Awake infant functional MRI for visual neuroscience: reflections from the Foundations of Cognition Project

Introduction Awake infant functional MRI (fMRI) has considerable potential to provide a window into the function of the developing brain. In visual neuroscience, NeuroAI approaches combining fMRI with computational models, such as deep neural networks, offer powerful tools for characterizing representations along the ventral visual stream (VVS), a hierarchy of cortical regions that subserve object recognition in adults. Extending these approaches to infancy could provide insights into the trajectories and mechanisms of visual development and learning. A prerequisite, however, is the acquisition of suitable data. Through the Foundations of Cognition Project, we therefore investigated the feasibility of awake infant fMRI for characterizing developing visual representations at scale. Here, we reflect on how effectively this goal was achieved, reporting acquisition rates, data quality metrics, and practical insights from the design process that may inform future infant fMRI research. Methods We developed two passive viewing paradigms to balance competing objectives: capturing and maintaining infant attention, eliciting broad VVS responses, and limiting the number of experimental conditions to ensure sufficient data per condition given variable participation times across infants. The first comprised six naturalistic videos, chosen to maintain attention and encourage stillness. The second comprised 36 object images, which were expected to be less engaging but, if acquired, would enable fine-grained characterization of VVS representations and direct comparison with computational models. Results We acquired awake fMRI from 130 2-month-old and 65 9-month-old infants. Nearly all 2-month-olds (97%) participated in an awake scan, contributing a median of 18.63 min. At 9 months, 63% participated, with a median duration of 13.02 min. Although 9-month-olds contributed fewer minutes, they moved less during scanning, so fewer runs were excluded due to motion. Both age groups attended to the stimuli, although 2-month-olds were interested for a greater proportion of time (median 75.27%) than 9-month-olds (48.52%). Both paradigms elicited reliable stimulus-driven responses across the developing brain, particularly within the VVS, and noise-ceiling analyses showed increasingly robust group-level representational patterns with larger samples. Discussion Despite practical challenges, we show that awake infant fMRI is a viable approach for characterizing visual representations at scale, providing a foundation for computational and NeuroAI studies of visual development.

Authors

Institutions

Publication Details

Journal
Frontiers in Neuroscience
Published
2026-11-20
DOI
https://doi.org/10.3389/fnins.2026.1852835
Citations
1
Primary Topic
Visual perception and processing mechanisms
Type
article
Field-Weighted Citation Impact
1.69

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Awake infant functional MRI for visual neuroscience: reflections from the Foundations of Cognition Project

Chiara Caldinelli, Adrienne Foran, Josué Rico‐Picó, Tamrin Holloway et al.
1 citations
Frontiers in Neuroscience
Visual perception and processing mechanisms
1.69
article

Awake infant functional MRI for visual neuroscience: reflections from the Foundations of Cognition Project

Chiara Caldinelli, Adrienne Foran, Josué Rico‐Picó, Tamrin Holloway, Cliona O'Doherty, Anna Truzzi, Deirdre O’Hara, Amy Burke, Lorijn Zaadnoordijk, Aoibhe De Burca, Alex R. Wade, Enna-Louise D’Arcy, Áine T Dineen, Anna Kravchenko, Anisha Wadhwa, Graham King, Jessica White, Claire Ambre, Kevin Campion, Katie Herbert, Ailbhe Tarrant, Oscar Markey, Maebh Healy, Cameron Murphy, Eleanor J. Molloy, Rhodri Cusack, Angela T. Byrne
article en
1 citations

Abstract

Introduction Awake infant functional MRI (fMRI) has considerable potential to provide a window into the function of the developing brain. In visual neuroscience, NeuroAI approaches combining fMRI with computational models, such as deep neural networks, offer powerful tools for characterizing representations along the ventral visual stream (VVS), a hierarchy of cortical regions that subserve object recognition in adults. Extending these approaches to infancy could provide insights into the trajectories and mechanisms of visual development and learning. A prerequisite, however, is the acquisition of suitable data. Through the Foundations of Cognition Project, we therefore investigated the feasibility of awake infant fMRI for characterizing developing visual representations at scale. Here, we reflect on how effectively this goal was achieved, reporting acquisition rates, data quality metrics, and practical insights from the design process that may inform future infant fMRI research. Methods We developed two passive viewing paradigms to balance competing objectives: capturing and maintaining infant attention, eliciting broad VVS responses, and limiting the number of experimental conditions to ensure sufficient data per condition given variable participation times across infants. The first comprised six naturalistic videos, chosen to maintain attention and encourage stillness. The second comprised 36 object images, which were expected to be less engaging but, if acquired, would enable fine-grained characterization of VVS representations and direct comparison with computational models. Results We acquired awake fMRI from 130 2-month-old and 65 9-month-old infants. Nearly all 2-month-olds (97%) participated in an awake scan, contributing a median of 18.63 min. At 9 months, 63% participated, with a median duration of 13.02 min. Although 9-month-olds contributed fewer minutes, they moved less during scanning, so fewer runs were excluded due to motion. Both age groups attended to the stimuli, although 2-month-olds were interested for a greater proportion of time (median 75.27%) than 9-month-olds (48.52%). Both paradigms elicited reliable stimulus-driven responses across the developing brain, particularly within the VVS, and noise-ceiling analyses showed increasingly robust group-level representational patterns with larger samples. Discussion Despite practical challenges, we show that awake infant fMRI is a viable approach for characterizing visual representations at scale, providing a foundation for computational and NeuroAI studies of visual development.

Frontiers in NeuroscienceVol. 20
Queen's University Belfast (GB), Universitat Pompeu Fabra (ES), Universidad de Granada (ES), Trinity College Dublin (IE), Rotunda Hospital (IE), Children's Health Ireland at Crumlin (IE), Coombe Women & Infants University Hospital (IE), St. James's Hospital (IE), Centro de Investigación Mente, Cerebro y Comportamiento (ES), University of York (GB), Columbia University (US), Stanford University (US)
Marie Curie, European Research Council, Research Ireland
Good health and well-being
Openalex Percentile: Top 24%
Visual perception and processing mechanisms
1.69
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon