Be careful what you say around a pregnant woman ‒ there's a good chance her unborn child is listening in on the conversation.
In a 1988 issue of The Lancet, you'll find a short research letter published under the odd title 'Fetal soap addiction.' Peter Hepper, a psychologist at Queens University Belfast, had studied babies whose mothers were fans of a particular Australian soap opera. He wrote:
"The response to the theme tune of the programme Neighbours was studied 4-5 days after birth in infants whose mothers had watched it daily during pregnancy ("soap" group, n = 7) and in those whose mothers had not watched it (control, n = 8). The theme tune was played to the infants individually and their behaviour observed."
The results of this experiment were striking. The children of Neighbours fans would inexplicably become calm upon hearing the show's upbeat musical theme, whereas control babies would not. It didn't seem to be anything about the music itself; instead, it was almost as though the Neighbours babies remembered all those weeknights at 6:30pm when a familiar song would waft through the womb.
For Hepper, these findings were not a great surprise. In his previous animal experiments, offspring could recognise their siblings even if they had been separated at birth, a behaviour he attributed to prenatal learning. Extending this research to humans was a natural step. "I was working with newborn human recognition at the time," says Hepper, "and simply planned and did the Neighbours study to see if human newborns had the ability to recognise tunes heard before birth."
That humans might hear and learn sounds in utero isn't as fantastical as it seems. After all, the growing fetus is constantly immersed in audio, from the continuous thump of the mother's heartbeat to the rasp of her expanding and contracting lungs.
"A few intrepid women in labour have allowed obstetrician researchers to listen to intrauterine sounds by inserting a tiny hydrophone through the cervix," says Christine Moon, a developmental psychologist at Pacific Lutheran University and University of Washington. "On some recordings, the heartbeat sound is loud, and on others it is hardly audible... Also audible are low frequency rumbles of the digestive system, called borborygmi."
Sounds from the outside world also find their way in. The most frequent and prominent voice is the mother's, albeit very muffled. The tissues and fluids of the womb filter out higher pitched noises, but sounds below 300 Hz are audible. Vowels penetrate better than consonants, and male voices do better than female ones.
"You can discern who is talking and the voice's loudness, rhythm, and melody," says Moon, "but it is hard to make out the words. In intelligibility tests of speech on the recordings, about 30-40% of the individual vowels and consonants are recognisable."
So sound is certainly present in the womb, but can fetuses hear it? And if so, can they learn anything from it?
This question goes back a long way. English philosopher John Locke, in An Essay Concerning Human Understanding (1689), hinted that humans might begin learning inside the womb, describing newborns as blank slates, "[except] perhaps for some faint ideas of hunger, and thirst, and warmth, and some pains, which they may have felt in the womb."
But it wasn't until the early 20th century that scientists first got involved. Chinese experimental psychologist Zing-Yang Kuo disputed the popular idea that animals had natural instinct, instead arguing that seemingly 'innate' behaviours were learnt. "Habits," he wrote in 1921, "begin to be formed at birth, or even in the embryo."
Kuo began testing this idea in chicken embryos in the 1930s. He would carefully cut away the shell and outer membranes of a chicken egg, and then with a Chinese writing brush would gently daub the inner membrane with melted vaseline. This turned the membrane transparent, allowing Kuo to study the behaviour and movement of a living embryo as it grew. He found that, even before hatching, chicks were responsive to external stimuli, including touch, light, electricity and sound.
Around this time, other scientists were beginning to apply these ideas to humans. A series of experiments by Albrecht Peiper, Wilbert Ray, and others showed that human fetuses will kick and increase their heart rates when startled by a sudden loud noise. Without a doubt, fetuses could hear sounds from the outside world. But evidence for learning was harder to pin down.
One solution was to test newborn babies for traces of fetal memory. Beginning in the late 1970s, Anthony DeCasper and colleagues pioneered a clever method for testing voice preferences in human newborns. Babies were placed on their backs, and sucked on a non-nutritive nipple while listening to audio through a pair of headphones. Based on how frequently they sucked, babies could choose which of two audio recordings they heard.
"Newborns don’t have a lot of motor control, but they can control sucking to some extent," says Moon. "With hearing, we give them a pacifier that is connected to a computer that measures their sucking pressure, and the computer is programmed to deliver sound into headphones based on sucking pressure and its patterns.
"After a few minutes of experience with the sucking = sound connection, even the youngest newborns learn to change sucking patterns to activate some sounds more than others and to keep sucking to prolong them."
Armed with this experimental setup, it turns out that newborn babies have a surprising capacity for voice discrimination, and will change their sucking rate to prolong familiar sounds. For instance, babies much prefer to hear their mother's voice over that of an unfamiliar female, and their mother's language over a foreign one.
"A unique feature of the mother's voice [in the womb] is that it is accompanied by synchronous motion," says Moon. "When we talk, our abdomen and chest moves, and we shift our bodies and make gestures with our heads, arms and hands. In this way, the maternal voice is different from all others.
"Her voice is also unique because it is the one that is most often available. Throughout the third trimester... every single time a pregnant woman talks, her voice is available to be heard whether she is talking to her unborn child or not. The sheer amount of exposure to mother’s voice as well as the synchronised motion may contribute to its special status after birth."
Although these newborn findings were intriguing, they didn't necessarily prove that embryos were learning voices in utero. These experiments were conducted several hours or, more typically, several days after birth, so learning during the brief postnatal period could not be completely ruled out. For direct evidence, we needed to test the fetus itself.
In a 1994 study, DeCasper and colleagues instructed pregnant women to recite one of two children's rhymes (La Poulette or Le Petit Crapaud) daily, between weeks 33 and 37 of pregnancy. At week 37, fetuses heard a recorded version of both rhymes, spoken by an unrelated woman. The results were impressive: fetuses showed a lower heartbeat only during the familiar rhyme. This suggests that their brains were recognising the previously-heard rhymes and responding to them in a distinct way.
Today, there is consensus among researchers that individuals do, indeed, learn prior to birth. This holds true not only for humans, but also a host of other species, great and small. "We have directly tested representatives from all vertebrate groups ‒ fish, amphibians, reptiles, birds and mammals ‒ and in all have found evidence of prenatal learning," says Hepper.
This learning can involve olfactory, visual or auditory stimuli. For instance, rats exposed to apple juice in the womb will later show a preference for apple juice as pups. Cuttlefish that can see crabs through their transparent egg envelopes prior to hatching prefer to eat crabs over shrimp. In one Australian songbird, the superb fairy-wren (Malurus cyaneus), embryos in the egg learn elements from their mother's incubation calls and later repurpose them as begging calls, allowing their parents to distinguish fairy-wren nestlings from imposter cuckoos.
From these three examples, it's easy to see that prenatal learning might have different functions across species. In some cases, there might even be no function at all ‒ early learning could simply be a side-effect from the maturation of the brain and central nervous system.
"Fundamentally, the underlying aim of prenatal learning is to ensure survival, and this may be by assisting in learning about safe foods or recognising the prime caregiver who will keep the individual safe," says Hepper. "This seems to apply across all vertebrates. For humans, there is the added function of beginning to acquire language abilities."
A baby's first exposure to language occurs inside the womb. True, they can only hear a subset of the syllables spoken around them, but this early exposure may give vital clues about the identity of their tutors and to which vocal cues they should be paying attention. This could offer a head-start in the complex business of language acquisition.
Not surprisingly, research in the last few decades has prompted a surge in so-called 'prenatal education' products like Bellybuds. These are flat speakers that attach to the mother's belly and play sound directly to the womb. Such products are claimed to increase parent‒child bonding and kickstart the baby's language development, but there's little convincing evidence that they work. Prenatal learning programs like the Eastman Project and Prenatal University often provide extra stimulation to the baby both before and after birth, so their findings are difficult to interpret.
On this matter, Moon is unequivocal: "There are no well-controlled studies of children who have received extra stimulation only during the prenatal period and who have been tested later for the effects."
On top of this, she says, there may be very real dangers in exposing fetuses to artificial noise. For one, it's impossible to predict how different frequencies will be dampened or intensified inside the womb, and unintentionally excessive sounds can damage the developing fetal ear and brain. Another concern is that artificial audio may disrupt the fetus' natural sleep‒wake patterns, especially in the final months of pregnancy. We simply don't know what long-term consequences this might have for a developing fetus.
In a 2000 review of prenatal learning, Moon and her colleague William Fifer encouraged parents to speak and sing to their unborn children, but warned against pumping additional audio into the womb. She stands by this warning today.
"Some of the claims of people who market prenatal stimulation programs may be correct," she says, "but they may also be tragically incorrect. Why take the chance with your baby’s development?"
Edited by Ellie Michaelides, and supported by Eilidh Southren.