Combined functional and causal connectivity analyses of language networks in children: A feasibility study

Instead of assessing activation in distinct brain regions, approaches to investigating the networks underlying distinct brain functions have come into the focus of neuroscience research. Here, we provide a completely data-driven framework for assessing functional and causal connectivity in functional magnetic resonance imaging (fMRI) data, employing Granger’s causality. We investigate the networks underlying story processing in 17 healthy children (8f, 9m, 10.4 ± 2.8 years, 6.5–15.4 years). Extensive functional connectivity exists between brain regions, including some not detected in standard random effects analyses. Causal connectivity analyses demonstrate a clear dominance of left-sided language regions for both forward and backward interactions with other network nodes. We believe our approach to be useful in helping to assess language networks in the normal or pathological setting; it may also aid in providing better starting estimates for the more hypothesis-driven approaches like structural equation or dynamic causal modeling.