RP’s moral weights and analysis of cage-free campaigns suggest that the average cost-effectiveness of cage-free campaigns is on the order of 1000x that of GiveWell’s top charities.[5] Even if the campaigns’ marginal cost-effectiveness is 10x worse than the average, that would be 100x.
This seems to be the key claim of the piece, so why isn’t the “1000x” calculation actually spelled out?
The “cage-free campaigns analysis” estimates
how many chickens will be affected by corporate cage-free and broiler welfare commitments won by all charities, in all countries, during all the years between 2005 and the end of 2018
This analysis gives chicken years affected per dollar as 9.6-120 (95%CI), with 41 as the median estimate.
The moral weights analysis estimates “welfare ranges”, ie, the difference in moral value between the best possible and worst possible experience for a given species. This doesn’t actually tell us anything about the disutility of caging chickens. For that you would need to make up some additional numbers:
Welfare ranges allow us to convert species-relative welfare assessments, understood as percentage changes in the portions of animals’ welfare ranges, into a common unit. To illustrate, let’s make the following assumptions:
Chickens’ welfare range is 10% of humans’ welfare range.
Over the course of a year, the average chicken is about half as badly off as they could be in conventional cages (they’re at the ~50% mark in the negative portion of their welfare range).
Over the course of a year, the average chicken is about a quarter as badly off as they could be in a cage-free system (they’re at the ~25% mark in the negative portion of their welfare range).
Anyway, the 95%CI for chicken welfare ranges (as a fraction of human ranges) is 0.002-0.869, with 0.332 as the median estimate.
So if we make the additional assumptions that:
All future animal welfare interventions will be as effective as past efforts (which seems implausible given diminishing marginal returns)
Cages cause chickens to lose half of their average welfare (a totally made up number)
Then we can multiply these out to get:
The “DALYs / $ through GiveWell charities” comes from the fact that it costs ~$5000 to save the life of a child. Assming “save a life” means adding ~50 years to the lifespan, that means $100 / DALY, or 0.01 DALYs / $.
A few things to note here:
There is huge uncertainty here. The 95% CI in the table indicates that chicken interventions could be anywhere from 10,000x to 0.1x as effective as human charities. (Although I got the 5th and 95th percentiles of the output by simply multiplying the 5th and 95th percentiles of the inputs. This is not correct, but I’m not sure there’s a better approach without more information about the input distributions.)
To get these estimates we had to make some implausible assumptions and also totally make up some numbers.
The old cortical neuron count proxy for moral weight says that one chicken life year is worth 0.003, which is 1/100th of the RP welfare range estimate of 0.33. This number would mean chicken interventions are only 0.7x as effective as human interventions, rather than 700x as effective. [edit: oops, maths wrong here. see Michael’s comment below.]
But didn’t RP prove that cortical neuron counts are fake?
Hardly. They gave a bunch of reasons why we might be skeptical of neuron count (summarised here). But I think the reasons in favour of using cortical neuron count as a proxy for moral weight are stronger than the objections. And that still doesn’t give us any reason to think RP’s has a better methodology for calculating moral weights. It just tells us to not take cortical counts to literally.
Points in favour of cortical neuron counts as a proxy for moral weight:
Neuron counts correlate with our intuitions of moral weights. Cortical counts would say that ~300 chicken life years are morally equivalent to one human life year, which sounds about right.
There’s a common sense story of: more neurons → more compute power → more consciousness.
It’s a simple and practical approach. Obtaining the moral weight of an arbitrary animal only requires counting neurons.
Compare with the RP moral weights:
If we interpret the welfare ranges as moral weights, then 3 chicken life years are worth one human life year. This is not a trade I would make.
If we don’t interpret welfare ranges as moral weights, then the RP numbers tell us literally nothing.
The methodology is complex, difficult to understand, expensive, and requires reams zoological observation to be applied to new animals.
And let’s not forget second order effects. Raising people out of poverty can increase global innovation and specialisation and accelerate economic development which could have benefits centuries from now. It’s not obvious that helping chickens has any real second order effects.
In conclusion:
It’s not obvious to me that RP’s research actually tells us anything useful about the effectiveness of animal charities compared to human charities.
There are tons of assumptions and simplifications that go into these RP numbers, so any conclusions we can draw must be low confidence.
Cortical neuron counts still looks like a pretty good way to compare welfare across species. Under cortical neuron count, human charities come out on top.
Thanks for all this, Hamish. For what it’s worth, I don’t think we did a great job communicating the results of the Moral Weight Project.
As you rightly observe, welfare ranges aren’t moral weights without some key philosophical assumptions. Although we did discuss the significance of those assumptions in independentposts, we could have done a much better job explaining how those assumptions should affect the interpretation of our point estimates.
Speaking of the point estimates, I regret leading with them: as we said, they’re really just placeholders in the face of deep uncertainty. We should have led with our actual conclusions, the basics of which are that the relevant vertebrates are probably within an OOM of humans and shrimps and the relevant adult insects are probably within two OOMs of the vertebrates. My guess is that you and I disagree less than you might think about the range of reasonable moral weights across species, even if the centers of my probability masses are higher than yours.
I agree that our methodology is complex and hard to understand. But it would be surprising if there were a simple, easy-to-understand way to estimate the possible differences in the intensities of valenced states across species. Likewise, I agree that “there are tons of assumptions and simplifications that go into these RP numbers, so any conclusions we can draw must be low confidence.” But there are also tons of assumptions and biases that go into our intuitive assessments of the relative moral importance of various kinds of nonhuman animals. So, a lot comes down to how much stock you put in your intuitions. As you might guess, I think we have lots of reasons not to trust them once we take on key moral assumptions like utilitiarianism. So, I take much of the value of the Moral Weight Project to be in the mere fact that it tries to reach moral weights from first principles.
It’s time to do some serious surveying to get a better sense of the community’s moral weights. I also think there’s a bunch of good work to do on the significance of philosophical / moral uncertainty here. I If anyone wants to support this work, please let me know!
Thanks for responding to my hot takes with patience and good humour!
Your defenses and caveats all sound very reasonable.
the relevant vertebrates are probably within an OOM of humans
So given this, you’d agree with the conclusion of the original piece? At least if we take the “number of chickens affected per dollar” input as correct?
I agree with Ariel that OP should probably be spending more on animals (and I really appreciate all the work he’s done to push this conversation forward). I don’t know whether OP should allocate most neartermist funding to AW as I haven’t looked into lots of the relevant issues. Most obviously, while the return curves for at least some human-focused neartermist options are probably pretty flat (just think of GiveDirectly), the curves for various sorts of animal spending may drop precipitously. Ariel may well be right that, even if so, the returns probably don’t fall off so much that animal work loses to global health work, but I haven’t investigated this myself. The upshot: I have no idea whether there are good ways of spending an additional $100M on animals right now. (That being said, I’d love to see more extensive investigation into field building for animals! If EA field building in general is cost-competitive with other causes, then I’d expect animal field building to look pretty good.)
I should also say that OP’s commitment to worldview diversification complicates any conclusions about what OP should do from its own perspective. Even if it’s true that a straightforward utilitarian analysis would favor spending a lot more on animals, it’s pretty clear that some key stakeholders have deep reservations about straightforward utilitarian analyses. And because worldview diversification doesn’t include a clear procedure for generating a specific allocation, it’s hard to know what people who are committed to worldview diversification should do by their own lights.
The upshot: I have no idea whether there are good ways of spending an additional $100M on animals right now.
I haven’t read this in a ton of detail, but I liked this post from last year trying to answer this exact question (what are potentially effective ways to deploy >$10M in projects for animals).
The old cortical neuron count proxy for moral weight says that one chicken life year is worth 0.003, which is 1/100th of the RP welfare range estimate of 0.33. This number would mean chicken interventions are only 0.7x as effective as human interventions, rather than 700x as effective.
ok, animal charities still come out an order of magnitude ahead of human charities given the cage-free campaigns analysis and neuron counts
but the broader point is that the RP analyses seem far from conclusive and it would be silly to use them unilaterally for making huge funding allocation decisions, which I think still stands
Others have enumerated many reservations with this critique, which I agree with. Here I’ll give several more.
why isn’t the “1000x” calculation actually spelled out?
As you’ve seen, given Rethink’s moral weights, many plausible choices for the remaining “made-up” numbers give a cost-effectiveness multiple on the order of 1000x. Vasco Grilo conducted a similar analysis which found a multiple of 1.71k. I didn’t commit to a specific analysis for a few reasons:
I agree with your point that uncertainty is really high, and I don’t want to give a precise multiple which may understate the uncertainty.
Reasonable critiques can be made of pretty much any assumptions made which imply a specific multiple. Though these critiques are important for robust methodology, I wanted the post to focus specifically upon how difficult it seems to avoid the conclusion of prioritizing animal welfare in neartermism. I believe that given Rethink’s moral weights, a cost-effectiveness multiple on the order of 1000x will be found by most plausible choices for the additional assumptions.
(Although I got the 5th and 95th percentiles of the output by simply multiplying the 5th and 95th percentiles of the inputs. This is not correct, but I’m not sure there’s a better approach without more information about the input distributions.)
Sadly, I don’t think that approach is correct. The 5th percentile of a product of random variables is not the product of the 5th percentiles—in fact, in general, it’s going to be a product of much higher percentiles (20+).
To see this, imagine if a bridge is held up 3 spokes which are independently hammered in, and each spoke has a 5% chance of breaking each year. For the bridge to fall, all 3 spokes need to break. That’s not the same as the bridge having a 5% chance of falling each year—the chance is actually far lower (0.01%). For the bridge to have a 5% chance of falling each year, each spoke would need to have a 37% chance of breaking each year.
As you stated, knowledge of distributions is required to rigorously compute percentiles of this product, but it seems likely that the 5th percentile case would still have the multiple several times that of GiveWell top charities.
let’s not forget second order effects
This is a good point, but the second order effects of global health interventions on animals are likely much larger in magnitude. I think some second-order effects of many animal welfare interventions (moral circle expansion) are also positive, and I have no idea how it all shakes out.
Sadly, I don’t think that approach is correct. The 5th percentile of a product of random variables is not the product of the 5th percentiles—in fact, in general, it’s going to be a product of much higher percentiles (20+).
As something of an aside, I think this general point was demonstrated and visualised well here.
wanted the post to focus specifically upon how difficult it seems to avoid the conclusion of prioritizing animal welfare in neartermism
I wasn’t familiar with these other calculations you mention. I thought you were just relying on the RP studies which seemed flimsy. This extra context makes the case much stronger.
Sadly, I don’t think that approach is correct. The 5th percentile of a product of random variables is not the product of the 5th percentiles—in fact, in general, it’s going to be a product of much higher percentiles (20+).
I don’t think that’s true either.
If you’re multiplying noramlly distributed distributions, the general rule is that you add the percentage variances in quadrature.
Which I don’t think converges to a specific percentile like 20+. As more and more uncertainties cancel out the relative contribution of any given uncertainty goes to zero.
IDK. I did explicitly say that my calculation wasn’t correct. And with the information on hand I can’t see how I could’ve done better. Maybe I should’ve fudged it down by one OOD.
On the percentile of a product of normal distributions, I wrote this Python script which shows that the 5th percentile of a product of normally distributed random variables will in general be a product of much higher percentiles (in this case, the 16th percentile):
import random
MU = 100
SIGMA = 10
N_SAMPLES = 10 ** 6
TARGET_QUANTILE = 0.05
INDIVIDUAL_QUANTILE = 83.55146375 # From Google Sheets NORMINV(0.05,100,10)
samples = []
for _ in range(N_SAMPLES):
r1 = random.gauss(MU, SIGMA)
r2 = random.gauss(MU, SIGMA)
r3 = random.gauss(MU, SIGMA)
sample = r1 * r2 * r3
samples.append(sample)
samples.sort()
# The sampled 5th percentile product
product_quantile = samples[int(N_SAMPLES * TARGET_QUANTILE)]
implied_individual_quantile = product_quantile ** (1/3)
implied_individual_quantile # ~90, which is the *16th* percentile by the empirical rule
I apologize for overstating the degree to which this reversion occurs in my original reply (which claimed an individual percentile of 20+ to get a product percentile of 5), but I hope this Python snippet shows that my point stands.
I did explicitly say that my calculation wasn’t correct. And with the information on hand I can’t see how I could’ve done better.
This is completely fair, and I’m sorry if my previous reply seemed accusatory or like it was piling on. If I were you, I’d probably caveat your analysis’s conclusion to something more like “Under RP’s 5th percentile weights, the cost-effectiveness of cage-free campaigns would probably be lower than that of the best global health interventions”.
I think your BOTEC is unlikely to give meaningful answers because it treats averting a human death as equivalent to moving someone from the bottom of their welfare range to the top of their welfare range. At least to me, this seems plainly wrong—I’d vastly prefer shifting someone from receiving the worst possible torture to the greatest possible happiness for an hour to extending someone’s ordinary life for an hour.
The objections you raise are still worth discussing, but I think the best starting place for discussing them is Duffy (2023)’s model (Causal model, report), rather than your BOTEC.
But didn’t RP prove that cortical neuron counts are fake?
Hardly. They gave a bunch of reasons why we might be skeptical of neuron count (summarised here). But I think the reasons in favour of using cortical neuron count as a proxy for moral weight are stronger than the objections.
I don’t think the reasons in favour of using neuron counts provide much support for weighing by neuron counts or any function of them in practice. Rather, they primarily support using neuron counts to inform missing data about functions and capacities that do determine welfare ranges (EDIT: or moral weights), in models of how welfare ranges (EDIT: or moral weights) are determined by functions and capacities. There’s a general trend that animals with more neurons have more capacities and more sophisticated versions of some capacities.
However, most functions and capacities seem pretty irrelevant to welfare ranges, even if relevant for what welfare is realized in specific circumstances. If an animal can already experience excruciating pain, presumably near the extreme of their welfare range, what do humans have that would make excruciating pain far worse for us in general, or otherwise give us far wider welfare ranges? And why?
“If an animal can already experience excruciating pain, presumably near the extreme of their welfare range, what do humans have that would make excruciating pain far worse for us in general, or otherwise give us far wider welfare ranges? And why?”
We have a far more advanced consciousness and self awareness, that may make our experience of pain orders of magnitude worse (or at least different) than for many animals—or not.
I think there is far more uncertainty in this question than many ackhnowledge—RP acknowledge the uncertainty but I don’t think present it as clearly as they could. Extreme pain for humans could be a wildly different experience than it is for animals, or it could be quite similar. Even if we assume hedonism (which I don’t), we can oversimplify the concepts of “Sentience” and “welfare ranges” to feel like we have more certainty over these numbers than we do.
We have a far more advanced consciousness and self awareness, that may make our experience of pain orders of magnitude worse (or at least different) than for many animals—or not.
I agree that that’s possible and worth including under uncertainty, but it doesn’t answer the “why”, so it’s hard to justify giving it much or disproportionate weight (relative to other accounts) without further argument. Why would self-awareness, say, make being in intense pain orders of magnitude worse?
And are we even much more self-aware than other animals when we are in intense pain? One of the functions of pain is to take our attention, and it does so more the more intense the pain. That might limit the use of our capacities for self-awareness: we’d be too focused on and distracted by the pain. Or, maybe our self-awareness or other advanced capacities distract us from the pain, making it less intense than in other animals.
(My own best guess is that at the extremes of excruciating pain, sophisticated self-awareness makes little difference to the intensity of suffering.)
They won’t be literally identical: they’ll differ in many ways, like physical details, cognitive expression and behavioural influence. They’re separate instantiations of the same broad class of functions or capacities.
I would say the number of times a function or capacity is realized in a brain can be relevant, but it seems pretty unlikely to me that a person can experience suffering hundreds of times simultaneously (and hundreds of times more than chickens, say). Rethink Priorities looked into these kinds of views here. (I’m a co-author on that article, but I don’t work at Rethink Priorities anymore, and I’m not speaking on their behalf.)
FWIW, I started very pro-neuron counts (I defended them here and here), and then others at RP, collaborators and further investigation myself moved me away from the view.
FWIW, I started very pro-neuron counts (I defended them here and here), and then others at RP, collaborators and further investigation myself moved me away from the view.
There are other simple methodologies that make vaguely plausible guesses (under hedonism), like:
welfare ranges are generally similar or just individual-relative across species capable of suffering and pleasure (RP’s Equality Model),
the intensity categories of pain defined by Welfare Footprint Project (or some other functionally defined categories) have similar ranges across animals that have them, and assign numerical weights to those categories, so that we should weigh “disabling pain” similarly across animals, including humans,
the pain intensity scales with the number of just-noticeable differences in pain intensities away from neutral across individuals, so we just weigh by their number (RP’s Just Noticeable Difference Model[1]).
In my view, 1, 2 and 3 are more plausible and defensible than views that would give you (cortical or similar function) neuron counts as a good approximation. I also think the actually right answer, if there’s any (so excluding the individual-relative interpretation for 1), will look like 2, but more complex and with possibly different functions. RP explicitly considered 1 and 3 in its work. These three models give chickens >0.1x humans’ welfare ranges:
Model 1 would give the same welfare ranges across animals, including humans, conditional on capacity for suffering and pleasure.
Model 2 would give the same sentience-conditional welfare ranges across mammals (including humans) and birds, at least. My best guess is also the same across all vertebrates. I’m less sure that invertebrates can experience similarly intense pain even conditional on sentience, but it’s not extremely unlikely.
Model 3 would probably pretty generally give nonhuman animals welfare ranges at least ~0.1x humans’, conditional on sentience, according to RP.[2]
You can probably come up with some models that assign even lower welfare ranges to other animals, too, of course, including some relatively simple ones, although not simpler than 1.
Note that using cortical (or similar function) neuron counts also makes important assumptions about which neurons matter and when. Not all plausibly conscious animals have cortices, so you need to identify which structures have similar roles, or else, chauvinistically, rule these animals out entirely regardless of their capacities. So this approach is not that simple, either. Just counting all neurons would be simpler.
(I don’t work for RP anymore, and I’m not speaking on their behalf.)
‘There’s a common sense story of: more neurons → more compute power → more consciousness.’
I think it is very unclear what “more consciousness” even means. “Consciousness” isn’t “stuff” like water that you can have a greater weight or volume of.
Several influential EAs have suggested using neuron counts as rough proxies for animals’ relative moral weights. We challenge this suggestion.
We take the following ideas to be the strongest reasons in favor of a neuron count proxy:
neuron counts are correlated with intelligence and intelligence is correlated with moral weight,
additional neurons result in “more consciousness” or “more valenced consciousness,” and
increasing numbers of neurons are required to reach thresholds of minimal information capacity required for morally relevant cognitive abilities.
However:
in regards to intelligence, we can question boththe extent to which more neurons are correlated with intelligence and whether more intelligence in fact predicts greater moral weight;
many ways of arguing that more neurons results in more valenced consciousness seem incompatible with our current understanding of how the brain is likely to work; and
there is no straightforward empirical evidence or compelling conceptual arguments indicating that relative differences in neuron counts within or between species reliably predicts welfare relevant functional capacities.
Overall, we suggest that neuron counts should not be used as a sole proxy for moral weight, but cannot be dismissed entirely. Rather, neuron counts should be combined with other metrics in an overall weighted score that includes information about whether different species have welfare-relevant capacities.
I think it’s very unclear for sure. Why could consciousness not be like water that you could have more or less volume of? When I was a child I was perhaps conscious but less so than now?
Could a different species with a different brain structure could have a different “nature” of consciousness while not necessarily being more or less?
I agree it’s very unclear, but there could be directionality unless I’m missing some of the point of the concept...
I’m not saying it’s impossible to make sense of the idea of a metric of “how conscious” something is, just that it’s unclear enough what this means that any claim employing the notion without explanation is not “commonsense”.
Also part (although not all) of the attraction of “more neurons=more consciousness” is I think a picture that comes from “more input=more of a physical stuff”, which is wrong in this case. I actually do (tentatively!) think that consciousness is sort of a cluster-y concept, where the more of a range of properties a mind has, the more true* it is to say it is conscious, but none of those properties definitively is “really” what being conscious requires. (i.e. sensory input into rational belief, ability to recognize your own sensory states, some sort of raw complexity requirement to rule out very simple systems with the previous 2 features etc.) And I think larger neuron counts will rough correlate with having more of these sorts of properties. But I doubt this will lead to a view where something with a trillion neurons is a thousand times more conscious than something with a billion. *Degrees of truth are also highly philosophically controversial though.
Points in favour of cortical neuron counts as a proxy for moral weight:
Neuron counts correlate with our intuitions of moral weights. Cortical counts would say that ~300 chicken life years are morally equivalent to one human life year, which sounds about right.
That neuron counts seem to correlate with intuitions of moral weight is true, but potentially misleading. We discuss these results, drawing on our own data here.
I would quite strongly recommend that more survey research be done (including more analysis, as well as additional surveys: we have some more unpublished data of our own on this), if before taking the correlations as a reason to prefer using neuron count as a proxy (in contrast to a holistic assessment of different capacities).
This seems to be the key claim of the piece, so why isn’t the “1000x” calculation actually spelled out?
The “cage-free campaigns analysis” estimates
This analysis gives chicken years affected per dollar as 9.6-120 (95%CI), with 41 as the median estimate.
The moral weights analysis estimates “welfare ranges”, ie, the difference in moral value between the best possible and worst possible experience for a given species. This doesn’t actually tell us anything about the disutility of caging chickens. For that you would need to make up some additional numbers:
Anyway, the 95%CI for chicken welfare ranges (as a fraction of human ranges) is 0.002-0.869, with 0.332 as the median estimate.
So if we make the additional assumptions that:
All future animal welfare interventions will be as effective as past efforts (which seems implausible given diminishing marginal returns)
Cages cause chickens to lose half of their average welfare (a totally made up number)
Then we can multiply these out to get:
The “DALYs / $ through GiveWell charities” comes from the fact that it costs ~$5000 to save the life of a child. Assming “save a life” means adding ~50 years to the lifespan, that means $100 / DALY, or 0.01 DALYs / $.
A few things to note here:
There is huge uncertainty here. The 95% CI in the table indicates that chicken interventions could be anywhere from 10,000x to 0.1x as effective as human charities. (Although I got the 5th and 95th percentiles of the output by simply multiplying the 5th and 95th percentiles of the inputs. This is not correct, but I’m not sure there’s a better approach without more information about the input distributions.)
To get these estimates we had to make some implausible assumptions and also totally make up some numbers.
The old cortical neuron count proxy for moral weight says that one chicken life year is worth 0.003, which is 1/100th of the RP welfare range estimate of 0.33. This number would mean chicken interventions are only 0.7x as effective as human interventions, rather than 700x as effective. [edit: oops, maths wrong here. see Michael’s comment below.]
But didn’t RP prove that cortical neuron counts are fake?
Hardly. They gave a bunch of reasons why we might be skeptical of neuron count (summarised here). But I think the reasons in favour of using cortical neuron count as a proxy for moral weight are stronger than the objections. And that still doesn’t give us any reason to think RP’s has a better methodology for calculating moral weights. It just tells us to not take cortical counts to literally.
Points in favour of cortical neuron counts as a proxy for moral weight:
Neuron counts correlate with our intuitions of moral weights. Cortical counts would say that ~300 chicken life years are morally equivalent to one human life year, which sounds about right.
There’s a common sense story of: more neurons → more compute power → more consciousness.
It’s a simple and practical approach. Obtaining the moral weight of an arbitrary animal only requires counting neurons.
Compare with the RP moral weights:
If we interpret the welfare ranges as moral weights, then 3 chicken life years are worth one human life year. This is not a trade I would make.
If we don’t interpret welfare ranges as moral weights, then the RP numbers tell us literally nothing.
The methodology is complex, difficult to understand, expensive, and requires reams zoological observation to be applied to new animals.
And let’s not forget second order effects. Raising people out of poverty can increase global innovation and specialisation and accelerate economic development which could have benefits centuries from now. It’s not obvious that helping chickens has any real second order effects.
In conclusion:
It’s not obvious to me that RP’s research actually tells us anything useful about the effectiveness of animal charities compared to human charities.
There are tons of assumptions and simplifications that go into these RP numbers, so any conclusions we can draw must be low confidence.
Cortical neuron counts still looks like a pretty good way to compare welfare across species. Under cortical neuron count, human charities come out on top.
Thanks for all this, Hamish. For what it’s worth, I don’t think we did a great job communicating the results of the Moral Weight Project.
As you rightly observe, welfare ranges aren’t moral weights without some key philosophical assumptions. Although we did discuss the significance of those assumptions in independent posts, we could have done a much better job explaining how those assumptions should affect the interpretation of our point estimates.
Speaking of the point estimates, I regret leading with them: as we said, they’re really just placeholders in the face of deep uncertainty. We should have led with our actual conclusions, the basics of which are that the relevant vertebrates are probably within an OOM of humans and shrimps and the relevant adult insects are probably within two OOMs of the vertebrates. My guess is that you and I disagree less than you might think about the range of reasonable moral weights across species, even if the centers of my probability masses are higher than yours.
I agree that our methodology is complex and hard to understand. But it would be surprising if there were a simple, easy-to-understand way to estimate the possible differences in the intensities of valenced states across species. Likewise, I agree that “there are tons of assumptions and simplifications that go into these RP numbers, so any conclusions we can draw must be low confidence.” But there are also tons of assumptions and biases that go into our intuitive assessments of the relative moral importance of various kinds of nonhuman animals. So, a lot comes down to how much stock you put in your intuitions. As you might guess, I think we have lots of reasons not to trust them once we take on key moral assumptions like utilitiarianism. So, I take much of the value of the Moral Weight Project to be in the mere fact that it tries to reach moral weights from first principles.
It’s time to do some serious surveying to get a better sense of the community’s moral weights. I also think there’s a bunch of good work to do on the significance of philosophical / moral uncertainty here. I If anyone wants to support this work, please let me know!
Thanks for responding to my hot takes with patience and good humour!
Your defenses and caveats all sound very reasonable.
So given this, you’d agree with the conclusion of the original piece? At least if we take the “number of chickens affected per dollar” input as correct?
I agree with Ariel that OP should probably be spending more on animals (and I really appreciate all the work he’s done to push this conversation forward). I don’t know whether OP should allocate most neartermist funding to AW as I haven’t looked into lots of the relevant issues. Most obviously, while the return curves for at least some human-focused neartermist options are probably pretty flat (just think of GiveDirectly), the curves for various sorts of animal spending may drop precipitously. Ariel may well be right that, even if so, the returns probably don’t fall off so much that animal work loses to global health work, but I haven’t investigated this myself. The upshot: I have no idea whether there are good ways of spending an additional $100M on animals right now. (That being said, I’d love to see more extensive investigation into field building for animals! If EA field building in general is cost-competitive with other causes, then I’d expect animal field building to look pretty good.)
I should also say that OP’s commitment to worldview diversification complicates any conclusions about what OP should do from its own perspective. Even if it’s true that a straightforward utilitarian analysis would favor spending a lot more on animals, it’s pretty clear that some key stakeholders have deep reservations about straightforward utilitarian analyses. And because worldview diversification doesn’t include a clear procedure for generating a specific allocation, it’s hard to know what people who are committed to worldview diversification should do by their own lights.
I haven’t read this in a ton of detail, but I liked this post from last year trying to answer this exact question (what are potentially effective ways to deploy >$10M in projects for animals).
700/100=7, not 0.7.
oh true lol
ok, animal charities still come out an order of magnitude ahead of human charities given the cage-free campaigns analysis and neuron counts
but the broader point is that the RP analyses seem far from conclusive and it would be silly to use them unilaterally for making huge funding allocation decisions, which I think still stands
Hi Hamish! I appreciate your critique.
Others have enumerated many reservations with this critique, which I agree with. Here I’ll give several more.
As you’ve seen, given Rethink’s moral weights, many plausible choices for the remaining “made-up” numbers give a cost-effectiveness multiple on the order of 1000x. Vasco Grilo conducted a similar analysis which found a multiple of 1.71k. I didn’t commit to a specific analysis for a few reasons:
I agree with your point that uncertainty is really high, and I don’t want to give a precise multiple which may understate the uncertainty.
Reasonable critiques can be made of pretty much any assumptions made which imply a specific multiple. Though these critiques are important for robust methodology, I wanted the post to focus specifically upon how difficult it seems to avoid the conclusion of prioritizing animal welfare in neartermism. I believe that given Rethink’s moral weights, a cost-effectiveness multiple on the order of 1000x will be found by most plausible choices for the additional assumptions.
Sadly, I don’t think that approach is correct. The 5th percentile of a product of random variables is not the product of the 5th percentiles—in fact, in general, it’s going to be a product of much higher percentiles (20+).
To see this, imagine if a bridge is held up 3 spokes which are independently hammered in, and each spoke has a 5% chance of breaking each year. For the bridge to fall, all 3 spokes need to break. That’s not the same as the bridge having a 5% chance of falling each year—the chance is actually far lower (0.01%). For the bridge to have a 5% chance of falling each year, each spoke would need to have a 37% chance of breaking each year.
As you stated, knowledge of distributions is required to rigorously compute percentiles of this product, but it seems likely that the 5th percentile case would still have the multiple several times that of GiveWell top charities.
This is a good point, but the second order effects of global health interventions on animals are likely much larger in magnitude. I think some second-order effects of many animal welfare interventions (moral circle expansion) are also positive, and I have no idea how it all shakes out.
As something of an aside, I think this general point was demonstrated and visualised well here.
Disclaimer: I work RP so may be biased.
I wasn’t familiar with these other calculations you mention. I thought you were just relying on the RP studies which seemed flimsy. This extra context makes the case much stronger.
I don’t think that’s true either.
If you’re multiplying noramlly distributed distributions, the general rule is that you add the percentage variances in quadrature.
Which I don’t think converges to a specific percentile like 20+. As more and more uncertainties cancel out the relative contribution of any given uncertainty goes to zero.
IDK. I did explicitly say that my calculation wasn’t correct. And with the information on hand I can’t see how I could’ve done better. Maybe I should’ve fudged it down by one OOD.
Thanks for being charitable :)
On the percentile of a product of normal distributions, I wrote this Python script which shows that the 5th percentile of a product of normally distributed random variables will in general be a product of much higher percentiles (in this case, the 16th percentile):
I apologize for overstating the degree to which this reversion occurs in my original reply (which claimed an individual percentile of 20+ to get a product percentile of 5), but I hope this Python snippet shows that my point stands.
This is completely fair, and I’m sorry if my previous reply seemed accusatory or like it was piling on. If I were you, I’d probably caveat your analysis’s conclusion to something more like “Under RP’s 5th percentile weights, the cost-effectiveness of cage-free campaigns would probably be lower than that of the best global health interventions”.
I think your BOTEC is unlikely to give meaningful answers because it treats averting a human death as equivalent to moving someone from the bottom of their welfare range to the top of their welfare range. At least to me, this seems plainly wrong—I’d vastly prefer shifting someone from receiving the worst possible torture to the greatest possible happiness for an hour to extending someone’s ordinary life for an hour.
The objections you raise are still worth discussing, but I think the best starting place for discussing them is Duffy (2023)’s model (Causal model, report), rather than your BOTEC.
I don’t think the reasons in favour of using neuron counts provide much support for weighing by neuron counts or any function of them in practice. Rather, they primarily support using neuron counts to inform missing data about functions and capacities that do determine welfare ranges (EDIT: or moral weights), in models of how welfare ranges (EDIT: or moral weights) are determined by functions and capacities. There’s a general trend that animals with more neurons have more capacities and more sophisticated versions of some capacities.
However, most functions and capacities seem pretty irrelevant to welfare ranges, even if relevant for what welfare is realized in specific circumstances. If an animal can already experience excruciating pain, presumably near the extreme of their welfare range, what do humans have that would make excruciating pain far worse for us in general, or otherwise give us far wider welfare ranges? And why?
“If an animal can already experience excruciating pain, presumably near the extreme of their welfare range, what do humans have that would make excruciating pain far worse for us in general, or otherwise give us far wider welfare ranges? And why?”
We have a far more advanced consciousness and self awareness, that may make our experience of pain orders of magnitude worse (or at least different) than for many animals—or not.
I think there is far more uncertainty in this question than many ackhnowledge—RP acknowledge the uncertainty but I don’t think present it as clearly as they could. Extreme pain for humans could be a wildly different experience than it is for animals, or it could be quite similar. Even if we assume hedonism (which I don’t), we can oversimplify the concepts of “Sentience” and “welfare ranges” to feel like we have more certainty over these numbers than we do.
I agree that that’s possible and worth including under uncertainty, but it doesn’t answer the “why”, so it’s hard to justify giving it much or disproportionate weight (relative to other accounts) without further argument. Why would self-awareness, say, make being in intense pain orders of magnitude worse?
And are we even much more self-aware than other animals when we are in intense pain? One of the functions of pain is to take our attention, and it does so more the more intense the pain. That might limit the use of our capacities for self-awareness: we’d be too focused on and distracted by the pain. Or, maybe our self-awareness or other advanced capacities distract us from the pain, making it less intense than in other animals.
(My own best guess is that at the extremes of excruciating pain, sophisticated self-awareness makes little difference to the intensity of suffering.)
by that logic, two chickens have the same moral weight as one chicken because they have the same functions and capacities, no?
They won’t be literally identical: they’ll differ in many ways, like physical details, cognitive expression and behavioural influence. They’re separate instantiations of the same broad class of functions or capacities.
I would say the number of times a function or capacity is realized in a brain can be relevant, but it seems pretty unlikely to me that a person can experience suffering hundreds of times simultaneously (and hundreds of times more than chickens, say). Rethink Priorities looked into these kinds of views here. (I’m a co-author on that article, but I don’t work at Rethink Priorities anymore, and I’m not speaking on their behalf.)
FWIW, I started very pro-neuron counts (I defended them here and here), and then others at RP, collaborators and further investigation myself moved me away from the view.
Oh, interesting. That moves my needle.
As I see it, we basically have a choice between:
simple methodology to make vaguely plausible guesses about the unknowable phenomology of chickens (cortical neuron count)
complex methodology to make vaguely plausible guesses about the unknowable phenomology of chickens (other stuff)
I much prefer the simple methodology where we can clearly see what assumptions we’re making and how that propagates out.
There are other simple methodologies that make vaguely plausible guesses (under hedonism), like:
welfare ranges are generally similar or just individual-relative across species capable of suffering and pleasure (RP’s Equality Model),
the intensity categories of pain defined by Welfare Footprint Project (or some other functionally defined categories) have similar ranges across animals that have them, and assign numerical weights to those categories, so that we should weigh “disabling pain” similarly across animals, including humans,
the pain intensity scales with the number of just-noticeable differences in pain intensities away from neutral across individuals, so we just weigh by their number (RP’s Just Noticeable Difference Model[1]).
In my view, 1, 2 and 3 are more plausible and defensible than views that would give you (cortical or similar function) neuron counts as a good approximation. I also think the actually right answer, if there’s any (so excluding the individual-relative interpretation for 1), will look like 2, but more complex and with possibly different functions. RP explicitly considered 1 and 3 in its work. These three models give chickens >0.1x humans’ welfare ranges:
Model 1 would give the same welfare ranges across animals, including humans, conditional on capacity for suffering and pleasure.
Model 2 would give the same sentience-conditional welfare ranges across mammals (including humans) and birds, at least. My best guess is also the same across all vertebrates. I’m less sure that invertebrates can experience similarly intense pain even conditional on sentience, but it’s not extremely unlikely.
Model 3 would probably pretty generally give nonhuman animals welfare ranges at least ~0.1x humans’, conditional on sentience, according to RP.[2]
You can probably come up with some models that assign even lower welfare ranges to other animals, too, of course, including some relatively simple ones, although not simpler than 1.
Note that using cortical (or similar function) neuron counts also makes important assumptions about which neurons matter and when. Not all plausibly conscious animals have cortices, so you need to identify which structures have similar roles, or else, chauvinistically, rule these animals out entirely regardless of their capacities. So this approach is not that simple, either. Just counting all neurons would be simpler.
(I don’t work for RP anymore, and I’m not speaking on their behalf.)
Although we could use a different function of the number instead, for increasing or diminishing marginal returns to additional JNDs.
Maybe lower for some species RP didn’t model, e.g. nematodes, tiny arthropods?
‘There’s a common sense story of: more neurons → more compute power → more consciousness.’
I think it is very unclear what “more consciousness” even means. “Consciousness” isn’t “stuff” like water that you can have a greater weight or volume of.
Hi David,
Relatedly, readers may want to check Why Neuron Counts Shouldn’t Be Used as Proxies for Moral Weight. Here are the key takeaways:
I think it’s very unclear for sure. Why could consciousness not be like water that you could have more or less volume of? When I was a child I was perhaps conscious but less so than now?
Could a different species with a different brain structure could have a different “nature” of consciousness while not necessarily being more or less?
I agree it’s very unclear, but there could be directionality unless I’m missing some of the point of the concept...
I’m not saying it’s impossible to make sense of the idea of a metric of “how conscious” something is, just that it’s unclear enough what this means that any claim employing the notion without explanation is not “commonsense”.
100% agree nice one
Also part (although not all) of the attraction of “more neurons=more consciousness” is I think a picture that comes from “more input=more of a physical stuff”, which is wrong in this case. I actually do (tentatively!) think that consciousness is sort of a cluster-y concept, where the more of a range of properties a mind has, the more true* it is to say it is conscious, but none of those properties definitively is “really” what being conscious requires. (i.e. sensory input into rational belief, ability to recognize your own sensory states, some sort of raw complexity requirement to rule out very simple systems with the previous 2 features etc.) And I think larger neuron counts will rough correlate with having more of these sorts of properties. But I doubt this will lead to a view where something with a trillion neurons is a thousand times more conscious than something with a billion.
*Degrees of truth are also highly philosophically controversial though.
That neuron counts seem to correlate with intuitions of moral weight is true, but potentially misleading. We discuss these results, drawing on our own data here.
I would quite strongly recommend that more survey research be done (including more analysis, as well as additional surveys: we have some more unpublished data of our own on this), if before taking the correlations as a reason to prefer using neuron count as a proxy (in contrast to a holistic assessment of different capacities).