Hide table of contents

I wonder whether improving the welfare of arthropods (e.g. insects and crustaceans) and nematodes[1] (roundworms) is sufficiently important and neglected to greatly outweigh the low tractability, thus being underrated.


In What We Owe to the Future, William McAskill suggests using the number of neurons as “a very rough heuristic” for the “capacity for wellbeing”[2], presenting in Table 9.2 the total population and neurons of humans, farmed animals excluding arthropods, and wild fish. The table below extends that table to terrestrial and marine arthropods, and nematodes[3]. I have also included additional columns for the neurons per animal, and total neurons relative to humans.

The capacity for welfare of arthropods and nematodes dominates based on neuron counts. Moreover, as a 1st approximation, it looks like animal welfare is really marine arthropod welfare.


Total population

Neurons per animal

Total neurons

Total neurons relative to humans






Farmed animals excluding arthropods




2.86 %

Wild fish





Terrestrial arthropods


316 k



Marine arthropods


316 k


45.2 k







Wild Animal Initiative and Rethink Priorities (RP) have conducted research on the welfare of arthropods. For instance, 6 of the 11 species analysed in RP's Welfare Range Table are arthropods. Furthermore, the Insect Welfare Project (introduced here) and Shrimp Welfare Project aim to improve the welfare of farmed insects and shrimps, which are terrestrial and marine arthropods, respectively. These are great first steps, but they do not seem to quite reflect the scope of the capacity for welfare of arthropods. Terrestrial and marine arthropods have 20 k and 2 M times as many neurons as farmed animals excluding arthropods.

Nematodes have been mentioned by Wild Animal Initiative, Rethink Priorities and Open Philanthropy, but I did not find research on improving their welfare. So nematode welfare is potentially even more neglected than that of arthropods, while still accounting for 20 k times as many neurons as farmed animals excluding arthropods.


I understand improving the welfare of arthropods and nematodes is quite difficult, especially because virtually all of them are wild[5]. However, given it is so important and neglected, I suppose trying is worth it, at least for the sake of the value of information.

  1. ^

     I encourage you to watch this video from Journey to the Microcosmos for a quick visual introduction to the world of nematodes.

  2. ^

     As far as I understand, the Moral Weight Project from Rethink Priorities also plans to assume this. The title of the 4th post of the sequence of the project is “Relative Neuron Count as a Proxy for Moral Weight”.

  3. ^

     I took the population size of terrestrial and marine arthropods, and nematodes from the 1st table of this post from rosehadshar. I computed the number of neurons of humans, farmed animals, and wild fish from the ratio between the total neurons and population size. I computed the number of neurons of terrestrial and marine arthropods from the geometric mean between the lower and upper bounds of 100 k and 1 M provided here by Brian Tomasik. I calculated the number of neurons of nematodes from the geometric mean between the lower and upper bounds of 100 and 1 k provided here by Animal Ethics.

  4. ^

     Many models of consciousness requiring less than 10 neurons are described in Herzog 2007.

  5. ^

    For example, according to this article from Abraham Rowe:

    1 trillion to 1.2 trillion insects are raised on farms annually for food and animal feed.

    These correspond to only 10^-9 (= 10^(9 - 18)) of the total number of terrestrial arthropods.

New Answer
New Comment

2 Answers sorted by

It's definitely interesting to consider the welfare of all sorts of living creatures, but with respect to neuron count, I think it is really the neuron count per organism that is the key thing to consider. As far as I understand it, the neuron count is taken as a proxy for the capacity for valenced/subjective experience of which the specific organism is capable. For example, greater neuron counts enable increasingly complex forms of processing such as self-representation. But it's a potential indicator at the level of the organism, not at the collective of all the organisms of that type.

I could put 10000 nematodes together but they don't create a collective consciousness at the same level as a single animal with 3160000 neurons, because their processing is separate. Hence, you want to weight the individual animal in some way, in terms of its capacities for valenced experience, before you multiply by the number of organisms.

Hi Jamie,

Thanks for commenting!

As far as I understand it, the neuron count is taken as a proxy for the capacity for valenced/subjective experience of which the specific organism is capable.

That is my understanding too.

I could put 10000 nematodes together but they don't create a collective consciousness at the same level as a single animal with 3160000 neurons, because their processing is separate.

I agree that putting lots of organisms together does not necessarily create a collective consciousness. However, if we interpret welfare like water as illustrated... (read more)

Jamie Elsey
Hi Vasco, I get what you are saying. I don't have a strong opinion on what kind of function of neuron count it would be but I just am not sure it is likely to be directly proportional. I expect there might be some irregular transitions along the scale of conscious experience that different numbers of neurons get you, and that is not including the differences in arrangement/how the number of neurons within the individual enable orders of magnitude of different and more complex circuitry, which could pay out as wildly different capacities for conscious experience than just the linear increase in neuron number might suggest. Of course, it could also shake out in the other direction where tons more neurons aren't actually getting you correspondingly more capacities and that would make an even stronger case for the nematodes...but it could also be that the level of experience that nematode neurons get you barely registers or doesn't even cross the threshold of having any subjective experience. In the extreme, I imagine there is some level of neuronal number/complexity that we don't think has any experience at all, and just having tons of those could never add up to a single conscious being. (I'm not saying that is the case for nematodes though)
Vasco Grilo
All of that makes sense, thanks for clarifying!
Robi Rahman
Further, consider that perhaps most nematode qualia are alike, in which case we may want to discount them to the extent they are just duplicating identical experiences.

What Jamie said -- the number of neurons across a population is irrelevant. What matters is the capacity for suffering, and that is dependent on the number (and arrangement) of neurons in an individual. This is my favorite discussion:


I fear that thinking in the terms above (total neurons in some group) does significant harm.

Hi Matt,

Thanks for commenting!

I agree the capacity for welfare of one individual does not depend much on the number of individuals. However, when prioritising solutions to improve animal welfare, I think we should take into account the number of individuals (amongst many other factors). 

For example, developping a cheap and tasty meat subsitute for chicken is arguably more pressing than for turkeys. Factory-farmed (FF) chickens and turkeys have roughly the same welfare according to the Weighted Animal Welfare Index from Charity Entrepeneurship (-56 and... (read more)

Hey Vasco, As a founder of One Step for Animals, you don't need to convince me we should be looking to help chickens.  :-) It is when we say that X chickens = 1 human, or Y mosquitoes = 1 human, or Z electrons = 1 human -- that's where I get off the train (as I lay out in Losing My Religions). Thanks again and keep up the great work!
Vasco Grilo
Ah, I know I need not convince you of that! I think the relevant chapter from Losing My Religions is "Biting the Philosophical Bullet". From I understand, you think the Repugnant Conclusion (RC) is sufficiently against your intuitions for the total view (which implies the RC) to be wrong. The RC follows from 3 premises (see here). I would be curious to know the extent to which (and why) you disagree with each of them.
Sorted by Click to highlight new comments since:

You should also take the probability of sentience into account. There are some estimates from my colleagues at RP here: https://forum.effectivealtruism.org/posts/T5fSphiK6sQ6hyptX/opinion-estimating-invertebrate-sentience

We also have a post about the value/usefulness of neuron counts coming soon: https://forum.effectivealtruism.org/s/y5n47MfgrKvTLE3pw

Hi Michael,

Thanks for sharing those estimates.

I prefer to think about sentience as non-binary. In expectation, any being is arguably sentient to some extent, even if very little, so there is a sense in which "probability of sentience" is always very close to 1. Alas, we cannot rule out suffering in fundamental physics.

However, I guess we can still use "moral weight | sentience"*"probability of sentience" to estimate moral weights. In this case, "sentience" would mean something more restrictive than what I am referring to above. In the post, I am assuming this product is directly proportional to the number of neurons. Do you think neurons are a better proxy for "moral weight | ("restrictive") sentience" than for the product?

Looking forward to that post!

Curated and popular this week
Relevant opportunities