Low weight at birth is associated with all sorts of health troubles later in life, so it seems a great idea to give nutritional supplements to pregnant women in developing nations, to add some heft to their babies. Yet the results aren't impressive. (The anthropologist Christopher W. Kuzawa notes, for instance, that this review of 13 such programs found the average weight improvement for the babies was a paltry one ounce.) Which illustrates the state of work on "fetal origins"—the theory that pre-birth experiences in the womb have a powerful effect decades later on the adult mind and body. On the one hand, as Annie Murphy Paul writes in Origins, her fine new book about the field, the idea suggests that we should ensure that developing fetuses have a healthy environment. On the other hand, the work so far can't say how to do that.

The other day, at this conference I heard Kuzawa propose an explanation for some of this befuddlement: We can "tell" the baby-to-be that food is abundant by making sure its mother ate well last month; but its development may depend instead on how that mother ate through her entire life.

A keystone of much "fetal origins" work is that the developing infant responds to cues about the kind of world it will have to join. Is food scarce or plentiful? Is life anxiety-filled or calm? Do people live to be 90 around here, or drop by 62? The mother's experience of life, translated into hormones, blood sugar, blood pressure and other physiological signals, helps determine which of the developing baby's genes are activated and how much, molding it to fit its future environment.

New parents easily imagine this fetus as a clueless investor, reacting every hour to the latest news flash. (Hence their neurotic fears about that one sip of alcohol, bite of sushi, or late-night fight that will ruin the budding child's life.) Instead, Kuzawa proposes, we should see the developing baby as a long-term player, looking for clues about its world on different time scales: months, years or even generations.

That perspective could provide a framework to organize and explain disparate pieces of data that Murphy Paul mentions. For instance, some research has found that a mother's malnourishment late in pregnancy puts her child at higher risk, decades later, for diabetes, while malnourishment in early gestation is a risk for heart disease. And the stresses of the Arab-Israeli war in 1967 seem to increase the risk of schizophrenia in adults who were gestated then—if their mothers were in the second month of pregnancy, but not the fourth or fifth. All this suggests there are distinct periods in a pregnancy, each particularly sensitive to one environmental influence, but not others.

Kuzawa proposes to use evolutionary reasoning to find the hidden logic of these different windows. Doing so, he says, will probably require figuring out the time-frame of the each cue that an embryo uses to prepare itself for its world.

Which brings us back to nutritional supplements: If fetuses were "tuned" to today's cues about their environment, then they should respond to maternal nutritional supplements much more than they do. But it stands to reason, Kuzawa said, that a creature that will live for 30, 40 or even 90 years should not prepare itself for the environment of next month. Next month may be way out of line with typical conditions. To prepare for eating in the mother's world, the fetus ought to find out what her whole life was like.

Some eerie facts of fetal-growth research seem to line up with the idea. Here are a couple Kuzawa cited. In this study of mothers and children in 1930's England, a woman's adult height was not a great predictor of her daughter's birth weight. Much better was the mother's height back when she herself was 7. And this study in Guatemala found that infants who grew faster in their first three years of life were those whose mothers had eaten better as children.

Kuzawa thinks our species may have evolved a mechanism for "telling" the fetus what to expect on average over decades or even generations. By basing its development on its mother's childhood condition (which means, Kuzawa points out, that it's reflecting its grandmother's experience, and so on backwards in time), the fetus takes a long-term average of conditions in its world. It can't be "fooled" by one rich harvest or a plague year. That makes sense from an evolutionary point of view. But it also means the fetus can't be "fooled" by our well-meaning attempts to guide it for the few months of its mother's pregnancy.

How does this square with evidence that disasters and wars do have a big and immediate effect on the children gestated during the crisis? As Murphy Paul recounts, these "natural experiments" suggest that bad experiences have immediate effects. (Douglas Almond has found (pdf) that children whose mothers were pregnant during the 1918 flu pandemic were 15 percent more likely than near-peers to drop out of high school; they earned lower wages throughout life, and as older adults were 20 percent more likely to be disabled.)

Perhaps severe shocks overwhelm the usual pathways by which environment communicates with embryo, Kuzawa suggests. If that's so, then extreme cases like the 1918 flu pandemic or the Dutch Hunger Winter are a mixed blessing for fetal-origins research. On the one hand, they show dramatic effects which helped convince skeptics; but, on the other, they may not represent the way the system usually works.

For parents and policy wonks, too, Kuzawa's idea is a mix of good news and bad news. If the developing fetus is impervious to day-to-day ups and downs, our minor scrapes and flubs can't harm it. But that also means that our well-intentioned efforts won't help it much, either. At least, not until we take its own long view of life.

Kuzawa, C. (2005). Fetal origins of developmental plasticity: Are fetal cues reliable predictors of future nutritional environments? American Journal of Human Biology, 17 (1), 5-21 DOI: 10.1002/ajhb.20091

Martin RM, Smith GD, Frankel S, & Gunnell D (2004). Parents' growth in childhood and the birth weight of their offspring. Epidemiology (Cambridge, Mass.), 15 (3), 308-16 PMID: 15097011

Stein AD, Barnhart HX, Wang M, Hoshen MB, Ologoudou K, Ramakrishnan U, Grajeda R, Ramirez-Zea M, & Martorell R (2004). Comparison of linear growth patterns in the first three years of life across two generations in Guatemala. Pediatrics, 113 (3 Pt 1) PMID: 14993588