Peter Singer's drowning child argument is one of the clearest pieces of moral reasoning in modern philosophy. Its premise is that suffering is bad and that we are obligated to prevent it when we can. Most people who encounter it cannot find a satisfying objection.
But Singer never quite defines what suffering is. He takes it as obvious — the drowning child is suffering, the child dying of preventable disease is suffering, and we should stop both. The argument works because we nod along to each step.
The harder question is: when you actually press on it, what are we talking about?
What can suffer? How much? Can a shrimp suffer? Can the version of you that doesn't exist yet? Does reducing suffering even work the way we assume? These questions don't undermine Singer's conclusion. They complicate it — and they point toward forms of suffering that are more hidden, more intractable, and in some cases far larger in scale than the ones we usually see.
Three definitions that change everything
The word "suffering" does a lot of work. It covers three quite different things, and which definition you accept changes what you spend your moral energy on.
Suffering as pain — the simplest and oldest view. Suffering is what it feels like to be hurt, sick, or dying. This is Jeremy Bentham's formulation: "the question is not, can they reason? nor, can they talk? but, can they suffer?" Pure and direct. A child with a disease suffers. A person in a war zone suffers. Your job is to stop it.
Suffering as frustrated desire — a wider view. Suffering includes not just pain but the experience of having your will blocked: being imprisoned when you want to be free, being silenced when you need to be heard, living a life that is safe but small and controlled. A person in a care home with excellent medical attention but no autonomy, no choices, no agency — on this definition, that person suffers even if they feel no physical pain. An animal in a cage with adequate food and no cruelty still suffers if its entire nature is thwarted.
Suffering as resistance to what is — the hardest to grasp but possibly the most accurate. On this account, suffering is not a thing that happens to you but a relationship you have with your experience. It is the clinging, the wishing things were otherwise, the gap between how things are and how you need them to be. Two people in identical circumstances — same illness, same loss, same circumstances — can suffer entirely differently, because of what they do with the experience. A monk in chronic pain may suffer less than a healthy person in paralysing anxiety about the future.
The Buddhist concept here is dukkha, often translated as suffering but more precisely translated as unsatisfactoriness — the deep friction of wanting things to be other than they are. On this account, you could eliminate disease, poverty, and violence and still have a planet full of suffering, because the suffering would relocate to whatever gap remained between what people have and what they want.
This is not an excuse for inaction. It is a warning that suffering has roots we can't easily reach from the outside.
The measurement problem — and what can suffer at all
Once you try to measure suffering, you run into a wall.
Health economists use QALYs — Quality-Adjusted Life Years. You ask people how many years of perfect health they would exchange for ten years with a given condition. If people say they'd trade eight years of health for ten years with depression, depression gets a weight of 0.8. These numbers decide which treatments governments fund, how we allocate scarce medical resources, what a life is worth in a courtroom.
The problem: people who actually live with a condition almost always rate it as less bad than people who imagine having it. Deaf people, on average, rate deafness as barely reducing their quality of life. Hearing people rate imagined deafness as catastrophic. The QALY system mainly uses outsider estimates. Which means the official measure of suffering systematically misrepresents the experience of the people doing the suffering.
Then there is the question of what can suffer at all — and here the ground becomes genuinely unstable.
In 2021, the UK government reviewed the scientific evidence on whether crustaceans feel pain. The review concluded: strong evidence for octopuses, moderate evidence for crabs and lobsters. The Animal Welfare (Sentience) Act was extended to cover them. A legal, political, and scientific determination that a lobster probably experiences something when it is dropped into boiling water.
Researchers at Rethink Priorities are attempting something stranger still: a cross-species welfare comparison, trying to assign empirically grounded moral weights to different animals. Does a chicken suffer as much as a human? Their current best estimates range from roughly 1/5 to 1/50, depending on the domain and the method. The numbers carry enormous uncertainty — but the exercise matters because it makes explicit what we've always been doing implicitly. We already have moral weights for animals. They're just unexamined, inconsistent, and probably wrong.
The wall you hit, eventually, is the hard problem of consciousness. You can measure stress hormones, nociceptors, avoidance behaviour — all the external signatures of pain. What you cannot measure is whether there is something it is like to be that organism. Whether the lights are on. Whether the shrimp's recoil is accompanied by an experience, or is just a biological reflex running in the dark.
Nobody has solved this. Not for shrimp, not for fish, not fully even for other humans — we infer it from behavioural similarity and shared evolutionary history. Where the inference is strong (primates, mammals), we act accordingly. Where it weakens (fish, crustaceans, insects), we mostly act as if the answer is zero, not because the evidence supports zero but because the alternative is inconvenient.
The inconvenient alternative: there are approximately ten thousand billion billion insects on Earth. If insects experience anything at all — even one ten-thousandth of what a human experiences — the aggregate suffering occurring in the natural world at any given moment dwarfs everything in human history. Either insects don't suffer in any morally relevant sense, or we are surrounded by suffering at a scale we've never seriously engaged with.
Where suffering actually hides
Here is the finding that has changed how serious researchers think about harm reduction more than almost any other: the ACE study.
In the 1990s, Dr Vincent Felitti at Kaiser Permanente was running a successful obesity programme. Patients were losing significant weight. They were also dropping out at high rates — just before reaching their goals. He interviewed the dropouts and expected to find failure of willpower. He found, instead, that most of them described childhoods of abuse, neglect, or violence. Overeating, many of them said, was a solution: it made them feel safe, numb, or large enough to be left alone.
He contacted the CDC. Together they designed the Adverse Childhood Experiences study — the largest investigation of its kind. Over 17,000 middle-class, college-educated adults with jobs and healthcare were asked about ten categories of childhood trauma: physical, emotional, and sexual abuse; neglect; domestic violence; household substance abuse; household mental illness; parental separation; and a household member in prison.
The results were linear and steep. Each additional category of childhood trauma increased the likelihood of heart disease, cancer, chronic lung disease, depression, suicide attempt, alcoholism, and drug use — substantially. People with ACE scores of 6 or higher had life expectancies roughly 20 years shorter than those with scores of zero.
This was independent of income, race, education, and healthcare access.
The mechanism: chronic childhood stress doesn't just leave psychological marks. It physically rewires the developing brain, alters the stress response system, changes inflammatory pathways, and modifies gene expression in ways that persist into adulthood and can be passed to the next generation. The effects are biological, measurable, and long.
The implication for suffering is this: most of the suffering we spend resources trying to manage — addiction, mental illness, chronic disease, violence, poverty — is substantially downstream of something that happened in childhood. We are, as one researcher put it, mopping up floods rather than turning off taps.
An intervention that reduces domestic violence in a household with a three-year-old doesn't just help the adult. It changes the ACE score of that child. Which changes their brain development. Which changes their adult health trajectory. Which changes the ACE scores of their children. The ripple effect is generational, and the return on investment is almost certainly higher than almost any other form of harm reduction we know of.
The ACE study has been known since 1998. It has barely changed policy. Political systems optimise for electoral cycles. The payoff here is in decades.
Does any of this work?
There is a challenge to the entire project of reducing suffering that is rarely addressed: the hedonic treadmill.
In the 1970s, researchers Philip Brickman and Donald Campbell described the tendency of humans to return to a roughly stable level of wellbeing regardless of circumstances. Their most famous study compared lottery winners to people who had become paraplegic. One year later, both groups had largely returned to their pre-event baseline. The lottery winners were not significantly happier. The paraplegic individuals were not significantly less happy.
If people just adapt, does any of this matter?
The more careful answer: it depends on the type of suffering. We adapt well to material gains, improved comfort, and most circumstantial changes. We do not adapt well to chronic pain, loss of close relationships, social rejection, lack of autonomy, or extreme deprivation.
This matters because it points toward where harm reduction actually works. Lifting people out of extreme poverty — where the suffering is in the chronic, relentless category — produces lasting reductions in wellbeing that don't simply erode. Giving someone a better television does not. Ending a child's experience of domestic violence changes their developmental trajectory in ways that are not undone by later adaptation. Reducing loneliness — which increases mortality by roughly as much as smoking 15 cigarettes a day, and which is rising in most developed nations — is in the slow-adaptation category: something people don't simply get used to.
The treadmill also has a hopeful implication. If material improvement above a threshold fades, then wellbeing doesn't ultimately depend on accumulating more. The research consistently finds that what people cannot adapt away from — in the positive direction — is deep social connection, meaning, and a sense of agency over their own lives. These are not things that require wealth to produce. They are also, not coincidentally, exactly what extreme poverty, trauma, and isolation destroy.
What we're actually fighting
Singer's argument is that suffering is bad and we should prevent it. That remains true.
But the more you press on what suffering is, the more its real geography reveals itself. It hides in childhoods. It compounds across generations. It exists, in some form we don't fully understand, in animals we've never thought to consider. It doesn't always respond to the interventions we aim at it, because we've been aiming at symptoms while the roots go deeper.
None of this makes the drowning child thought experiment less powerful. It makes the pond larger.
Further reading
- The ACE study: Felitti et al. (1998), "Relationship of Childhood Abuse and Household Dysfunction to Many of the Leading Causes of Death in Adults," American Journal of Preventive Medicine — the original paper, freely available.
- David Chalmers, "Facing Up to the Problem of Consciousness" (1995) — the paper that named the hard problem.
- Rethink Priorities' Moral Weights Project (rethinkpriorities.org) — the most serious ongoing attempt to quantify welfare across species.
- Better Never to Have Been by David Benatar (2006) — the most rigorous philosophical argument that bringing any person into existence is a harm. Not for the faint-hearted.