What do babies look at?

Babies and toddlers learn incredibly quickly. Any parent will know that just a few weeks can make a huge difference to their development. Language is especially exciting, but equally challenging: how do you study how language begins when the language learners don’t speak or gesture yet?

Luckily, years of developmental research have provided solutions to this puzzle. We now know that babies and toddlers look for longer at things they find interesting. We can use this “novelty preference” to tell us about what they’re learning. To do this, we show them pictures on a computer screen record where they look, and for how long. The most recent research uses eye-trackers, which measure not just for how long babies look at something, but where they look. So, eye-trackers can tell us in the finest detail what babies and toddlers are using to learn from.

In these studies, children either sit on their parent’s lap or in a high-chair in front of a computer screen. The eye-tracker is a small box that sits underneath the screen. Because the pupils of the eye are darker than the surrounding iris, the eye-tracker can compute where children are looking and for how long. These studies last for a few minutes at most, and allow us to explore what babies have learned from what we show them, and what they find interesting. 

Eye-tracking has revolutionised research in child development. For example, this work from the University of Pennsylvania showed that babies as young as six months old already know what some common words mean. Our recent work shows that knowing what something is called affects how ten-month-old babies interact with it – even before they can talk. Research centres all over the world now use eye-tracking to understand what babies and toddlers are thinking, which in turn helps us develop new ways of supporting learning. We’ve also shown that two-year-old children learn words for new things when they see those things in different contexts, suggesting that giving children a range of learning opportunities is important for early language acquisition.

This research in language learning has been carried out by Shirly Ma (Leverhulme Trust Doctoral Scholar, Lancaster University), Charlotte Rothwell (BSc Psychology student, Lancaster University), and Elle Hart (BSc Psychology graduate, Lancaster University).

What do toddlers play with?

Recording where children look in this way has told us a lot about how what they see determines what they learn. But learning in the real world is different: once they learn how to reach and grasp, children have control over what they are learning from. They can move objects closer, turn them around for a different view, hold more than one object at once – and in doing so, they construct their own learning environment. However, as yet we don’t know what children learn best from: for example, when you let a toddler play freely, will s/he prefer to learn from challenging toys, or simple ones? Because almost all of children’s day-to-day learning happens through play, it’s important for us to understand what happens during this curiosity-based exploration, and how language affects it.

We use new technology to conduct some of the first studies into toddler’s real-world, curiosity-based learning. We give toddlers specially-designed, 3D printed toys and let them play. First, we first record what they play with and for how long, to explore whether they pick toys up in a particular order; for example, do they start off with the simple toys then work their way up to the complex ones? At the same time, we use head-mounted eye-trackers to shed light on exactly what toddlers see during play. This exciting new technology consists of a swimming cap-type hat with two small cameras that record the child’s eye and their visual scene. Linking these two video feeds allows us to calculate exactly what toddlers are looking at, giving us a detailed picture of what toddlers choose to learn from.

This research in curiosity-based learning was (or is being!) carried out by Han Ke (Postdoctoral Research Associate, Lancaster University), Marina Loucaides (Leverhulme Trust Doctoral Scholar, Lancaster University) and Ben Malem (BSc Psychology graduate, Lancaster University).

What do children need to know in order to learn?

There are many competing theories as to how much knowledge children are born with. Learning language is extremely challenging; it’s a complex and infinitely creative system, and (perhaps luckily for parents!) children’ don’t simply learn it from directly teaching from adults. Some argue that from birth children have rich, complex knowledge of language, grammar and/or concepts which allow them to make sense of the world and learn language fast. Others argue that children are born with much simpler, general abilities to learn to associate things in the world – such as sounds and objects – but that these learning abilities are astonishingly rapid and powerful.

There is still no consensus, and recently researchers have used computer and robotic simulations to test learning both with and without complex, pre-existing knowledge. In collaboration with colleagues at the Plymouth University, UK, we have used the iCub robot to show that a relatively simple learning system with no built-in information about the world or language can learn to recognise objects and learn words for those objects, reflecting the results of a recent study with two-year-old children. Importantly, this work suggests that  young children can depend on simple but effective mechanisms to learn about the world, rather than depending on detailed inborn knowledge.

This research was carried out in collaboration with Dr Anthony Morse and Professor Angelo Cangelosi from the Cognitive Robotics and Neural Systems group at Plymouth University, UK.