Skip to main content

Bayesian Learning in the Infant Brain

Periodic Reporting for period 1 - BabyBayes (Bayesian Learning in the Infant Brain)

Reporting period: 2019-05-01 to 2020-10-31

The unique cognitive abilities observed in humans do not suddenly emerge in adulthood, but are rather deeply rooted in early infancy. After centuries of considering infants as essentially defective adults, research in cognitive development is now of central interest for not only understanding our unique human abilities, but also developing efficient machine learning algorithms, as well as devising scientifically informed educational programs. Over the past decades, research in developmental psychology has repeatedly evidenced infants’ amazing learning abilities, dispelling the original vision of a neurologically insufficient and cognitively confused infant. However, developmental psychology still lacks an algorithmic understanding of early learning mechanisms, integrating cognitive computations with brain processes. Crucially, this knowledge gap has been holding the field back from having the resources to design theory-driven screening procedures with young infants in order to identify early markers of atypical development, before these deficits lead to profound delays. The development of non-invasive brain imaging techniques has recently opened the black-box of the infant brain, and, we are now at a turning point where we can move from the description of what infants can learn to the exploration of how the typically developing brain implements its learning strategies. The present project aims at bringing together developmental psychology, neuroscience and computational modelling to unravel the brain mechanisms underlying typically developing infants’ learning skills. To this end, during the implementation of this project, I learnt a novel neuroimaging technique compatible with infant research, and implemented a novel research paradigm to address the localization and dynamics of neural computations in the developing brain.
This first reporting period covers the first work package (WP) of this project. The objective of this WP was to capitalize on the expertise of the outgoing host to acquire transferable knowledge with a novel and emerging neuroimaging technique compatible with infant research: functional near-infrared spectroscopy (fNIRS), and use this technique to investigate the presence of anticipatory predictive signals in the infant brain as a viable neural learning strategy. The project also involved the adaptation of this experimental protocol to a more classic brain imaging technique: EEG, in order to investigate the dynamics of prediction signals in the infant brain. The project started at the beginning of summer 2019, in the outgoing institution at Yale university.
Over the course of this first WP, I was able to conceptualize and design a novel experimental paradigm to investigate predictive processing in the infant brain using fNIRS. I successfully implemented this experimental paradigm to test infant participants. In parallel, I also developed and implemented the EEG counterpart of this study.
I participated in setting up a fully functional environment to welcome families and test infant participants. Unfortunately, I had to face unforeseen hardware issues preventing us from testing infants with fNIRS. In December 2019, the lab was able to resume infant testing with the assistance of the company providing the neuroimaging fNIRS hardware system. In the meantime, I was able to get further hands-on training with this technique, by running an alternative fNIRS experiment with adult participants.
I was also confronted with insufficient numbers of recruited infants, delaying the start of the testing period. In order to overcome this, the lab hired two full-time research assistants, launched a website and increased presence on social media, and local institutions.
I was able to start piloting the fNIRS study in late December 2019, and I started collecting the actual experimental data in February 2020. I collected data from 15 infants up to mid-March 2020. In parallel, I started piloting the EEG experiment with infants at the beginning of March 2020. I was able to collect pilot data from three babies up to mid-Mard 2020. Unfortunately, due to COVID restrictions, the lab stopped infant testing in the middle of March 2020.
I conducted preliminary data analysis on the small fNIRS database but the small sample size did not allow me to perform statistical analyses. Instead, I focused on the adult fNIRS dataset in order to perfect my data analysis skills and train on novel methods. A manuscript is currently under preparation.
Overall, I was able to develop an innovative experimental paradigm to investigate predictive processing in the infant brain using fNIRS. The unforeseen COVID crisis significantly impacted the progress of the first work package. With the progress of the vaccination campaign, I hope to resume testing in a relatively near future. This experiment is expected to provide crucial insights in our understanding of infant learning abilities. This type of results has the potential of revealing the neural mechanisms underlying typical development, and therefore hold exciting promises for the early detection of atypical development.
The adult dataset acquired instead of the planned infant study aimed at investigating the brain networks involved in face and speech processing using fNIRS. Importantly our results show that speech and face specific responses can be recorded using fNIRS. The technique and/or the proposed experimental design however did not allow us to pinpoint interactions between the two processing streams. These results document audiovisual processing using fNIRS.
fNIRS experimental design