Brains are often described as cost-efficient communication networks that optimally balance long-connection costs against fast communication. Inspired by the “use it or lose it” principle, we present a game-theoretic model of self-organizing neural units showing the brain is suboptimal in both regards. Regional competition for connectivity under propagative dynamics yields networks resembling the human cortex yet more efficient and economical. In addition, using a reservoir computing framework, we find comparable information processing capacity, but synthetic optimal communication networks show lower computational reliability. Last, virtual lesions reveal why these networks are fragile: To optimize communication, they funnel information through a spatially clustered “oligarchy” of transmodal hubs. The human brain instead uses a distributed “rich-club” backbone that better resists targeted attacks, despite higher wiring costs and less efficient communication. Cortical networks thus trade both cost and efficiency for reliable computation, highlighting computational reliability as an overlooked and perhaps even more prominent driver of brain connectivity than wiring cost or communication efficiency.

The creation and curation of synaptic-level neuronal networks, or connectomes, enables the study of the relationship between neuronal structure and function. Topological characteristics of neuronal networks have been studied extensively. Separately, there have been considerable efforts to classify the morphology, cell types, and lineages of neurons. Here, we introduce a network metric that combines topological analysis with node metadata. This entropic quantity measures the diversity of incoming or outgoing connections to a node in terms of the metadata distribution. We find that in Caenorhabditis elegans, the top-scoring neurons (PVR, RMGL/R, DVA, CEP, ADE, URXR, RIGL, BAG, SMBDL) have known functions that integrate and disseminate multimodal information involved in sensorimotor functions. In the nerve cord of Drosophila melanogaster, we find that top-scoring neurons are embryonic neurons located in the abdominal neuropil, where sensorimotor coordination is required for complex innate behavior such as mating.

The unique and intricate pattern of human cortical folding is rooted in fetal neurodevelopmental processes and can now be comprehensively quantified by new neuroimaging-derived measures of sulcal complexity. Here, we provide the first genetic maps of human sulcal complexity. Beginning with large effects of rare variants, we survey nine different neurogenetic syndromes (n=615), detecting visible changes in sulcal complexity on a shared axis of sulcal change coupled to the prenatal timing of sulcation. Turning to common genetic variants, we use genome-wide association studies of complexity scores for 40 sulci in the UK Biobank (n~29,000) to (i) resolve variable heritability across sulci, (ii) reveal both local and remote shared genetic effects with cortical morphology, and (iii) identify complexity-associated genes and their embedding in brain maps of prenatal gene expression. These reference genetic maps uncover multiple new mechanistic pathways for cortical morphogenesis in health and disease.

Dysconnectivity in schizophrenia is a pervasive concept across various levels of systems biology. To better understand disrupted patterns of molecular connectivity in schizophrenia, we apply innovative approaches to gene co-expression networks starting from multiple regions of postmortem brain (bulk RNA-Seq from dorso-lateral prefrontal cortex- (DLPFC) (Ndonors=297; sex: M/F = 212/85), hippocampus (Ndonors=250; sex: M/F = 181/69) and caudate (Ndonors=349; sex: M/F = 242/107). Here we identify differentially connected genes (DCG) in schizophrenia networks that deviate from architectural relationships characteristic of neurotypical gene networks based on three network metrics- total connectivity (kTot), clustering coefficient (C), and intra-module degree (kIn). We find multiple DCG consistent across all brain regions, most of which we then independently confirm in brain single nuclei (snRNAseq) data and in four independent human iPSC-derived brain organoids. DCG specific for each network parameter shows enrichment in schizophrenia genetic signal, in pathways prominently related to neurons and oligodendroglia functionality, and in cell-type specific co-expression patterns that differ between neurotypical and schizophrenia, implicating neuronal-oligodendroglia incoordination.

Understanding how early life experiences shape brain network development is a key challenge in the neuroscience of mental health disorders. To address this, we used magnetic resonance imaging (MRI) similarity network analysis to study the effects of stress in the rat, an important animal model in neuropsychiatry. We measured magnetization transfer ratio (MTR) at each of 53 distinct cortical areas and estimated a cortical similarity network for each individual scan, in two independent experimental datasets. We first characterized normative network development in rats scanned repeatedly between postnatal days 20 (weanling) and 290 (mid-adulthood) (N=47), and then contrasted these findings with a cohort exposed to early life stress in the form of repeated maternal separation (RMS, N=40). The normative rat cortical similarity network exhibited biologically meaningful organization, aligning with prior cytoarchitectonic and tract-tracing data, and displayed complex topological features, including rich club organization. During postnatal and adolescent development, brain regions became more similar, including an early phase of fronto-hippocampal convergence. Early increases in inter-areal similarity were reversed in a later phase of fronto-hippocampal divergence in mid-adulthood. RMS exposure altered inter-areal similarity, especially between frontal and parahippocampal regions, that were also most active developmentally and in aging. Our results reveal how normative cortical network changes in the developing brain are influenced by early life stress. These findings suggest a new translational framework for elucidating how environmental risk factors lead to atypical development of cortical networks.

One avenue to better understand brain evolution is to map molecular patterns of evolutionary changes in neuronal cell types across entire nervous systems of distantly related species. Generating whole-animal single-cell transcriptomes of three nematode species from the Caenorhabditis genus, we observed a remarkable stability of neuronal-cell-type identities over more than 45 million years of evolution. Conserved patterns of combinatorial expression of homeodomain transcription factors are among the best classifiers of homologous neuron classes. Unexpectedly, we discover an extensive divergence in neuronal signaling pathways. Although identities of neurotransmitter-producing neurons (glutamate, acetylcholine, γ-aminobutyric acid [GABA], and several monoamines) remain stable, expression of ionotropic and metabotropic receptors for all these neurotransmitter systems shows substantial divergence, resulting in more than half of all neuron classes changing their capacity to be receptive to specific neurotransmitters. Neuropeptidergic signaling is also remarkably divergent, both at the level of neuropeptide expression and receptor expression, yet the overall dense network topology of the wireless neuropeptidergic connectome remains stable. Novel neuronal signaling pathways are suggested by our discovery of small secreted proteins that show no obvious hallmarks of conventional neuropeptides but show similar patterns of highly neuron-type-specific and highly evolvable expression profiles. In conclusion, by investigating the evolution of entire nervous systems at the resolution of single-neuron classes, we uncover patterns that may reflect basic principles governing evolutionary novelty in neuronal circuits.

Dysconnectivity in schizophrenia is a pervasive trait across various levels of systems biology. To better understand disrupted patterns of molecular connectivity distinguishing schizophrenia from control non-clinical populations, we applied novel approaches to gene co-expression networks in large samples of postmortem brains from multiple regions relevant to schizophrenia: the dorso-lateral prefrontal cortex- (DLPFC) (Ndonors=297), hippocampus (Ndonors=250) and caudate (Ndonors=349). We identified differentially connected genes (DCGs) in schizophrenia networks that deviated from architectural relationships characteristic of control gene co-expression networks, by assessing three network metrics - total connectivity (K), clustering coefficient (C), and intra-module degree (kIn) determined by projecting the modular community structure of the control networks onto the schizophrenia co-expression networks. Genes showing significant absolute case-control differences for these metrics (i.e., irrespective of difference directionality) were then tested for their relationships with common genetic variants conferring risk of schizophrenia and their biological significance through post-GWAS analyses (stratified LDSC and MAGMA), gene ontology annotations and enrichment in schizophrenia-relevant gene sets. We identified multiple DCGs, with case-control differences of connectivity metrics, consistent across brain regions. When parsed by parameter specificity, these genes show shared and specific enrichment in schizophrenia genetic signal, biological ontologies and selected cell-type markers. Notably these findings revealed widespread disturbances in co-expression connectivity affecting both neuronal and glial cells, particularly oligodendrocytes. Overall, our results highlight disrupted co-expression network architecture in schizophrenia, implicating disrupted neuronal-glial crosstalk and its effect on synaptic transmission. ### Competing Interest Statement The authors have declared no competing interest.

Many genes are linked to psychiatric disorders, but genome-wide association studies (GWAS) and differential gene expression (DGE) analyses in post-mortem brain tissue often implicate distinct gene sets. This disconnect impedes therapeutic development, which relies on integrating genetic and genomic insights. We address this issue using a novel multivariate technique that reduces DGE bias by leveraging gene co-expression networks and controlling for confounds such as drug exposure. Deep RNA sequencing was performed in bulk post-mortem sgACC from individuals with bipolar disorder (BD; N=35), major depression (MDD; N=51), schizophrenia (SCZ; N=44), and controls (N=55). Toxicology data dimensionality was reduced using multiple correspondence analysis; case-control gene expression was then analyzed using 1) traditional DGE and 2) group regularized canonical correlation analysis (GRCCA) - a multivariate regression method that accounts for feature interdependence. Gene set enrichment analyses compared results with established neuropsychiatric risk genes, gene ontology pathways, and cell type enrichments. GRCCA revealed a significant association with SCZ (Pperm =0.001; no significant BD or MDD association), and the resulting gene weight vector correlated with DGE SCZ-control t-statistics (R=0.53; P<0.05). Both methods indicated down-regulation of immune and microglial genes and upregulation of ion transport and excitatory neuron genes. However, GRCCA - at both the gene and transcript level - showed stronger enrichments (FDR<0.05). Notably, GRCCA results were enriched for SCZ GWAS-implicated genes (FDR<0.05), while DGE results were not. These findings identify a SCZ-specific sgACC gene expression pattern that highlights SCZ risk genes and implicates neuro-immune pathways, thus demonstrating the utility of multivariate approaches to integrate genetic and genomic signals.

Economic efficiency has been a popular explanation for how networks self-organize within the developing nervous system. However, the precise nature of the economic negotiations governing this putative organizational principle remains unclear. Here, we address this question further by combining large-scale electrophysiological recordings to characterize the functional connectivity of developing neuronal networks in vitro, with a generative modeling approach capable of simulating network formation. We find that the best fitting model uses a homophilic generative wiring principle in which neurons form connections to other neurons which are spatially proximal and have similar connectivity patterns to themselves. Homophilic generative models outperform more canonical models in which neurons wire depending upon their spatial proximity either alone or in combination with the extent of their local connectivity. This homophily-based mechanism for neuronal network emergence accounts for a wide range of observations that are described, but not sufficiently explained, by traditional analyses of network topology. Using rodent and human neuronal cultures, we show that homophilic generative mechanisms can accurately recapitulate the topology of emerging cellular functional connectivity, representing an important wiring principle and determining factor of neuronal network formation in vitro.

We developed a computational pipeline (now provided as a resource) for measuring morphological similarity between cortical surface sulci to construct a sulcal phenotype network (SPN) from each magnetic resonance imaging (MRI) scan in an adult cohort (n = 34,725; 45-82 years). Networks estimated from pairwise similarities of 40 sulci on 5 morphological metrics comprised two clusters of sulci, represented also by the bimodal distribution of sulci on a linear-to-complex dimension. Linear sulci were more heritable and typically located in unimodal cortex, and complex sulci were less heritable and typically located in heteromodal cortex. Aligning these results with an independent fetal brain MRI cohort (n = 228; 21-36 gestational weeks), we found that linear sulci formed earlier, and the earliest and latest-forming sulci had the least between-adult variation. Using high-resolution maps of cortical gene expression, we found that linear sulcation is mechanistically underpinned by trans-sulcal gene expression gradients enriched for developmental processes.

Childhood maltreatment (CM) leads to a lifelong susceptibility to mental ill-health which might be reflected by its effects on adult brain structure, perhaps indirectly mediated by its effects on adult metabolic, immune, and psychosocial systems. Indexing these systemic factors via body mass index (BMI), C-reactive protein (CRP), and rates of adult trauma (AT), respectively, we tested three hypotheses: (H1) CM has direct or indirect effects on adult trauma, BMI, and CRP; (H2) adult trauma, BMI, and CRP are all independently related to adult brain structure; and (H3) childhood maltreatment has indirect effects on adult brain structure mediated in parallel by BMI, CRP, and AT. Using path analysis and data from N = 116,887 participants in UK Biobank, we find that CM is related to greater BMI and AT levels, and that these two variables mediate CM’s effects on CRP [H1]. Regression analyses on the UKB MRI subsample ( N = 21,738) revealed that greater CRP and BMI were both independently related to a spatially convergent pattern of cortical effects (Spearman’s ρ = 0.87) characterized by fronto-occipital increases and temporo-parietal reductions in thickness. Subcortically, BMI was associated with greater volume, AT with lower volume and CPR with effects in both directions [H2]. Finally, path models indicated that CM has indirect effects in a subset of brain regions mediated through its direct effects on BMI and AT and indirect effects on CRP [H3]. Results provide evidence that childhood maltreatment can influence brain structure decades after exposure by increasing individual risk toward adult trauma, obesity, and inflammation.

Adolescent development of human brain structural and functional networks is increasingly recognized as fundamental to emergence of typical and atypical adult cognitive and emotional proodal magnetic resonance imaging (MRI) data collected from N [Formula: see text] 300 healthy adolescents (51%; female; 14 to 26 y) each scanned repeatedly in an accelerated longitudinal design, to provide an analyzable dataset of 469 structural scans and 448 functional MRI scans. We estimated the morphometric similarity between each possible pair of 358 cortical areas on a feature vector comprising six macro- and microstructural MRI metrics, resulting in a morphometric similarity network (MSN) for each scan. Over the course of adolescence, we found that morphometric similarity increased in paralimbic cortical areas, e.g., insula and cingulate cortex, but generally decreased in neocortical areas, and these results were replicated in an independent developmental MRI cohort (N [Formula: see text] 304). Increasing hubness of paralimbic nodes in MSNs was associated with increased strength of coupling between their morphometric similarity and functional connectivity. Decreasing hubness of neocortical nodes in MSNs was associated with reduced strength of structure-function coupling and increasingly diverse functional connections in the corresponding fMRI networks. Neocortical areas became more structurally differentiated and more functionally integrative in a metabolically expensive process linked to cortical thinning and myelination, whereas paralimbic areas specialized for affective and interoceptive functions became less differentiated, as hypothetically predicted by a developmental transition from periallocortical to proisocortical organization of the cortex. Cytoarchitectonically distinct zones of the human cortex undergo distinct neurodevelopmental programs during typical adolescence.

Structural similarity networks provide insights into neurodevelopment, but are primarily described in humans, thus limiting mechanistic insights. Rats are a valuable model due to their complex behaviors and biological similarities to humans. We present the first characterization of individual structural similarity networks in rats, elucidating developmental patterns and environmental susceptibility.

The connection patterns of neural circuits form a complex network. How signaling in these circuits manifests as complex cognition and adaptive behaviour remains the central question in neuroscience. Concomitant advances in connectomics and artificial intelligence open fundamentally new opportunities to understand how connection patterns shape computational capacity in biological brain networks. Reservoir computing is a versatile paradigm that uses high-dimensional, nonlinear dynamical systems to perform computations and approximate cognitive functions. Here we present conn2res: an open-source Python toolbox for implementing biological neural networks as artificial neural networks. conn2res is modular, allowing arbitrary network architecture and dynamics to be imposed. The toolbox allows researchers to input connectomes reconstructed using multiple techniques, from tract tracing to noninvasive diffusion imaging, and to impose multiple dynamical systems, from spiking neurons to memristive dynamics. The versatility of the conn2res toolbox allows us to ask new questions at the confluence of neuroscience and artificial intelligence. By reconceptualizing function as computation, conn2res sets the stage for a more mechanistic understanding of structure-function relationships in brain networks.

The cerebral cortex underlies many of our unique strengths and vulnerabilities - but efforts to understand human cortical organization are challenged by reliance on incompatible measurement methods at different spatial scales. Macroscale features such as cortical folding and functional activation are accessed through spatially dense neuroimaging maps, whereas microscale cellular and molecular features are typically measured with sparse postmortem sampling. Here, we integrate these distinct windows on brain organization by building upon existing postmortem data to impute, validate and analyze a library of spatially dense neuroimaging-like maps of human cortical gene expression. These maps allow spatially unbiased discovery of cortical zones with extreme transcriptional profiles or unusually rapid transcriptional change which index distinct microstructure and predict neuroimaging measures of cortical folding and functional activation. Modules of spatially coexpressed genes define a family of canonical expression maps that integrate diverse spatial scales and temporal epochs of human brain organization - ranging from protein-protein interactions to large-scale systems for cognitive processing. These module maps also parse neuropsychiatric risk genes into subsets which tag distinct cyto-laminar features and differentially predict the location of altered cortical anatomy and gene expression in patients. Taken together, the methods, resources and findings described here advance our understanding of human cortical organization and offer flexible bridges to connect scientific fields operating at different spatial scales of human brain research.

Gene expression in the human cortex is shown to exhibit a generalizable three-component architecture that reflects neuronal, metabolic, and immune programmes of healthy brain development. The three components have distinct associations with autism spectrum disorder and schizophrenia, revealing connections between previously unrelated results from studies of case-control neuroimaging, differential gene expression, and genetic risk.