Evolutionary cost-balancing arguments in welfare biology (the study of animal well-being in the wild) assume that producing suffering and happiness has metabolic or neural costs, and that natural selection allocated these states efficiently. A specific version of this reasoning – which I'll call the "Evening Out Argument" – holds that if a type of suffering is highly likely and largely unavoidable for an animal, evolution would reduce its intensity. The logic is that, if suffering functions to motivate learning and avoidance, then intense suffering provides little behavioral benefit when the animal can barely avoid the bad outcome anyway – it's just cost without payoff.

I argue this logic fails for large categories of animal experience. Most notably, it doesn't apply to background motivational states like hunger or anxiety, which function differently from discrete learning signals. I'll also discuss the prevalence of chronic and maladaptive suffering, concerns about the biological plausibility of some assumptions behind the Evening Out Argument, and why high infant mortality doesn't imply reduced suffering in long-lived species.

For context, many people familiar with the topic have explicitly stated that they do not consider cost-balancing arguments to be strong; however, I have seen some people in EA use “Evening Out” logic in discussions and place a fair bit of weight on it.

A brief history

In his famous 1995 paper “Towards Welfare Biology: Evolutionary economics of animal consciousness and suffering,” Yew-Kwang Ng formulated the “Buddhist Premise”: that under reasonable conditions, suffering should exceed enjoyment for the average wild animal. In 2019, Zach Groff co-authored an update to that paper with Ng himself, where they pointed out a math mistake in Ng’s earlier work. As Groff explained in a summary post, the math encapsulates an initially missed consideration about animal well-being:

The mistake in the original paper may appear technical, but it is suggestive of an aspect of wild animal suffering that prevailing intuitions in the space may miss. [...] Most wild animals have far more offspring than can survive to maturity, so the experience of an average animal is to be born and then nearly immediately suffer a horrible death. Based on this, Ng speculated that the Buddhist Premise should hold before offering a proof of it based on the axioms. But the intuitive argument misses a potential evolutionary pressure the math picks up. Because the costs (e.g. resource usage) of suffering depend on the probability of experiencing suffering, when the probability of suffering increases, the severity of suffering should decrease. In other words, if the probability of being born and then immediately dying is sufficiently high, then increasing the amount of suffering is less advantageous for genetic reproduction.

I propose calling the argument Groff makes at the end of that passage the “Evening Out Argument.” It’s an evolutionary cost-balancing argument which, this time, pushes against the Buddhist premise.

To restate the argument the way I understand it, the idea is that if a certain type of suffering is particularly likely for an animal, this high likelihood for suffering evens out the severity of that suffering. Why would this happen? Well, our brains producing hedonic states like suffering and happiness is assumed to have certain costs (e.g., metabolic or neural costs), so natural selection is trying to balance those costs and allocate suffering efficiently. If the function of suffering is to motivate us to learn things and avoid things, and if a negative event is so frequent as to be nearly unavoidable, then intense suffering provides diminishing behavioral returns and incurs disproportionate costs. Over time, natural selection would “economize” to tone down the suffering. If we were to imagine suffering as an alarm system, we can afford to make the alarm very loud if it rarely sounds, but a loud alarm that sounds nearly all the time would cost too much without providing a proportionate benefit.

Groff also added:

Note well: suffering may very well dominate enjoyment in nature. We cannot arrive at a conclusion on that. Our point is that it does not necessarily dominate.

Based on the “note well” and other comments, it seems that Groff primarily wanted to correct a mistake and prevent people from overupdating based on the claims and framings in the 1995 paper. Groff himself explains that he wouldn’t put a lot of stock into evolutionary cost-balancing reasoning:

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.^[1]

As others^[2] have also pointed out, I think we’d get the best sense of net wild animal welfare not from abstract arguments but by studying individual animals up close. I don’t think anyone who works on these topics really disagrees (my post is directed more towards non-experts than experts). Still, I have seen versions of the Evening Out Argument come up here and there in discussions, and I got the impression that some people put a lot more weight on these sorts of considerations than I would.

So, I’m writing this post to prevent a situation where non-experts overupdate on Groff’s correction or the Evening Out Argument in particular,^[3] similar to how, earlier on, some people may have^[4] overupdated on the clean results and the authoritative-sounding economics modelling in Ng’s foundational paper.

The upshot is that math and economic reasoning tell us little about the distribution of happiness and suffering in nature. Instead, we should either remain uncertain about net wild animal welfare (if we know of no better considerations than the cost-balancing arguments), or we should mostly form our impressions from stronger/more direct considerations (e.g., from studying the lives of individual representative animals).

In the rest of this post, I’ll give several arguments and examples to illustrate how the Evening Out Argument makes implausible predictions or rests on dubious assumptions.^[5]

Mayflies vs baby turtles

I believe that the Evening Out Argument, that animals for whom early death is a near certainty might suffer less from injuries than we would otherwise assume, has some plausibility. The most convincing supporting example I can think of is mayflies:

Mayflies only live 1-2 days in their adult form. Using armchair evolutionary reasoning, I’d indeed consider it plausible that they wouldn’t suffer much when they get injured, because avoiding injury is a comparatively low priority to them. (Instead, their adult life is all about mating.)

The literature on insect pain behavior, which I admittedly don’t know well, even contains some examples where insects like praying mantises continue feeding or mating despite severe injuries. However, these cases are mostly anecdotal and could also be consistent with those insects modulating or suppressing their pain rather than simply not feeling any (see this source and a summary of it). So, I don’t know if mayflies display an absence of pain behavior; still, since they wouldn’t benefit evolutionarily from having their attention directed towards limb protection, it would make sense if they experienced less pain^[6] than animals that need their limbs for much longer.

Overall, the Evening Out Argument seems plausible to me when it comes to (e.g.) mayflies. However, what matters isn't the probability of suffering and early death, but whether pain-guided learning would be adaptive given the organism's potential remaining lifespan if it survives. The crucial point here is that mayflies aren’t going to need their limbs for long – their objective is to reproduce quickly and then die.

Compare that to sea turtles: Sea turtles can have offspring in the hundreds, but the baby turtles get to live 50-80 additional years, sometimes even longer, if they make it all the way into old age. Since sea turtles are going to need their limbs for such a long time, they probably benefit from learning to look after them. As a result, I don’t see why baby sea turtles should experience lessened suffering when they get eaten by birds, crocodiles, etc, even though such an outcome is unfortunately very likely for them. Natural selection shapes traits based on what's adaptive for individuals who survive to reproduce – all reproducing sea turtles are ones who survived early mortality and went on to benefit from pain-guided learning for decades. Evolution lacks compassion, so it won’t turn on the pain switch only after the turtles made it through the most difficult stage.

Hedonic accounting: discrete events vs background states

Evolutionary cost-balancing arguments seem plausible to me when I consider the hedonic reward (positive or negative) from discrete, fitness-relevant events in an animal’s life – for instance, injury, social ostracism, mating, or eating a high-quality meal. Animals learn from positive or negative rewards that are tied to these events. Accordingly, it seems at least plausible that an animal would experience more or less reward depending on how common or uncommon the relevant events are among their species (in the ancestral equilibrium), and how much they increase or decrease the individual’s fitness.

However, a lot of the brain’s hedonic accounting is not about the reward from discrete events, at least not directly. Instead, animals have background motivational states like hunger, horniness, boredom, fear, etc. These background states are less about long-term learning and more about directing our attention in the moment. Negative background states (hunger, thirst, anxiety, discomfort) signal “something needs to change.” Positive background states (satiation, safety, contentment) signal “continue as you are.”

Importantly, when we apply the logic of the Evening Out Argument to background motivational states, the argument fails to apply. The function of hunger, for instance, is not to learn a lesson that, once learned, makes the signal unnecessary. Instead, it is to seek food now (and sustain ongoing motivation). So, even if an animal’s typical experience involved hunger very often and nearly-unavoidably-so, that in itself would be no reason for natural selection to tone down the hunger signal. An animal that habituated to chronic hunger would stop foraging – so the signal can't fade.

Not only does the Evening Out Argument fail to apply to background motivational states, but we should also expect that these states more often lean towards the negative, because there are structurally more ways for an organism to be in a state that requires correction than for it to be just right. Relatedly, there’s the Anna Karenina principle: there are many ways to be suboptimal, few ways to be optimal.

Anni Leskelä's framing of suffering as primarily being a motivational state linked to homeostasis reinforces this point (from a blogpost on whether social animals suffer more):

Contrary to the standard biology textbook view, suffering is more than just a signal of a harmful situation. [...] Suffering as a motivational state is typically the mental component of an animal’s homeostatic regulation, i.e. the processes that keep all the relevant physiological variables between healthy parameters. [...] [T]hirst when your blood volume starts to drop, pain when a wound opens and leaves you vulnerable to pathogens and blood loss, sickness when you have ingested toxins and need to expel them. When the threat isn’t currently actual but can pretty reliably be predicted to come true unless you take physiological or behavioural precautions, your species will evolve predictive homeostatic processes. Many of these predictive processes are cognitive or emotional in nature, e.g. people often feel distress in darkness and high places – things that cause absolutely no damage in themselves, but correlate with future homeostatic disturbances.

In other words, even if there’s no discrete pain-causing event happening right now, animals may be thinking about risks or threats in the future (prompting vigilance), or experiencing discomfort from hunger or the cold (prompting correcting actions). Being “exactly right” in all relevant aspects is more rare.

In that light, one feature of these background states/motivational predictive processes I find worth highlighting is prey animal psychology (or relatedly, the landscape of fear). Prey animals in areas with predators, even when no predators are currently present, experience a baseline level of anxiety, hypervigilance, heightened or chronic stress, and other PTSD-like effects. Again, the Evening Out Argument doesn’t apply to these levels of background anxiety and stress because the whole point of them is for the animal to stay vigilant.

Lastly, because energy is transferred up the food chain in an inefficient way (see: the trophic pyramid), wild animal prey species have more biomass than predator species, and many prey animals have large numbers of offspring. This means prey animal psychology accounts for a large proportion of wild animal experience.^[7]

Chronic suffering and old age

So far, my discussion of unpleasant/uncomfortable experiences has covered mental states that are still biologically adaptive. However, it is not uncommon for animals to suffer for maladaptive reasons. When this happens, it’s not the suffering that is prohibitively costly – on the contrary, it would be costlier to make the organism safe from those kinds of malfunctions.

For instance, consider pain and discomfort from parasitism where the host has no behavioral defenses. I’m particularly thinking of internal parasites, where no amount of behavioral response can remove them (unlike the evolved responses to external parasites, such as grooming in primates or rolling in dirt or dust for certain mammals). Since parasites often benefit from keeping their host alive, the chronic pain and general weakening can last for a long time.

Also, old animals can suffer enormously with minimal fitness consequences, because selection pressure is weak when it comes to events that happen later in life, after an animal has already reproduced. For instance, many longer-lived animals develop arthritis as they age, which we know from humans (and from behavioral evidence in non-humans) is a painful chronic condition.

The cost-balancing logic of “Evening out” assumes that suffering is like an alarm that evolution allocated efficiently. But if chronic suffering is just a byproduct of crude nociceptive machinery, there's no reason to expect the resultant suffering to be limited to things the animal can do something about, and no reason to expect that suffering to shut itself off if the alarm stays triggered constantly. Unfortunately, this means that once things are out of control for the animal in question, the suffering involved can get a lot more severe, or can go on for a lot longer, than would make “economic sense” evolutionarily.

Lastly, it is worth noting that there’s no equally common counterbalancing phenomenon where things coincidentally go well and an animal ends up chronically “extra happy.” This asymmetry isn't by accident: an animal stuck in chronic suffering can often still reproduce, whereas an animal stuck in chronic bliss would likely stop doing what's needed to survive. Evolution tolerates the former but selects against the latter.

Other biological plausibility concerns: path dependencies and modularity

The previous section showed that evolution tolerates chronic suffering when it doesn't impair reproduction. A related point is that evolution also can't easily fix suboptimal designs, even when they cause suffering and even if the suffering is maladaptive, because evolution is constrained by path dependencies.

Optimizing assumptions make sense in economics, but they aren’t straightforwardly applicable to biology. Evolution cannot go back to the drawing board, so innovation in biology is path-dependent and constrained because it must happen in incremental steps of small mutations.^[8]

To illustrate how animals carry genetic baggage, the wild animal welfare researcher Mal Graham gave the example^[9] of the human backbone/spine, which we inherited from quadruped ancestors. Our back is in many ways poorly designed, which leads to various complications (e.g., lower back pain or disc hernias).

Similarly, we shouldn't assume that hedonic systems are optimally calibrated just because they evolved – they may carry their own 'genetic baggage' from earlier evolutionary contexts. (For instance, an alien who doesn't understand biology might think that the existence of superstimuli like refined sugar or drugs is a thing to be celebrated, but in reality, they often lead to addiction, where the reward system gets hijacked in ways that produce more suffering than pleasure.)

A separate concern is that suffering may not be governed by a single dial. There is evidence that human motivation, for instance, is modular, with multiple (semi-)independent systems governing reward-seeking and avoidance of punishment or threat. Since the avoidance system relates to anxiety and fear, tuning it down doesn't eliminate all suffering – in fact, it can even increase exposure to situations that cause other forms of suffering (physical injury, social conflict). If something like this is going on, it would immediately complicate the logic of the Evening Out Argument. (See the puzzle about fast life history strategies in humans in the appendix for a more detailed example and discussion of modularity in motivation.)

Ancestral environments and environmental mismatch

Animal populations are almost never in their ancestral equilibrium. This likely affects their welfare negatively, on average.

Quoting Michael St Jules, who originally made this argument:

Whatever your beliefs about the expected value of the average welfare of animals in the wild under constant conditions and at equilibrium, I think you should expect, a priori (weakly, without further evidence), their average welfare to be lower after conditions change.

As Michael notes, the argument is another application of the Anna Karenina principle. Animals have evolved to live in a specific ecological situation in their ancestral environment. They are adapted to a specific temperature range, a specific level of intraspecies competition over food or mates, specific diseases or parasites, etc. Changes to the environment can sometimes be good for animal welfare^[10], but more often than not, random changes to a precise adaptive equilibrium are going to be negative.^[11]

Even granting the Evening Out logic, it only applies when animals are near their ancestral equilibrium. But environments change constantly, and human activity has accelerated this dramatically (to the point that many biologists are calling our era the “Anthropocene”). When conditions shift faster than evolution can respond, animals face novel stressors their hedonic systems weren't calibrated for – precisely the situations where we shouldn't expect suffering to be “evened out.”

Concluding thoughts and implications

If I’m correct that the Evening Out Argument is very weak, what follows?

The main upshot is that we should either remain uncertain about net wild animal welfare (if we know of no better considerations than the cost-balancing arguments), or we should mostly form our impressions from stronger/more direct considerations (e.g., from studying the lives of individual representative animals).^[12]

Some readers may worry that undermining the Evening Out Argument amounts to pushing people toward viewing nature as net-negative, which could lead to radical or harmful conclusions. However, I don’t think that any such conclusions would be warranted from someone concluding that. More generally, I don’t think disagreements among effective altruists on net animal welfare matter much in practice, and they shouldn’t cause any tensions.

• From a longtermist perspective, preventing unnecessary suffering remains a priority either way, no matter one’s view about net wild animal welfare. Likewise, since there are better ways of making the most out of the future than spreading Darwinian ecosystems, spreading Darwinian ecosystems would never be a top priority even for people who think happiness in nature outweighs suffering.^[13]
• For "shorttermists,"^[14] opinions on net animal welfare could affect how they evaluate small-scale interventions that people already regularly engage in (converting lawns to gravel, not opposing certain construction). If the stakes are low, people should just do their thing. Anyone convinced of the Buddhist premise still has strong reasons – irreversibility, uncertainty from knock-on effects^[15], norms against unilateral action and fanaticism – to avoid large-scale interventions (such as, e.g., advocating for destroying the rainforest).
• From a practical movement- and coalition-building perspective, it seems best for advocates of any cause, including wild-animal advocates, to avoid unnecessary in-fighting and to signal to outsiders that they are reasonable people who can be worked with.

To summarize, I don't think the question is action-relevant for most EAs. But going back to the Evening Out Argument, I think it’s good practice to criticize overrated arguments that are actually weak. I also wouldn’t want people to think things that are “too good to be true” (specifically, evening out applying to many of the most severe sources of suffering) about a topic that is really quite large in scope and touches on big philosophical questions (e.g., “Is nature good?”). Even if there are few immediate practical upshots, it seems important to think about things accurately and carefully.

Appendix: A puzzle for “Evening Out”

Here's a puzzle involving variation in life history strategies among humans. The following is somewhat contested, but I think there’s “something there.”

Among humans, some people appear to be pursuing what we could call "fast" life history strategies (earlier reproduction, higher risk-taking, shorter time horizons, and less investment per offspring). These strategies are associated with unpredictable or harsh early environments and, in their more extreme manifestations, with cluster B personality traits.

People pursuing fast life history strategies experience negative events (relationship instability, conflicts, consequences of risk-taking) more frequently. By the cost-balancing logic of the Evening Out Argument, we might expect that the hedonic intensity of each setback should be lower for them – that evolution should have "economized" on their suffering (and incentivized risk-taking).

However, interestingly enough, this seems empirically backwards, at least for some cluster B presentations. Borderline personality disorder, for instance, is characterized by intense emotional reactivity rather than dampened responses. Both borderline personality disorder and anti-social personality disorder are linked to increased suicide rates.^[16]

Maybe fast life history strategies come with specific forms of suffering being reduced. For instance, there's evidence that antisocial traits are associated with reduced fear conditioning and lower behavioral inhibition (in the BIS/BAS framework from personality psychology). In other words, people with these traits are less deterred by the chance of bad things happening. So perhaps fast life history diminishes anticipatory anxiety while leaving other forms of suffering intact or even amplified?

But if that's true, it complicates the simple cost-balancing picture. As already mentioned in the section on “Other biological plausibility concerns,” it suggests that “suffering” isn't a single dial that evolution turns up or down based on event frequency. Instead, different regulatory systems (social pain, physical pain, emotional volatility, fear of punishment/negative reward) can be tuned somewhat independently, and tuning one down doesn't necessarily reduce overall suffering – it might even increase it by leading to more situations that activate other systems.

This example is admittedly speculative: it involves within-species variation shaped by developmental plasticity, not cross-species differences shaped over evolutionary time. Still, the puzzle illustrates how quickly cost-balancing reasoning can run into complications when applied to real-life biological systems.

Acknowledgments

Thanks to my wife Lydia for helpful comments and edits, and to Claude 4.5 for useful critiques and editing suggestions.

---

1. The quoted comment was in reply to Brian Tomasik, who had commented: “When I first discussed Ng (1995)'s mathematical proof with some friends in 2006, they said they didn't find it very convincing because it's too speculative and not very biologically realistic. Other people since then have said the same, and I agree.
   [...]
   While I think these kinds of arguments are interesting, I give them relatively low epistemic weight because they're so theoretical. I think the best way to assess the net hedonic balance of wild animals is to watch videos and read about their lives, seeing what kinds of emotions they display, and then come up with our own subjective opinions about how much pain and pleasure they feel.” ↩︎

2. For instance, Brian Tomasik, as quoted in the previous endnote. ↩︎

3. As I said in the intro, this isn’t just a speculative concern. I have seen some people in EA place significant weight on this argument in discussions. ↩︎

4. I say “may have” because it could be that most people who became concerned about suffering in nature had the same experience as Brian, where they mostly updated from having their attention drawn to the high infant mortality in wild animals, but didn’t place much stock on any particular mathematical model. ↩︎

5. Thanks to the EA forum user Linch for prompting me to think about these topics – see the discussion in the comment section inspired by his post “Eating Honey is (Probably) Fine, Actually”. ↩︎

6. Note that mayflies are the adult stage of the insect. Mayfly nymphs – the baby stage – live a lot longer before they morph into adult mayflies, and their nymph lifestyle involves less seeking and risk-taking behavior and more maintenance and avoidance behavior. ↩︎

7. Of course, some large animals like elephants are neither prey animals (apart from getting hunted by poachers) nor predators, and elephants actually make up a surprisingly large percentage of wild (non-arthropod) land animal biomass. Still, the point about prey animal psychology being the common wild animal psychology remains valid. ↩︎

8. Of course, there is also no forward-planning in evolution, so if a mutation doesn’t confer a benefit right away, the allele is unlikely to increase its frequency in the population, even if it were paving the way for further mutations that would eventually stack together to produce something greatly advantageous. ↩︎

9. See the discussion section of the video here, starting at 18mins20 ↩︎

10. E.g., in humans, many changes were positive and even though modern life makes more people addicted to things or leads to more upwards social comparisons, it’s hard to see how we are worse off than in the distant past. ↩︎

11. An illustrative example here is cats often being slightly at risk of being a bit cold indoors in ordinary homes. Domestic cats descend from the African wildcat, which evolved in the warm climate of the Near East and North Africa. Their baseline body temperature is also higher than ours, which means they need to work harder to maintain it. That’s why cats are notorious heat-seekers. It gets especially bad for feral cats, who have a reduced lifespan compared to indoor cats at least partly due to weather exposure. ↩︎

12. I also suspect that there isn’t a uniquely correct answer to when a given amount of suffering is “outweighed” by a given amount of happiness. This would mean that experts could reach different conclusions (even under ideal reasoning conditions) because, while they would agree about all the facts, they might start out with different normative intuitions about the features of suffering and happiness they deem the most relevant and worth caring about. ↩︎

13. Some people may value untouched nature as an intrinsic good. Ideally, we can satisfy their values with a compromise of (e.g.) keeping a limited number of nature reserves intact but not spreading Darwinian life on an astronomical scale. ↩︎

14. I’m primarily thinking of people with the epistemic belief that affecting the long-run future is too intractable, so they choose to restrict their altruistic attention to actions that have clear, easily overseeable benefits. ↩︎

15. Especially if someone’s argument for “shorttermism” is that long-run effects make our altruistic actions meaningless because we cannot predict the sign of any major long-run consequences, that would be a strong argument to focus on small-scale interventions with contained effects, not large-scale irreversible interventions with knock-on effects ↩︎

16. Narcissistic personality disorder doesn’t obviously fit the picture, but it is associated with suffering and conflict in one’s social surroundings, and, at least in some presentations, grandiosity is not always a shield from neuroticism (see, “narcissistic collapse” or “vulnerable narcissism”). ↩︎