Most Animals Have Bad Lives

Imagine waking up tomorrow in a world very different from our own. There is no society, nor do you have any friends or acquaintances. You must fend for yourself, acquiring your own food. Disease runs rampant. The world is filled with monsters that are trying to kill and eat you, often slowly over the course of many minutes. You’re constantly hungry, thirsty, and in discomfort from temperature extremes. Then, after about a week of this brutal existence, you die painfully—from starvation, being eaten alive, disease, thirst, or being crushed to death.

Question: was your life worth living?

Obvious answer: no!

But this is the life that nearly every conscious animal on Earth will ever experience. For every human alive today, there are hundreds of millions of creatures consigned to this brief, brutal, and hellish existence. This is the life of nearly every animal, and it’s why I’m generally in favor of decreasing wild animal populations.

For every living human, there are around 100,000 fish and about 100 million insects. Almost all potentially-conscious animals are arthropods (a wide class of organisms that includes insects, shrimp, and lobsters), and of the ones that aren’t arthropods, the vast majority are fish. Reptiles and amphibians are also quite numerous, though less so than fish. Many of these simple creatures can probably suffer intensely.

The most numerous animals are small prey animals! Thus, to determine whether animals tend to live good lives, one should primarily look at the lives of small prey animals. Outside of our human bubble, almost everyone is a short-lived insect, decapod, amphibian, or fish!

The most numerous species are all or nearly all R-strategists, meaning they give birth to enormous numbers of offspring few of whom survive for very long. Tuna lay about 10 million eggs (those don’t all become live offspring, but many do) while salmon lay thousands. The ocean sunfish lays about 800 million!

Populations tend to be roughly stable long run. If an organism has 100,000 offspring, on average about two of them will survive (otherwise populations grow exponentially until returning to stability). Thus, nearly every conscious creature who will ever live is a short-lived R-strategist, who will probably die after days or weeks. Adult mayflies, at the high end, only live a few hours; fish mostly starve to death after a few days of consciousness.

This brief life is characterized by struggle. Food is scarce, there are constant dangers from predators who tear their victims limb from limb, starvation and disease are common, and temperature extremes cause constant discomfort. Because animals’ lives are so brief, they don’t have enough welfare to outweigh the badness of death. If you live for a week and then get eaten alive or starve to death, it’s very unlikely that you’d accrue enough welfare to outweigh the badness of death.

This is the basic argument for wild animal suffering predominating welfare. Most animals live short lives, with the main source of enjoyments being food and drink, and then die painfully. In such brief lives, they don’t have enough welfare to outweigh the badness of death. Just like you wouldn’t trade an extra week of life for even ten minutes of torture, these animals don’t have enough welfare to outweigh the badness of their deaths—often gradual and drawn out affairs.

I find this basic argument quite persuasive, but there are a number of objections to it. Thus, I thought in this article I’d rebut all the major objections. If I am correct, the conclusions are dramatic; the world is filled with horrific suffering. If nearly every creature who has ever lived experiences a hellish existence that you’d be wrong to wish on your worse enemy, you should support diminishing wild animal populations. It is bad to sustain the conditions that bring more creatures into hell!

We are biased against this conclusion. Because we only see pretty and sublime nature, and live cushy lives infinitely better than wild animals, we can scarcely fathom just how bad life is for nearly every creature who has ever lived.

In addressing the objections, I’ll first quote my summary of them, and then explain why I disagree.

But animals value being alive!

Animals, in typical circumstances, value being alive and want to continue being alive. Thus, some people argue that this means their lives are mostly positive. This argument, however, is extremely dubious.

1. An animal can have a net negative life but want to continue living. In theory, a creature could experience an arbitrarily large amount of suffering yet still want to live. This doesn’t tell us whether its life is good. It would certainly be wrong to bring a creature into existence if it experienced arbitrarily vast amounts of torture but never wanted to die.
2. Most animals don’t understand suicide. They don’t grasp death or realize it’s an option. You can’t use the fact that a creature doesn’t kill itself as evidence it has a nice life if it doesn’t understand suicide. That’s like taking the fact that someone doesn’t leave a play as evidence that they’re enjoying it, even if they don’t know that there are any exits.
3. The basic argument for net negative wild-animal welfare comes primarily from the badness of their painful deaths. It could very well be that animals’ existences are mostly nice, but their deaths are painful enough to outweigh them (if you get tortured for the last ten minutes of a day, most of the day could be pleasant, but that doesn’t mean the day overall is pleasant). Lots of badness can be concentrated into just a brief period of time.
4. Evolution, for obvious reasons, strongly programs animals not to want to commit suicide. Those that commit suicide can’t pass on their genes. Thus, we should expect simple animals not to commit suicide no matter how miserable they are.

The simple animals are barely conscious—of the significantly conscious animals, most have nice lives.

Another objection claims that though most animals have bad lives, they’re mostly not very conscious. Fish, for instance, probably don’t have very intense pain. Of the creatures that have lives that matter significantly, most are longer-lived.

First of all, the factual premise is likely wrong. Fish very likely feel pain (~80% confidence) and probably insects and other arthropods mostly do as well (~60% confidence). In fact, there are various strong theoretical reasons to suspect they might feel pain including:

1. When these animals face negative stimuli, they behave as if they’re in a lot of pain. A lobster being boiled in a pot struggles with all its might. It looks like you or I might look if we were boiled in a pot. When fish are in pain, they rock back and forth, rub their injured area, and struggle in exactly the way one would expect if they were in intense pain. When flies are given capsaicin (the chemical that makes food spicy, which most animals find very unpleasant to eat) laced food, they pull at their mouths, thrash around, and even go so far as to starve themselves to death.^[1] This behavior certainly doesn’t seem like how one feels when only in mild discomfort.
2. Evolution would program animals to experience lots of pain during death, because a dead organism can’t pass on its genes.
3. The evolutionary function of pain is to teach organisms a lesson (e.g. avoid predators because they’ll hurt you). But simpler organisms with simpler cognitive architecture might thus need more pain to be taught a lesson.
4. During the most intense kinds of human experiences, pain envelops the totality of what we experience. But for a simple animal, intense pain by default takes in the totality of its experience. This is one reason why pain in simple organisms might be more intense (in humans, when someone’s cortex is rendered largely non-functional, sometimes they feel vastly more intense pain).
5. Historically, humans have underestimated the sentience of others. Until the 1980s, it was commonly thought that babies and dogs weren’t conscious. Thus, there’s a decent probability we’re underestimating consciousness in other animals, even simple ones.
6. While we normally assume that animals with simple brains don’t feel very intensely, it’s not clear what justifies that assumption. Certainly if a human responded aversively the way animals do but had serious brain damage that made their brain much simpler, we wouldn’t automatically confidently assume they didn’t feel intense pain.
7. Plausibly how bad pain is depends partially on relative signal intensity. The most intense pain occurs when the main stimulus the brain is processing is pain—others are drowned out. But for simple organisms, pain is often the only signal their brains take in.
8. Various surprisingly robust behavior can be achieved with only a simple brain. Simple creatures can have a fairly robust visual cortex—creatures with 1/100,000th the neurons that we have don’t have 1/100,000th the ability to see. What reason is there to be confident that pain is any different? And note, even if you have uncertainty about pain in simple creatures, they’re so numerous that nearly all the world’s pain is had by simple creatures.

The most detailed report ever done on this subject estimated that even simple animals likely felt pain within an order of magnitude or two as intensely as we do (read the report for defenses of their methodology or see my defense of it here). Even if one goes with lower-end estimates of pain in fish in arthropods and fish, nearly all suffering in the world is likely to be had by fish and arthropods. Given considerable uncertainty, nearly all expected very intense suffering in the world is had by simple creatures.

Even if one were outlandishly certain that fish and arthropods aren’t conscious, amphibians and reptiles most certainly are, as even skeptical scientists tend to admit. Thus, even on maximally skeptical assumptions, most animals would be R-strategists! Hell, even if you thought only mammals were conscious, you’d still end up concluding that most animals ever born were R-strategists.

Thus, on any reasonable weighting, nearly all the expected suffering and welfare in the world is experienced by small animals like fish and insects. Statistically, larger vertebrates are a rounding error even on generous assumptions.

How do we know what animals experience? They’re so different from us.

It’s true, of course, that animals are very different from us. This should be cause for humility and uncertainty. But if a creature lives a week and then dies painfully, and behaves when it’s dying as if it’s in extreme pain, then the most reasonable assumption is that it lived a bad life. The same evolutionary pressures that make death unpleasant in humans also make death unpleasant in animals. Also, even if you’re uncertain about whether suffering predominates in nature, given the sheer amount of possible suffering, you should be hesitant about expanding nature.

Mathematical models show that under certain assumptions, animals in nature live good lives.

In an early paper, Ng produced a mathematical model showing animals lived mostly net-negative lives. Zach Groff updated this model, showing that when one does the math correctly, whether animals live good lives or not depends on various factual assumptions. Thus, he concludes we should be uncertain about net animal welfare.

I don’t think that highly oversimplified models tell us very much about animal life, particularly given how assumption dependent they are. If I’m trying to figure out if a baby who lives a week and then starves to death had an overall positive life, the way I should do that isn’t by constructing a model of the evolutionary function of pleasure and pain. Instead, it’s by seeing what happened in its life and making an assessment. If you know that nearly every animal ever born lives a week or so of constant struggle and then dies painfully, your judgment of whether they had a nice life shouldn’t be much affected by your assessment of context-sensitive formal models.

In addition, Groff’s model suggests that under some assumptions, animals with shorter lives will generally have less evolutionary effort spent making them feel intense pain. But we don’t have to speculate: we can just look at how intensely animals seem, based on their behavior, to suffer before death. The answer is: quite a lot! Behaviorally, even simple animals like fish behave like they’re in very intense pain shortly before death!

For the record, I think Groff agrees with this. He says in a comment:

I agree that this sort of argument deserves relatively low epistemic weight and that the argument is very speculative, as I tried to emphasize in the paper but am worried that not everybody picked up.

The core intuition behind the argument seems to be that if most creatures will die shortly after birth, then it makes less sense to spend a lot of resources making sure they feel lots of pain. If you are going to make 100,000 model cars, you won’t invest too many resources in any one of them. But this on its own doesn’t blunt the core argument because:

1. It might be very evolutionarily cheap to produce pain. It may be that even simple organisms can feel lots of pain (see above for reasons to think that). Even simple creatures can have pretty robust abilities to see and hear. To give an analogy from Bjorn Merker, even simple DNA structures can self-replicate—it’s plausible that pain can be aptly sustained in simple creatures. Especially because there’s strong empirical evidence that neuron count doesn’t correlate robustly with intensity of experience.
2. It may be that more complicated brains produce more muted responses to pain. A more complicated brain might be able to richly modulate pain, to make pain signals less intense and more calibrated. Just as a defective radio might make a loud and hideous sound, it might be that a simple animal has less well-calibrated pain signals, and instead just feels unfiltered very intense pain. On this assumption, simpler animals might feel more pain than more complicated animals.
3. The same factors that lead to diminished pain should also, it seems, lead to diminished pleasure. If animals don’t evolve much pain capacity, because most pain signals are wasted, then it would also be inefficient for them to derive much pleasure capacity (because most pleasure is wasted). My understanding is that the response to this is that if pain and pleasure signals are costly, then they’ll be developed later—but as far as I can tell, there isn’t evidence that vastly greater ability to experience pleasure develops later in life.

Overall, I just don’t think it makes sense to put much weight in a model as speculative as this, when the thing it implies seems to run directly contrary to behavior in various organisms. The best evidence concerning how much pain fish feel when they die is not broad, evolutionary considerations, but behavioral and physiological evidence for profound stress and panic in fish who die.

Death doesn’t take very long.

Some, like Michael Plant and Kyle York, have noted that death doesn’t take very long while adults can live long pleasant lives. If a creature lives a year, and then dies painfully, even if its death takes 10 minutes, only about one fifty-thousandth of its life will be spent dying. Presumably, then, the pleasure it gets in the rest of its life can offset. However:

1. Even if this were true, extreme suffering is bad enough that it can be tens of thousands of times more significant than ordinary life. If you offered me an extra month of life, that would culminate in me being slowly boiled alive for a minute, it’s not clear that the month of life would contain enough pleasure to offset—even though ten minutes is only about one fifty-thousandth of a month. Extreme suffering can be tens of thousands of times more intense than mild suffering. Look at some of the ways that animals die in nature, even ones that take only a few minutes, and ask yourself: how many days of happy life would you give up to not have to experience it? For me, it would certainly be at least in the thousands. It’s easy to underestimate intense pain’s badness when it’s remote and far away—when it’s present, I think most people would give up many years of happy life to make it stop.

2. Nearly all animals are very short lived—insects and fish. If a creature lives a week and then dies painfully, even if its death takes ten minutes (predation is usually shorter, starvation and dehydration are longer) then still a full .1% of its life is spent dying painfully. Certainly I would not trade a week of life for having to spend ten minutes being boiled alive—and my life is much nicer than the life of the average animal. And even of the fish larvae that reach the point when they’re probably conscious, well over 99% starve to death or get eaten alive within their first few days of life. The picture is even more dismal if we assign even non-trivial probability to sentience in early larval stage (though that is controversial).

3. Relying on York’s numbers, it looks like for every batch of egg-laying, about a third as much time is spent in the process of actively dying—starving or being eaten alive—as is lived by adult fish. Adjusting for sentience, by conservative estimates, about 540 trillion years of adult fish consciousness go by annually, and about 190 trillion years of pain experience while actively dying occur annually. Given that the pain of death is likely hundreds or thousands of times more intense than the pleasure of ordinary existence, to my mind, this makes quite clear the case for most fish having negative lives. Most of them die very shortly after birth in a way that is very painful.

4. A more reasonable estimate looks like the following—fish lay on average about 1 million eggs (this is conservative, because some like tuna lay vastly more). 2% become larvae, and 10% of those become conscious. Thus, for every adult fish who reproduces (if we assume to make the math easy that two are needed, ignoring hermaphrodites) there are ~1000 fish that die in the larval stage after becoming conscious. If adult fish live around a year, and it takes a day for a larval fish to die on average (probably most starve to death which takes multiple days), then about three times as much time is spent dying while conscious in the larval stage than as adults. Probably from this about 10 fish progress past the larval stage, and eight of them die before reproducing, likely after just a few weeks. (I’ll put the more full calculation in a footnote, which adds up to about 1.8 times as much time spent dying as being an adult).^[2]

5. If the amount of time a species spends starving to death is commensurate with the amount of time it spends living as an adult, its life is not overall worth living! If the human population spent about a third as much time starving to death as living as adults, humans would not have overall positive lives. And as we’ve just seen, by the estimate that strikes me as more reasonable, probably more time is spent starving to death than being an adult fish. Of all the moments of conscious life experienced by fish, a substantial portion, maybe about 10%, is spent starving to death.

1. Often death takes quite a while, if it is from starvation, dehydration, or disease. Some animals, for instance, are eaten alive from the inside by parasites.
2. Even aside from death, given the constant struggle of life, I’d guess most animals’ everyday lives aren’t very pleasant.

Doesn’t concern about this assume utilitarianism?

No! Extreme suffering is bad. It’s bad to be eaten alive, ripped apart by predators, and to starve to death. Utilitarianism has not cornered the market on thinking extreme suffering is bad. Even if you’re a non-utilitarian, you should be concerned that there are biological systems that create short-lived animals who experience intense suffering—just like you’d be concerned if a machine created 10 billion puppies and left them all to starve to death. Have some empathy; any sane calculation will hold that a creature desperately trying to escape from unbearable agony—being ripped limb from limb by predators—is ethically serious. If you would give anything to make it stop if it was done to you, you should try hard to make sure it doesn’t happen to others!

Nature is a deeply cruel place. Most animals live short lives and die painfully. We should regard this situation as tragic and support actions that bring fewer animals—especially insects—into existence. I would not want to be born as an insect or fish, struggle for a few days, and then be eaten alive. If offered the chance to experience the life of every conscious creature on Earth so far, I would confidently turn it down. But if you find an insect’s life unpleasant enough that you wouldn’t choose it over non-existence, you shouldn’t support creating huge numbers of insects that live their lives rather than non-existence. You shouldn’t force unto others an experience that you yourself would be wise not to endure.

The jury is still out on this conclusion to some degree. It’s certainly possible that animals in nature mostly live pleasant lives. But overall, I think this is fairly unlikely. Most likely, wild animal suffering outweighs wild animal welfare by quite a sizeable margin.

(The article below discusses in more detail what we should do about this tragic situation, and see here for explanations of why the views according to which we should preserve nature even if animals mostly live miserable lives make little sense.)

1. ^{^}

   If genetically modified to have capsaicin receptors.

2. ^{^}

   For the more full calculation: assume conservatively that adult fish are 1.5 times as likely to be conscious as the later stage fish I previously assumed were conscious. Let’s say they have a 75% chance of consciousness and the later stage larvae (called precocial larvae) have a 50% chance of consciousness. Assume additionally that the larvae I previously counted as non-conscious (altricial larvae) have a 10% chance of being conscious.

   Now, York assumed it takes larvae ten seconds to die because they have a feeding pouch. But that seems conservative because: 1) some probably suffocate; 2) some probably still starve; and 3) parasites might kill them. Even if only one in a thousand starve, this will majorly skew the 10 second numbers. For this reason, assume that they take on average 10 minutes to die.

   Taking this into account, per adult fish there are 500,000 eggs. From this, there are 10,000 altricial larvae that take 10 minutes to die. Then, there are 1,000 precocial larvae who take about a day to die. Adjusting for sentience, we get 3/4 of a year of conscious adult fish life among the fish who survive long enough to reproduce, 500 days (about 1.37 years) of precocial larvae death which typically occurs a few days after sentience, and ~6 days of death among the altricial larvae. Overall, about 1.8 times as much time is spent dying as a juvenile as living as an adult.

   If rather than taking into account time, we simply take into account number of individuals and adjust for probability of sentience, out of 500,000 eggs, then in expectation about 1,000 individuals die in the altricial stage, 500 die in the larval stage, 10 progress past the larval stage, and one goes on to reproduce. This means that of the expected conscious individuals born, probably about .67% don’t die almost immediately as larvae, and even of the lucky .67%, most die shortly after.