Wiring cost and topological participation of the mouse brain connectome

Brain connectomes are topologically complex systems, anatomically embedded in 3D space. Anatomical conservation of “wiring cost” explains many but not all aspects of these networks. Here, we examined the relationship between topology and wiring cost in the mouse connectome by using data from 461 systematically acquired anterograde-tracer injections into the right cortical and subcortical regions of the mouse brain. We estimated brain-wide weights, distances, and wiring costs of axonal projections and performed a multiscale topological and spatial analysis of the resulting weighted and directed mouse brain connectome. Our analysis showed that the mouse connectome has small-world properties, a hierarchical modular structure, and greater-than-minimal wiring costs. High-participation hubs of this connectome mediated communication between functionally specialized and anatomically localized modules, had especially high wiring costs, and closely corresponded to regions of the default mode network. Analyses of independently acquired histological and gene-expression data showed that nodal participation colocalized with low neuronal density and high expression of genes enriched for cognition, learning and memory, and behavior. The mouse connectome contains high-participation hubs, which are not explained by wiring-cost minimization but instead reflect competitive selection pressures for integrated network topology as a basis for higher cognitive and behavioral functions.Network organization of the brain is fundamental to the emergence of complex neuronal dynamics, cognition, learning, and behavior. Modern concepts of anatomical network connectivity originated in the 19th and early 20th century with the ascendancy of the neuron theory: the concept of discrete nerve cells contiguously connected via axonal projections and synaptic junctions (1, 2). In the last decade, the connectome has emerged as a new word to define the complete structural “wiring diagram” of a nervous system or brain (3). At the small scale of synaptically connected neurons, the connectome has only been completely mapped for the 302-neuron nervous system of the roundworm Caenorhabditis elegans, using serial electron microscopy and painstaking visual synaptic reconstruction (4). At the large scale of axonally connected brain regions, draft connectomes have been mapped for the cat and macaque, by collation of primary tract-tracing studies (5–7), and for the human, using in vivo diffusion-weighted magnetic resonance imaging measures of white matter tract organization (8), or interregional covariation measures of cortical thickness or volume (9).Topological analyses of these connectomes have consistently demonstrated a repertoire of complex network properties, including the simultaneous presence of modules and hubs (10). The seemingly ubiquitous appearance of these topological features, e.g., both at the cellular scale of the worm brain and at the areal scale of the human brain, supports scale- and species- invariant organizational principles of nervous systems, consistent with Ramón y Cajal’s seminal “laws of conservation for time, space and material” (1, 11–13). Anatomically localized and functionally specialized modules conserve space and (biological) material by reducing the average length of axonal projections, or wiring cost; anatomically distributed and functionally integrative hubs conserve (conduction) time by reducing the average axonal delay, or speed of interneuronal communication. The simultaneous presence of modules and hubs supports a contemporary reformulation of Ramón y Cajal’s laws as a trade-off between minimization of wiring cost and maximization of topological integration.Magnetic resonance imaging (MRI) allowed for testing such organizational principles in large-scale mammalian connectomes with high throughput whole-brain imaging. However, MRI methods measure anatomical connectivity indirectly and at low (millimeter scale) spatial resolution (14). In contrast, tract tracing methods measure anatomical connectivity directly, by detecting axonally mediated propagation of injected tracer, and at higher (micrometer scale) spatial resolution. Tract-tracing methods represent the current “gold standard” for mapping mammalian connectomes. However, most tract-tracing connectome studies to date have been limited to metaanalyses of primary datasets with limited brain coverage and variable definitions of brain regions and interregional connections (6, 7). Tract-tracing methods for comprehensive and systematic mapping of the connectome did not exist until recently (15–18).The recent step change in the quality and quantity of available tract-tracing measurements in mammalian species, such as the macaque and the mouse, provides a crucial opportunity to test theories of connectome organization more rigorously. Some of the first systematic high-quality tract tracing studies in the macaque have revealed many previously unreported weak and long-range axonal projections (19, 20). These studies have also shown that spatial constraints on wiring cost, modeled by an exponential decay weight–distance relationship, can account for many important aspects of the macaque connectome (21, 22).We therefore considered it important to comprehensively evaluate the design principles of the mouse connectome in a systematically acquired dataset of axonal tract-tracing experiments (17). We measured the topological and spatial properties of this connectome and compared these properties to equivalent properties of reference lattice and random graphs. We hypothesized that the connectome would have a complex topology and include integrative hubs inexplicable by minimization of wiring cost. We also explored the neurobiological substrates of the mouse connectome by correlating topological properties with histological and gene-expression properties quantified from independently acquired datasets.     

From the Cover: Shifting gradients of macroscale cortical organization mark the transition from childhood to adolescence

The transition from childhood to adolescence is marked by pronounced shifts in brain structure and function that coincide with the development of physical, cognitive, and social abilities. Prior work in adult populations has characterized the topographical organization of the cortex, revealing macroscale functional gradients that extend from unimodal (somatosensory/motor and visual) regions through the cortical association areas that underpin complex cognition in humans. However, the presence of these core functional gradients across development as well as their maturational course have yet to be established. Here, leveraging 378 resting-state functional MRI scans from 190 healthy individuals aged 6 to 17 y old, we demonstrate that the transition from childhood to adolescence is reflected in the gradual maturation of gradient patterns across the cortical sheet. In children, the overarching organizational gradient is anchored within the unimodal cortex, between somatosensory/motor and visual territories. Conversely, in adolescence, the principal gradient of connectivity transitions into an adult-like spatial framework, with the default network at the opposite end of a spectrum from primary sensory and motor regions. The observed gradient transitions are gradually refined with age, reaching a sharp inflection point in 13 and 14 y olds. Functional maturation was nonuniformly distributed across cortical networks. Unimodal networks reached their mature positions early in development, while association regions, in particular the medial prefrontal cortex, reached a later peak during adolescence. These data reveal age-dependent changes in the macroscale organization of the cortex and suggest the scheduled maturation of functional gradient patterns may be critically important for understanding how cognitive and behavioral capabilities are refined across development.

The cerebral cortex is comprised of a complex web of large-scale networks that are central to its information processing capabilities (1, 2). Although the topographic organization of this distributed network architecture has been characterized in adulthood (3–6), there is a growing consensus that macroscale patterns of functional connectivity are not static across the lifespan (7–12). In particular, the transformative brain changes occurring throughout childhood and adolescence are critical for supporting the emergence and development of physical, cognitive, and social abilities (13–17). Yet, despite clear evidence that patterns of brain organization from neural circuits through large-scale cortical networks relate to behavior (18–20), the age-dependent changes that underpin hierarchical shifts across the functional connectome have yet to be systematically investigated.Technical advances in the field of connectomics have provided researchers with tools to map the spatial organization of large-scale distributed networks in the human brain (3, 5, 6). The cortical sheet is comprised of areal units, traditionally defined through their embryological origins, cytoarchitecture, and evoked functions (21–23). Critically, the brain is a multiscale system, and these areal parcels are embedded within segregated processing streams and corresponding networks that are evident through both anatomical projections and patterns of coherent neural activity at rest (2, 24–26). This converging evidence suggests the presence of a broad division separating the unimodal somatosensory/motor (somato/motor) and visual territories that form domain-specific hierarchical connections (22) and the heteromodal association areas that integrate long-distance projections from widely distributed sets of brain systems (27). However, the distinction between the unimodal and heteromodal cortex is not reflected in abrupt transitions along the cortical surface. Recent work has revealed that this macroscale property of brain organization is evident in the presence of a principal functional gradient that situates discrete large-scale networks and associated areal parcels along a continuous spectrum, extending from the unimodal systems that underpin perception and action through the association territories implicated in more abstract cognitive functions (12, 28, 29).Cortical gradients capture the topography of large-scale networks, offering a complementary approach to standard brain parcellation techniques. Here, rather than marking discrete boundaries at locations of abrupt change in cortical microstructure, function, and/or connectivity, spatial variation across the cortex is examined continuously along overlapping organizing axes (28, 30). As such, the gradient approach provides an organizing spatial framework for linking multiple large-scale networks and functions in separate domains while avoiding potential biases induced by the inconsistent size and shape of parcels (31). In both humans and nonhuman primates, the association cortex end of the principal gradient is anchored within the default network, including portions of the ventral and dorsal medial prefrontal, posterior/retrosplenial, and inferior parietal cortex (32–34). Converging evidence from tract tracing data in marmosets, for instance, highlights the presence of a gradient of sequential networks radiating outward from the primary cortex to a transmodal network that has many parallels with the human default network (35). The default network acts as a hub that integrates representational information across the cortex (1, 6). As such, it constitutes a functional system hypothesized to underpin self-referential processing and core aspects of mental simulation (36). It has recently been proposed that the default network sits at the apex of cortical hierarchy, situated as the most distant association network from the sensory cortex (37). This feature of brain organization may reduce the strong constraints of sensory/motor input on default network function, facilitating the emergence of abstract cognition through the integration of information across modalities.Suggesting cognitive abilities are closely associated with these macroscale gradients, the spatial framework evident in profiles of intrinsic brain activity mirrors the patterns of functional specialization and flexibility emerging through extrinsic (task-evoked) studies of the human brain (20). These functional gradients have been well characterized in adult populations, providing a framework for describing the integration of local processing streams throughout cortical (28) and subcortical systems (30). The use of low-dimensional representations of functional connectivity provides a unified perspective to efficiently explain core organizing properties of the human cerebral cortex, linking specific regions, networks, and functions. Recent work examining gradients in cortical microstructures has revealed a gradual differentiation of the myelin content of cortical systems from adolescence through young adulthood (38). However, the presence of this overarching organizational structure across development as well as its maturational course have yet to be established.Human brain development is influenced by a complex series of dynamic processes across biological systems, ranging from shifting profiles of gene expression (39) to hierarchical changes in brain structure and function (14). Although still in its early stages, work in this area has characterized several core organizational principles underlying large-scale network development (40–44). In human and nonhuman primates, for instance, there is an earlier maturational plateau in gray matter (45–47) as well as in synaptic formation and subsequent pruning within unimodal sensory/motor and subcortical territories relative to aspects of the association cortex (46, 48). This developmental sequence is reflected in patterns of network activity. For example, brain functions in infancy are characterized by the predominance of short-range connectivity (49, 50), which gradually transitions through childhood and adolescence as long-range network connections become increasingly evident (11, 51–56). Collectively, these results suggest that brain development may entail a shift from a sensory organization (unimodal gradient) to a globally distributed spatial framework (association gradient) across childhood and adolescence. Whole-brain connectome-wide neurodevelopmental studies have identified patterns of functional network organization that are predictive of behavioral traits (57) and associated with mental health outcomes in adolescence and young adulthood (7). However, to date, there have been few opportunities to directly explore the age-dependent maturation of functional gradients across the cortex. The characterization of age-dependent changes in the macroscale organization of the human brain would provide a tremendous opportunity to understand how connectome development shapes the evolving expression of individual differences in behavior across health and disease.In examining age-dependent shifts in the macroscale functional organization of the cortex, we applied the dimensionality reduction approach of diffusion map embedding (28, 58) to resting-state functional MRI (fMRI) data to extract a global framework that accounts for the dominant connectome-level connectivity patterns in a population of children and adolescents. The resulting components, or gradients, reflect the distillation of complex patterns of local and global functional connectivity across the connectome into simple and interpretable spatial architectures. This approach allowed us to establish the extent to which functional maturation is nonuniformly distributed across cortical networks. In doing so, our analyses revealed the presence of a developmental change from a functional motif first dominated by the distinction between sensory and motor systems and then balanced through interactions with later-maturing connectivity within the association cortex.     

From the Cover: Structural differences in amyloid-β fibrils from brains of nondemented elderly individuals and Alzheimer's disease patients

Although amyloid plaques composed of fibrillar amyloid-β (Aβ) assemblies are a diagnostic hallmark of Alzheimer''s disease (AD), quantities of amyloid similar to those in AD patients are observed in brain tissue of some nondemented elderly individuals. The relationship between amyloid deposition and neurodegeneration in AD has, therefore, been unclear. Here, we use solid-state NMR to investigate whether molecular structures of Aβ fibrils from brain tissue of nondemented elderly individuals with high amyloid loads differ from structures of Aβ fibrils from AD tissue. Two-dimensional solid-state NMR spectra of isotopically labeled Aβ fibrils, prepared by seeded growth from frontal lobe tissue extracts, are similar in the two cases but with statistically significant differences in intensity distributions of cross-peak signals. Differences in solid-state NMR data are greater for 42-residue amyloid-β (Aβ42) fibrils than for 40-residue amyloid-β (Aβ40) fibrils. These data suggest that similar sets of fibril polymorphs develop in nondemented elderly individuals and AD patients but with different relative populations on average.

Amyloid plaques in brain tissue, containing fibrils formed by amyloid-β (Aβ) peptides, are one of the diagnostic pathological signatures of Alzheimer''s disease (AD). Clear genetic and biomarker evidence indicates that Aβ is key to AD pathogenesis (1). However, Aβ is present as a diverse population of multimeric assemblies, ranging from soluble oligomers to insoluble fibrils and plaques, and may lead to neurodegeneration by a number of possible mechanisms (2–7).One argument against a direct neurotoxic role for Aβ plaques and fibrils in AD is the fact that plaques are not uncommon in the brains of nondemented elderly people, as shown both by traditional neuropathological studies (8, 9) and by positron emission tomography (10–13). On average, the quantity of amyloid is greater in AD patients (10) and (at least in some studies) increases with decreasing cognitive ability (12, 14, 15) or increasing rate of cognitive decline (16). However, a high amyloid load does not necessarily imply a high degree of neurodegeneration and cognitive impairment (11, 13, 17).A possible counterargument comes from studies of the molecular structures of Aβ fibrils, which show that Aβ peptides form multiple distinct fibril structures, called fibril polymorphs (18–20). Polymorphism has been demonstrated for fibrils formed by both 40-residue amyloid-β (Aβ40) (19, 21–24) and 42-residue amyloid-β (Aβ42) (22, 25–29) peptides, the two main Aβ isoforms. Among people with similar total amyloid loads, variations in neurodegeneration and cognitive impairment may conceivably arise from variations in the relative populations of different fibril polymorphs. As a hypothetical example, if polymorph A was neurotoxic but polymorph B was not, then people whose Aβ peptides happened to form polymorph A would develop AD, while people whose Aβ peptides happened to form polymorph B would remain cognitively normal. In practice, brains may contain a population of different propagating and/or neurotoxic Aβ species, akin to prion quasispecies or “clouds,” and the relative proportions of these and their dynamic interplay may affect clinical phenotype and rates of progression (30).Well-established connections between molecular structural polymorphism and variations in other neurodegenerative diseases lend credence to the hypothesis that Aβ fibril polymorphism plays a role in variations in the characteristics of AD. Distinct strains of prions causing the transmissible spongiform encephalopathies have been shown to involve different molecular structural states of the mammalian prion protein PrP (30–32). Distinct tauopathies involve different polymorphs of tau protein fibrils (33–37). In the case of synucleopathies, α-synuclein has been shown to be capable of forming polymorphic fibrils (38–40) with distinct biological effects (41–43).Experimental support for connections between Aβ polymorphism and variations in characteristics of AD comes from polymorph-dependent fibril toxicities in neuronal cell cultures (19), differences in neuropathology induced in transgenic mice by injection of amyloid-containing extracts from different sources (44–46), differences in conformation and stability with respect to chemical denaturation of Aβ assemblies prepared from brain tissue of rapidly or slowly progressing AD patients (47), and differences in fluorescence emission spectra of structure-sensitive dyes bound to amyloid plaques in tissue from sporadic or familial AD patients (48, 49).Solid-state NMR spectroscopy is a powerful method for investigating fibril polymorphism because even small, localized changes in molecular conformation or structural environment produce measurable changes in ¹³C and ¹⁵N NMR chemical shifts (i.e., in NMR frequencies of individual carbon and nitrogen sites). Full molecular structural models for amyloid fibrils can be developed from large sets of measurements on structurally homogeneous samples (21, 25, 26, 29, 38, 50). Alternatively, simple two-dimensional (2D) solid-state NMR spectra can serve as structural fingerprints, allowing assessments of polymorphism and comparisons between samples from different sources (22, 51).Solid-state NMR requires isotopic labeling and milligram-scale quantities of fibrils, ruling out direct measurements on amyloid fibrils extracted from brain tissue. However, Aβ fibril structures from autopsied brain tissue can be amplified and isotopically labeled by seeded fibril growth, in which fibril fragments (i.e., seeds) in a brain tissue extract are added to a solution of isotopically labeled peptide (21, 22, 52). Labeled “daughter” fibrils that grow from the seeds retain the molecular structures of the “parent” fibrils, as demonstrated for Aβ (19, 21, 24, 53) and other (54, 55) amyloid fibrils. Solid-state NMR measurements on the brain-seeded fibrils then provide information about molecular structures of fibrils that were present in the brain tissue at the time of autopsy. Using this approach, Lu et al. (21) developed a full molecular structure for Aβ40 fibrils derived from one AD patient with an atypical clinical history (patient 1), showed that Aβ40 fibrils from a second patient with a typical AD history (patient 2) were qualitatively different in structure, and showed that the predominant brain-derived Aβ40 polymorph was the same in multiple regions of the cerebral cortex from each patient. Subsequently, Qiang et al. (22) prepared isotopically labeled Aβ40 and Aβ42 fibrils from frontal, occipital, and parietal lobe tissue of 15 patients in three categories, namely typical long-duration Alzheimer''s disease (t-AD), the posterior cortical atrophy variant of Alzheimer''s disease (PCA-AD), and rapidly progressing Alzheimer''s disease (r-AD). Quantitative analyses of 2D solid-state NMR spectra led to the conclusions that Aβ40 fibrils derived from t-AD and PCA-AD tissue were indistinguishable, with both showing the same predominant polymorph; that Aβ40 fibrils derived from r-AD tissue were more structurally heterogeneous (i.e., more polymorphic); and that Aβ42 fibrils derived from all three categories were structurally heterogeneous, with at least two prevalent Aβ42 polymorphs (22).In this paper, we address the question of whether Aβ fibrils that develop in cortical tissue of nondemented elderly individuals with high amyloid loads are structurally distinguishable from fibrils that develop in cortical tissue of AD patients. As described below, quantitative analyses of 2D solid-state NMR spectra of brain-seeded samples indicate statistically significant differences for both Aβ40 and Aβ42 fibrils. Differences in the 2D spectra are subtle, however, indicating that nondemented individuals and AD patients do not develop entirely different Aβ fibril structures. Instead, data and analyses described below suggest overlapping distributions of fibril polymorphs, with different relative populations on average.     

The evolution of self-control

Cognition presents evolutionary research with one of its greatest challenges. Cognitive evolution has been explained at the proximate level by shifts in absolute and relative brain volume and at the ultimate level by differences in social and dietary complexity. However, no study has integrated the experimental and phylogenetic approach at the scale required to rigorously test these explanations. Instead, previous research has largely relied on various measures of brain size as proxies for cognitive abilities. We experimentally evaluated these major evolutionary explanations by quantitatively comparing the cognitive performance of 567 individuals representing 36 species on two problem-solving tasks measuring self-control. Phylogenetic analysis revealed that absolute brain volume best predicted performance across species and accounted for considerably more variance than brain volume controlling for body mass. This result corroborates recent advances in evolutionary neurobiology and illustrates the cognitive consequences of cortical reorganization through increases in brain volume. Within primates, dietary breadth but not social group size was a strong predictor of species differences in self-control. Our results implicate robust evolutionary relationships between dietary breadth, absolute brain volume, and self-control. These findings provide a significant first step toward quantifying the primate cognitive phenome and explaining the process of cognitive evolution.Since Darwin, understanding the evolution of cognition has been widely regarded as one of the greatest challenges for evolutionary research (1). Although researchers have identified surprising cognitive flexibility in a range of species (2–40) and potentially derived features of human psychology (41–61), we know much less about the major forces shaping cognitive evolution (62–71). With the notable exception of Bitterman’s landmark studies conducted several decades ago (63, 72–74), most research comparing cognition across species has been limited to small taxonomic samples (70, 75). With limited comparable experimental data on how cognition varies across species, previous research has largely relied on proxies for cognition (e.g., brain size) or metaanalyses when testing hypotheses about cognitive evolution (76–92). The lack of cognitive data collected with similar methods across large samples of species precludes meaningful species comparisons that can reveal the major forces shaping cognitive evolution across species, including humans (48, 70, 89, 93–98).To address these challenges we measured cognitive skills for self-control in 36 species of mammals and birds (Fig. 1 and Tables S1–S4) tested using the same experimental procedures, and evaluated the leading hypotheses for the neuroanatomical underpinnings and ecological drivers of variance in animal cognition. At the proximate level, both absolute (77, 99–107) and relative brain size (108–112) have been proposed as mechanisms supporting cognitive evolution. Evolutionary increases in brain size (both absolute and relative) and cortical reorganization are hallmarks of the human lineage and are believed to index commensurate changes in cognitive abilities (52, 105, 113–115). Further, given the high metabolic costs of brain tissue (116–121) and remarkable variance in brain size across species (108, 122), it is expected that the energetic costs of large brains are offset by the advantages of improved cognition. The cortical reorganization hypothesis suggests that selection for absolutely larger brains—and concomitant cortical reorganization—was the predominant mechanism supporting cognitive evolution (77, 91, 100–106, 120). In contrast, the encephalization hypothesis argues that an increase in brain volume relative to body size was of primary importance (108, 110, 111, 123). Both of these hypotheses have received support through analyses aggregating data from published studies of primate cognition and reports of “intelligent” behavior in nature—both of which correlate with measures of brain size (76, 77, 84, 92, 110, 124).Open in a separate windowFig. 1.A phylogeny of the species included in this study. Branch lengths are proportional to time except where long branches have been truncated by parallel diagonal lines (split between mammals and birds ∼292 Mya).With respect to selective pressures, both social and dietary complexities have been proposed as ultimate causes of cognitive evolution. The social intelligence hypothesis proposes that increased social complexity (frequently indexed by social group size) was the major selective pressure in primate cognitive evolution (6, 44, 48, 50, 87, 115, 120, 125–141). This hypothesis is supported by studies showing a positive correlation between a species’ typical group size and the neocortex ratio (80, 81, 85–87, 129, 142–145), cognitive differences between closely related species with different group sizes (130, 137, 146, 147), and evidence for cognitive convergence between highly social species (26, 31, 148–150). The foraging hypothesis posits that dietary complexity, indexed by field reports of dietary breadth and reliance on fruit (a spatiotemporally distributed resource), was the primary driver of primate cognitive evolution (151–154). This hypothesis is supported by studies linking diet quality and brain size in primates (79, 81, 86, 142, 155), and experimental studies documenting species differences in cognition that relate to feeding ecology (94, 156–166).Although each of these hypotheses has received empirical support, a comparison of the relative contributions of the different proximate and ultimate explanations requires (i) a cognitive dataset covering a large number of species tested using comparable experimental procedures; (ii) cognitive tasks that allow valid measurement across a range of species with differing morphology, perception, and temperament; (iii) a representative sample within each species to obtain accurate estimates of species-typical cognition; (iv) phylogenetic comparative methods appropriate for testing evolutionary hypotheses; and (v) unprecedented collaboration to collect these data from populations of animals around the world (70).Here, we present, to our knowledge, the first large-scale collaborative dataset and comparative analysis of this kind, focusing on the evolution of self-control. We chose to measure self-control—the ability to inhibit a prepotent but ultimately counterproductive behavior—because it is a crucial and well-studied component of executive function and is involved in diverse decision-making processes (167–169). For example, animals require self-control when avoiding feeding or mating in view of a higher-ranking individual, sharing food with kin, or searching for food in a new area rather than a previously rewarding foraging site. In humans, self-control has been linked to health, economic, social, and academic achievement, and is known to be heritable (170–172). In song sparrows, a study using one of the tasks reported here found a correlation between self-control and song repertoire size, a predictor of fitness in this species (173). In primates, performance on a series of nonsocial self-control control tasks was related to variability in social systems (174), illustrating the potential link between these skills and socioecology. Thus, tasks that quantify self-control are ideal for comparison across taxa given its robust behavioral correlates, heritable basis, and potential impact on reproductive success.In this study we tested subjects on two previously implemented self-control tasks. In the A-not-B task (27 species, n = 344), subjects were first familiarized with finding food in one location (container A) for three consecutive trials. In the test trial, subjects initially saw the food hidden in the same location (container A), but then moved to a new location (container B) before they were allowed to search (Movie S1). In the cylinder task (32 species, n = 439), subjects were first familiarized with finding a piece of food hidden inside an opaque cylinder. In the following 10 test trials, a transparent cylinder was substituted for the opaque cylinder. To successfully retrieve the food, subjects needed to inhibit the impulse to reach for the food directly (bumping into the cylinder) in favor of the detour response they had used during the familiarization phase (Movie S2).Thus, the test trials in both tasks required subjects to inhibit a prepotent motor response (searching in the previously rewarded location or reaching directly for the visible food), but the nature of the correct response varied between tasks. Specifically, in the A-not-B task subjects were required to inhibit the response that was previously successful (searching in location A) whereas in the cylinder task subjects were required to perform the same response as in familiarization trials (detour response), but in the context of novel task demands (visible food directly in front of the subject).     

Vitamin D prevents cognitive decline and enhances hippocampal synaptic function in aging rats

Vitamin D is an important calcium-regulating hormone with diverse functions in numerous tissues, including the brain. Increasing evidence suggests that vitamin D may play a role in maintaining cognitive function and that vitamin D deficiency may accelerate age-related cognitive decline. Using aging rodents, we attempted to model the range of human serum vitamin D levels, from deficient to sufficient, to test whether vitamin D could preserve or improve cognitive function with aging. For 5–6 mo, middle-aged F344 rats were fed diets containing low, medium (typical amount), or high (100, 1,000, or 10,000 international units/kg diet, respectively) vitamin D3, and hippocampal-dependent learning and memory were then tested in the Morris water maze. Rats on high vitamin D achieved the highest blood levels (in the sufficient range) and significantly outperformed low and medium groups on maze reversal, a particularly challenging task that detects more subtle changes in memory. In addition to calcium-related processes, hippocampal gene expression microarrays identified pathways pertaining to synaptic transmission, cell communication, and G protein function as being up-regulated with high vitamin D. Basal synaptic transmission also was enhanced, corroborating observed effects on gene expression and learning and memory. Our studies demonstrate a causal relationship between vitamin D status and cognitive function, and they suggest that vitamin D-mediated changes in hippocampal gene expression may improve the likelihood of successful brain aging.Vitamin D, a secosteroid hormone known for its role in bone and calcium homeostasis, is now well recognized for its many diverse functions and actions on a variety of tissues and cell types (1, 2). Vitamin D typically refers to the precursor forms of the hormone obtained through the skin’s exposure to sunlight [vitamin D3 (VitD3)] or from dietary sources (VitD3 or VitD2). A metabolite of vitamin D, 25-hydroxyvitamin D (25OHD), is a serum biomarker of vitamin D status or repletion. In recent years, there is particular concern that large segments of the population may have low levels of 25OHD, and therefore are vitamin D-deficient (3). Due to factors such as reduced intake, absorption, and decreased exposure to sunlight, aging adults (≥50 y of age) are especially susceptible (3–6). Notably, this predisposition for lower 25OHD levels in the elderly has been linked to higher risk for numerous age-related disorders, including cancer and metabolic and vascular diseases (7–10).Inadequate vitamin D status also correlates with a greater risk for cognitive decline in the elderly (4, 11–15), suggesting that optimal levels may promote healthy brain aging (16, 17). Because the brain expresses vitamin D receptors (VDRs) and can synthesize the active form of the hormone, the possible cognitive enhancing effects of vitamin D may reflect a primary action in the brain rather than a result of secondary systemic effects (18–22). Indeed, we and others have shown that vitamin D, as well as the biologically active form of the hormone, 1,25-dihydroxyvitamin D, has direct neuroprotective actions and can reduce some biomarkers of brain aging (20, 23–28).Given that the aging population is projected to increase dramatically in the near future (29), along with estimates that a significant proportion of the elderly are vitamin D-deficient (3), there is a critical need to determine whether efforts to improve vitamin D status can reduce age-related cognitive decline. Despite calls for more definitive research along these lines (30), few long-term intervention studies have examined the impact of manipulating vitamin D on cognitive function with advancing age. To test the hypothesis that higher vitamin D levels improve cognitive function in aging animals, middle-aged male F344 rats were placed on diets containing low, medium [National Research Council (NRC)-required], or high levels of VitD3 (or cholecalciferol) for 5–6 mo. The middle-age period was chosen because it increasingly appears to be an important window of time at which to initiate interventions designed to preserve cognitive function into the geriatric period. At midlife, subtle cognitive impairments begin to appear, along with structural and genomic changes associated with brain aging (31–34). Our results show that higher than normal dietary VitD3 may improve the chances of successful brain aging and that changes in neuronal synaptic function in the hippocampus may underlie its protective effects against age-related cognitive decline.     

The neural architecture of language: Integrative modeling converges on predictive processing

The neuroscience of perception has recently been revolutionized with an integrative modeling approach in which computation, brain function, and behavior are linked across many datasets and many computational models. By revealing trends across models, this approach yields novel insights into cognitive and neural mechanisms in the target domain. We here present a systematic study taking this approach to higher-level cognition: human language processing, our species’ signature cognitive skill. We find that the most powerful “transformer” models predict nearly 100% of explainable variance in neural responses to sentences and generalize across different datasets and imaging modalities (functional MRI and electrocorticography). Models’ neural fits (“brain score”) and fits to behavioral responses are both strongly correlated with model accuracy on the next-word prediction task (but not other language tasks). Model architecture appears to substantially contribute to neural fit. These results provide computationally explicit evidence that predictive processing fundamentally shapes the language comprehension mechanisms in the human brain.

A core goal of neuroscience is to decipher from patterns of neural activity the algorithms underlying our abilities to perceive, think, and act. Recently, a new “reverse engineering” approach to computational modeling in systems neuroscience has transformed our algorithmic understanding of the primate ventral visual stream (1–8) and holds great promise for other aspects of brain function. This approach has been enabled by a breakthrough in artificial intelligence (AI): the engineering of artificial neural network (ANN) systems that perform core perceptual tasks with unprecedented accuracy, approaching human levels, and that do so using computational machinery that is abstractly similar to biological neurons. In the ventral visual stream, the key AI developments come from deep convolutional neural networks (DCNNs) that perform visual object recognition from natural images (1, 2, 4, 9, 10), widely thought to be the primary function of this pathway. Leading DCNNs for object recognition have now been shown to predict the responses of neural populations in multiple stages of the ventral stream (V1, V2, V4, and inferior temporal [IT] cortex), in both macaque and human brains, approaching the noise ceiling of the data. Thus, despite abstracting away aspects of biology, DCNNs provide the basis for a first complete hypothesis of how the brain extracts object percepts from visual input.Inspired by this success story, analogous ANN models have now been applied to other domains of perception (11, 12). Could these models also let us reverse-engineer the brain mechanisms of higher-level human cognition? Here we show how the modeling approach pioneered in the ventral stream can be applied to a higher-level cognitive domain that plays an essential role in human life: language comprehension, or the extraction of meaning from spoken, written, or signed words and sentences. Cognitive scientists have long treated neural network models of language processing with skepticism (13, 14), given that these systems lack (and often deliberately attempt to do without) explicit symbolic representation—traditionally seen as a core feature of linguistic meaning. Recent ANN models of language, however, have proven capable of at least approximating some aspects of symbolic computation and have achieved remarkable success on a wide range of applied natural language processing (NLP) tasks. The results presented here, based on this new generation of ANNs, suggest that a computationally adequate model of language processing in the brain may be closer than previously thought.Because we build on the same logic in our analysis of language in the brain, it is helpful to review why the neural network-based integrative modeling approach has proven so powerful in the study of object recognition in the ventral stream. Crucially, our ability to robustly link computation, brain function, and behavior is supported not by testing a single model on a single dataset or a single kind of data, but by large-scale integrative benchmarking (4) that establishes consistent patterns of performance across many different ANNs applied to multiple neural and behavioral datasets, together with their performance on the proposed core computational function of the brain system under study. Given the complexities of the brain’s structure and the functions it performs, any one of these models is surely oversimplified and ultimately wrong—at best, an approximation of some aspects of what the brain does. However, some models are less wrong than others, and consistent trends in performance across models can reveal not just which model best fits the brain but also which properties of a model underlie its fit to the brain, thus yielding critical insights that transcend what any single model can tell us.In the ventral stream specifically, our understanding that computations underlying object recognition are analogous to the structure and function of DCNNs is supported by findings that across hundreds of model variants, DCNNs that perform better on object recognition tasks also better capture human recognition behavior and neural responses in IT cortex of both human and nonhuman primates (1, 2, 4, 15). This integrative benchmarking reveals a rich pattern of correlations among three classes of performance measures—1) neural variance explained in IT neurophysiology or functional MRI (fMRI) responses (brain scores), 2) accuracy in predicting hits and misses in human object recognition behavior or human object similarity judgments (behavioral scores), and 3) accuracy on the core object recognition task (computational task score)—such that for any individual DCNN model we can predict how well it would score on each of these measures from the other measures. This pattern of results was not assembled in a single paper but in multiple papers across several laboratories and several years. Taken together, they provide strong evidence that the ventral stream supports primate object recognition through something like a deep convolutional feature hierarchy, the exact details of which are being modeled with ever-increasing precision.Here we describe an analogous pattern of results for ANN models of human language, establishing a link between language models, including transformer-based ANN architectures that have revolutionized NLP in AI systems over the last 3 y, and fundamental computations of human language processing as reflected in both neural and behavioral measures. Language processing is known to depend causally on a left-lateralized frontotemporal brain network (16–22) (Fig. 1) that responds robustly and selectively to linguistic input (23, 24), whether auditory or visual (25, 26). Yet, the precise computations underlying language processing in the brain remain unknown. Computational models of sentence processing have previously been used to explain both behavioral (27–41) and neural responses to linguistic input (42–64). However, none of the prior studies have attempted large-scale integrative benchmarking that has proven so valuable in understanding key brain–behavior–computation relationships in the ventral stream; instead, they have typically tested one or a small number of models against a single dataset, and the same models have not been evaluated on all three metrics of neural, behavioral, and objective task performance. Previously tested models have also left much of the variance in human neural/behavioral data unexplained. Finally, until the rise of recent ANNs (e.g., transformer architectures), language models did not have sufficient capacity to solve the full linguistic problem that the brain solves—to form a representation of sentence meaning capable of performing a broad range of real-world language tasks on diverse natural linguistic input. We are thus left with a collection of suggestive results but no clear sense of how close ANN models are to fully explaining language processing in the brain, or what model features are key in enabling models to explain neural and behavioral data.Open in a separate windowFig. 1.Comparing ANN models of language processing to human language processing. We tested how well different models predict measurements of human neural activity (fMRI and ECoG) and behavior (reading times) during language comprehension. The candidate models ranged from simple embedding models to more complex recurrent and transformer networks. Stimuli ranged from sentences to passages to stories and were 1) fed into the models and 2) presented to human participants (visually or auditorily). Models’ internal representations were evaluated on three major dimensions: their ability to predict human neural representations (brain score, extracted from within the frontotemporal language network [e.g., Fedorenko et al. (71)]; the network topography is schematically illustrated in red on the template brain above); their ability to predict human behavior in the form of reading times (behavioral score); and their ability to perform computational tasks such as next-word prediction (computational task score). Consistent relationships between these measures across many different models reveal insights beyond what a single model can tell us.Our goal here is to present a systematic integrative modeling study of language in the brain, at the scale necessary to discover robust relationships between neural and behavioral measurements from humans, and performance of models on language tasks. We seek to determine not just which model fits empirical data best but also what dimensions of variation across models are correlated with fit to human data. This approach has not been applied in the study of language or any other higher cognitive system, and even in perception has not been attempted within a single integrated study. Thus, we view our work more generally as a template for how to apply the integrative benchmarking approach to any perceptual or cognitive system.Specifically, we examined the relationships between 43 diverse state-of-the-art ANN language models (henceforth “models”) across three neural language comprehension datasets (two fMRI, one electrocorticography [ECoG]), as well as behavioral signatures of human language processing in the form of self-paced reading times, and a range of linguistic functions assessed via standard engineering tasks from NLP. The models spanned all major classes of existing ANN language approaches and included simple embedding models [e.g., GloVe (65)], more complex recurrent neural networks [e.g., LM1B (66)], and many variants of transformers or attention-based architectures—including both “unidirectional-attention” models [trained to predict the next word given the previous words, e.g., GPT (67)] and “bidirectional-attention” models [trained to predict a missing word given the surrounding context, e.g., BERT (68)].Our integrative approach yielded four major findings. First, models’ relative fit to neural data (neural predictivity or “brain score”)—estimated on held-out test data—generalizes across different datasets and imaging modality (fMRI and ECoG), and certain architectural features consistently lead to more brain-like models: Transformer-based models perform better than recurrent networks or word-level embedding models, and larger-capacity models perform better than smaller models. Second, the best models explain nearly 100% of the explainable variance (up to the noise ceiling) in neural responses to sentences. This result stands in stark contrast to earlier generations of models that have typically accounted for at most 30 to 50% of the predictable neural signal. Third, across models, significant correlations hold among all three metrics of model performance: brain scores (fit to fMRI and ECoG data), behavioral scores (fit to reading time), and model accuracy on the next-word prediction task. Importantly, no other linguistic task was predictive of models’ fit to neural or behavioral data. These findings provide strong evidence for a classic hypothesis about the computations underlying human language understanding, that the brain’s language system is optimized for predictive processing in the service of meaning extraction. Fourth, intriguingly, the scores of models initialized with random weights (prior to training, but with a trained linear readout) are well above chance and correlate with trained model scores, which suggests that network architecture is an important contributor to a model’s brain score. In particular, one architecture introduced just in 2019, the generative pretrained transformer (GPT-2), consistently outperforms all other models and explains almost all variance in both fMRI and ECoG data from sentence-processing tasks. GPT-2 is also arguably the most cognitively plausible of the transformer models (because it uses unidirectional, forward attention) and performs best overall as an AI system when considering both natural language understanding and natural language generation tasks. Thus, even though the goal of contemporary AI is to improve model performance and not necessarily to build models of brain processing, this endeavor appears to be rapidly converging on architectures that might capture key aspects of language processing in the human mind and brain.     

INAUGURAL ARTICLE by a Recently Elected Academy Member:Fluid intelligence and the locus coeruleus–norepinephrine system

The last decade has seen significant progress identifying genetic and brain differences related to intelligence. However, there remain considerable gaps in our understanding of how cognitive mechanisms that underpin intelligence map onto various brain functions. In this article, we argue that the locus coeruleus–norepinephrine system is essential for understanding the biological basis of intelligence. We review evidence suggesting that the locus coeruleus–norepinephrine system plays a central role at all levels of brain function, from metabolic processes to the organization of large-scale brain networks. We connect this evidence with our executive attention view of working-memory capacity and fluid intelligence and present analyses on baseline pupil size, an indicator of locus coeruleus activity. Using a latent variable approach, our analyses showed that a common executive attention factor predicted baseline pupil size. Additionally, the executive attention function of disengagement––not maintenance––uniquely predicted baseline pupil size. These findings suggest that the ability to control attention may be important for understanding how cognitive mechanisms of fluid intelligence map onto the locus coeruleus–norepinephrine system. We discuss how further research is needed to better understand the relationships between fluid intelligence, the locus coeruleus–norepinephrine system, and functionally organized brain networks.

In this article, we outline what we see as a potentially important relationship for understanding the biological basis of intelligence: that is, the relationship between fluid intelligence and the locus coeruleus–norepinephrine system. This is largely motivated by our findings that baseline pupil size is related to fluid intelligence (1, 2); the larger the pupils, the higher the fluid intelligence. The connection to the locus coeruleus is based on research showing that the size of the pupil can be used as an indicator of locus coeruleus activity (3 –8). A large body of research on the locus coeruleus–norepinephrine system in animal and human studies has shown how this system is critical for an impressively wide range of behaviors and cognitive processes, from regulating sleep/wake cycles, to sensation and perception, attention, learning and memory, decision making, and more (9 –12). The locus coeruleus–norepinephrine system achieves this primarily through its widespread projection system throughout the cortex, strong connections with the prefrontal cortex, and the effect of norepinephrine at many levels of brain function (10). Given the broad role of this system in behavior, cognition, and brain function, we propose that the locus coeruleus–norepinephrine system is essential for understanding the biological basis of intelligence.     

Modular origins of high-amplitude cofluctuations in fine-scale functional connectivity dynamics

The topology of structural brain networks shapes brain dynamics, including the correlation structure of brain activity (functional connectivity) as estimated from functional neuroimaging data. Empirical studies have shown that functional connectivity fluctuates over time, exhibiting patterns that vary in the spatial arrangement of correlations among segregated functional systems. Recently, an exact decomposition of functional connectivity into frame-wise contributions has revealed fine-scale dynamics that are punctuated by brief and intermittent episodes (events) of high-amplitude cofluctuations involving large sets of brain regions. Their origin is currently unclear. Here, we demonstrate that similar episodes readily appear in silico using computational simulations of whole-brain dynamics. As in empirical data, simulated events contribute disproportionately to long-time functional connectivity, involve recurrence of patterned cofluctuations, and can be clustered into distinct families. Importantly, comparison of event-related patterns of cofluctuations to underlying patterns of structural connectivity reveals that modular organization present in the coupling matrix shapes patterns of event-related cofluctuations. Our work suggests that brief, intermittent events in functional dynamics are partly shaped by modular organization of structural connectivity.

Structural and functional brain networks exhibit complex topology, and functional brain networks display rich temporal dynamics (1–3). The topological organization of structural connectivity (SC; the connectome) is characterized by broad degree distributions, hubs linked into cores and rich clubs, and multiscale modularity (4–6). Functional connectivity (FC), as measured with resting-state functional MRI (fMRI), displays consistent system-level architecture (7–9) as well as fluctuating dynamics (10–12) and complex spatiotemporal state transitions (13, 14). Resting brain dynamics exhibit metastable behavior. The lack of a fixed attractor allows for exploration of a large repertoire of network states and configurations (15–17).Recent work has uncovered fine-scale dynamics of FC as measured with fMRI during rest and passive movie watching (18, 19). The approach leverages an exact decomposition of averaged FC estimates into patterns of edge cofluctuations resolved at the timescale of single image frames (20). These studies reveal that ongoing activity is punctuated by brief, intermittent, high-amplitude bursts of brain-wide cofluctuations of the blood-oxygenation level–dependent (BOLD) signal. The approach is reminiscent of an earlier approach proposed for electroencephalography (EEG) data in which an exact frame-wise analysis of modeled and human scalp EEG data using the Hilbert transform revealed brief large-scale desynchronous bursts (21). In the BOLD literature, episodes of high-amplitude cofluctuations, referred to as “events,” drive long-time estimates of FC and represent patterns with consistent topography across time and across individuals (18, 19, 22). The occurrence of events appears unrelated to nonneuronal physiological processes, head motion, or acquisition artifacts. A better understanding of how events originate may illuminate the basis for individual differences in FC and its variation across cognitive state, development, and disorders. Here, we aim to provide a generative model for the origin of events in neuronal time series and uncover potential structural bases for their emergence in fine-scale dynamics.The relationship of structure to function has been a central objective of numerous empirical and computational studies, leveraging cellular population recordings (23, 24), electrophysiological (25), and neuroimaging techniques (26–28). While there is broad consensus that “structure shapes function” on long timescales (29, 30), relating specific dynamic features to the topology of the underlying structural network is an open problem. Computational models have made important contributions to understanding how SC (31, 32), time delays, and noisy fluctuations (33) contribute to patterns of FC as estimated over long and short timescales. Model implementations range from biophysically based neural mass models to much simpler phase oscillators such as the Kuramoto model (34). Despite their overt simplicity, phase oscillator models can generate a wide range of complex synchronization and coordination states, and they reproduce patterns of empirical FC (35), including temporal dynamics at intermediate timescales (36). These modeled dynamics reproduce ongoing fluctuations between integrated (less modular) and segregated (more modular) network states (37, 38), a key characteristic of empirical fMRI resting-state dynamics (39).Here, we pursue a computational modeling approach that seeks to relate high-amplitude cofluctuations to whole-brain network structure. We simulate spontaneous BOLD signal dynamics on an empirical SC matrix of the human cerebral cortex using an implementation of a coupled phase oscillator model incorporating phase delays, the Kuramoto–Sakaguchi (KS) model (40). The KS model is well suited for this purpose because its parsimonious parametrization allows for drawing specific links between network structure and synchronization patterns. The KS model also allows simulation focused on a specific frequency band of interest so that it can more closely replicate the oscillatory behavior of neural populations often found in the gamma band (41). We find that over broad parameter ranges, BOLD signals exhibit significant high-amplitude network-wide fluctuations strongly resembling intermittent events observed in empirical data. Model dynamics reproduce several key characteristics of empirical events, including their strong contribution to long-time averages of FC as well as recurrent patterns across time. Simulated events are significantly related to network structure. They fall into distinct clusters aligned with different combinations of modules in underlying SC. Disruption of structural modules largely abolishes the occurrence of events in BOLD dynamics. These findings suggest a modular origin of high-amplitude cofluctuations in fine-scale FC dynamics.     

Epigenetic MRI: Noninvasive imaging of DNA methylation in the brain

Both neuronal and genetic mechanisms regulate brain function. While there are excellent methods to study neuronal activity in vivo, there are no nondestructive methods to measure global gene expression in living brains. Here, we present a method, epigenetic MRI (eMRI), that overcomes this limitation via direct imaging of DNA methylation, a major gene-expression regulator. eMRI exploits the methionine metabolic pathways for DNA methylation to label genomic DNA through ¹³C-enriched diets. A ¹³C magnetic resonance spectroscopic imaging method then maps the spatial distribution of labeled DNA. We validated eMRI using pigs, whose brains have stronger similarity to humans in volume and anatomy than rodents, and confirmed efficient ¹³C-labeling of brain DNA. We also discovered strong regional differences in global DNA methylation. Just as functional MRI measurements of regional neuronal activity have had a transformational effect on neuroscience, we expect that the eMRI signal, both as a measure of regional epigenetic activity and as a possible surrogate for regional gene expression, will enable many new investigations of human brain function, behavior, and disease.

The brain is ever-changing in structure and function as a result of development, aging, environmental influence, and disease. Two fundamental mechanisms underpin these changes: neuronal activation, which occurs over relatively short time scales (milliseconds, seconds, and minutes), and gene expression, which occurs over longer time scales (hours, days, or even longer) (1–3). Advances in imaging technology have transformed how we investigate these mechanisms.Functional MRI (fMRI) has revolutionized our understanding of the human brain by providing a powerful nondestructive method to image neural activity (4–7). In contrast, the technologies to image gene expression have been limited to methods that require invasive sampling and tissue processing (8–11). Although these techniques have provided tremendous knowledge about gene expression and gene regulation in the brain, especially in animal models, their destructive nature makes longitudinal studies of the same samples impossible, thus limiting our ability to translate and expand scientific discoveries to human brains. This is especially unfortunate because longer-term changes in brain function play critical roles in both brain diseases and responses of the brain to environmental change (1–3). The ability to measure and visualize gene expression and regulation in the brain noninvasively would revolutionize the study of brain function, behavior, and disease (12).Efforts to map brain gene expression and regulation in living organisms to date have involved imaging reporter genes or associated enzymes using optical techniques, positron emission tomography (PET), or MRI (13–17). These methods are either limited to model organisms or require transgenic animals engineered to express a particular reporter gene and an exogenous contrast probe interacting with the reporter gene to produce the desired images (15, 16). Therefore, such methods have no clear path for translation to humans. Furthermore, these methods are limited to just a few genes and, therefore, cannot provide a comprehensive portrait of gene expression.PET imaging of histone deacetylases (HDACs) in the human brain was recently demonstrated, using a radioactive tracer that can pass the blood–brain barrier (BBB) and target HDAC isoforms (18, 19). This probes a major form of epigenetic gene regulation, histone acetylation, but PET requires introducing radioactive materials into the body, and it only targets one of the enzymes that regulates histone acetylation, rather than histone acetylation itself. Moreover, PET lacks the specificity to distinguish between the target molecule and downstream metabolic products (20–22). Other imaging epigenetics approaches have studied correlations between brain MRI and gene expression or methylation in postmortem tissues, saliva, or blood (23–28); although useful, they can only provide indirect insights into the brain gene expression and regulation.We present successful direct imaging of brain DNA methylation using an approach we call epigenetic MRI (eMRI), which integrates stable isotope 5-methyl-2′-deoxycytidine (5mdC) labeling through diet and magnetic resonance spectroscopic imaging (MRSI). Using pigs fed by a customized diet enriched in ¹³C-methionine (¹³C-Met) and innovations to ¹³C-MRSI, we report robust mapping in intact brain hemispheres that revealed strong regional differences in DNA methylation. We chose pigs as a surrogate to assess feasibility of translation to humans because of stronger similarities in brain size and anatomy than rodent models (29). Significant eMRI signal differences were observed in animals fed with enriched diet for different numbers of days, demonstrating the dynamic nature of this signal. Given the noninvasiveness of our method, these results provide a path toward a global DNA-methylation brain-imaging paradigm for humans. Because DNA methylation is one of the major regulators of gene expression, eMRI promises to become a powerful tool to understand the molecular basis of brain function and disease.     

Lactate supply overtakes glucose when neural computational and cognitive loads scale up

Neural computational power is determined by neuroenergetics, but how and which energy substrates are allocated to various forms of memory engram is unclear. To solve this question, we asked whether neuronal fueling by glucose or lactate scales differently upon increasing neural computation and cognitive loads. Here, using electrophysiology, two-photon imaging, cognitive tasks, and mathematical modeling, we show that both glucose and lactate are involved in engram formation, with lactate supporting long-term synaptic plasticity evoked by high-stimulation load activity patterns and high attentional load in cognitive tasks and glucose being sufficient for less demanding neural computation and learning tasks. Indeed, we show that lactate is mandatory for demanding neural computation, such as theta-burst stimulation, while glucose is sufficient for lighter forms of activity-dependent long-term potentiation (LTP), such as spike timing–dependent plasticity (STDP). We find that subtle variations of spike number or frequency in STDP are sufficient to shift the on-demand fueling from glucose to lactate. Finally, we demonstrate that lactate is necessary for a cognitive task requiring high attentional load, such as the object-in-place task, and for the corresponding in vivo hippocampal LTP expression but is not needed for a less demanding task, such as a simple novel object recognition. Overall, these results demonstrate that glucose and lactate metabolism are differentially engaged in neuronal fueling depending on the complexity of the activity-dependent plasticity and behavior.

Brain activity and performance are tightly constrained by neurovasculature–neuroenergetic coupling (1–3). Neuroenergetics, that is, brain energy metabolism, relies on the blood supply of glucose from the circulation. Evidence accrued over the last two decades has indicated that blood glucose is taken up during synaptic activity (4, 5), mainly by glial cells (astrocytes and oligodendrocytes), and metabolized by aerobic glycolysis, resulting in the release of lactate before transport to neurons as an energy substrate (6–13) necessary for optimized neuronal coding and memory consolidation (14–22). When astrocytes constitute the source of lactate, this process is known as the astrocyte–neuron lactate shuttle in which lactate is transferred from astrocytes to neurons through monocarboxylate transporters, providing an energy substrate for neurons (7). Indeed, lactate can be rapidly metabolized to pyruvate, enter the tricarboxylic acid cycle, and feed the mitochondrial respiratory chain to produce ATP. Other fates of glucose include its glial storage in the form glycogen (7, 23, 24); some degree of glucose uptake occurs in neurons via transporters mainly aimed at feeding the pentose phosphate shunt to produce reducing equivalents (25–27), which is involved in olfactory memory in Drosophila (28). Nevertheless, the nature of the energy substrate, glucose versus lactate, allocated to various forms of memory engram and cognitive load is not known.Here, we tested various forms of activity patterns (rate- and time-coding) for Hebbian long-term synaptic plasticity expression in rat cornu ammonis 1 (CA1) hippocampal pyramidal cells and behavioral tasks with increasing cognitive loads to determine under which conditions glucose and/or lactate are crucial for engram formation and memory. To this end, using brain slice and in vivo electrophysiology, two-photon imaging, mathematical modeling, and recognition memory tasks, we show that astrocytic lactate is mandatory for demanding neural computation, while glucose is sufficient for lighter forms of activity-dependent long-term potentiation (LTP) and that subtle variations of action potential amount or frequency are sufficient to direct the energetic dependency from glucose to lactate. Furthermore, we demonstrate that lactate is necessary for a cognitive task requiring high attentional load (object-in-place [OiP] task) and for the corresponding in vivo hippocampal potentiation but is not needed for a less demanding task (novel object recognition [NOR]). Our results demonstrate that glucose and lactate metabolism are differentially engaged in neuronal fueling depending on the complexity of the activity-dependent plasticity and behavior. Beyond reconciling a decades-long debate (7, 11, 26, 27), our results demonstrate the importance of distinguishing specific cellular and molecular mechanisms because the corresponding cognitive perturbations might depend on whether lactate or glucose metabolism is perturbed.     

The spontaneous emergence of conventions: An experimental study of cultural evolution

How do shared conventions emerge in complex decentralized social systems? This question engages fields as diverse as linguistics, sociology, and cognitive science. Previous empirical attempts to solve this puzzle all presuppose that formal or informal institutions, such as incentives for global agreement, coordinated leadership, or aggregated information about the population, are needed to facilitate a solution. Evolutionary theories of social conventions, by contrast, hypothesize that such institutions are not necessary in order for social conventions to form. However, empirical tests of this hypothesis have been hindered by the difficulties of evaluating the real-time creation of new collective behaviors in large decentralized populations. Here, we present experimental results—replicated at several scales—that demonstrate the spontaneous creation of universally adopted social conventions and show how simple changes in a population’s network structure can direct the dynamics of norm formation, driving human populations with no ambition for large scale coordination to rapidly evolve shared social conventions.Social conventions are the foundation for social and economic life (1–7), However, it remains a central question in the social, behavioral, and cognitive sciences to understand how these patterns of collective behavior can emerge from seemingly arbitrary initial conditions (2–4, 8, 9). Large populations frequently manage to coordinate on shared conventions despite a continuously evolving stream of alternatives to choose from and no a priori differences in the expected value of the options (1, 3, 4, 10). For instance, populations are able to produce linguistic conventions on accepted names for children and pets (11), on common names for colors (12), and on popular terms for novel cultural artifacts, such as referring to junk email as “SPAM” (13, 14). Similarly, economic conventions, such as bartering systems (2), beliefs about fairness (3), and consensus regarding the exchangeability of goods and services (15), emerge with clear and widespread agreement within economic communities yet vary broadly across them (3, 16).Prominent theories of social conventions suggest that institutional mechanisms—such as centralized authority (14), incentives for collective agreement (15), social leadership (16), or aggregated information (17)—can explain global coordination. However, these theories do not explain whether, or how, it is possible for conventions to emerge when social institutions are not already in place to guide the process. A compelling alternative approach comes from theories of social evolution (2, 18–20). Social evolutionary theories maintain that networks of locally interacting individuals can spontaneously self-organize to produce global coordination (21, 22). Although there is widespread interest in this approach to social norms (6, 7, 14, 18, 23–26), the complexity of the social process has prevented systematic empirical insight into the thesis that these local dynamics are sufficient to explain universally adopted conventions (27, 28).Several difficulties have limited prior empirical research in this area. The most notable of these limitations is scale. Although compelling experiments have successfully shown the creation of new social conventions in dyadic and small group interactions (29–31), the results in small group settings can be qualitatively different from the dynamics in larger groups (Model), indicating that small group experiments are insufficient for demonstrating whether or how new conventions endogenously form in larger populations (32, 33). Important progress on this issue has been made using network-based laboratory experiments on larger groups (15, 24). However, this research has been restricted to studying coordination among players presented with two or three options with known payoffs. Natural convention formation, by contrast, is significantly complicated by the capacity of individuals to continuously innovate, which endogenously expands the “ecology” of alternatives under evaluation (23, 29, 31). Moreover, prior experimental studies have typically assumed the existence of either an explicit reward for universal coordination (15) or a mechanism that aggregates and reports the collective state of the population (17, 24), which has made it impossible to evaluate the hypothesis that global coordination is the result of purely local incentives.More recently, data science approaches to studying norms have addressed many of these issues by analyzing behavior change in large online networks (34). However, these observational studies are limited by familiar problems of identification that arise from the inability to eliminate the confounding influences of institutional mechanisms. As a result, previous empirical research has been unable to identify the collective dynamics through which social conventions can spontaneously emerge (8, 34–36).We addressed these issues by adopting a web-based experimental approach. We studied the effects of social network structure on the spontaneous evolution of social conventions in populations without any resources to facilitate global coordination (9, 37). Participants in our study were rewarded for coordinating locally, however they had neither incentives nor information for achieving large scale agreement. Further, to eliminate any preexisting bias in the evolutionary process, we studied the emergence of arbitrary linguistic conventions, in which none of the options had any a priori value or advantage over the others (3, 23). In particular, we considered the prototypical problem of whether purely local interactions can trigger the emergence of a universal naming convention (38, 39).     

Brain system for mental orientation in space,time, and person

Orientation is a fundamental mental function that processes the relations between the behaving self to space (places), time (events), and person (people). Behavioral and neuroimaging studies have hinted at interrelations between processing of these three domains. To unravel the neurocognitive basis of orientation, we used high-resolution 7T functional MRI as 16 subjects compared their subjective distance to different places, events, or people. Analysis at the individual-subject level revealed cortical activation related to orientation in space, time, and person in a precisely localized set of structures in the precuneus, inferior parietal, and medial frontal cortex. Comparison of orientation domains revealed a consistent order of cortical activity inside the precuneus and inferior parietal lobes, with space orientation activating posterior regions, followed anteriorly by person and then time. Core regions at the precuneus and inferior parietal lobe were activated for multiple orientation domains, suggesting also common processing for orientation across domains. The medial prefrontal cortex showed a posterior activation for time and anterior for person. Finally, the default-mode network, identified in a separate resting-state scan, was active for all orientation domains and overlapped mostly with person-orientation regions. These findings suggest that mental orientation in space, time, and person is managed by a specific brain system with a highly ordered internal organization, closely related to the default-mode network.Orientation in space, time, and person is a fundamental cognitive function and the bedrock of neurological and psychiatric mental status examination (1, 2). Orientation is defined as the “tuning between the subject and the internal representation he forms of the corresponding public reference system”: that is, the external world (1). Although the representation of the external world by means of a cognitive map has been widely investigated (3–5), the way in which the self refers to this map has yet to be understood. Moreover, the behaving self refers not only to spatial landmarks but also to remembered or imagined events, or to people around, yielding a “cognitive mapping” of the time and person domains of the mental world (2, 6–9). However, it is still unknown whether mental orientation in space, time, and person relies on similar or distinct neurocognitive systems.Several lines of research support the idea that similar neurocognitive systems underlie orientation in these three domains. Behavioral studies indicate a common psychological metric for proximity estimations (“cognitive distance”) in space, time, and person (7); for example, manipulation of stimuli’s distance in one orientation domain affects the perceived distance in the other two domains (10, 11). Accordingly, a recent neuroimaging study mapped cognitive distance estimations in the three domains to a single region in the inferior parietal lobe (IPL) (12). However, other neuroimaging studies that investigated processing of places, events, and people separately have found activation in brain regions besides the IPL, including the precuneus and posterior cingulate cortices, medial prefrontal cortex (mPFC), and lateral frontal and temporal lobes (6, 13–29). Notably, these regions constitute a part of the default-mode network (DMN), a system involved in self-referential processes (24, 30–35). These findings suggest a common brain system for orientation across domains, possibly related to the DMN.Clinical observations in patients with disorientation in space, time, and person are less clear: on the one hand, clinical syndromes may involve disorientation in several domains simultaneously, and disorientation disorders in the three domains involve lesions in similar brain regions, usually overlapping with the DMN (1, 2). On the other hand, disorientation may be limited to one specific domain – space, time, or person (2, 36–40). In addition, patients with traumatic brain injury or after electro-convulsive therapy regain their orientation gradually, from personal to spatial and temporal orientation (41, 42) whereas patients with Alzheimer’s disease typically lose orientation in time first, then in place, and then in person, suggesting that partially separate systems underlie orientation in each domain.Here, we investigated the neurocognitive system underlying orientation in space, time, and person and its relation to the DMN. To this aim, we used a mental-orientation task, with individually tailored stimuli in the space (places), time (events), and person (people) domains. To gain high anatomical specificity, we used high-resolution 7-Tesla functional MRI (fMRI). To capitalize on the spatial acuity of the high-resolution fMRI, we applied a strategy of analyzing each subject individually in native space and combined the results to compare activations for the three domains. Finally, we compared our results to the DMN as identified in each individual subject by analysis of resting-state fMRI. We hypothesized that orientation across different domains relies on a shared “core” brain system, in close relation to the DMN, yet orientation in specific domains may involve additional specialized subsystems.     

The evolution of brain neuron numbers in amniotes

Reconstructing the evolution of brain information-processing capacity is paramount for understanding the rise of complex cognition. Comparative studies of brain evolution typically use brain size as a proxy. However, to get a less biased picture of the evolutionary paths leading to high cognitive power, we need to compare brains not by mass but by numbers of neurons, which are their basic computational units. This study reconstructs the evolution of brains across amniotes by directly analyzing neuron numbers by using the largest dataset of its kind and including essential data on reptiles. We show that reptiles have not only small brains relative to body size but also low neuronal densities, resulting in average neuron numbers over 20 times lower than those in birds and mammals of similar body size. Amniote brain evolution is characterized by the following four major shifts in neuron–brain scaling. The most dramatic increases in brain neurons occurred independently with the appearance of birds and mammals, resulting in convergent neuron scaling in the two endotherm lineages. The other two major increases in the number of neurons happened in core land birds and anthropoid primates, which are two groups known for their cognitive prowess. Interestingly, relative brain size is associated with relative neuronal cell density in reptiles, birds, and primates but not in other mammals. This has important implications for studies using relative brain size as a proxy when looking for evolutionary drivers of animal cognition.

The evolution of cognitive capacity or “intelligence” and its underlying neural substrate has been of long-standing interest to biologists. Great strides have been made in understanding the evolution of brain size in vertebrates, with studies analyzing data on thousands of species (1–3). Since larger animals have larger brains but are not necessarily smarter, most studies of cognitive evolution use relative brain size (corrected for body size), which is thought to reflect extra neurons beyond those needed for controlling the body (4). We now have a good idea where major changes in brain–body scaling happened within birds (2) and mammals (3), and it is also clear that both mammals and birds have relatively larger brains than nonavian sauropsids (hereafter referred to as reptiles), although this has been rarely formally quantified because data on reptilian brain sizes are scarce (5).However, we still lack a clear picture of the evolution of actual brain processing capacity. This is because the same increase in relative brain size can be reached by different evolutionary paths, not always involving actual brain enlargement, and might often result from selection on body size (3). Moreover, similarly sized brains of distantly related species can harbor substantially different numbers of neurons overall and in major brain parts (6, 7). These two caveats invalidate the very idea that we can estimate extra neurons and glean information about cognitive capacity from absolute or relative brain size alone.This capacity is better determined by the number of neurons in the brain or specific brain parts (although their relative importance is still debated), their connections, interneuronal distance, and axonal conduction velocity (8, 9). Unlike brain size, though, these measures are not readily available for a sufficient number of species to be of practical use. Nevertheless, thanks to methodological advances (10), neuronal scaling rules (the allometric relationship between brain mass and neuron numbers) have now been determined for eight high-level mammalian clades (6, 11–13) as well as for a limited sampling of birds (14, 15).To get the big picture of amniote brain evolution, we have to include data on nonavian reptiles. The deepest split in amniote evolution occurred between the synapsid lineage, leading to mammals, and the sauropsid lineage, including reptiles and birds. We cannot tell if similarities between birds and mammals are due to shared ancestry or convergent evolution without considering reptiles. Yet, the dearth of quantitative data on reptile brains is striking—brain mass is available for 183 species (5, 16), compared to thousands for birds and mammals, and neuron numbers are known for a mere 4 reptile species (17–19).Taken together, to understand the evolution of brain processing capacity in amniotes, we need to include nonavian reptiles, consider changes in both brain–body and neuron–brain scaling, and examine the allocation of neurons to different brain parts. In this study, we provide these much needed data and reconstruct the big picture of brain evolution in amniotes in terms of neuron numbers.     

Structural and functional brain rewiring clarifies preserved interhemispheric transfer in humans born without the corpus callosum

Why do humans born without the corpus callosum, the major interhemispheric commissure, lack the disconnection syndrome classically described in callosotomized patients? This paradox was discovered by Nobel laureate Roger Sperry in 1968, and has remained unsolved since then. To tackle the hypothesis that alternative neural pathways could explain this puzzle, we investigated patients with callosal dysgenesis using structural and functional neuroimaging, as well as neuropsychological assessments. We identified two anomalous white-matter tracts by deterministic and probabilistic tractography, and provide supporting resting-state functional neuroimaging and neuropsychological evidence for their functional role in preserved interhemispheric transfer of complex tactile information, such as object recognition. These compensatory pathways connect the homotopic posterior parietal cortical areas (Brodmann areas 39 and surroundings) via the posterior and anterior commissures. We propose that anomalous brain circuitry of callosal dysgenesis is determined by long-distance plasticity, a set of hardware changes occurring in the developing brain after pathological interference. So far unknown, these pathological changes somehow divert growing axons away from the dorsal midline, creating alternative tracts through the ventral forebrain and the dorsal midbrain midline, with partial compensatory effects to the interhemispheric transfer of cortical function.The human brain connectome is sculpted out of intensive interaction between genes and environment during development (1). Subtle interference in this lengthy interactive process generates individual differences and peculiarities within the normal range of human variation (2). However, more drastic developmental disturbances impose changes that will greatly modify adult brain circuits, with adverse consequences for cognition and behavior (3). This is the case of callosal dysgenesis (CD), a condition well-known for generating gross morphological changes in the brain, and pronounced cognitive and behavioral consequences to the patients (4).CD differs greatly from callosotomy, however, because the latter is characterized by a clear disconnection syndrome (5), whereas the former displays a considerable degree of interhemispheric communication, despite absence of the corpus callosum (CC) (6, 7). The clinical presentation of CD is highly variable, ranging from subtle, subnormal, cognitive symptoms to severe mental retardation, epilepsy, and somatic deficits (4). Strikingly, however, all cases maintain some degree of interhemispheric transfer (8), possibly mediated by compensatory pathways (9) that would—at least partially—replace the role of the (absent or defective) CC. In fact, some evidence for white-matter circuit reorganization in CD has been produced (10–12), but no rewiring of any kind has been identified in adults subjected to callosal transection (13), although failure to interrupt interhemispheric transfer has been reported after callosotomy in humans and monkeys (14, 15). None of the anomalous tracts in CD, however, has been proven to explain the disconnection paradox observed in these patients, which has remained unsolved since its discovery by Sperry and his collaborators (6, 7).To approach this puzzle, we followed the hypothesis that the early infliction of the callosal defect would provide the developing brain with a set of anomalous connections that might preserve the crossed transfer of information between cortical areas, and thus assume—at least in part—the functions of the CC. We used multiple neuroimaging techniques and neuropsychological tests on subjects with total (agenesis) and partial CD, and compared them with normal controls. In addition to confirming the presence of the previously known long, aberrant tracts of the white matter, the Probst and the sigmoid bundles (10–12), we describe the trajectory of two so-far unreported homotopic interhemispheric tracts that course through the posterior or anterior commissures, and provide evidence suggestive of their functional role in enabling crossed-transfer of complex tactile function between the hemispheres of these subjects.     

Time-resolved resting-state brain networks

Neuronal dynamics display a complex spatiotemporal structure involving the precise, context-dependent coordination of activation patterns across a large number of spatially distributed regions. Functional magnetic resonance imaging (fMRI) has played a central role in demonstrating the nontrivial spatial and topological structure of these interactions, but thus far has been limited in its capacity to study their temporal evolution. Here, using high-resolution resting-state fMRI data obtained from the Human Connectome Project, we mapped time-resolved functional connectivity across the entire brain at a subsecond resolution with the aim of understanding how nonstationary fluctuations in pairwise interactions between regions relate to large-scale topological properties of the human brain. We report evidence for a consistent set of functional connections that show pronounced fluctuations in their strength over time. The most dynamic connections are intermodular, linking elements from topologically separable subsystems, and localize to known hubs of default mode and fronto-parietal systems. We found that spatially distributed regions spontaneously increased, for brief intervals, the efficiency with which they can transfer information, producing temporary, globally efficient network states. Our findings suggest that brain dynamics give rise to variations in complex network properties over time, possibly achieving a balance between efficient information-processing and metabolic expenditure.The coordination of brain activity between disparate neural populations is a dynamic and context-dependent process (1–3). Although dynamic patterns of neural synchronization may be evident in time-dependent measures of functional connectivity (4, 5), the temporal stability of high-level topological properties is unknown. The topology of large-scale cortical activity—such as its efficient network layout (6), community structure (7), network hubs (8), rich-club organization (9, 10), and small worldness (11, 12)—may reflect fundamental aspects of cortical computation. Temporal fluctuations in these graph-theoretic measures may hence speak to adaptive properties of neuronal information processing.With international connectome mapping consortia such as the Human Connectome Project (HCP) (13) and the developing Human Connectome Project in full swing, resting-state functional magnetic resonance imaging (rsfMRI) data of unprecedented temporal resolution are now available to map the time-resolved properties of functional brain networks. Imaging the brain at rest reveals spontaneous low-frequency fluctuations in brain activity that are temporally correlated between functionally related regions (14–17). Interregional correlations are referred to as functional connections, and they collectively form a complex network (18).Functional brain networks are typically mapped in a time-averaged sense, based on the assumption that functional connections remain relatively static (stationary) in the resting brain. However, recent investigations have furnished compelling evidence challenging the “static” conceptualization of resting-state functional connectivity (5). In particular, the application of time-resolved methodologies for analyzing time series data has consistently revealed fluctuations in resting-state functional connectivity at timescales ranging from tens of seconds to a few minutes (19–24). Furthermore, the modular organization of functional brain networks appears to be time-dependent in the resting state (25, 26) and modulated by learning (27) and cognitive effort (28, 29). It is therefore apparent that reducing fluctuations in functional connectivity to time averages has led to a very useful but static and possibly oversimplified characterization of the brain’s functional networks. For example, connections that toggle between correlated and anticorrelated states are reduced to zero in a time-averaged sense, assuming equal dwell times in each state.Conventionally, rsfMRI data are sampled at a resolution of 2 s or slower. Using multiband accelerated echo planar imaging, the HCP has acquired high-quality rsfMRI data at a subsecond resolution (30). This order of magnitude improvement in temporal resolution is highly advantageous to the feasibility of time-resolved functional connectomics. Faster sampling rates enable a richer temporal characterization of resting-state fluctuations, denser sampling of physiological confounds, and greater degrees of freedom (30).Using a sliding-window approach applied to HCP rsfMRI data, we mapped the evolution of functional brain networks over a continuous 15-min interval at a subsecond resolution. For each of 10 individuals, this yielded a time series of correlation matrices (regions × regions × time), where matrix elements quantified the functional connectivity at a given time instant between cortical and subcortical regions comprising established brain parcellation atlases. We developed a statistic to test the time-resolved connectome for evidence of nonstationary temporal dynamics and applied it to the 10 individuals as well as a replication data set and simulated rsfMRI data.Our main aim was to investigate the consequences of nonstationary fluctuations on the topological organization of functional brain networks. We hypothesized that dynamic behavior is coordinated across the brain so that transitions between distinct states are marked by reorganization of the brain’s functional topology. Evidence for this hypothesis is provided by the coordinated fluctuations in network measures, such as hub centrality (31), that have been observed in simulated rsfMRI data (32, 33).     

Single-neuron firing cascades underlie global spontaneous brain events

The resting brain consumes enormous energy and shows highly organized spontaneous activity. To investigate how this activity is manifest among single neurons, we analyzed spiking discharges of ∼10,000 isolated cells recorded from multiple cortical and subcortical regions of the mouse brain during immobile rest. We found that firing of a significant proportion (∼70%) of neurons conformed to a ubiquitous, temporally sequenced cascade of spiking that was synchronized with global events and elapsed over timescales of 5 to 10 s. Across the brain, two intermixed populations of neurons supported orthogonal cascades. The relative phases of these cascades determined, at each moment, the response magnitude evoked by an external visual stimulus. Furthermore, the spiking of individual neurons embedded in these cascades was time locked to physiological indicators of arousal, including local field potential power, pupil diameter, and hippocampal ripples. These findings demonstrate that the large-scale coordination of low-frequency spontaneous activity, which is commonly observed in brain imaging and linked to arousal, sensory processing, and memory, is underpinned by sequential, large-scale temporal cascades of neuronal spiking across the brain.

The brain at rest exhibits slow (<0.1 Hz) but highly organized spontaneous activity as measured by functional MRI (fMRI) (1, 2). Much research in this area has utilized the temporal coordination of these signals to assess the functional organization a large number of brain networks. In recent years, however, new attention has been directed to a less-studied aspect of this signal, namely the conspicuous and discrete spontaneous events that occur simultaneously across the brain (3–5). These global resting-state fMRI events appear to reflect transient arousal modulations at a timescale of ∼10 s (4, 6) and also to be closely related to activity among clusters of cholinergic projection neurons in the basal forebrain (4, 5).The nature of global brain events is of great interest, as is their spatiotemporal dynamics. Some evidence suggests they take the form of traveling waves, propagating coherently according to the principles of the cortical hierarchy (7, 8), and shaping functional connectivity measures important for assessing the healthy and diseased brain (7, 9, 10). Other work has linked such global activity to phenomena as varied as modulation of the autonomic nervous system (11–14), cleansing circulation of cerebrospinal fluid in the glymphatic system (15–19), and memory consolidation mediated by hippocampal sharp-wave ripples (20, 21). In general, the global activity measured through brain imaging appears coordinated over timescales of seconds with a range of other neural and physiological events (11, 12, 14, 21–23). In a few cases, the relationship between local and global neural events has been studied using simultaneous measurements. For example, brain-wide fMRI fluctuations and local field potential (LFP) power changes are locked to the issuance of hippocampal ripples (21, 24). However, very little is understood about the extent to which single neurons participate in the expression and coordination of global spontaneous events of the seconds timescale. To approach this topic, recent technological advances have made it possible to track and compare the spiking activity of a large number of isolated neurons recorded simultaneously across multiple brain areas.A few recent studies utilizing high-density neuronal recording (25) have accumulated initial evidence suggesting a close relationship between the brain state and neuronal population dynamics (26–28). A large proportion of neurons, regardless of their location, showed strong modulations in their discharging rate that were coordinated in time with physiological arousal measures (28), across thirsty and sated states (26), and during exploratory and nonexploratory behaviors (27). Nevertheless, these studies leave open the question of how neuronal population dynamics are organized at a finer timescale of seconds surrounding spontaneous global events during immobile rest, and whether and how such dynamics are coincident with arousal modulations, hippocampal ripples, and sensory excitability. To investigate this topic, we examine the spiking activity recorded from large neuronal populations of neurons in immobilized mice, focusing on their seconds-scale coordination with global events and with one another. We further studied the impact of this spontaneous spiking on the magnitude of visually evoked responses and its time locking with other physiological signals related to arousal, such as LFP changes, hippocampal ripples, and changes in pupil diameter.