Abstract
Attentional flexibility is a foundational component of human cognition and develops substantially throughout childhood. Although the neural activity supporting attentional flexibility has been well-characterised in adults, its developmental trajectory during childhood and adolescence remains poorly understood. Here, we used magnetoencephalography (MEG) to study brain network dynamics underlying attentional set-shifting. MEG data were collected from 63 participants aged 4 - 19 years, including 30 with a diagnosis of attention-deficit/hyperactivity disorder (ADHD). We applied dynamic network modes (DyNeMo), a multivariate generative modelling approach that infers a dynamic mixture of spectrally resolved "modes" that generate the observed data. We related the resulting mode dynamics to task condition (non-shift, implicit shift, explicit shift), reaction time (RT), accuracy, age and ADHD diagnosis. Greater attention shifting demands were associated with increased modulation of frontal theta and posterior alpha modes. Transient lapses in attention, indexed by slower RTs or incorrect responses, were likewise characterised by heightened modulation of these modes across all task conditions. Developmentally, older children showed higher overall activity in the frontal theta mode, and stronger post-stimulus modulation of both frontal theta and posterior alpha modes. No significant effects of ADHD diagnosis were observed on either behavioural performance or mode dynamics. Together, these findings provide mechanistic insight into large-scale network dynamics that support the maturation of attentional flexibility, and point to network signatures associated with momentary lapses in attention.