As a species, humans are incredibly smart. We tell stories, create magnificent art and astounding technology, build cities, and explore space. We haven’t been around nearly as long as many other species, but in many respects we’ve accomplished more than any have before us. We eat them and they don’t eat us. We even run scientific studies on them—and are thinking about re-creating some of those that have gone extinct. But our intelligence comes with a curious caveat: our babies are among the dumbest—or, rather, the most helpless—that exist. A baby giraffe can stand within an hour of birth, and can even potentially flee predators on its first day of life. A monkey can grasp its mother and hang on for protection and nourishment. A human infant can’t even hold up its own head.

The evolution of human intelligence isn’t something that Celeste Kidd had ever pondered. A developmental cognitive scientist who currently works at the University of Rochester, her work had focussed mostly on learning and decision-making in children. Over years of observing young children, she became impressed with the average child’s level of sophistication. But when she looked at the infants she encountered, she saw a baffling degree of helplessness: How could they be so incompetent one second and so bright so soon thereafter? One day, she posed the question to her colleague Steven Piantadosi. “Both of us wondered what could possibly justify the degree of helplessness human infants exhibit,” she told me recently. “Even other primate babies, like baby chimps, which are close in evolutionary terms, can cling onto their moms.” She began to see a contradiction: humans are born quite helpless, far more so than any other primate, but, fairly early on, we start becoming quite smart, again far more so than any other primate. What if this weren’t a contradiction so much as a causal pathway?

That’s the argument that Kidd and Piantadosi make in their new paper, published in the June issue of PNAS. Humans become so intelligent because human infants are so incredibly helpless, they argue; the one necessitates the other. The theory is startling, but it isn’t entirely new. Researchers have been pondering the peculiarities of our birth and its evolutionary significance for quite some time. Humans belong to the subset of mammals, called viviparous mammals, that give live birth to their young. This means that infants must grow to a mature enough state inside the body to be born, but they can’t be so big that they are unable to come out. This leads to a trade-off: the more intelligent an animal is, the larger its head generally is, but the birth canal imposes an upper limit on just how large that head can be before it gets stuck. The brain, therefore, must keep maturing, and the head must continue growing, long after birth. The more intelligent an animal will eventually be, the more relatively immature its brain is at birth.

Researchers have long known about this trade-off, and about the connection between brain size and neural density and intelligence. For instance, Robin Dunbar found that the ratio of neocortical volume to brain size can predict the social-group size in a number of species, including bats, cetaceans, and primates, while Simon Reader has demonstrated links in tool use and innovation to brain size in primates. Kidd and Piantadosi’s new idea is that increased helplessness in newborns mandates increased intelligence in parents—and that a runaway selection dynamic can account for both. Natural selection favors humans with large brains, because those humans tend to be smarter. This may create evolutionary incentives for babies that are born at an even earlier developmental stage, which require more intelligence to raise. This creates the dynamic: over time, helpless babies make parents more intelligent, which makes babies more helpless, which makes their parents more intelligent, and so on.

During their investigation, Kidd and Piantadosi realized something important that strengthened their theory. It turns out that another variable has an even higher correlation with intelligence than brain size—time to maturity, or weaning time. In other words, the time it takes to shepherd newborns through absolute helplessness to a point of relative self-sufficiency predicts primate intelligence more strongly than the best measure that has previously been proposed, namely, head circumference. Orangutans have smarter babies than baboons and they wean them longer. Baboon babies, in turn, are weaned longer, and are smarter, than lemur babies.

Putting these facts together helped Kidd and Piantadosi develop their hypothesis. The connection between head size and intelligence does create incentives for babies to arrive earlier. But it’s the connection between weaning time and intelligence that may really be driving the cycle. You need to be smarter to care for more helpless creatures, which means you need a larger brain—which means that babies have to enter the world at an even more helpless stage of development, since there is a finite size to their brain at birth, mandated by the physiology of live birth. And so the cycle continues.

Of course, the theory is just that—a model. Ideally, to prove it you would look at head size, birth time, and intelligence over the span of human evolution, to see if we were born earlier as we got smarter—data that are unavailable. (Kidd also stresses repeatedly that this theory supplements, but does not supplant, earlier ones: it can coexist quite naturally with both the social-group account of intelligence—the Dunbar approach—and what’s called the metabolic accounts of intelligence, which posits that our digestive system has allowed for our brain’s high metabolic needs, and that we grew smarter to be able to find and share difficult-to-gather food.) But there are some intriguing converging pieces of evidence. For one, other animals that are not viviparous have not evolved the same levels of intelligence, suggesting an inherent link between live birth and brainpower. And in modern humans, a few pieces of evidence appear to suggest that smarter parents are more likely to have offspring that survive. In one limited sample—two hundred and twenty-two Serbian Roma women—maternal I.Q. and child mortality were negatively correlated (that is, higher I.Q. meant lower mortality), even controlling for education, age, and a number of other factors. In a larger sample of Californian parents, in 1978, years of education were linked to infant-mortality rates. Global epidemiological studies suggest a decrease in mortality that equals between seven and nine per cent for each year of a mother’s education. None of this is decisive, of course, but it is suggestive.

There is, of course, one follow-up question: Why did this cycle happen to humans and not to lemurs? When I asked Kidd about this, she told me that their theory cannot offer an answer—likely as not, it’s a matter of pure genetic luck that became self-reinforcing. As we grew smarter, we were better able to take care of our infants, so they could be born more helpless and allow us to grow even smarter.

One intriguing way to test the hypothesis further: look at twins. Typically, twins are not carried to full term. Does that mean that twins are more intelligent than non-twins, since they require longer weaning time—and would a mother who is a twin, in turn, give birth to smarter children? “It’s a really good question, to look at the types of twins that run in families and whether that leads to higher intelligence,” Kidd said. “It’s a predictable hypothesis,” and one that they may well include in their ongoing research on train heritability in twins. And it’s a hypothesis that came, of all places, from a five-year-old—Kidd was being interviewed about twins for a podcast by the daughter of Sindya Bhanoo, a former columnist at the Times. Dumb babies, maybe, but oh so smart, oh so soon thereafter.