with the cost-effectiveness of interventions that benefit humans. To do that, we need to estimate the value of each kind of animal relative to humans. If we understand each individual’s value in terms of the welfare they generate, whether positive or negative, then this means we need to estimate the amount of welfare that each kind of animal realizes relative to the amount of welfare that humans realize.
How should we react if our method for generating these estimates produces a surprising result? Suppose, for instance, that we make various assumptions and generate a method for estimating how much welfare animals can realize relative to how much welfare humans can realize (which is the first step toward estimating how much welfare animals actually realize). And suppose that, when applied, our method suggests that chickens and humans can realize roughly the same amount of welfare at any given time. Call this “the Equality Result.” Would getting the Equality Result itself be a reason to think we made a mistake?
Let’s make this concrete. The Moral Weight Project assumes hedonism—i.e., that all and only positively valenced experiences are good for you and all and only negatively valenced experiences are bad for you. Moreover, it assumes that, in the absence of good direct measures of the intensities of valenced experiences, the best way to assess differences in the potential intensities of animals’ valenced experiences—i.e., their “hedonic potential”—is to look at differences in other capacities that might serve as proxies for differences in hedonic potential. If the Moral Weight Project delivers the Equality Result, would that alone be a reason to think that either hedonism or our methodology is mistaken?
We don’t think so. The purpose of this post is to explain why. To do that, we’ll proceed in two stages. First, we’ll consider and respond to three kinds of skepticism about the Equality Result. Then, we’ll offer some reasons to think that, given hedonism, the Equality Result wouldn’t be surprising.
Here are the three kinds of skepticism:
- Someone might balk at the implications of the Equality Result given certain independent theses. For instance, given utilitarianism, the Equality Result probably implies that there should be a massive shift in neartermist resources toward animals (and among animals, toward fish and invertebrates). Someone might find this unbelievable.
- Someone might balk at the assumptions behind the Equality Result. If they’re inclined to accept some theory of welfare that does not support the Equality Result, then they might reject the Equality Result on that basis.
- Someone might balk at the Equality Result even given hedonism. That is, they might think that even if we assume hedonism, we shouldn’t end up with the Equality Result. They might think it’s just implausible that a chicken can suffer as intensely as a human—or, more positively, that a chicken’s pleasures can be nearly as intense as a human’s.
Then, we’ll discuss some prominent theories of valence, explaining how each one could end up supporting the Equality Result.
Three Kinds of Skepticism
We begin with the three skeptical reactions.
Balking at the Implications
In this section, we’ll focus on balking at the implications of the Equality Result. Again, given utilitarianism, the Equality Result probably implies that there should be a massive shift in neartermist resources toward animals. Someone might find this unbelievable.
The fundamental problem is that this reaction amounts to shooting the messenger. If the Equality Result “has radical implications,” that isn’t the fault of the Equality Result per se. On its own, it doesn’t have any practical implications. To get some, you have to add—at a minimum—a normative theory like utilitarianism and a lot of empirical assumptions. But then the theory and the empirical assumptions deserve equal scrutiny for the counterintuitive practical implications.
The natural objection is something like:
We’ve got great evidence for utilitarianism and the relevant empirical assumptions; we don’t have great evidence for the Equality Result. So while it's true that other assumptions deserve some scrutiny for generating counterintuitive practical implications, we shouldn't focus on them. Instead, we should focus on the least-well-confirmed piece of the puzzle.
There are three problems with this objection. The first is circularity. That is, part of the evidence for utilitarianism is just its fit with common intuitions about what’s right and wrong. But if utilitarianism only fits with common intuitions given that animals have much smaller welfare ranges than humans (i.e., given that the difference between the best and worst welfare states that animals can realize is much smaller than the difference between the best and worst welfare states that humans can realize), then this objection is assuming what needs to be established. And once we notice this, it’s less plausible that utilitarianism is off the hook when it comes to counterintuitive practical implications.
Of course, if it were—or ought to be—obvious that animals have much smaller welfare ranges than humans, then assuming as much would be fine. But it isn’t. One way to appreciate this is to note that many theories of welfare seem to imply that humans and animals have the same welfare ranges—e.g., any theory where flourishing is assessed relative to species-specific standards, as in perfectionism. We get the same result from views that measure welfare in proportional rather than absolute terms: consider, for instance, a simple version of desire satisfactionism on which welfare is determined by the ratio of the percentage of satisfied desires to the percentage of unsatisfied desires. More radically, it’s possible that variablism is true, according to which different theories of welfare are true of different welfare subjects, which is most naturally developed in a way that supports the Equality Result. Moreover, note that none of these theories was designed to produce the Equality Result; it just falls out of independently-motivated parts of each view.
In any case, there’s a second, methodological problem with this objection. Essentially, it amounts to casting doubt on a broadly empirical conclusion—that particular animals have certain capacities that allow them to realize some amount of welfare—based on a moral conclusion. But we wouldn’t accept that inference in other contexts. It would be perverse to deny that people are dying of preventable diseases because, if they were, we might have strenuous obligations to provide aid. Likewise, insofar as welfare ranges are assessable via empirical methods, it seems perverse to deny that animals have particular welfare ranges based on the moral consequences of recognizing them.
Granted, differences in welfare ranges aren’t value-free: if there are any such differences, they’re generated by variation in animals’ ability to realize the determinants of welfare—i.e., given hedonism, the intensities of valenced states. But once we fix the theory of welfare, differences in welfare ranges should be grounded in empirical differences, not philosophical ones. Of course, someone could reject hedonism. That, however, isn’t the objection we’re considering here. Instead, we’re considering the objection that the Equality Result is implausible because of its moral implications. If those moral implications follow from the best empirical evidence available, then that seems like the wrong kind of authority to grant to our moral judgments.
Granted again, a methodology for assessing welfare ranges isn’t philosophically innocent: it includes various assumptions about what would constitute good evidence for differences in the intensities of valenced states across species. Here again, though, it’s strange to think that we should question methodological decisions because they lead to uncomfortable practical results. If the methodology is broken, it’s broken on its own terms—not because it’s part of an argument for allocating more money to animals.
The third problem—which goes back to the original, “Don’t shoot the messenger” response—is that if we’re singling out a party for blame, it ought to be utilitarianism. After all, utiltarianism’s consequentialism makes it outcomes-focused, its welfarism makes welfare realization its exclusive focus, its impartiality makes it impossible to bracket the concerns of animals, and its aggregationism allows the sheer number of animals to do the rest. We’re probably in for radical practical implications even if we have big differences in welfare ranges. To be clear: for all that, utilitarianism may be the best theory around! So, we aren’t suggesting that radical practical implications are a strike against utilitarianism. Instead, we’re suggesting that if radical practical implications are the problem, then the Equality Result is probably the wrong place to focus.
The upshot is that “Don’t shoot the messenger” remains the right response to someone balking at the practical results of the Equality Result. If you don’t like those results, there are lots of other places to look.
Balking at the Assumptions Behind the Equality Result
Second, someone might balk at the assumptions behind the Equality Result. In particular, someone might be inclined to accept some other theory of welfare that does not support the Equality Result.
Of course, as we’ve just seen, the Equality Result doesn’t depend on hedonism: it’s supported by—or at least compatible with—many theories of welfare and one meta-theory of welfare. But let’s set that aside, as surely there are theories of welfare that wouldn’t lead to the Equality Result.
This brings us to a concession and a retrenchment. Perhaps unsurprisingly, our view is that if you have good reason to accept some theory of welfare that doesn’t support the Equality Result, then you also have good reason not to believe the Equality Result. That’s the concession.
That being said, there are two points to note. First, as we’ve argued elsewhere, we doubt that abandoning hedonism will change the bottom line all that much. While hedonic goods and bads aren’t all of welfare, they’re surely a large part of it.
Second, note the conditional here: “if you have good reason to accept some theory of welfare that doesn’t support the Equality Result.” If you’re just reacting against the Equality Result itself—without any theoretical backing—then it’s crucial to assess the evidential value of the anti-Equality-Result intuition. Let’s do that in the context of assessing the third reason someone might balk at the Equality Result.
Balking at the Equality Result Even Given the Assumptions that Support It
Third, then, someone might balk at the Equality Result even given the assumptions that ostensibly support it. That is, they might think that even if we assume hedonism, we shouldn’t end up with the Equality Result. They might think it’s just implausible that a chicken can suffer as intensely as a human—or, more positively, that a chicken’s pleasures can be nearly as intense as a human’s.
The basic problem with this is that the intuition is uncalibrated, uncalibrated intuitions are vulnerable to various biases, and there are some highly relevant biases in the present context.
Intuitions are calibrated when they’re checked against some independent standard. Forecasters, for instance, calibrate their intuitions by making forecasts, seeing how they resolve, and learning from their successes and failures. But intuitions about hedonic capacity—the relative potential intensities of valenced states across organisms—aren’t going to “resolve” anytime soon; we’re a long way from a rigorous science of interspecies pain intensity comparisons.
Moreover, there’s a large literature on how uncalibrated intuitions can be affected by irrelevant factors. We won’t review it here. Instead, let’s consider some of the irrelevant factors that could affect intuitions about chickens specifically.
First, people can have the wrong target in mind. Most obviously, it’s easy to confuse greater diversity of valenced states with greater intensity. It’s probably true that humans have some affective states that chickens lack. Moreover, those human-specific affective states may be tied to aspects of human life that people value, such as human friendship, romantic love, parenthood, and so on. Still, unique affective states aren’t therefore more intense. The pain of losing a friend is significant, but at any given moment, it isn’t necessarily more intense than, say, the pain of breaking a limb. But those peaks are what matter for assessing welfare ranges.
Second, people can mistakenly think that:
- I’m unable to imagine what it’s like to be a chicken.
…is good evidence for:
- When chickens feel pain, it’s very different from what I feel.
…and then it’s a short step to:
- Chickens’ pain isn’t as intense as my pain.
When we make the inferences explicit, the problems are obvious. There are lots of differences between humans and animals that make it difficult to imagine their inner lives. This isn’t a reason to think that their pains have an entirely different character or that their pains are any less painful. Indeed, one feature of excruciating pain is how consuming it is: it “drowns out” all our other cognitive abilities, which suggests that when it comes to pain, our experiences may be most similar to those of other animals. Moreover, when it comes to the second inference, different isn’t the same as less intense.
Third, people seem to be influenced even by the fact that some animals are categorized as consumable. Bratanova et al. (2011), for instance, told participants about a marsupial native to Papa New Guinea, suggesting to some that the marsupial is a food animal that humans hunt; they didn’t give such hints to the control group. This difference alone was enough to lead participants to judge that the animal is less likely to have the capacity to suffer and less deserving of moral concern. (For related results, see Bastian et al. (2012), Loughnan, Haslam, and Bastian (2010), Ruby and Heine, 2012, and Kunst and Hohle, 2016.)
In short, we don’t have widely-accepted empirical reasons to think that other animals have less intense pleasures and pains than humans do. So, while we might have intuitions about the intensities of their valenced states, there’s no particular reason to trust them, as they aren’t calibrated. Moreover, there are positive reasons not to trust them, as there are various clear ways that these intuitions could go wrong.
If Hedonism Is True, Then the Equality Result Shouldn’t Be Surprising
In what remains, we’ll explain why the Equality Result wouldn’t be surprising if hedonism is true. This is not the same as arguing that if hedonism is true, then we ought to expect the Equality Result. The point here is more modest. We’re only arguing that the Equality Result wouldn’t be particularly notable given hedonism. In other words, conditional on hedonism but in advance of inquiry, we shouldn’t assign a low credence the hypothesis that humans and chickens have the same welfare ranges.
We assume that valenced experiences didn’t pop into existence: they evolved because they’re adaptive for organisms that have them. So, if there are differences in the possible intensities of valenced states across organisms, then, with various caveats, we should be able to explain those differences in terms of their adaptive value. Our basic suggestion is that there are plausible sets of hypotheses about the adaptive value of valenced states on which there aren’t any differences at all—hence, the Equality Result.
To see this, let’s consider three popular theories about the function of valenced experiences: they represent fitness-relevant information in a motivationally-salient way (Cutter & Tye 2011); they provide a common currency for decision making (Ginsburg & Jablonka 2019); they facilitate learning (Damasio & Carvalho 2013). These theories don’t just tell us what valenced states actually do for organisms that have them. Additionally, they suggest etiologies for valenced states. If the representational view is true, then, probably, organisms that could represent fitness-relevant information were more successful than organisms without that ability because the former, and not the latter, could act on that information. Likewise, if the common currency theory is true, then, probably, organisms with a single currency for decision making were more successful than organisms without that didn’t have a single currency. And so on.
The question, then, is whether these functions or etiologies could plausibly deliver the Equality Result. We think they could.
If the representational theory is true, then it might seem obvious that there would be variation in the abilities of organisms to represent fitness-relevant information. Presumably, humans can represent much more fitness-relevant information than chickens. However, there are two reasons why we shouldn’t make too much of this.
First, it doesn’t seem plausible that the intensities of pains scale directly with representational abilities. When we consider our own experiences, it’s notable that our pains don’t obviously provide us with information that other animals lack when they’re in pain. Do we learn more about the location of damage when we’re in pain? Do we learn more about the kind of damage? If not, then why would we expect to learn much more about the severity of the threat to fitness—which, presumably, is what the information about intensity conveys?
Second, even if we were to think that humans have more sophisticated pain perception without increasing intensity, it’s an open question whether cognitive sophistication allows organisms to make more fine-grained distinctions within an intensity range (call this the “compression hypothesis”) rather than increasing the intensity range of their experiences (call this the “expansion hypothesis”—names due to Henry Shevlin). That is, it could work out that the worst pain for any given sentient organism is as intense as the worst pain for any other sentient organism. However, perhaps some organisms can distinguish 99 grades of pain that fall short of that worst form of suffering, whereas other organisms can only distinguish 3. On such a view, representational sophistication adds nuances to an intensity scale; it doesn’t push the endpoints further from one another.
Let’s now consider the decision-making theory, where valenced states help organisms rank some options (e.g., fleeing) as better than others (e.g., fighting) in a motivationally-salient way (Ginsburg & Jablonka 2019). However, organisms don’t need to be able to rank all possible options at a given time; they need to be able to represent some salient set of them. How could we estimate the number of such options and, therefore, the intensity range of organisms’ valenced experiences? One possibility is to use evidence from the working memory literature about how many units of information organisms can keep in mind simultaneously. Roughly, working memory is a cognitive system that temporarily makes limited amounts of information available for tasks that require executive control. For humans, that number may be as low as four units (Cowan 2005), which doesn’t leave much room for variation between humans and chickens. So, if the evidence were to support the Equality Result, it wouldn’t be that surprising.
Finally, there’s the learning theory. On this view, the “key function [of pain] appears to be that the aversive experience of pain creates a strong and lasting motivation that enables the animal to avoid getting into a similar situation in the future” (Sneddon et al. 2014, 202). Correspondingly, a key function of pleasure is to create a strong and lasting motivation to get into similar situations in the future. Given this theory, we might expect some variation in the intensity of valenced states based on how much animals stand to gain, fitness-wise, from having such strong and lasting motivations.
However, it isn’t at all clear that chickens have less to gain, fitness-wise, from strong and lasting motivations. This is a difficult empirical question that it would take considerable work to assess. Moreover, it’s important to remember that any plausible theory of humans’ success is going to invoke many other abilities besides having strong and lasting motivations. If anything, those abilities may reduce the need for particularly intense states rather than increase it, as humans can reason about the value of avoiding future pains and securing future pleasures in a way that chickens, presumably, cannot. So, chickens might need more intense states to accomplish that same aims—a hypothesis entertained by more than one evolutionary biologist (see, e.g., Dawkins 2017 and Barash 2022).
Granting What We’ve Criticized
Nevertheless, let’s suppose that as representational capacity, working memory, and long-term memory increase, the potential intensity of valenced states increases. Why would this be adaptive? There are two basic stories. One of them is that the mechanisms that produce valenced states are somehow bound up with these capacities, such that it’s inherently the case that these changes produce more intensely valenced states. That could be true, but there’s no obvious reason to believe it. So, we set it aside.
The other possibility is that there’s some link between those abilities and the kinds of threats to fitness that those abilities help organisms ward off, such that organisms with these abilities developed more intensely valenced states to help them avoid those threats. But that isn’t what we see. Across species, the most intense pains are associated with bodily damage, hunger, thirst, and other threats to survival; the most intense pleasures are associated with food, sex, play, and other obvious boons to survival. We don’t see an obvious shift in the objects of the most intense valenced states—the things that organisms’ valenced states are about—which, presumably, is what we’d get if nature were introducing new intensity levels to take advantage of new cognitive abilities. It isn’t the case that people find it especially intensely painful nowto do things that will lead to them failing to produce offspring later, or that they find it especially intensely painful now to neglect responsibilities that would ensure adequate food later.
Granted, these kinds of failures can be negatively valenced. But that isn’t enough. The hypothesis in question is that these capacities created a selection pressure for more intensely valenced states because it was now advantageous to have those new intensity levels. The point, essentially, is that if those new intensity levels were advantageous because of the relevant abilities, we would expect those states to be somehow associated with the use of those abilities. But that isn’t what we see.
If the Moral Weight Project were to deliver the Equality Result, would that be a reason to think that there’s a mistake somewhere up the chain? In short, no. It’s true that if you have good reason to accept some theory of welfare that doesn’t support the Equality Result, then you also have good reason not to believe the Equality Result. However, we suspect that that situation is relatively rare. Most people will balk at either the practical implications of the Equality Result or at the Equality Result itself. But the practical implications aren’t really the Equality Result’s fault; you should blame utilitarianism if you don’t like them. And while the Equality Result may be counterintuitive, there are plenty of reasons not to make very much of its being counterintuitive. That anti-Equality-Result intuition isn’t calibrated, and so may well be vulnerable to significant biases.
On top of all that, conditional on hedonism, we shouldn’t assign a particularly low credence to the Equality Result. There are popular theories of valence that seem to fit fairly naturally with it. So, while that isn’t an argument for the Equality Result, it’s a check against knee-jerk skepticism.
Finally, let’s be clear: we are not claiming that the Equality Result is correct. Instead, our claim is that given the assumptions behind the Moral Weight Project (and perhaps even without them), we shouldn’t flinch at “animal-friendly” results.
This research is a project of Rethink Priorities. It was written by Bob Fischer. Thanks to Marcus Davis, Jason Schukraft, Adam Shriver, and Michael St. Jules for helpful feedback on earlier versions of this post. If you’re interested in RP’s work, you can learn more by visiting our research database. For regular updates, please consider subscribing to our newsletter.
Granted, these aren’t the only possible ways of resisting the Equality Result. Someone might be convinced that neuron counts are an excellent proxy for relative moral weights, that brains support many conscious subsystems, or that there are neurological reasons to think that some animals are basically split-brain patients. However, we've discussed those possibilities elsewhere and we bracket them here.
Granted, some think that fitting with intuitions about cases is a relatively small part of the case for utilitarianism, if a factor at all. Many EAs, for instance, might think that it's a good option among plausibly fully describable (though not actually described in full) expected-value-maximizing options, which they support based on views about the requirements of rationality, the nature of moral patienthood, and so on. Still, this leaves the remaining objections untouched. Thanks to Marcus Davis for this point.
For an overview of the heuristics and biases literature, see Kahneman 2011. For philosophical examples, see Machery et al. 2004; Swain, Alexander, & Weinberg 2008; Buckwalter & Stich 2014; Costa et al. 2014, Schwitzgebel & Cushman 2012, and De Cruz 2015.
I would make the same claim more strongly: "modus tollens" / "reductio ad absurdum" (as in, "this assumption gives a conclusion I don't like", rather than "this gives an internally inconsistent conclusion") style ethical reasoning is, broadly speaking, not good. Unless you believe standard 21st century morality is correct about everything, you should expect your ethical assumptions to lead to some unexpected results. Ozy wrote something about this that I really liked:
OP said something similar, which is a less general argument but, I think, harder to dispute:
I thought this was the most powerful sentence under the "Balking at the Implications" heading.
Thanks for this, Michael! I hadn't seen that line from Ozy. I really like it.
Do your arguments really support this? Or just the weaker S-Equality Result, that chickens and humans have roughly the same capacity to suffer negative welfare?
fwiw, only that weaker claim seems plausible to me, since the hedonic contribution to welfare strikes me as plausibly asymmetric: suffering is very bad, but mere pleasure is only mildly good. I think the vast majority of positive well-being comes from non-hedonic sources. So I don't think chickens or wireheading humans can realize much positive well-being at all -- like, orders of magnitude less than a well-lived human life (featuring love, genuine accomplishments, etc.).
Thanks for this question, Richard. You're right that I don't focus on positive affective states in the post, though I think most of the arguments would port over. In any case, since the MWP assumes hedonism, the result that chickens and wireheaded humans can realize the same amount of welfare is still pretty significant. Indeed, even the weaker S-Equality Result is significant if your asymmetry hypothesis is correct, as S-Equality would get you most of the way toward (plain old) Equality.
Separately, and as you might guess, I'm skeptical of the view that humans who are maxed out hedonically are still realizing orders of magnitude less welfare than humans who are flourishing by more conventional standards. I think the intuitions that support that view boil down to humans preferring the human way of life--a preference that doesn't strike me as having much evidential value. But I suppose that's a conversation for another time!
I think the whole issue of "one person's modus ponens is another's person's modus tolens" is not very well understood by most people and including most philosophers and myself. In fact, I'm don't think anyone knows quite how to think about these things. I guess it gets into Quinean holism and the intractability problems that accompany it.
But, presumably, it has something to do with Bayesian networks of beliefs and regularization in machine learning (~valuing simplicity) as well as Bayesian philosophy of science more generally. [Part IV of Itzhak Gilboa's decision theory book gets into some of this stuff, which seemed pretty interesting.]
I don't understand why much more attention is not paid to these things in philosophy, where formal epistemology seems to still be considered a pretty niche field.
I hope people think more about these issues.
Thank you for writing this. This is awesome.
Appreciate the support!