Ninety percent of crying bouts answered within twenty-five seconds.
And that's not a mother doing it alone. That's everyone within earshot.
Hannah sent us something this week. She's been reading about alloparenting, which is a word I'll let Herman define in a moment, and she noticed something about maternity leave. Her point is that in a hunter-gatherer society, a mother doesn't stay home alone with a baby for months and then suddenly go back to full-time work. The return is gradual. It starts earlier, it happens around other adults, other mothers, older children.
And she may spend less one-on-one time with the baby than a modern stay-at-home mother.
Right, but she's also not separated for eight or ten hours a day. So Hannah's question is whether our whole daycare model is built around the wrong assumption. Instead of the two options we argue about, mom stays home or baby goes to daycare all day, what would a system look like that recreates the social structure humans actually evolved for? Daycare near the workplace, parents in and out during the day, flexible drop-off, mixed ages, multiple familiar adults.
And she asked for both halves. What do we actually know about attention and ratios in hunter-gatherer societies, and then, if you were designing a modern daycare around those principles, what would it look like?
That second half is the one nobody has built. Which I find strange.
The reason it's strange is that the binary Hannah's pointing at is not ancient. It's about seventy years old.
Seventy years is being generous.
In the sense that the two-earner household with paid daycare as the default is a post-war, industrialized arrangement. The forager pattern is millions of years deep. And the two things being compared aren't even the same category of thing. One is a human universal, the other is a mid-twentieth-century labor-market solution.
Start with the concept, because alloparenting is the load-bearing word here.
Alloparenting is care from anyone other than the mother. Fathers, grandmothers, siblings, unrelated campmates, adolescents, older children. And the claim from the Cambridge group, Chaudhary and Swanepoel, is that this is not a convenience or a fallback. It's a core human adaptation. Which matters because the competing story, the intensive mothering narrative where one adult is supposed to meet every need, is the anomaly. Not the norm.
And the isolation that goes with that anomaly.
That's the emotional hook here, and I don't want to undersell it. Multiple researchers link the isolation of modern parenting to postpartum depression and maternal exhaustion. Emily Emmott at UCL put it as women being put in an impossible situation and set up to fail. Gul Deniz Salali, who studies the BaYaka, had a baby nine months before that NPR piece and said it had been really lonely, and that the BaYaka moms have so much support. That's a working anthropologist saying it.
So the arc is: what the attention data actually show, then the gradual return to work, then the design question. And the design question ends in a gap.
It ends in a gap, yeah.
Let's do the numbers first, because they're the part that bends your head.
The study to start with is the Mbendjele BaYaka one. Chaudhary and Salali, published in Developmental Psychology in November 2023. Eighteen children, ages zero to four, in the Republic of Congo. Twelve-hour focal follows, and care coded every twenty seconds. So a researcher is watching a specific child for twelve hours and writing down what's happening every twenty seconds.
That's a brutal protocol.
It's brutal and it's why the sample is eighteen. What they found is that infants received attentive care and physical contact for about nine hours a day, from up to fifteen different caregivers.
Nine hours of the twelve.
And it's not one person cycling through. The care split came out roughly: alloparents thirty-six percent, fathers six percent, mothers sixty percent.
So the mother is still the single largest provider.
Substantially the largest. That's the piece that gets lost when people hear "alloparenting" and assume it means the mother is a bystander. She's the majority of it. She's just not the only one, and she is rarely alone.
Sixty percent of a number that's spread across eight people is very different from sixty percent of a number that's one person.
If you imagine the mother's sixty percent as a solid block of time where she is the only adult present, you've got the modern picture and you've misread the data. The sixty percent is interleaved. She's the primary, but she's primary inside a field of other people, and the handoffs are constant and small. A minute here, five minutes there. The baby is never parked with one person and left.
So the mother's share is a share of attention, not a share of custody.
That's exactly the right way to put it. And then there's the crying data, which is the most striking thing in the paper. Of about two hundred and twenty crying bouts that didn't resolve on their own quickly, half were answered within ten seconds. Ninety percent within twenty-five seconds.
Twenty-five seconds.
And it's not always the mother answering. The children had two to three other people responding to their crying, on top of her.
Hold on. Say that last part again.
Two to three other people, beyond the mother, would respond when the child cried.
So a baby cries and there are four people moving.
Four people moving. And the Cambridge press release has the Botswana comparison, the !Kung. Infants aged ten to twenty weeks are in physical contact with someone about ninety percent of daylight hours. Almost a hundred percent of crying bouts are responded to, almost always with comforting or nursing. And scolding is extremely rare.
Ninety percent of daylight hours in contact. I want to sit with that for a second.
Whereas an infant in a crib in a room by themselves is in contact with someone close to zero percent of the time, and they're alone in the room.
I was in contact with a branch for about eleven hours a day and I turned out fine.
You were in contact with a branch and you never learned to walk upright, so I'm going to reserve judgment on the causal claim.
Upright is overrated. Get to the hands-on care, because I want to know what "attentive care" actually means in the coding.
Bathing, feeding, kissing, hugging, stroking. About eight people other than the mother provided that kind of regular hands-on care. And the children had roughly fifteen to twenty caregivers total over the study period, but only about eight of those were doing regular hands-on work. That distinction matters and it comes back later.
Total network of fifteen to twenty, core of about eight.
And then the ratios, which is where the policy argument lives.
Give me the ratios.
Caregiver-to-child ratios in the observed hunter-gatherer groups were greater than five to one, and can exceed ten to one. Now put that against UK nursery regulations: one carer to three children under two, one carer to four children aged two to three.
And that's the regulated floor, not the actual staffing.
It's the floor. And here's the bit that made me sit up. The UK recently relaxed the ratio for two-year-olds to one caregiver for five children. Which is the opposite direction from what the researchers observed in the field.
So the policy lever that the data point at is the one being turned the other way.
Chaudhary himself says it. He explicitly suggests raising nursery staff-to-child ratios toward hunter-gatherer levels.
That's a concrete, arguable claim from the researcher. Not "we should be nicer to mothers." A ratio.
And it's the kind of thing that costs money, which is why it goes nowhere. But let me add the constraint that the same paper flags, because it cuts against the naive version of this. The children may be adapted to a stable core of a handful of key carers within a larger network. So consistency matters, not just headcount.
Which is bad news for a model that just hires more bodies.
It kills the "staff up with agency temps" version of the idea. If a child has fifteen people rotating through and none of them is the same person twice, you've increased the headcount and destroyed the thing the headcount was supposed to do.
The child isn't counting heads. The child is tracking faces.
Exactly that. And there's a developmental reason for it, which is that an infant's attachment system is built to discriminate. It's not built to be soothed by whoever is nearest. It's built to recognize the specific people who have been reliable. So a room with a high adult-to-child ratio and total staff turnover is, from the infant's point of view, a room full of strangers. The ratio on the spreadsheet looks great. The experience is the opposite of the forager pattern.
Which means the ratio number and the consistency number have to be read together.
They have to be read together. If you raise ratios without stabilizing the staff, you've spent the money and missed the point. Now, the other thing I want to put on the record is that children in these societies are not just recipients of care. They're providers. In some communities, children begin providing some childcare from age four, and they're capable of sensitive caregiving. Pre-adolescents commonly look after infants. From about age two, hunter-gatherer children spend large parts of the day in mixed-age playgroups, ages two to sixteen, without adult supervision.
Two to sixteen, unsupervised.
For large parts of the day. And compare that to the UK's NSPCC recommendation that babysitters be at least in their late teens.
So our regulatory floor for a babysitter is fifteen years older than the age at which a forager child is doing the job competently.
Roughly, yes.
I'm not saying we should hand a two-year-old a baby.
Nobody's saying that.
I'm saying the gap between "four-year-olds can do sensitive care" and "we don't let anyone under seventeen watch a child" is a gap in the culture, not in the child.
And the cross-cultural data back up that the subadults are the big contributors. There's a 2024 study in Evolutionary Human Sciences, Chaudhary, Page, Salali, Migliano, that used wearable "motes" to build close-proximity networks for forty-nine children under four across Agta camps in the Philippines and BaYaka camps in Congo. The most involved people in both populations were unrelated campmates, siblings and subadults. And the aggregate allomaternal contributions came out: subadults fifty-five percent, unrelated campmates thirty-three percent, siblings about twenty-nine and a half, fathers eleven, grandmothers nine.
Fathers at eleven and grandmothers at nine is the surprise for me.
And that's the finding the authors highlight. Fathers' and grandmothers' involvement was the most variable between populations. So the people Western policy obsesses over, dad and grandma, are the two most inconsistent sources of care cross-culturally. The consistent ones are the unrelated campmates and the kids.
Which is a very different picture from the mom and dad and a nursery room model.
The Agta data specifically: alloparents provided nearly three quarters of care for babies under two, and nearly eighty percent of care for children aged two to six.
So past infancy it's overwhelmingly not the mother.
Past infancy it's mostly not the mother. And then there's Efe, in the DRC. Infants had fourteen alloparents a day by eighteen weeks old and were passed between caregivers about eight times an hour.
Eight times an hour.
Konner's chapter has the range as five to twenty-four different people, mean of about fourteen point two. Non-mothers accounted for thirty-nine percent of physical contact at three weeks and sixty percent at eighteen weeks.
So the mother's share of contact drops as the child gets older.
And the Aka, in the Central African Republic, are the extreme version. One-to-four-month-olds held by mothers less than forty percent of the time in camp, transferred to other caregivers an average of seven point three times per hour, with seven different caregivers on an average day.
Seven different caregivers a day as the norm.
Konner's summary line for the !Kung is the counterweight, though. !Kung mothers still accounted for seventy-five to eighty percent of all physical contact. Nonmaternal caregivers responded to about half of infant crying bouts, often in tandem with mothers.
So the picture is not distributed care replacing the mother. It's the mother plus a crowd.
It's not the mother handing off and leaving. It's the mother and two other people all responding at once.
That's the thing I keep coming back to, and I want to address the obvious objection before someone in the audience makes it: that this is a story about mothers being less attached.
Which the data directly contradict, and I'm glad you said it. Mothers are the largest single provider in every one of these datasets. In the Mbendjele study, sixty percent of close care. For the !Kung, seventy-five to eighty percent of physical contact.
The mother is not reduced. She's relieved.
She's relieved of the piece that was never supposed to be hers alone. And the alloparenting paper's framing is that the intensive mothering narrative, one adult meeting every need, can lead to maternal exhaustion and has dangerous consequences. That's Chaudhary and Swanepoel. Not "mothers are doing too little." "Mothers are being asked to do more than one person can."
And that gets us to work. Because the other half of the objection is "sure, but someone had to forage."
And that's the study I actually think is the centerpiece for Hannah's question. Visine and colleagues, Scientific Reports, October 2024. They tracked three hundred and forty-eight foraging trips by twenty-two BaYaka mothers in Congo, with GPS and heart-rate monitors.
Heart rate on a foraging trip.
To measure energy expenditure. So they can ask whether taking the baby along costs the mother anything in calories or in food brought back.
And the answer?
Mothers took infants along on fifty-nine point two percent of trips. They foraged alone only eleven point eight percent of the time, with a mean group size of about three point two.
So nine out of ten trips had someone else there.
Nine out of ten. Mothers reported receiving direct childcare from foraging-group members during sixty-nine percent of trips while collecting food, and thirty-one percent while walking. And when they foraged with husbands and infants, husbands provided direct childcare in ninety-seven percent of trips.
Ninety-seven.
So the father isn't a helper on an occasional basis. When he's on the trip, he's doing the care almost every time.
And the children.
Here's the one I want to land. Children in early or middle childhood provided direct childcare in sixty-three percent of cases. Eighty-five of one hundred and thirty-four caregivers reported.
The children are the caregivers in two thirds of trips.
And it changes the economics in a way that surprised me. The number of children in the foraging group increased the mothers' net food returns. More adults in the group decreased them.
Wait, more adults decreased returns.
Possible food competition. If you add another full-grown forager to the group, they're eating from the same patch. If you add a child, you get a pair of hands that helps with the baby and doesn't eat like an adult.
So the kids are net positive to the household economy.
On these measures, yes. And crucially: mothers maintained their energy expenditure and their food returns with or without infants. Taking the baby along didn't cost them.
That's the finding that dismantles the whole framing of Hannah's question, though.
How?
Because the modern debate treats the baby as a cost to productivity. The BaYaka data say the baby wasn't a cost. The group absorbed it and the mother's returns held steady. The baby is only a cost in a system where the mother is the only available pair of hands.
And that's the deep version of the point. There was no binary in these societies. Mothers never fully stopped working, and never fully separated. They interleaved.
If I'm designing a modern daycare around this, the first thing I have to admit is that I'm not copying a childcare arrangement. I'm copying a work arrangement.
Start with the physical structure, because that's the easiest and it's also the most revealing. Daycare in or near the workplace, so a parent can come in and out during the day. Drop-off and pickup flexible rather than fixed.
Because the fixed time is the abrupt part. Nine a.m. and five thirty are walls.
They're walls. And nothing in the forager data is a wall. It's continuous. You hand the baby to the next person and keep doing what you were doing.
The design principle is permeability.
Permeability of the boundary between work and care. And notice how badly the modern office fails at that. You can't bring a baby into an office even though most of what people do in an office is sit at a screen and type.
I want the ratio piece next, because that's the policy lever.
A hunter-gatherer-inspired daycare would have more adults per child than a UK nursery, but it would also have more children per adult in mixed-age groups, with older children playing a genuine caregiving role. So it's not just "hire more staff." It's "change the age mix."
Which is where the regulatory wall shows up.
The regulatory wall is real, and it's the reason nobody has built this. Mixed-age grouping is a licensing problem. Older children as caregivers is a liability nightmare. Flexible drop-off is a staffing problem because you can't predict headcount. Employer-sponsored on-site daycare is the closest thing that exists.
Name them.
KinderCare At Work, Bright Horizons, Vivvi, Albany Med Kids. And in India, corporate crèches like Feathertouch and Petals. These are co-located with work.
That's the thing they get right, co-location.
It's the one design principle from the list they actually implement. Parent can walk over at lunch. That's real.
The thing they get wrong is everything else on the list. They're age-segregated by regulatory requirement, staffed by rotating agency workers, structured around fixed drop-off and pickup. High turnover, which is exactly what the consistency finding rules out.
The turnover is not a small detail, it's the central failure. The average tenure of a childcare worker in the US is measured in months, not years. So the child who needs a stable core of a handful of familiar people gets a rotating cast instead, and the ratio on paper says there are plenty of adults, and the child's experience says there are no familiar ones.
The co-location is a real win and it's sitting on top of a structure that undoes it.
It's a real win sitting on top of a structure that undoes it. And I want to be honest about the limits of what I know here. I have not seen a single pilot, anywhere, that tries the whole thing. Not "alloparental daycare" as a brand, not a research nursery built to forager specification, not an experimental site.
Then let me ask the question the episode is really about. Why not?
Three constraints, and I think they're stacked.
Go.
Regulatory, first. Licensing ratios, staff-child ratios, background checks, the prohibition on mixed-age grouping for infants, liability for older children as caregivers. Every one of those is a law, and the law encodes the assumption that a small number of trained adults is the right structure. The forager pattern is many untrained adults and older children, and that is illegal in most jurisdictions. You cannot get a license for it.
Economic, second.
Economic is that co-location means the employer pays for or subsidizes childcare, and most employers have no reason to. A warehouse, a hospital, a restaurant — the marginal value of keeping a specific worker is low enough that the cost of the on-site nursery doesn't pencil out.
Except where it does. Hospitals and universities.
Hospitals and universities, which is where the existing on-site daycare clusters. Not a coincidence. Those are the employers where the workers are hard to replace and the workers happen to be disproportionately women of childbearing age, so the retention math changes.
And the third constraint.
The cultural one, and it's the deepest. The modern framework treats maternal employment and infant care as competing goods. Either mother stays home and the baby gets optimal care, or mother works and the baby is in daycare and we measure the damage. Every study, every op-ed, every policy proposal sits inside that frame. The forager data don't fit inside the frame because there was no trade-off to manage. She worked and the baby was with her, because the work was done in a group with three other people and a four-year-old.
The frame is the thing. The nursery isn't the bug, the binary is the bug.
The downstream effects of breaking the binary are interesting. If daycare is co-located and permeable, the "eight or ten hours of separation" figure collapses to a series of brief absences. If older children are caregivers in mixed-age groups, the childcare workforce problem gets partially solved by the children themselves, which is a sentence I know sounds unhinged to a modern licensing officer.
Say it anyway, it's true.
If the mother is not the sole care provider, then the mother's career isn't penalized in the way it currently is, because the career wasn't interrupted for eighteen months. Which is the gender-equity version of the argument, and it cuts differently from how people expect, because the policy some people reach for, long and fully paid leave, is actually a version of the same binary. It says: stop working completely, then resume. The forager pattern says: never stop, never fully leave.
Salali's line about her own baby is the ground truth for that. The BaYaka mothers have so much support. She had a baby and it was lonely.
Pasco Fearon at Cambridge, the developmental psychologist: it's almost certainly true that we weren't really adapted to raise children as a single parent or just as two-parent families.
Not "we've lost something nice." Not adapted. As in, the setup doesn't fit the organism.
Emmott again: the laws don't reflect it and the child-care system available doesn't reflect it, and many women are put in an impossible situation and set up to fail.
The design question is answered, in a way, by the fact that it hasn't been answered. We know the ratios, we know the consistency requirement, we know the age mix, we know the permeability. Every one of those is a checkbox you could write down. And nobody has. Which means the obstacle isn't that we don't know what to build. It's that the frame won't let us build it.
I want to add one caution, because the data deserve it. The Cambridge group is explicit that contemporary hunter-gatherers must not be treated as living fossils. The forager societies being observed are contemporary people with their own histories, not time capsules.
Which matters for what we're allowed to conclude.
It matters because the claim is not "copy them." It's "ask whether our design principles reduce the mismatch." And the answer to whether some of these structures would help doesn't depend on whether the observed societies are pristine. It depends on whether the human infant has expectations, and the responsiveness data say it does.
The caveat on the sample.
The Mbendjele study is eighteen children, and Kathryn Humphreys at Vanderbilt says it's hard to draw generalizations for the full community.
Small sample. Keep the numbers honest and keep the direction.
Keep the direction. The direction is the same across the BaYaka, the Agta, the Efe, the Aka, and the !Kung, from five different research groups. That's not one study. That's a research program.
Where does that leave someone actually living this? Because the data are useful and the design is blocked and the audience has a baby right now.
It leaves the individual version. The alloparenting finding doesn't require a licensed facility. It requires a handful of consistent people who know the child. The consistency research says one reliable grandmother beats five rotating helpers.
A small stable network beats a big transient one.
The design finding says proximity is not separation. A parent who works from home and hands the baby across the room every forty minutes is running a version of this.
It's a degraded version.
Of course. It's a degraded version. But the direction is right, and the direction is the only thing a person can move on their own.
Let's get Hilbert's take. He's been turning something over the whole time.
