A generative network model of neurodevelopment

Abstract

The emergence of large-scale brain networks, and their continual refinement, represent crucial developmental processes that can drive individual differences in cognition and which are associated with multiple neurodevelopmental conditions. But how does this organization arise, and what mechanisms govern the diversity of these developmental processes? There are many existing descriptive theories, but to date none are computationally formalized. We provide a mathematical framework that specifies the growth of a brain network over developmental time. Within this framework macroscopic brain organization, complete with spatial embedding of its organization, is an emergent property of a generative wiring equation that optimizes its connectivity by renegotiating its biological costs and topological values continuously over development. The rules that govern these iterative wiring properties are controlled by a set of tightly framed parameters, with subtle differences in these parameters steering network growth towards different neurodiverse outcomes. Regional expression of genes associated with the developmental simulations converge on biological processes and cellular components predominantly involved in synaptic signaling, neuronal projection, catabolic intracellular processes and protein transport. Together, this provides a unifying computational framework for conceptualizing the mechanisms and diversity of childhood brain development, capable of integrating different levels of analysis - from genes to cognition.

Bridging Brain and Cognition: A Multilayer Network Analysis of Brain Structural Covariance and General Intelligence in a Developmental Sample of Struggling Learners

“…Furthermore, results from different levels can more easily be interpreted if these datasets are analyzed and interpreted using a unified quantitative and conceptual framework, such as network science. Last, and perhaps most important, cognitive neuroscientists must formulate mechanistic (e.g., Bertolero et al 2018 ) and generative models (for instance, Akarca et al 2020 ) to gain further insights from the past and help guide future controlled experiments.…”

“… Bertolero et al ( 2018 ) found that a mechanistic model assuming that “connector hubs” (diverse club nodes, see Bertolero et al 2017 ), which regulate the activity of their neighboring communities to be more modular but maintain the capability of “task-appropriate information integration across communities”, significantly predicted higher cognitive performance on various tasks, including language and working memory. Furthermore, in the same sample studied here, Akarca et al ( 2020 ) applied a generative network modeling approach to simulate the growth of brain network connectomes, finding that it is possible to simulate structural networks with statistical properties mirroring the spatial embedding of those observed. The parameters of these generative models were shown to correlate with neuroimaging measures not used to train the models (including grey matter measures), cognitive performance (including vocabulary and mathematics), and relate to gene expression in the cortex.…”

Spatial and temporal autocorrelation weave complexity in brain networks

“…Brain networks share many properties with other complex biological and physical systems, so they can be analyzed using similar graph-theoretic methods. Recently, generative graph models have been used in neuroscience to model the human connectome [3, 104, 105], opening up new avenues of analysis of both functional [13, 14] and structural [15, 108, 109, 111, 112] connectivity data. One reason why generative graph models are particularly attractive in the field is that they appear, at first glance, to model the trade-off between wiring cost and topological efficiency in brain networks [113].…”

Bridging brain and cognition: A multilayer network analysis of brain structural covariance and general intelligence in a developmental sample of struggling learners

“…Furthermore, results from different levels can more easily be interpreted if these datasets are analyzed using a unified quantitative framework that combines strengths from various statistical techniques (such as pairwise and partial correlations to reveal causality in brain functional connectivity networks, see Reid et al 2019). Last, and perhaps most important, cognitive neuroscientists must formulate mechanistic (e.g., Bertolero et al 2018) and generative models (for instance, Akarca et al 2020) to gain further insights from past and help guide future controlled experiments. Researchers must not shy away from but rather embrace the complexity of the brain and cognition (see Fried and Robinaugh 2020 for similar argument for mental health research).…”

“…Bertolero et al 2018 found that a mechanistic model assuming that “connector hubs” (diverse club nodes, see Bertolero, Yeo, and D’Esposito 2017), which regulate the activity of their neighboring communities to be more modular but maintain the capability of “task appropriate information integration across communities”, significantly predicted higher cognitive performance on various tasks including language and working memory. Furthermore, in the same sample studied here, Akarca et al 2020 applied a generative network modelling approach to simulate the growth of brain network connectomes, finding that it is possible to simulate structural networks with statistical properties mirroring the spatially embedding of those observed. The parameters of these generative models were shown to correlate with neuroimaging measures not used to train the models (including grey matter measures), cognitive performance (including vocabulary and mathematics) and relate to gene expression in the cortex.…”