Early path dominance as a principle for neurodevelopment.

We perform targeted attack, a systematic computational unlinking of the network, to analyze its effects on global communication across the brain network through its giant cluster. Across diffusion magnetic resonance images from individuals in the UK Biobank, Adolescent Brain Cognitive Development Study and Developing Human Connectome Project, we find that targeted attack procedures on increasing white matter tract lengths and densities are remarkably invariant to aging and disease. Time-reversing the attack computation suggests a mechanism for how brains develop, for which we derive an analytical equation using percolation theory. Based on a close match between theory and experiment, our results demonstrate that tracts are limited to emanate from regions already in the giant cluster and tracts that appear earliest in neurodevelopment are those that become the longest and densest.

Achieving Occam's razor: Deep learning for optimal model reduction.

All fields of science depend on mathematical models. Occam's razor refers to the principle that good models should exclude parameters beyond those minimally required to describe the systems they represent. This is because redundancy can lead to incorrect estimates of model parameters from data, and thus inaccurate or ambiguous conclusions. Here, we show how deep learning can be powerfully leveraged to apply Occam's razor to model parameters. Our method, FixFit, uses a feedforward deep neural network with a bottleneck layer to characterize and predict the behavior of a given model from its input parameters. FixFit has three major benefits. First, it provides a metric to quantify the original model's degree of complexity. Second, it allows for the unique fitting of data. Third, it provides an unbiased way to discriminate between experimental hypotheses that add value versus those that do not. In three use cases, we demonstrate the broad applicability of this method across scientific domains. To validate the method using a known system, we apply FixFit to recover known composite parameters for the Kepler orbit model and a dynamic model of blood glucose regulation. In the latter, we demonstrate the ability to fit the latent parameters to real data. To illustrate how the method can be applied to less well-established fields, we use it to identify parameters for a multi-scale brain model and reduce the search space for viable candidate mechanisms.

Metabolism modulates network synchrony in the aging brain.

Brain aging is associated with hypometabolism and global changes in functional connectivity. Using functional MRI (fMRI), we show that network synchrony, a collective property of brain activity, decreases with age. Applying quantitative methods from statistical physics, we provide a generative (Ising) model for these changes as a function of the average communication strength between brain regions. We find that older brains are closer to a critical point of this communication strength, in which even small changes in metabolism lead to abrupt changes in network synchrony. Finally, by experimentally modulating metabolic activity in younger adults, we show how metabolism alone-independent of other changes associated with aging-can provide a plausible candidate mechanism for marked reorganization of brain network topology.

Diet modulates brain network stability, a biomarker for brain aging, in young adults.

Epidemiological studies suggest that insulin resistance accelerates progression of age-based cognitive impairment, which neuroimaging has linked to brain glucose hypometabolism. As cellular inputs, ketones increase Gibbs free energy change for ATP by 27% compared to glucose. Here we test whether dietary changes are capable of modulating sustained functional communication between brain regions (network stability) by changing their predominant dietary fuel from glucose to ketones. We first established network stability as a biomarker for brain aging using two large-scale (n = 292, ages 20 to 85 y; n = 636, ages 18 to 88 y) 3 T functional MRI (fMRI) datasets. To determine whether diet can influence brain network stability, we additionally scanned 42 adults, age < 50 y, using ultrahigh-field (7 T) ultrafast (802 ms) fMRI optimized for single-participant-level detection sensitivity. One cohort was scanned under standard diet, overnight fasting, and ketogenic diet conditions. To isolate the impact of fuel type, an independent overnight fasted cohort was scanned before and after administration of a calorie-matched glucose and exogenous ketone ester (d-ß-hydroxybutyrate) bolus. Across the life span, brain network destabilization correlated with decreased brain activity and cognitive acuity. Effects emerged at 47 y, with the most rapid degeneration occurring at 60 y. Networks were destabilized by glucose and stabilized by ketones, irrespective of whether ketosis was achieved with a ketogenic diet or exogenous ketone ester. Together, our results suggest that brain network destabilization may reflect early signs of hypometabolism, associated with dementia. Dietary interventions resulting in ketone utilization increase available energy and thus may show potential in protecting the aging brain.

Ion gradient-driven bifurcations of a multi-scale neuronal model.

Metabolic limitations within the brain frequently arise in the context of aging and disease. As the largest consumers of energy within the brain, ion pumps that maintain the neuronal membrane potential are the most affected when energy supply becomes limited. To characterize the effects of such limitations, we analyze the ion gradients present in a conductance-based (Morris-Lecar) neural mass model. We show the existence and locations of Neimark-Sacker and period-doubling bifurcations in the sodium, calcium, and potassium reversal potentials and demonstrate that these bifurcations form physiologically relevant bounds of ion gradient variability. Within these bounds, we show how depolarization of the gradients causes decreased neural activity. We also show that the depolarization of ion gradients decreases inter-regional coherence, causing a shift in the critical point at which the coupling occurs and thereby inducing loss of synchrony between regions. In this way, we show that the Larter-Breakspear model captures ion gradient variability present at the microscale level and propagates these changes to the macroscale effects such as those observed in human neuroimaging studies.

Ground-truth "resting-state" signal provides data-driven estimation and correction for scanner distortion of fMRI time-series dynamics.

The fMRI community has made great strides in decoupling neuronal activity from other physiologically induced T2* changes, using sensors that provide a ground-truth with respect to cardiac, respiratory, and head movement dynamics. However, blood oxygenation level-dependent (BOLD) time-series dynamics are also confounded by scanner artifacts, in complex ways that can vary not only between scanners but even, for the same scanner, between sessions. Unfortunately, the lack of an equivalent ground truth for BOLD time-series has thus far stymied the development of reliable methods for identification and removal of scanner-induced noise, a problem that we have previously shown to severely impact detection sensitivity of resting-state brain networks. To address this problem, we first designed and built a phantom capable of providing dynamic signals equivalent to that of the resting-state brain. Using the dynamic phantom, we then compared the ground-truth time-series with its measured fMRI data. Using these, we introduce data-quality metrics: Standardized Signal-to-Noise Ratio (ST-SNR) and Dynamic Fidelity that, unlike currently used measures such as temporal SNR (tSNR), can be directly compared across scanners. Dynamic phantom data acquired from four "best-case" scenarios: high-performance scanners with MR-physicist-optimized acquisition protocols, still showed scanner instability/multiplicative noise contributions of about 6-18% of the total noise. We further measured strong non-linearity in the fMRI response for all scanners, ranging between 8-19% of total voxels. To correct scanner distortion of fMRI time-series dynamics at a single-subject level, we trained a convolutional neural network (CNN) on paired sets of measured vs. ground-truth data. The CNN learned the unique features of each session's noise, providing a customized temporal filter. Tests on dynamic phantom time-series showed a 4- to 7-fold increase in ST-SNR and about 40-70% increase in Dynamic Fidelity after denoising, with CNN denoising outperforming both the temporal bandpass filtering and denoising using Marchenko-Pastur principal component analysis. Critically, we observed that the CNN temporal denoising pushes ST-SNR to a regime where signal power is higher than that of noise (ST-SNR > 1). Denoising human-data with ground-truth-trained CNN, in turn, showed markedly increased detection sensitivity of resting-state networks. These were visible even at the level of the single-subject, as required for clinical applications of fMRI.

The Refractory Period Matters: Unifying Mechanisms of Macroscopic Brain Waves.

The relationship between complex brain oscillations and the dynamics of individual neurons is poorly understood. Here we utilize maximum caliber, a dynamical inference principle, to build a minimal yet general model of the collective (mean field) dynamics of large populations of neurons. In agreement with previous experimental observations, we describe a simple, testable mechanism, involving only a single type of neuron, by which many of these complex oscillatory patterns may emerge. Our model predicts that the refractory period of neurons, which has often been neglected, is essential for these behaviors.

Inferring a network from dynamical signals at its nodes.

We give an approximate solution to the difficult inverse problem of inferring the topology of an unknown network from given time-dependent signals at the nodes. For example, we measure signals from individual neurons in the brain, and infer how they are inter-connected. We use Maximum Caliber as an inference principle. The combinatorial challenge of high-dimensional data is handled using two different approximations to the pairwise couplings. We show two proofs of principle: in a nonlinear genetic toggle switch circuit, and in a toy neural network.

Making Sense of Computational Psychiatry.

In psychiatry we often speak of constructing "models." Here we try to make sense of what such a claim might mean, starting with the most fundamental question: "What is (and isn't) a model?" We then discuss, in a concrete measurable sense, what it means for a model to be useful. In so doing, we first identify the added value that a computational model can provide in the context of accuracy and power. We then present limitations of standard statistical methods and provide suggestions for how we can expand the explanatory power of our analyses by reconceptualizing statistical models as dynamical systems. Finally, we address the problem of model building-suggesting ways in which computational psychiatry can escape the potential for cognitive biases imposed by classical hypothesis-driven research, exploiting deep systems-level information contained within neuroimaging data to advance our understanding of psychiatric neuroscience.

Oxytocin attenuates trust as a subset of more general reinforcement learning, with altered reward circuit functional connectivity in males.

Oxytocin (OT) is an endogenous neuropeptide that, while originally thought to promote trust, has more recently been found to be context-dependent. Here we extend experimental paradigms previously restricted to de novo decision-to-trust, to a more realistic environment in which social relationships evolve in response to iterative feedback over twenty interactions. In a randomized, double blind, placebo-controlled within-subject/crossover experiment of human adult males, we investigated the effects of a single dose of intranasal OT (40 IU) on Bayesian expectation updating and reinforcement learning within a social context, with associated brain circuit dynamics. Subjects participated in a neuroeconomic task (Iterative Trust Game) designed to probe iterative social learning while their brains were scanned using ultra-high field (7T) fMRI. We modeled each subject's behavior using Bayesian updating of belief-states ("willingness to trust") as well as canonical measures of reinforcement learning (learning rate, inverse temperature). Behavioral trajectories were then used as regressors within fMRI activation and connectivity analyses to identify corresponding brain network functionality affected by OT. Behaviorally, OT reduced feedback learning, without bias with respect to positive versus negative reward. Neurobiologically, reduced learning under OT was associated with muted communication between three key nodes within the reward circuit: the orbitofrontal cortex, amygdala, and lateral (limbic) habenula. Our data suggest that OT, rather than inspiring feelings of generosity, instead attenuates the brain's encoding of prediction error and therefore its ability to modulate pre-existing beliefs. This effect may underlie OT's putative role in promoting what has typically been reported as 'unjustified trust' in the face of information that suggests likely betrayal, while also resolving apparent contradictions with regard to OT's context-dependent behavioral effects.

Abnormal hippocampal structure and function in clinical anxiety and comorbid depression.

Given the high prevalence rates of comorbidity of anxiety and depressive disorders, identifying a common neural pathway to both disorders is important not only for better diagnosis and treatment, but also for a more complete conceptualization of each disease. Hippocampal abnormalities have been implicated in anxiety and depression, separately; however, it remains unknown whether these abnormalities are also implicated in their comorbidity. Here we address this question by testing 32 adults with generalized anxiety disorder (15 GAD only and 17 comorbid MDD) and 25 healthy controls (HC) using multimodal MRI (structure, diffusion and functional) and automated hippocampal segmentation. We demonstrate that (i) abnormal microstructure of the CA1 and CA2-3 is associated with GAD/MDD comorbidity and (ii) decreased anterior hippocampal reactivity in response to repetition of the threat cue is associated with GAD (with or without MDD comorbidity). In addition, mediation-structural equation modeling (SEM) reveals that our hippocampal and dimensional symptom data are best explained by a model describing a significant influence of abnormal hippocampal microstructure on both anxiety and depression-mediated through its impact on abnormal hippocampal threat processing. Collectively, our findings show a strong association between changes in hippocampal microstructure and threat processing, which together may present a common neural pathway to comorbidity of anxiety and depression.

Acute psychological stress induces short-term variable immune response.

In spite of advances in understanding the cross-talk between the peripheral immune system and the brain, the molecular mechanisms underlying the rapid adaptation of the immune system to an acute psychological stressor remain largely unknown. Conventional approaches to classify molecular factors mediating these responses have targeted relatively few biological measurements or explored cross-sectional study designs, and therefore have restricted characterization of stress-immune interactions. This exploratory study analyzed transcriptional profiles and flow cytometric data of peripheral blood leukocytes with physiological (endocrine, autonomic) measurements collected throughout the sequence of events leading up to, during, and after short-term exposure to physical danger in humans. Immediate immunomodulation to acute psychological stress was defined as a short-term selective up-regulation of natural killer (NK) cell-associated cytotoxic and IL-12 mediated signaling genes that correlated with increased cortisol, catecholamines and NK cells into the periphery. In parallel, we observed down-regulation of innate immune toll-like receptor genes and genes of the MyD88-dependent signaling pathway. Correcting gene expression for an influx of NK cells revealed a molecular signature specific to the adrenal cortex. Subsequently, focusing analyses on discrete groups of coordinately expressed genes (modules) throughout the time-series revealed immune stress responses in modules associated to immune/defense response, response to wounding, cytokine production, TCR signaling and NK cell cytotoxicity which differed between males and females. These results offer a spring-board for future research towards improved treatment of stress-related disease including the impact of stress on cardiovascular and autoimmune disorders, and identifies an immune mechanism by which vulnerabilities to these diseases may be gender-specific.

Hyper-reactive human ventral tegmental area and aberrant mesocorticolimbic connectivity in overgeneralization of fear in generalized anxiety disorder.

The ventral tegmental area (VTA) has been primarily implicated in reward-motivated behavior. Recently, aberrant dopaminergic VTA signaling has also been implicated in anxiety-like behaviors in animal models. These findings, however, have yet to be extended to anxiety in humans. Here we hypothesized that clinical anxiety is linked to dysfunction of the mesocorticolimbic circuit during threat processing in humans; specifically, excessive or dysregulated activity of the mesocorticolimbic aversion circuit may be etiologically related to errors in distinguishing cues of threat versus safety, also known as "overgeneralization of fear." To test this, we recruited 32 females with generalized anxiety disorder and 25 age-matched healthy control females. We measured brain activity using fMRI while participants underwent a fear generalization task consisting of pseudo-randomly presented rectangles with systematically varying widths. A mid-sized rectangle served as a conditioned stimulus (CS; 50% electric shock probability) and rectangles with widths of CS ±20%, ±40%, and ±60% served as generalization stimuli (GS; never paired with electric shock). Healthy controls showed VTA reactivity proportional to the cue's perceptual similarity to CS (threat). In contrast, patients with generalized anxiety disorder showed heightened and less discriminating VTA reactivity to GS, a feature that was positively correlated with trait anxiety, as well as increased mesocortical and decreased mesohippocampal coupling. Our results suggest that the human VTA and the mesocorticolimbic system play a crucial role in threat processing, and that abnormalities in this system are implicated in maladaptive threat processing in clinical anxiety.

Circuit-wide structural and functional measures predict ventromedial prefrontal cortex fear generalization: implications for generalized anxiety disorder.

The ventromedial prefrontal cortex (vmPFC) plays a critical role in a number of evaluative processes, including risk assessment. Impaired discrimination between threat and safety is considered a hallmark of clinical anxiety. Here, we investigated the circuit-wide structural and functional mechanisms underlying vmPFC threat-safety assessment in humans. We tested patients with generalized anxiety disorder (GAD; n = 32, female) and healthy controls (n = 25, age-matched female) on a task that assessed the generalization of conditioned threat during fMRI scanning. The task consisted of seven rectangles of graded widths presented on a screen; only the midsize one was paired with mild electric shock [conditioned stimulus (CS)], while the others, safety cues, systematically varied in width by ±20, 40, and 60% [generalization stimuli (GS)] compared with the CS. We derived an index reflecting vmPFC functioning from the BOLD reactivity on a continuum of threat (CS) to safety (GS least similar to CS); patients with GAD showed less discrimination between threat and safety cues, compared with healthy controls (Greenberg et al., 2013b). Using structural, functional (i.e., resting-state), and diffusion MRI, we measured vmPFC thickness, vmPFC functional connectivity, and vmPFC structural connectivity within the corticolimbic systems. The results demonstrate that all three factors predict individual variability of vmPFC threat assessment in an independent fashion. Moreover, these neural features are also linked to GAD, most likely via an vmPFC fear generalization. Our results strongly suggest that vmPFC threat processing is closely associated with broader corticolimbic circuit anomalies, which may synergistically contribute to clinical anxiety.

Using network dynamic fMRI for detection of epileptogenic foci.

BACKGROUND: Epilepsy is one of the most prevalent neurological disorders. It remains medically intractable for about one-third of patients with focal epilepsy, for whom precise localization of the epileptogenic zone responsible for seizure initiation may be critical for successful surgery. Existing fMRI literature points to widespread network disturbances in functional connectivity. Per previous scalp and intracranial EEG studies and consistent with excessive local synchronization during interictal discharges, we hypothesized that, relative to same regions in healthy controls, epileptogenic foci would exhibit less chaotic dynamics, identifiable via entropic analyses of resting state fMRI time series. METHODS: In order to first validate this hypothesis on a cohort of patients with known ground truth, here we test individuals with well-defined epileptogenic foci (left mesial temporal lobe epilepsy). We analyzed voxel-wise resting-state fMRI time-series using the autocorrelation function (ACF), an entropic measure of regulation and feedback, and performed follow-up seed-to-voxel functional connectivity analysis. Disruptions in connectivity of the region exhibiting abnormal dynamics were examined in relation to duration of epilepsy and patients' cognitive performance using a delayed verbal memory recall task. RESULTS: ACF analysis revealed constrained (less chaotic) functional dynamics in left temporal lobe epilepsy patients, primarily localized to ipsilateral temporal pole, proximal to presumed focal points. Autocorrelation decay rates differentiated, with 100 % accuracy, between patients and healthy controls on a subject-by-subject basis within a leave-one-subject out classification framework. Regions identified via ACF analysis formed a less efficient network in patients, as compared to controls. Constrained dynamics were linked with locally increased and long-range decreased connectivity that, in turn, correlated significantly with impaired memory (local left temporal connectivity) and epilepsy duration (left temporal - posterior cingulate cortex connectivity). CONCLUSIONS: Our current results suggest that data driven functional MRI methods that target network dynamics hold promise in providing clinically valuable tools for identification of epileptic regions.

Influence of the BDNF genotype on amygdalo-prefrontal white matter microstructure is linked to nonconscious attention bias to threat.

Cognitive processing biases, such as increased attention to threat, are gaining recognition as causal factors in anxiety. Yet, little is known about the anatomical pathway by which threat biases cognition and how genetic factors might influence the integrity of this pathway, and thus, behavior. For 40 normative adults, we reconstructed the entire amygdalo-prefrontal white matter tract (uncinate fasciculus) using diffusion tensor weighted MRI and probabilistic tractography to test the hypothesis that greater fiber integrity correlates with greater nonconscious attention bias to threat as measured by a backward masked dot-probe task. We used path analysis to investigate the relationship between brain-derived nerve growth factor genotype, uncinate fasciculus integrity, and attention bias behavior. Greater structural integrity of the amygdalo-prefrontal tract correlates with facilitated attention bias to nonconscious threat. Genetic variability associated with brain-derived nerve growth factor appears to influence the microstructure of this pathway and, in turn, attention bias to nonconscious threat. These results suggest that the integrity of amygdalo-prefrontal projections underlie nonconscious attention bias to threat and mediate genetic influence on attention bias behavior. Prefrontal cognition and attentional processing in high bias individuals appear to be heavily influenced by nonconscious threat signals relayed via the uncinate fasciculus.

Network connectivity modulates power spectrum scale invariance.

Measures of complexity are sensitive in detecting disease, which has made them attractive candidates for diagnostic biomarkers; one complexity measure that has shown promise in fMRI is power spectrum scale invariance (PSSI). Even if scale-free features of neuroimaging turn out to be diagnostically useful, however, their underlying neurobiological basis is poorly understood. Using modeling and simulations of a schematic prefrontal-limbic meso-circuit, with excitatory and inhibitory networks of nodes, we present here a framework for how network density within a control system can affect the complexity of signal outputs. Our model demonstrates that scale-free behavior, similar to that observed in fMRI PSSI data, can be obtained for sufficiently large networks in a context as simple as a linear stochastic system of differential equations, although the scale-free range improves when introducing more realistic, nonlinear behavior in the system. PSSI values (reflective of complexity) vary as a function of both input type (excitatory, inhibitory) and input density (mean number of long-range connections, or strength), independent of their node-specific geometric distribution. Signals show pink noise (1/f) behavior when excitatory and inhibitory influences are balanced. As excitatory inputs are increased and decreased, signals shift towards white and brown noise, respectively. As inhibitory inputs are increased and decreased, signals shift towards brown and white noise, respectively. The results hold qualitatively at the hemodynamic scale, which we modeled by introducing a neurovascular component. Comparing hemodynamic simulation results to fMRI PSSI results from 96 individuals across a wide spectrum of anxiety-levels, we show how our model can generate concrete and testable hypotheses for understanding how connectivity affects regulation of meso-circuits in the brain.

Small-world network properties in prefrontal cortex correlate with predictors of psychopathology risk in young children: a NIRS study.

Near infrared spectroscopy (NIRS) is an emerging imaging technique that is relatively inexpensive, portable, and particularly well suited for collecting data in ecological settings. Therefore, it holds promise as a potential neurodiagnostic for young children. We set out to explore whether NIRS could be utilized in assessing the risk of developmental psychopathology in young children. A growing body of work indicates that temperament at young age is associated with vulnerability to psychopathology later on in life. In particular, it has been shown that low effortful control (EC), which includes the focusing and shifting of attention, inhibitory control, perceptual sensitivity, and a low threshold for pleasure, is linked to conditions such as anxiety, depression and attention deficit hyperactivity disorder (ADHD). Physiologically, EC has been linked to a control network spanning among other sites the prefrontal cortex. Several psychopathologies, such as depression and ADHD, have been shown to result in compromised small-world network properties. Therefore we set out to explore the relationship between EC and the small-world properties of PFC using NIRS. NIRS data were collected from 44 toddlers, ages 3-5, while watching naturalistic stimuli (movie clips). Derived complex network measures were then correlated to EC as derived from the Children's Behavior Questionnaire (CBQ). We found that reduced levels of EC were associated with compromised small-world properties of the prefrontal network. Our results suggest that the longitudinal NIRS studies of complex network properties in young children hold promise in furthering our understanding of developmental psychopathology.

Challenges and Frontiers in Computational Metabolic Psychiatry.

One of the primary challenges in metabolic psychiatry is that the disrupted brain functions that underlie psychiatric conditions arise from a complex set of downstream and feedback processes spanning across multiple spatiotemporal scales. Importantly, the same circuit can have multiple points of failure, each of which results in a different type of dysregulation, and thus elicits distinct cascades downstream that produce divergent signs and symptoms. Here, we illustrate this challenge by examining how subtle differences in circuit perturbations can lead to divergent clinical outcomes. We also discuss how computational models can perform the spatially heterogenous integration and bridge in vitro and in vivo paradigms. By leveraging recent methodological advances and tools, computational models can integrate relevant processes across scales (e.g., TCA-cycle, ion channel, neural microassembly, whole-brain macro-circuit) and across physiological systems (e.g., neural, endocrine, immune, vascular), providing a framework that can unite these mechanistic processes in a manner that goes beyond the conceptual and descriptive, to the quantitative and generative. These hold the potential to sharpen our intuitions towards circuit-based models for personalized diagnostics and treatment.

D-ß-hydroxybutyrate stabilizes hippocampal CA3-CA1 circuit during acute insulin resistance.

1.The brain primarily relies on glycolysis for mitochondrial respiration but switches to alternative fuels such as ketone bodies (KBs) when less glucose is available. Neuronal KB uptake, which does not rely on glucose transporter 4 (GLUT4) or insulin, has shown promising clinical applicability in alleviating the neurological and cognitive effects of disorders with hypometabolic components. However, the specific mechanisms by which such interventions affect neuronal functions are poorly understood. In this study, we pharmacologically blocked GLUT4 to investigate the effects of exogenous KB D-P-hydroxybutyrate (D-ßHb) on mouse brain metabolism during acute insulin resistance (AIR). We found that both AIR and D-ßHb had distinct impacts across neuronal compartments: AIR decreased synaptic activity and long-term potentiation (LTP) and impaired axonal conduction, synchronization, and action potential (AP) properties, while D- PHb rescued neuronal functions associated with axonal conduction, synchronization and LTP.