An Introduction to the Moral Weight Project

By Bob Fischer @ 2022-10-31T09:13 (+179)

Key Takeaways

• Our objective: provide “moral weights” for 11 farmed species.
• To make this tractable, we made four assumptions: utilitarianism, hedonism, valence symmetry, and unitarianism.
• Given these assumptions, an animal’s “moral weight” is that animal’s capacity for welfare—the total amount of welfare that the animal could realize.
• Capacity for welfare = welfare range (the difference between the best and worst welfare states the individual can realize at a time) × lifespan.
• Given welfare ranges, we can convert welfare improvements into DALY-equivalents averted, making cross-species cost-effectiveness analyses possible.

An Introduction to the Moral Weight Project

If we want to do as much good as possible, we have to compare all the ways of doing good—including ways that involve helping members of different species. But do benefits to humans count the same as benefits to chickens? What about chickens vs. carp? Carp vs. honey bees? In 2020, Rethink Priorities published the Moral Weight Series—a collection of five reports about these and related questions. The first introduces different theories of welfare and moral status and their interrelationships. The second compares two ways of estimating differences in capacity for welfare and moral status. The third explores the rate of subjective experience, its importance, and potential variation in the rate of subjective experience across taxa. The fourth considers critical flicker-fusion frequency as a proxy for the rate of subjective experience. The fifth assesses whether there’s variation in the intensity ranges of valenced experiences from species to species.

In May 2021, Rethink Priorities launched the Moral Weight Project, which extended and implemented the research program that those initial reports discussed. This post is the first 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. The aim of this post is to introduce the project and explain how EAs could use its results.
 

DALYs-averted and “moral weight discounts” 

If we want to do the most good per dollar spent—that is, if we want to maximize cost-effectiveness—we need a common currency for comparing very different interventions. GiveWell, Founders Pledge, Open Philanthropy, and many other EA organizations currently have one: namely, the number of “disability-adjusted life years (DALYs) averted.” A DALY is a health measure with two parts: years of human life lost and years of human life lost to disability. The former measures the extent to which a condition shortens a human’s life; the latter measures the health impact of living with a condition in terms of years of life lost. Together, these values represent the overall burden of the condition. So, averting a DALY is averting a loss—namely, the loss of a single year of human life that’s lived at full health.

Historically, some EAs used a “moral weight discount” to convert changes in animals’ welfare levels directly into DALYs-averted. That is, they understood the basic question to be:

At what point should we be indifferent between (say) improving chickens’ welfare and preventing the loss of a year of healthy human life?

Then, the task is to identify the correct “moral weight discount rate” to apply to the value of some quantity of chicken welfare to make it equal to the value of averting a DALY.

This framing makes it tempting to think in terms of the value that (certain groups of) people assign to chicken welfare (or the welfare of the members of whatever species) relative to the value that (certain groups of) people assign to preventing the loss of a year of healthy human life. This amounts to reinterpreting the question as follows:

At what point are we indifferent between (say) improving chickens’ welfare and preventing the loss of a year of healthy human life?

We can tackle this new question by surveying people. Originally, though, we weren't asking a sociological question, but a normative one. The task is to determine when people ought to be indifferent between improving chickens’ welfare and preventing the loss of a year of healthy human life. 

Ultimately, this requires sorting through the many combinations of plausible moral assumptions, sorting through solutions to the problem of moral uncertainty, sorting through responses to uncertainty about how to handle moral uncertainty, and so on. It’s a gargantuan task. In the interest of making some progress, then, we’re going to make some moral assumptions and proceed accordingly. In particular, we assume:

• Utilitarianism, according to which you ought to maximize (expected) utility.
• Hedonism, according to which welfare is determined wholly by positively and negatively valenced experiences (roughly, experiences that feel good and bad to the subject).
• Valence symmetry, according to which positively and negatively valenced experiences of equal intensities have symmetrical impacts on welfare.
• Unitarianism, according to which equal amounts of welfare count equally, regardless of whose welfare it is.

There are familiar arguments for and against these assumptions. We won’t discuss such arguments here. Instead, we’ll simply flag that while some aspects of the Moral Weight Project depend on these assumptions, much of it doesn’t. In fact, we devote an entire post to discussing the impact of rejecting hedonism. So, we hope that our work still proves useful to those with other philosophical views.

In any case, given these assumptions, people ought to be indifferent between improving chickens’ welfare and preventing the loss of a year of healthy human life when the gain in welfare is equivalent. And if that’s the case, then the task is to determine when an improvement in animal welfare results in as much welfare as you’d produce by extending a healthy human life by one year.

Or, rather, that’s how we’d answer the original, "when should we be indifferent" question. Again, though, what matters is that we find a way to express a change in animals’ welfare levels in the same common currency that people already use to compare very different interventions: namely, as some number of DALY-equivalents averted. That’s what matters for cost-effectiveness analyses across human and animal interventions, so that’s the mandate of the Moral Weight Project.

Crucial concepts: capacity for welfare and welfare ranges

Given the moral assumptions mentioned above, we should reinterpret the concept of a “moral weight” for members of particular species. Instead of being a measure of how much people value members of [insert species name], a “moral weight” is a measure of how much welfare animals of different kinds can potentially realize, produce, or generate—their capacity for welfare. That is, our assumptions entail that everyone’s welfare counts the same and that all welfare improvements count equally; still, members of some species may be able to realize more welfare than others. In essence, if welfare is like water, some individuals may be larger “buckets.” (See Figure 1.) An individual doesn’t matter more because they could realize more welfare than another; indeed, if unitarianism is true, individuals as such never matter more or less than others. Instead, if an individual’s interests matter more, it’s because more welfare is affected when those interests go satisfied or unsatisfied.

Figure 1: If welfare is like water, some individuals may be larger “buckets.”

An individual’s capacity for welfare has two components: their welfare range (which is just their capacity for welfare at a time) and their lifespan. Since lifespans are relatively easy to estimate—and, in any case, are often irrelevant because we’re looking at welfare impacts over fixed periods—we’ll ignore them here. 

An individual’s welfare range is the difference between the best and worst welfare states the individual can realize at a time. In other words, assume we can assign a positive number to the best welfare state the individual can realize and a negative number to the worst welfare state the individual can realize. The difference between them is their welfare range. (See Figure 2.)

Figure 2: In this diagram, the x-axis represents lifespan; the y-axis represents the amounts of welfare that an individual could realize at a time. The total area shaded red represents Red’s capacity for welfare; the total area shaded blue represents Blue’s capacity or welfare. The distance between A and B is Red’s welfare range; the distance between C and D, Blue’s welfare range.

Ultimately, the goal is to be able to compare welfare improvements and DALYs-averted. Welfare improvements are changes in realized welfare that we estimate using welfare ranges. So, while an individual’s welfare range is the difference between the “best” and “worst” welfare states the individual can realize, we aren’t interested in the best and worst in principle, as many possibilities are simply irrelevant to comparing welfare improvements and DALYs-averted. Instead, we’re interested in an interpretation of “best” and “worst” that lends itself to expressing welfare level changes in terms of DALYs-averted. 

With this in mind, we take the best human welfare state to be the average welfare level of the average human-in-full-health, whatever that happens to be. Plainly, there are many other senses of “best” where humans realize more welfare still. The average welfare level of the average human-in-full-health is hardly the pinnacle of welfare realization. But if we indexed to the maximum amount of welfare that a human-in-full-health could realize—however we interpret the modal force of “could” in that statement—we’d be left with difficult questions about the relationship between animals’ welfare ranges and a DALY-averted. By indexing to the average welfare level of the average human-in-full-health, the relationship is straightforward. 

Moreover, it’s the relationship we want to capture. The goal is to estimate the welfare gain from, say, moving layer hens from cages to a cage-free system relative to the welfare gain from extending human lives—not from creating perfect human lives. So, while it’s possible that various animals can realize more positive welfare at a time than the average welfare level of the average human-in-full-health, we think the greater risk here is underestimating total welfare impacts on animals by indexing to a welfare level that’s too high. 

On that note, we should flag that the DALY framework doesn’t allow for states worse than death. Effectively, it assumes that humans’ net welfare range is exclusively positive, not in the sense that humans can’t have any negative welfare states, but in the sense that their welfare is net positive even when in negative welfare states. We can understand why people chose to bake this assumption into the framework, given inevitable disagreements about which lives are worth living and the DALY framework’s use in resource allocation. However, whatever the merits of the net positivity assumption, it’s controversial as a theoretical claim. Moreover, it will affect our estimates of the relative values of different interventions. So, while we can get away with granting the assumption here, as this post is about the conceptual connection between animal welfare and a DALY-averted, the issue requires further investigation before actually doing cost-effectiveness analyses.

Whatever we conclude about the DALY framework, though, there’s no obvious moral reason for assuming that many nonhuman animals have net positive lives. Indeed, it’s quite plausible that many chickens, for instance, have net negative welfare states in standard intensive farming systems. So, while we’ll grant the DALY framework’s assumption that there are no states worse than death for humans, we won’t assume that nonhuman animals’ lives are invariably net positive.

If we allow that animals can be in states worse than death, we face the question of the skews of their welfare ranges. Can they be worse off than they can be well off? Better off than they can be badly off? Or are their welfare ranges symmetrical around the neutral point, such that they can be equally well off and badly off? Though this topic deserves considerable attention, it would take us too far afield here. (See this report for a preliminary investigation.) So, let’s assume symmetry. We’ll say that the relevant best animal welfare states are the average welfare levels of the average animal-(of a given species)-in-full-health. If we represent that welfare state with some arbitrary positive number, then the “worst” welfare state is represented by its negation.

Welfare ranges + species-relative welfare assessments = DALY-equivalents averted

Now suppose that we can estimate differences in welfare ranges. Then, we can convert species-relative welfare assessments—understood as assessments of animals’ welfare relative to how well off or poorly off they could be—into a common unit. Suppose, for example, that chickens’ welfare range is 10% of humans’ welfare range. Let’s also suppose that over the course of a year, the average chicken is about half as badly off as they could be in a conventional cage (i.e., their average welfare level is roughly halfway between their neutral point and their worst welfare state) and a quarter as badly off as they could be in an aviary system (i.e., their average welfare level is roughly a quarter of the way from their neutral point toward their worst welfare state). Then we can calculate the equivalent amount of human welfare and, by extension, the number of DALY-equivalents averted: 

1. Chickens’ welfare range is 10% of human’s welfare range.
2. Assuming symmetry around the neutral point, the negative portion of chickens’ welfare range is 5% of human’s positive welfare range. (For instance, if humans’ welfare range is 100 and chickens’ welfare range is 10, chickens range from -5 to 5, and the negative portion of that is 5% of humans’ welfare range.)
3. Given our assumptions about the welfare impacts of the two production systems, the move from conventional cages to aviary systems averts an amount of welfare equivalent to 25% of the average chicken’s negative welfare range. (Continuing with the numbers mentioned in the previous step, it moves chickens from -2.5 to -1.25).
4. So, assuming symmetry around the neutral point, 25% of chickens’ negative welfare range is equivalent to 1.25% (5% × 25%) of humans’ positive welfare range. 
5. By definition, averting a DALY averts the loss of an amount of welfare equivalent to the positive portion of humans’ welfare range for a year.
6. So, assuming symmetry around the neutral point, the move from conventional cages to aviary systems averts the equivalent of 0.0125 DALYs per chicken per year on average.

Two caveats. First, we want to stress that we are using these numbers for purely illustrative purposes. We are not suggesting that chickens' welfare range is 10% of humans' welfare range; we are not suggesting that the average chicken is about half as badly off as they could be in a conventional cage and a quarter as badly off as they could be in an aviary system. We'll give our estimate of chickens' welfare range toward the end of this series, but specific welfare assessments are beyond the scope of the project.

Second and relatedly, the Moral Weight Project focuses on questions about animals’ welfare ranges, not on questions about how to make species-relative welfare assessments. We simply assume that it’s possible to assess—however coarsely—animals’ welfare relative to how well off or poorly off they could be, leaving the details for future work. We grant the details will be controversial, as they require saying something about the location of the neutral point—that is, the point at which any increase in welfare makes the animal net positive overall and any decrease in welfare makes the animal net negative overall. For many practical purposes, though, we only need the roughest estimates. And those, we submit, are probably within reach.  

How could we estimate relative welfare ranges?

Over the next several posts, we lay the groundwork for estimating differences in welfare ranges. First, we describe the Welfare Range Table—a literature review of over 90 qualitative and quantitative proxies for variation in welfare ranges across 11 farmed species.

Then, we discuss the impact of different theories of welfare on welfare range estimates. Is all the action in the philosophy rather than the empirical evidence? We argue against this view.

After that, we turn to the hypothesis that we should use estimates of the total number of neurons in an animal’s brain as a proxy for its moral weight.

In the wake of our discussion of neuron counts, we take up two issues that aren’t related to welfare ranges per se, but which would affect the total amount of welfare that we ascribe to any particular individual: first, the possibility that various subsystems within brains are conscious in addition to the whole brain; second, the possibility that some animals don’t have a single, unified conscious perspective on the world.

With those conceptual issues behind us, we describe a set of models that we used to estimate differences in welfare ranges across taxa. 

Finally, we conclude the series with an application of all the above, investigating potential differences in welfare ranges across insect life stages—that is, from their larval to adult stages.

Acknowledgments

This research is a project of Rethink Priorities. It was written by Bob Fischer. Thanks to Rachel Norman, Martina Schiestl, Alex Schnell, and Anna Trevarthen 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.
 

Vasco Grilo @ 2022-11-05T19:27 (+1)

Thanks for sharing, I am looking forward to the series!

To make this tractable, we made four assumptions: utilitarianism, hedonism, valence symmetry, and unitarianism.

I am glad you made these assumptions! They seem the most reasonable and action-guiding.

So, let’s assume symmetry. We’ll say that the relevant best animal welfare states are the average welfare levels of the average animal-(of a given species)-in-full-health. If we represent that welfare state with some arbitrary positive number, then the “worst” welfare state is represented by its negation.

I have estimated here the welfare of broilers in a conventional scenario is 2.21 times as bad as their full health is good, and the welfare of hens in conventional cages is 2.19 times as bad as their full health is good. These estimates are pretty close to 1 given the high uncertainty, but I am curious about whether you will elaborate on the symmetry assumption in the series.

Bob Fischer @ 2022-11-06T11:46 (+2)

Thanks for your comment, Vasco! We don't discuss the symmetry assumption in other posts. Our only discussion is in the supplementary report that's linked above, which is focused on axiological asymmetries (where the theory of welfare itself posits a fundamental asymmetry), not contingent asymmetries (where contingent physiological facts about organisms explain the asymmetry). But to be clear: we aren't attached to the symmetry assumption and agree that it's controversial. We make it purely to simplify the project, as it makes it much easier to define a lower welfare bound (negating the number that represents full health) and, for that reason, makes cross-species comparisons more straightforward. That being said, we've built a BOTEC for doing cost-effectiveness analyses with welfare ranges (not yet released), and that tool allows you to factor in different welfare range "skews" (i.e., the appropriate asymmetry for a species). We're also interested in doing more work on asymmetry in the future. So I agree with you about the importance of the issue!