Interview with Joëlle Provasi and Bérengère Guillery : Part 1

Hello Joëlle Provasi and Bérengère Guillery

Joëlle Provasi, you are a senior lecturer at EPHE-PSL. Your research focuses on the rhythmic synchronisations of very young children in relation to their environment. You are particularly interested in the temporal dimension of actions and interactions in the development of infants and young children.

Bérengère Guillery, you are Director of Studies at EPHE-PSL and you work on memory and transmission in children.

Joëlle Provasi, you have studied the development of babies during the neonatal period. How is the memory process set up, given that the child has to assimilate a phenomenal amount of information and yet retains no long-term memory of this period?

I’m a teacher-researcher at the Ecole Pratique des Hautes Etudes in the CHART laboratory, which stands for “Laboratoire de Cognition Humaine et Artificielle. In this context, I’m interested in cognitive activities at the very beginning of life.

If we want to understand the genesis of cognition in infants, we need to take into account the foetal period and the many learning processes that already take place during this period. Research into foetal cognition is fairly recent, having begun in the 1980s, but we now know with certainty that the foetal sensory systems are functional at least during the last trimester of pregnancy. At this stage, the foetus feels tactile, kinaesthetic, vestibular, olfactory, gustatory and auditory stimuli, but above all it processes this sensory information, memorises it and discriminates between it and new sensory stimuli.

In 1980, De Casper and Fifer demonstrated, using an operant conditioning method usually used in animals, that a 3-day-old newborn baby could discriminate between the voice of its mother and the voice of another woman, and was capable of modifying its non-nutritive sucking behaviour to hear its mother’s voice in preference to the voice of another woman. Following this discovery, a number of other studies have highlighted the learning abilities of newborn babies. Newborns are able to discriminate between their mother tongue (heard in utero) and a foreign language if the latter has a different rhythmic class (Nazzi, Bertoncini, Mehler, 1998). This ability to discriminate is already present in utero between English and Mandarin for foetuses whose mothers are Anglo-Saxon (Kisilevsky et al., 2009). The prosody of language, i.e. accent, tone, intonation, articulation, pause, rhythm, tempo and flow, are the characteristics of language that best pass through the intrauterine wall and are therefore best preserved in utero. These characteristics will continue to be present ex-utero and this is what researchers will call transnatal continuity, which will enable the newborn child to rediscover the characteristics of stimuli already learned in utero. The many things the foetus learns in utero will serve as a basis for learning after birth. If the foetus is already able, in utero, to discriminate its mother’s voice from another woman’s voice (Kisilevsky & Hains, 2011), this ability will persist after birth even though the mother’s voice no longer has exactly the same acoustic properties in utero and ex utero. The aquatic context will have changed to an aerial one. Nevertheless, the newborn will recognise its mother’s voice and prefer it to that of another woman. This fact highlights another very important cognitive ability of the newborn, what researchers call categorisation, the mother’s voice, whether it is broadcast in an aquatic or aerial environment, of low or high amplitude, hoarse, clear, depressed or joyful, will always be recognised and discriminated by the newborn because, since it can be highly variable, it will always be part of the “mother’s voice” category.

The same applies to many other learning processes, such as olfaction and taste. After birth, newborns will recognise odours that they have learnt in utero, and they will prefer these odours, showing that they have discriminated between them and odours not learnt in utero. Very recently, researchers have shown (Ustun-Elavan et al., 2025) that while a kale capsule could trigger a crying facies observed on ultrasound in foetuses at 32 and 36 weeks’ gestation, repeated exposure to kale capsules during the last three weeks of gestation triggered laughter facies in the newborn between 2 and 4 weeks of life, indicating significant acceptance of this odour learned in utero. Benoist Schaal and his team also demonstrated this transnatal continuity between amniotic fluid, colostrum and breast milk. Here again, despite variations in the taste and smell of liquids depending on what the mother has eaten, the foetus and then the newborn will be able to categorise these smells as “mother smells” and will preferentially turn to these smells, which they will distinguish from the smells of other women.

The hypothesis concerning transnatal continuity would therefore be that neonatal acquisitions would contribute to a more rapid stabilisation of postnatal acquisitions through relearning. Many maternal stimuli continue to be present and stable ex utero. They could be the source of reactivation and consolidation of knowledge initiated during prenatal life.

Joëlle Provasi, you talk about transnatal continuity, yet we have no memory of anything before the age of 3 and very little up to the age of six. Bérengère Guillery: How do you explain this?

Once you’re an adult, it’s virtually impossible to remember events in your life that happened before the age of 3. Try, for example, to remember your first memory, avoiding events that are often mentioned in the family or elsewhere. How old were you at the time? Now try to recall experiences you had a little later, between the ages of 3 and 6: your memories are few and often charged with a particular emotional connotation.

All these memories linked to ourselves and which cause us to experience emotions, thoughts, reactions and reflections, form autobiographical memory. It is made up of episodic memory, memories of events that happened only once, with details, perceptions and emotions. It also includes memories of events that are general because they are repeated – my holiday by the sea every summer – or decontextualised by the loss of specific details – knowing that you went hiking in the mountains on your last holiday without being able to describe it in detail. Finally, it incorporates all the personal knowledge we accumulate throughout our lives, such as the names of our friends, the towns we have visited or the dates of birth of our nearest and dearest.

So why can’t we remember the events of our early years? According to Freud, this phenomenon, which he called “infantile amnesia”, lasts until the age of 6 or 8 and is due to the fact that, once we are adults, we repress these memories. Since then, a number of neuropsychological studies have been carried out, showing that the first memories emerge from the age of 2 to 3 and that autobiographical memory is really operational by the age of 5, although it must be admitted that there is considerable individual variability. These studies have identified several factors at the root of childhood amnesia: neurobiological, cognitive, language and perceptual.

The first, neurodevelopmental hypothesis, is based on the fact that the brain regions essential to the formation of autobiographical memory, and in particular the prefrontal regions, do not mature sufficiently during the first years of life to organise and maintain traces of past experiences. The very young child would therefore not be capable of encoding and restoring the memory and its source (where and when the event took place). The prefrontal regions undergo slow maturation until the end of adolescence. This late maturation could also explain why the younger the child, the more likely it is to forget. The first studies to reveal this phenomenon were carried out in rats in 1962 by Byron Campbell and Enid Hobart Campbell of Princeton University and Trenton State College. These authors carried out fear conditioning (delivery of an electric shock when the animal passed through a part of the cage) on several hundred rats aged between 18 and 100 days, then tested whether this conditioning was maintained after 7, 21 and 42 days. They observed that after 42 days, only the oldest rats continued to avoid the area where they had received the electric shocks. In the younger rats, the conditioning had been erased from their memory. This result has since been extended to several species, including humans. The more immature the brain, the more fragile the links between the different elements that make up the memory (place, date, context, etc.), which would explain why it is more easily forgotten.

Other hypotheses from the cognitive sciences can also be put forward. The first essential factor in the development of autobiographical memory is language. The appearance of conversational exchanges around the age of 2 marks an important stage in the development of social communication and memory. Storytelling or “narrative memory” enables social interaction, plays a part in the construction of identity and provides a cognitive architecture for recalling events in adulthood. Before the emergence of language, children’s experiences would not be accessible to their consciousness as mental representations. As soon as language is sufficiently developed, children can share their memories with others, enabling them to be consolidated in the long term and retrieved at a later date. For example, when parents talk to their children in the evening about the school day, they usually start by suggesting a temporal structure by talking about the morning’s activities, then the meal and finally the afternoon. Conversational exchanges therefore enable memories to be organised according to a certain chronology, with the action beginning and then ending, and to be associated with details or impressions. They provide a structure that will be used throughout life.

This greater forgetfulness in very young children also reflects another biological phenomenon, neurogenesis, i.e. the permanent birth of new neurons in the brain. In most regions of the brain, neurogenesis is largely complete at birth, except in two regions, including the hippocampus, the key region for memory, where production peaks during the first few months of life and then declines until adulthood. This excessive neurogenesis, described in 2012 by Sheena Josselyn and Paul Frankland, modifies the architecture of neuronal networks and disrupts the integrity of previously formed synapses, thus preventing the stabilisation of memory traces. As new neurons proliferate, memories of previous events would be attenuated or even erased. It would therefore be necessary to wait for the level of neurogenesis to decrease before stable memories could be formed. Another hypothesis is based on the activity of certain neurotransmitters, such as gamma-aminobutyric acid (GABA), whose role is to reduce the activity of neurons. Excessive production of GABA, such as that observed in the first years of life, inhibits the recall of memories during this period, thus preventing the long-term anchoring of memories.

Moreover, recalling autobiographical events involves mentally travelling back in time to the moment of experience in order to reconstruct all the details, perceptions and sensations of the moment. This mental journey through time is based on “autonotic consciousness”, defined as self-awareness through time. The “me” of today incorporates my personality, my life goals, my physical appearance and current perceptions, my knowledge and experience linked to my age and my professional and family situation. It also takes into account my impressions, thoughts and psychological state at the moment. However, this “me” of today is linked to the “me” of yesterday, even if I have changed and gained a few centimetres and no longer have the same interests as I had as a child. So this ‘me’ at the origin of our identity is extremely dynamic and evolves throughout our lives. At the same time, this autonotic awareness ensures a certain continuity of our identity over time. It allows us to project ourselves into the future, to anticipate future events and our needs.

Self-awareness is said to be immature in very young children, which explains why they are unable to organise and reconstruct their autobiographical memories. Self-awareness develops gradually: small children do not recognise themselves in a mirror until they are around 18 months old, and cannot describe themselves physically until they are around 4 years old, and a little later for character traits. The same applies to the Theory of Mind (ToM), i.e. the ability to attribute and understand one’s own mental states and those of others: sufficient development of ToM between the ages of 4 and 6 is necessary to remember a memory as personally experienced and not just to retain a feeling of familiarity.

Lastly, recalling an event involves relying on cues that provide access to the memory as a whole. Consequently, the greater the similarity between encoding and retrieval, the easier it is to access the memory. This was conceptualised in the 1970s by Endel Tulving under the term encoding specificity. But when the encoding goes back to childhood, the recovery cues that enable the adult to bring the memory back to consciousness are more limited, because a child’s reality differs from that of an adult: perception is different. Similarly, an adult’s interests are different from those of a child. The cues used by the adult to retrieve the memory will therefore not be the same as those used to encode the memory in childhood, making it more difficult to recall.

So both the formation of a memory and its retrieval are complex phenomena that go beyond the question of perception!

End of part 1. Discover part 2 in clicking here.