Posted on: 22 February 2019
In his inaugural lecture as the Thomas Mitchell Professor of Cognitive Neuroscience, Professor Rhodri Cusack, addressed the important question – Why are human infants so slow to develop?
The lecture, titled ‘Uncovering the Hidden Foundations of Cognition in Infancy with Neuroimaging and Artificial Intelligence’ looked at how human infants are born helpless and their behaviour remains primitive for a year. They do not crawl until nine months or speak their first words until they are one year old. This stands in stark contrast to the early mature behaviours of many animals such as lambs, which can walk and act immediately after birth. So why are human infants so slow to develop?
One theory is that they are born early with immature brains because their heads must be small to fit through the mother’s pelvis, which is in turn constrained by her need to walk efficiently. To test this theory, Professor Cusack recently used magnetic resonance imaging (MRI) to characterise the infant brain. Surprisingly, he and his team found that many systems were mature from birth, which is inconsistent with the immature brain theory. Why, therefore, are human infants helpless? Might it be instead that they need some period of “hidden learning” before they can begin to act? To address this, he and his team examined the relationship across species between brain complexity and the duration of the helpless period.
They found that for both birds and mammals, smarter animals are helpless for longer during infancy. This led them to propose that there may be something about learning in complex brains that mandates a slow initial period of development – in other words, we must “lie down before we can walk”. But what might this be? Here, they found inspiration from artificial intelligence. In the last decade deep neural networks, which were inspired by the architecture of the brain, have allowed machines to perform complex tasks such as driving at a similar level of performance to humans. And, it turns out, that deep neural networks also have a helpless period of “infancy”, which is longer in duration in smarter networks.
“Put simply, the cause for this helpless period is to allow the network to develop good representations of its problem space before it tries to solve any problems. Going forwards, with support from the European Research Council and Science Foundation Ireland, my team at Trinity is aiming to bring together computational models from artificial intelligence with neuroimaging in infants, to uncover the hidden foundations of cognition in infancy,” explained Professor Cusack.
ABOUT RHODRI CUSACK
Rhodri Cusack is the Thomas Mitchell Professor of Cognitive Neuroscience, a Fellow of Trinity College Dublin, and a Principal Investigator of the SFI Infant Centre. After reading Natural Sciences (specialising in physics) at Pembroke College, Cambridge he received a PhD in psychology from the University of Birmingham. He was then a postdoctoral fellow and subsequently group leader at the Medical Research Council Cognition and Brain Sciences Unit in Cambridge, and then an Associate Professor at the Brain and Mind Institute of the University of Western Ontario in Canada.
