When we tell children stories about seven-headed dragons, frogs that turn into princes or giant trees that are ladders to the sky, their imagination brings the fiction to life. Even though they have never seen magical creatures or transformations before, following stories like these is not a struggle for them, quite the opposite: it is a source of joy and wonder. In other words, when children encounter new combinations of concepts that they already know, they can readily integrate the separate elements and picture the new unit they form. This ability gives them endless possibilities to invent novel combinations of concepts: it is the foundation of humans’ unparalleled achievements.  

Despite its importance, we don’t yet know when and how this impressive combinatorial skill emerges. New research from cognitive scientists Barbara Pomiechowska, Ágnes Melinda Kovács, Gábor Bródy and Ernő Téglás reveals that we have to go back to the beginning of language acquisition to solve this puzzle: their results show that one year old infants already understand expressions combining known concepts with new ones they acquired during the study.  

 The research performed at Central European University used a baby-friendly task to tackle infants’ combinatorial skills. All the babies had to do was to watch a few simple videos – but in order to make sense of them, their minds had to work hard the whole time. Their first task was to learn two novel labels for “one” and “two”. The researchers helped them achieve this by showing them familiar objects, such as cars or apples, in sets of one and two. The babies saw a hand pointing at each set while a voice named them (e.g. “Look, mize car!” for one car and “Look, padu car!” for two cars). After just a few presentations with a handful of familiar items, babies deciphered what “mize” and “padu” mean. When they were shown two new sets of objects each made of the same familiar kind, dogs for example, they reliably looked at the set with one item when asked about “mize dog”, and the set with two dogs upon hearing the phrase “padu dog” even though the hand stopped pointing to the correct referent.

 This word learning task simply ensured that the researchers could then test infants’ ability to combine concepts in real time rather than to recall combinations that they already know from experience. These two processes can be hard to disentangle. When we ask a young child to show us the brown bear in a picture book, they can succeed even if they don’t know what “brown” means: by consistently hearing this combination of words whenever seeing a brown bear, they might have learned the phrase as a unit. But babies in this study only ever heard the words mize and padu coupled with a few specific familiar nouns (such as car and apple), and any other combination would be completely new to them. To make sense of such novel combinations, they would have to recruit their knowledge about both concepts and apply them together – and this is exactly what their task was in the second phase of the study. Although learning two new abstract words in such a short time is an impressive feat from one-year olds in and of itself, the next study challenged them further. 

 Babies had to find a match for new, complex phrases such as “padu duck” (referring to two ducks) while presented with four different options. One of these would only match the category (one duck), one of them only the quantity (two balls), potentially distracting infants from the third candidate, the correct match (two ducks). The fourth candidate was a “mixed” set (a duck paired with a different item, say, a ball).

If babies didn’t integrate the concepts of “two” and “duck” but processed them both separately, they might think this set also checks all the boxes: it has the set size of two and has a duck in it. The results suggest otherwise: after hearing the question “Where is padu duck?”, babies turned their gaze towards the set of two ducks. This tells us that they combined the two concepts in an adult-like manner as opposed to treating them as two separate search criteria: they seem to have understood that “padu” should behave like “two” and apply to a homogenous set of items, specified by the second word (i.e. duck). 

 With this study, researchers Barbara Pomiechowska, Ágnes Melinda Kovács, Gábor Bródy and Ernő Téglás demonstrate that combinatorial thought is already present at one year of age, expanding babies’ world beyond their past experience. While in infancy, combining concepts most likely facilitates learning from and communicating with others, later in life, it allows us to move past everything that’s already been thought of, enriching the human mind with endless possibilities.

‍

Original paper: Pomiechowska, B., Bródy, G., Téglás, E., & Kovács, Á. M. (2024b). Early-emerging combinatorial thought: Human infants flexibly combine kind and quantity concepts. Proceedings of the National Academy of Sciences of the United States of America, 121(29). https://doi.org/10.1073/pnas.2315149121