How many neurons are there?

Image from NOAA, in the public domain.

I just finished a large project trying to estimate that. I’ve posted it on its own page here. Here’s the abstract:

We estimate that there are between about 10^24 neurons on earth, with about an order of magnitude uncertainty. Most of these are from insects, with significant contributions from nematodes and fish. For insects, we multiplied the apparent number of insects on earth by the number of neurons in a small insect, the fruit fly. Most other classes of animal contribute 10^22 neurons at most, and so are unlikely to change the final analysis. For nematodes, we looked at studies that provide an average number of nematodes per square meter of soil or the ocean floor, and multiplied them by the number of neurons in Caenorhabditis elegans, an average-sized nematode. Fish may also play a significant role. We neglected a few categories that probably aren’t significant, but could conceivably push the estimate up.

Using a similar but less precise process based on evolutionary history and biomass over time, we also estimate that there have been 10^33 neuron-years of work over the history of life, again with around an order of magnitude of uncertainty.

Advertisements

Male dairy calves, male chicks, and relative suffering from animal foods

Or: Do “byproduct” animals of food animal production significantly affect estimates of comparative suffering caused by those foods? No.

[Image adapted from this image by Flickr user Sadie_Girl, under a CC BY-SA 2.0 license.]

See, relatedly: What happens to cows in the US?

Short version

There’s a shared belief in animal-welfare-oriented effective altruism that eggs and chicken meat cause a great deal more suffering than beef or dairy (1). You can make big strides towards reducing the amount of suffering caused in your diet by eating fewer eggs and chicken, even if you don’t go fully vegetarian or vegan.

Julia Galef, Brian Tomasik, and William MacAskill have made different versions of this calculation, with different metrics, and have come to the same conclusion. These three calculations include only the animal used directly for production. (Details about the calculations and my modifications are in the long version below.) But the production of several kinds of animal product require bringing into existence animals that aren’t used for that product – like male calves born to lactating dairy cows, or male chicks born when producing egg-laying hens. I wondered if including these animals would significantly change the amount of suffering in various animal foods.

It turns out that even accounting for these other animals indirectly created during production, the amount of suffering relative to other animal foods doesn’t change very much. If you buy the premises of these quantitative ethical comparisons, beef and dairy make so much product using so few animals that they’re still 1-3 orders of magnitude better than eggs or chicken. Or rather, the message of “eat less chicken” and “if you’re going to eat animal products, eat dairy and beef” still makes sense even if we account for the maximum number of other animals created incidental to production of each food animal. I’m going to call these the “direct and incidental animals” (DIA) involved in a single animal’s worth of product.

The question is complicated by the fact that “incidental” animals still go into another part of the system. Day-old male chicks are used for pet and livestock food, and male dairy calves are raised for meat.

Given that these male calves are tied to dairy production, it seems unlikely that production of dairy and meat is what it would be if they weren’t connected. For instance, if there is less demand for dairy and thus fewer male dairy calves, it seems like one of the following should happen:

  1. No change to meat calf supply, less meat will be produced (DIA estimates seem correct)
  2. Proportionally more meat calves will be raised (original estimates seem correct)
  3. Something between the above (more likely)

Reframed: It depends whether demand for dairy increases the meat supply and makes it less profitable to raise meat cows, or whether demand for meat makes it more profitable to raise dairy cows, or both. I’m not an economist and don’t go into which one of these is the case. (I tried to figure this out and didn’t make much headway.) That said, it seems likely that the actual expected number of animal lives or days of suffering is somewhere between the initial numbers and my altered values for each source.

The most significant change I find from the original findings suggest that meat cows cause a fair bit more suffering over a longer period of time than the original calculations predict, only if demand for meat is significantly propping up the dairy industry. But even if that’s true, the suffering caused by beef is a little smaller than that caused by pork, and nowhere near as much as smaller animals.

Modifications to other estimates including direct and incidental animals (DIA)

Tomasik’s original estimate DIA Tomasik’s estimate Galef’s orginal estimate DIA Galef’s estimate
Milk 0.12 equivalent days of suffering caused per kg demanded 0.14 equivalent days of suffering caused per kg demanded 0.000057 max lives per 1000 calories of milk 0.00013 max lives per 1000 calories of milk
Beef 1.9 max equivalent days of suffering caused per kg demanded 4.74 max equivalent days of suffering caused per kg demanded 0.002469 max lives per 1000 calories 0.0029 max lives per 1000 calories
Eggs 110 equivalent days of suffering caused per kg demanded 125 equivalent days of suffering caused per kg demanded 0.048485 lives per 1000 calories 0.048485 lives per 1000 calories

That’s basically it. For a little more info and how I came to these conclusions, read on.

Longer version

On the topic of effectively helping animals, one thing I’ve heard a few times is that eating dairy and beef aren’t terribly harmful, since they come from such large animals that a serving of beef or milk is a very small part of the output of the animal. On the other hand, chickens are very small – an egg is a day’s output of an animal, and a family can eat an entire chicken in one dinner. Compare that with the fact that most chickens are raised in extremely unnatural and unpleasant conditions, and you have a framework for directly comparing the suffering that goes into different animal products.

This calculation has been made by three people I’m aware of – Brian Tomasik on his website, William MacAskill in his book Doing Good Better, and Julia Galef on her blog. The organization One Step for the Animals also recommends people stop eating chickens, on these grounds, but I didn’t find a similar breakdown on their website after a few minutes of looking. It’s still worth checking out, though. (Did you know chicken consumption, in pounds/year, has surpassed beef consumption and is still climbing, but only over the last 20 years?)

Galef compares calories per life. She includes the male chicks killed for each egg-laying hen.

Tomasik looks at “days of suffering caused per kg demanded”.

Macaskill briefly examines three factors: the number of animal years and lives that go into a year of eating in the average omnivorous American diet, and also numerical “quality of life” estimates from Bailey Norwood. (He doesn’t combine these factors numerically so much as use them to establish a basis for recommending people avoid chicken. I didn’t do an in-depth analysis of his, but safe to say that like the others, adding in other animal lives doesn’t seem to change his conclusions significantly.)

With pigs and meat chickens, the case is straightforward – both sexes are raised for meat, and suppliers breed animals to sell them and retain enough to continue breeding. The aged animals are eventually slaughtered as meat as well.

But only female hens lay eggs. Meat chickens and egg chickens raised at scale in the USA are two different breeds, so when a breeder produces laying hens, they wind up with more male chicks than are needed for breeding. Similarly, dairy cows have to give birth to a calf every season they produce milk. The average dairy cow gives 2.4 calves in her lifetime, and slightly less than 1.2 of those are male. The male egg chicks and male dairy calves are used for meat.

Aged dairy cows and egg-laying chickens are also sold as meat. “Spent hens” that are no longer commercially profitable, at 72 weeks old, are sold for ‘processed chicken meat’. (Other sources claim pet food or possibly just going to landfills. Pet food sounds reasonable, but landfills seem unlikely to me, at least for large operations.) There aren’t as many of these as either cows or chickens raised directly for meat, so they’re a comparatively small fraction, but they’re clearly still feeding into the meat system.

🐔

When talking about this, we quickly run into some economic questions, like “perhaps if the demand for dairy dropped, the meat industry would start raising more calves for meat instead?”

My intuition says it ought to shake out one way or the other – either decreasing demand for dairy cows results in the price of meat going up, or decreasing demand for meat results in demand for dairy cows going down.

In the egg case, male chicks aren’t literally put in a landfill, they’re ground and sold for pet food. Without this otherwise unused source of protein, would pet food manufacturers increase demand for some other kind of meat? It seems possible that both this would happen and that the price of pet food would increase. Then, maybe less would be bought to make up for the difference, at least in the long term – cheap pet food must be somewhat inelastic, at least in the short term?

My supply and demand curves suggest that both demand should decline and price should increase. That said, we’re leaving the sphere of my knowledge and I don’t know how to advise you here. For the moment, I’m comfortable folding in both animals produced in the supply chain for a product, and animals directly killed or used for a product. But based on the economic factors above, these still don’t equate to “how many animal lives / days are expected to be reduced in the long term by avoiding consumption of a given product.”

At the most, though, dairy cows bring an extra 1.2 meat cow into existence, meat cows bring an extra .167 dairy cow,  and each egg-laying hen brings an extra 1 male chicken that is killed around the first day. These are the “direct and incidental animals” created for each animal directly used during productive.

 

Some notes on the estimates below:

I ignored things like fish and krill meal that go into production. Tomasik notes that 37% of the global fish harvest (by mass) is ground and used for animal feed for farmed fish, chickens, and pigs. But this seems to be mostly from wild forage fish, not farmed fish, and wild populations are governed by a different kind of population optimum – niches. We’d guess that each fish removed from the environment frees up resources that will be eaten by, on average, one new fish. (Of course, populations we’re fishing seem to be declining, so something is happening, but it’s certainly not one-to-one.)

I also only looked at egg-laying chickens, meat cows, and dairy cows. This is because pork and other industries aren’t sex-segregated – all babies born are raised for the same thing. A few will be kept aside and used to produce more babies, but even the breeding ones will eventually be turned into meat. The amount of days these animals live probably affect Tomasik’s calculations somewhat, but the breeding animals are still the minority.

I also didn’t include a detailed analysis because if you’re concerned about animal welfare, you probably already don’t eat veal. (I’m going to assert that if you want to eat ethically treated food, avoid a meat whose distinguishing preparation characteristic is “force-feed a baby”.) Veal is a byproduct of the dairy industry, but a minority of the calves. Foie gras does have a multiplier effect because female geese don’t fatten up as much, and are killed early, so for each goose turned into foie gras, another goose is killed young.

Old dairy cows and laying hens are used for meat, but it’s a minority of the meat production. I didn’t factor this in. See What happens to cows in the US for more on cows.

Modifications to other estimates including direct and incidental animals (DIA)

Tomasik’s original estimate DIA Tomasik’s estimate Galef’s orginal estimate DIA Galef’s estimate
Milk 0.12 equivalent days of suffering caused per kg demanded 0.14 equivalent days of suffering caused per kg demanded 0.000057 max lives per 1000 calories of milk 0.00013 max lives per 1000 calories of milk
Beef 1.9 max equivalent days of suffering caused per kg demanded 4.74 max equivalent days of suffering caused per kg demanded 0.002469 max lives per 1000 calories 0.0029 max lives per 1000 calories
Eggs 110 equivalent days of suffering caused per kg demanded 125 equivalent days of suffering caused per kg demanded 0.048485 lives per 1000 calories 0.048485 lives per 1000 calories

DIA modifications to Tomasik’s estimate

(Days of equivalent suffering / kg)

To adjust this estimate, I added the extra “equivalent days of suffering caused per kg demanded” for the other animals:

Egg-laying chickens
(4 suffering per day of life in egg-laying chickens * 501 days of life) + 1 * (3 suffering per days of life in meat chickens * 1 day of life) / 16 kg of edible product over life of egg-laying chicken = 125 max equivalent days of suffering caused per kg demanded (vs 110)

Dairy cows
(2 suffering per day of life in milk cows * 1825 days of life) + 1.2 * (1 suffering per day of life in meat cows * 395 days of life) / 30000 kg of edible product over life of dairy cow = 0.14 max equivalent days of suffering caused per kg demanded (vs 0.12)

Meat cows
(1 suffering per day of life in meat cows * 395 days of life) + 0.167 * (2 suffering per day of life in dairy cows * 1825 days of life) / 212 kg of edible product over life of meat cow = 4.74 max equivalent days of suffering caused per kg demanded (vs 1.9)

The meat cow number is the only very different one here.

DIA modifications to Galef’s estimate

I adjusted this by adding other lives to Galef’s estimate of lives per 1000 calories:

Egg-laying chicken
Galef included this in her calculation of 0.048485 lives per 1000 calories of eggs.

Dairy cows
[0.000057 lives per 1000 calories of milk] * 2.2 = 0.00013 max lives per 1000 calories of milk
[0.000075 lives per 1000 calories of cheese] * 2.2 = 0.00017 max lives per 1000 calories of cheese

Meat cows
[0.002469 lives per 1000 calories of beef] * 1.167 = 0.0029 max lives per 1000 calories of beef

Other economic notes

I’m hoping someone who knows more here will be able to make use of the information I found.

The number of meat cows in the US has been broadly decreasing since 1970. The number of dairy cows has also been decreasing since at least 1965, but dairy consumption is increasing, because those cows are giving far more milk.

When dairy prices drop, dairy farmers are known to kill some of their herds and sell them for meat, leading to a drop in meat prices.

We would also expect dairies and beef farms to compete with each other for some of the same resources, like land and feed.

A friend wondered whether dairy steers are much smaller than beef cows, so if shifting the same volume of meat production to these steers would mean more animal lives. It turns out that dairy steers and beef cows are about the same weight at slaughter.


(1) With fish perhaps representing much more suffering than eggs or chickens, and other large meat sources like pigs somewhere in the middle.)


 

This blog has a Patreon. If you like what you’ve read, consider giving it a look.