Neural Circuits Research Articles

Desegregation of neuronal predictive processing.

Neural circuits construct internal 'world-models' to guide behavior. The predictive processing framework posits that neural activity signaling sensory predictions and concurrently computing prediction-errors is a signature of those internal models. Here, to understand how the brain generates predictions for complex sensorimotor signals, we investigate the emergence of high-dimensional, multi-modal predictive representations in recurrent networks. We find that robust predictive processing arises in a network with loose excitatory/inhibitory balance. Contrary to previous proposals of functionally specialized cell-types, the network exhibits desegregation of stimulus and prediction-error representations. We confirmed these model predictions by experimentally probing predictive-coding circuits using a rich stimulus-set to violate learned expectations. When constrained by data, our model further reveals and makes concrete testable experimental predictions for the distinct functional roles of excitatory and inhibitory neurons, and of neurons in different layers along a laminar hierarchy, in computing multi-modal predictions. These results together imply that in natural conditions, neural representations of internal models are highly distributed, yet structured to allow flexible readout of behaviorally-relevant information. The generality of our model advances the understanding of computation of internal models across species, by incorporating different types of predictive computations into a unified framework.

Prefrontal neuronal dynamics in the absence of task execution

Prefrontal cortical activity represents stimuli in working memory tasks in a low-dimensional manifold that transforms over the course of a trial. Such transformations reflect specific cognitive operations, so that, for example, the rotation of stimulus representations is thought to reduce interference by distractor stimuli. Here we show that rotations occur in the low-dimensional activity space of prefrontal neurons in naïve male monkeys (Macaca mulatta), while passively viewing familiar stimuli. Moreover, some aspects of these rotations remain remarkably unchanged after training to perform working memory tasks. Significant training effects are still present in population dynamics, which further distinguish correct and error trials during task execution. Our results reveal automatic functions of prefrontal neural circuits allow transformations that may aid cognitive flexibility.

AROMHA Brain Health Test: A Remote Olfactory Assessment as a Screen for Cognitive Impairment.

Cost-effective, noninvasive screening methods for preclinical Alzheimer's disease (AD) and other neurocognitive disorders remain an unmet need. The olfactory neural circuits develop AD pathological changes prior to symptom onset. To probe these vulnerable circuits, we developed the digital remote AROMHA Brain Health Test (ABHT), an at-home odor identification, discrimination, memory, and intensity assessment. The ABHT was self-administered among cognitively normal (CN) English and Spanish speakers (n=127), participants with subjective cognitive complaints (SCC; n=34), and mild cognitive impairment (MCI; n=19). Self-administered tests took place remotely at home under unobserved (among interested CN participants) and observed modalities (CN, SCC, and MCI), as well as in-person with a research assistant present (CN, SCC, and MCI). Olfactory performance was similar across observed and unobserved remote self-administration and between English and Spanish speakers. Odor memory, identification, and discrimination scores decreased with age, and olfactory identification and discrimination were lower in the MCI group compared to CN and SCC groups, independent of age, sex, and education. The ABHT revealed age-related olfactory decline, and discriminated CN older adults from those with cognitive impairment. Replication of our results in other populations would support the use of the ABHT to identify and monitor individuals at risk for developing dementia.

Signatures of Bayesian inference emerge from energy-efficient synapses.

Biological synaptic transmission is unreliable, and this unreliability likely degrades neural circuit performance. While there are biophysical mechanisms that can increase reliability, for instance by increasing vesicle release probability, these mechanisms cost energy. We examined four such mechanisms along with the associated scaling of the energetic costs. We then embedded these energetic costs for reliability in artificial neural networks (ANNs) with trainable stochastic synapses, and trained these networks on standard image classification tasks. The resulting networks revealed a tradeoff between circuit performance and the energetic cost of synaptic reliability. Additionally, the optimised networks exhibited two testable predictions consistent with pre-existing experimental data. Specifically, synapses with lower variability tended to have (1) higher input firing rates and (2) lower learning rates. Surprisingly, these predictions also arise when synapse statistics are inferred through Bayesian inference. Indeed, we were able to find a formal, theoretical link between the performance-reliability cost tradeoff and Bayesian inference. This connection suggests two incompatible possibilities: evolution may have chanced upon a scheme for implementing Bayesian inference by optimising energy efficiency, or alternatively, energy-efficient synapses may display signatures of Bayesian inference without actually using Bayes to reason about uncertainty.

Signatures of Bayesian inference emerge from energy-efficient synapses

Biological synaptic transmission is unreliable, and this unreliability likely degrades neural circuit performance. While there are biophysical mechanisms that can increase reliability, for instance by increasing vesicle release probability, these mechanisms cost energy. We examined four such mechanisms along with the associated scaling of the energetic costs. We then embedded these energetic costs for reliability in artificial neural networks (ANNs) with trainable stochastic synapses, and trained these networks on standard image classification tasks. The resulting networks revealed a tradeoff between circuit performance and the energetic cost of synaptic reliability. Additionally, the optimised networks exhibited two testable predictions consistent with pre-existing experimental data. Specifically, synapses with lower variability tended to have (1) higher input firing rates and (2) lower learning rates. Surprisingly, these predictions also arise when synapse statistics are inferred through Bayesian inference. Indeed, we were able to find a formal, theoretical link between the performance-reliability cost tradeoff and Bayesian inference. This connection suggests two incompatible possibilities: evolution may have chanced upon a scheme for implementing Bayesian inference by optimising energy efficiency, or alternatively, energy-efficient synapses may display signatures of Bayesian inference without actually using Bayes to reason about uncertainty.

A therapeutic small molecule enhances γ-oscillations and improves cognition/memory in Alzheimer’s disease model mice

Brain rhythms provide the timing for recruitment of brain activity required for linking together neuronal ensembles engaged in specific tasks. The γ-oscillations (30 to 120 Hz) orchestrate neuronal circuits underlying cognitive processes and working memory. These oscillations are reduced in numerous neurological and psychiatric disorders, including early cognitive decline in Alzheimer's disease (AD). Here, we report on a potent brain-permeable small molecule, DDL-920 that increases γ-oscillations and improves cognition/memory in a mouse model of AD, thus showing promise as a class of therapeutics for AD. We employed anatomical, in vitro and in vivo electrophysiological, and behavioral methods to examine the effects of our lead therapeutic candidate small molecule. As a novel in central nervous system pharmacotherapy, our lead molecule acts as a potent, efficacious, and selective negative allosteric modulator of the γ-aminobutyric acid type A receptors most likely assembled from α1β2δ subunits. These receptors, identified through anatomical and pharmacological means, underlie the tonic inhibition of parvalbumin (PV) expressing interneurons (PV+INs) critically involved in the generation of γ-oscillations. When orally administered twice daily for 2 wk, DDL-920 restored the cognitive/memory impairments of 3- to 4-mo-old AD model mice as measured by their performance in the Barnes maze. Our approach is unique as it is meant to enhance cognitive performance and working memory in a state-dependent manner by engaging and amplifying the brain's endogenous γ-oscillations through enhancing the function of PV+INs.

Crystal structure identification with 3D convolutional neural networks with application to high-pressure phase transitions in SiO2

Efficient, reliable and easy-to-use structure recognition of atomic environments is essential for the analysis of atomic scale computer simulations. In this work, we train two neuronal network (NN) architectures, namely PointNet and dynamic graph convolutional NN (DG-CNN) using different hyperparameters and training regimes to assess their performance in structure identification tasks of atomistic structure data. We show benchmarks on simple crystal structures, where we can compare against established methods. The approach is subsequently extended to structurally more complex SiO2 phases. By making use of this structure recognition tool, we are able to achieve a deeper understanding of the crystallization process in amorphous SiO2 under shock compression. Lastly, we show how the NN based structure identification workflows can be integrated into OVITO using its python interface.

Real-time hardware emulation of neural cultures: A comparative study of in vitro, in silico and in duris silico models

Biological neural networks are well known for their capacity to process information with extremely low power consumption. Fields such as Artificial Intelligence, with high computational costs, are seeking for alternatives inspired in biological systems. An inspiring alternative is to implement hardware architectures that replicate the behavior of biological neurons but with the flexibility in programming capabilities of an electronic device, all combined with a relatively low operational cost. To advance in this quest, here we analyze the capacity of the HEENS hardware architecture to operate in a similar manner as an in vitro neuronal network grown in the laboratory. For that, we considered data of spontaneous activity in living neuronal cultures of about 400 neurons and compared their collective dynamics and functional behavior with those obtained from direct numerical simulations (in silico) and hardware implementations (in duris silico). The results show that HEENS is capable to mimic both the in vitro and in silico systems with high efficient-cost ratio, and on different network topological designs. Our work shows that compact low-cost hardware implementations are feasible, opening new avenues for future, highly efficient neuromorphic devices and advanced human–machine interfacing.

Exploring the globoid cell leukodystrophy protein network and therapeutic interventions

Globoid cell leukodystrophy is a severe rare disorder characterized by white matter degradation, resulting in a progressive loss of physical and mental abilities and has extremely limited therapeutic interventions. Therefore, this study aimed to delve into the Globoid cell leukodystrophy associated intricate network of differentially expressed genes (p < 0.05, |Fc|> 1) to identify potential druggable targets and possible therapeutic interventions using small molecules. The disease-associated neuronal protein circuit was constructed and analyzed, identifying 53 nodes (minimum edge cutoff 1), among which five (FOS, FOSB, GDNF, GFRA1, and JUN) were discerned as potential core protein nodes. Although our research enumerates the potential small molecules to target various protein nodes in the proposed disease network, we particularly underscore T-5224 to inhibit c-Jun activity as JUN was identified as one of the pivotal elements within the disease-associated neuronal protein circuit. The evaluation of T-5224 binding energy (− 11.0 kcal/mol) from docking study revealed that the compound to exhibit a notable affinity towards Jun/CRE complex. Moreover, the structural integrity of complex was affirmed through comprehensive molecular dynamics simulations, indicating a stable hydrophilic interaction between T-5224 and the Jun/CRE complex, thereby enhancing protein compactness and reducing solvent accessibility. This binding energy was further substantiated by free binding analysis, revealing a substantial thermodynamics complex state (− 448.00 ± 41.73 kJ/mol). Given that this investigation is confined to a computational framework, we additionally propose a hypothetical framework to ascertain the feasibility of inhibiting the Jun/CRE complex with T-5224 against Globoid cell leukodystrophy, employing a combination of in vitro and in vivo methodologies as a prospective avenue of this study.

Multiplexed volumetric CLEM enabled by scFvs provides insights into the cytology of cerebellar cortex

Mapping neuronal networks is a central focus in neuroscience. While volume electron microscopy (vEM) can reveal the fine structure of neuronal networks (connectomics), it does not provide molecular information to identify cell types or functions. We developed an approach that uses fluorescent single-chain variable fragments (scFvs) to perform multiplexed detergent-free immunolabeling and volumetric-correlated-light-and-electron-microscopy on the same sample. We generated eight fluorescent scFvs targeting brain markers. Six fluorescent probes were imaged in the cerebellum of a female mouse, using confocal microscopy with spectral unmixing, followed by vEM of the same sample. The results provide excellent ultrastructure superimposed with multiple fluorescence channels. Using this approach, we documented a poorly described cell type, two types of mossy fiber terminals, and the subcellular localization of one type of ion channel. Because scFvs can be derived from existing monoclonal antibodies, hundreds of such probes can be generated to enable molecular overlays for connectomic studies.

Huntingtin lowering impairs the maturation and synchronized synaptic activity of human cortical neuronal networks derived from induced pluripotent stem cells

Despite growing descriptions of wild-type Huntingtin (wt-HTT) roles in both adult brain function and, more recently, development, several clinical trials are exploring HTT-lowering approaches that target both wt-HTT and the mutant isoform (mut-HTT) responsible for Huntington's disease (HD). This non-selective targeting is based on the autosomal dominant inheritance of HD, supporting the idea that mut-HTT exerts its harmful effects through a toxic gain-of-function or a dominant-negative mechanism. However, the precise amount of wt-HTT needed for healthy neurons in adults and during development remains unclear. In this study, we address this question by examining how wt-HTT loss affects human neuronal network formation, synaptic maturation, and homeostasis in vitro. Our findings establish a role of wt-HTT in the maturation of dendritic arborization and the acquisition of network-wide synchronized activity by human cortical neuronal networks modeled in vitro. Interestingly, the network synchronization defects only became apparent when more than two-thirds of the wt-HTT protein was depleted. Our study underscores the critical need to precisely understand wt-HTT role in neuronal health. It also emphasizes the potential risks of excessive wt-HTT loss associated with non-selective therapeutic approaches targeting both wt- and mut-HTT isoforms in HD patients.

Changes in Neuroimmunological Synapses During Cerebral Ischemia.

The direct interplay between the immune and nervous systems is now well established. Within the brain, these interactions take place between neurons and resident glial cells, i.e., microglia and astrocytes, or infiltrating immune cells, influenced by systemic factors. A special form of physical cell-cell interactions is the so-called "neuroimmunological (NI) synapse." There is compelling evidence that the same signaling pathways that regulate inflammatory responses to injury or ischemia also play potent roles in brain development, plasticity, and function. Proper synaptic wiring is as important during development as it is during disease states, as it is necessary for activity-dependent refinement of neuronal circuits. Since the process of forming synaptic connections in the brain is highly dynamic, with constant changes in strength and connectivity, the immune component is perfectly suited for the regulatory task as it is in constant turnover. Many cellular and molecular players in this interaction remain to be uncovered, especially in pathological states. In this review, we discuss and propose possible communication hubs between components of the adaptive and innate immune systems and the synaptic element in ischemic stroke pathology.

Memristive Hodgkin–Huxley Neurons with Diverse Firing Patterns for High‐Order Neuromorphic Computing

The rich firing behaviors of biological neurons enable the nervous system to execute complex computations, emulating which in hardware is advantageous for constructing advanced intelligent machines. Hodgkin–Huxley (H–H) neurons based on memristors feature great merits of high bio‐plausibility and low hardware cost. However, a universal design rule of memristive H–H neurons is still lacking, hindering its development and applications. Herein, a universal H–H neuron circuit structure is proposed and its feasibility based on NbOx memristors is demonstrated. The constructed neuron achieves 23 types of firing behaviors observed in biological neurons, simplifying the communication between neurons. To better understand the correlation between circuit parameters and firing patterns, the firing patterns into three classes according to the switching cycle ratio of two memristors are categorized. The circuit design rules of each category of firing patterns are deeply elucidated and universal regularities for tuning circuit parameters to implement the switch between different firing behaviors are presented. Finally, the potential applications of different firing behaviors in neuromorphic intelligence systems are discussed. This work provides theoretical guidance for engineering memristive H–H neuron circuits, assisting in building high‐order neuromorphic systems based on firing patterns.

Neuromodulation with transcranial direct current stimulation contributes to motor function recovery via microglia in spinal cord injury

Spinal cord injury (SCI) is damage or trauma to the spinal cord, which often results in loss of function, sensation, or mobility below the injury site. Transcranial direct current stimulation (tDCS) is a non-invasive and affordable brain stimulation technique used to modulate neuronal circuits, which changes the morphology and activity of microglia in the cerebral cortex. However, whether similar morphological changes can be observed in the spinal cord remains unclear. Therefore, we evaluated neuronal population activity in layer 5 (L5) of M1 following SCI and investigated whether changes in the activities of L5 neurons affect microglia-axon interactions using C57BL/6J mice. We discovered that L5 of the primary motor cortex (corticospinal neurons) exhibited reduced synchronized activity after SCI that correlates with microglial morphology, which was recovered using tDCS. This indicates that tDCS promotes changes in the morphological properties and recovery of microglia after SCI. Combining immunotherapy with tDCS may be effective in treating SCI.

The role of functional emotion circuits in distinct dimensions of psychopathology in youth

Several mental disorders emerge during childhood or adolescence and are often characterized by socioemotional difficulties, including alterations in emotion perception. Emotional facial expressions are processed in discrete functional brain modules whose connectivity patterns encode emotion categories, but the involvement of these neural circuits in psychopathology in youth is poorly understood. This study examined the associations between activation and functional connectivity patterns in emotion circuits and psychopathology during development. We used task-based fMRI data from the Philadelphia Neurodevelopmental Cohort (PNC, N = 1221, 8–23 years) and conducted generalized psycho-physiological interaction (gPPI) analyses. Measures of psychopathology were derived from an independent component analysis of questionnaire data. The results showed positive associations between identifying fearful, sad, and angry faces and depressive symptoms, and a negative relationship between sadness recognition and positive psychosis symptoms. We found a positive main effect of depressive symptoms on BOLD activation in regions overlapping with the default mode network, while individuals reporting higher levels of norm-violating behavior exhibited emotion-specific lower functional connectivity within regions of the salience network and between modules that overlapped with the salience and default mode network. Our findings illustrate the relevance of functional connectivity patterns underlying emotion processing for behavioral problems in children and adolescents.

The serotonergic psychedelic N,N-dipropyltryptamine alters information-processing dynamics in in vitro cortical neural circuits

Abstract Most of the recent work in psychedelic neuroscience has been done using non-invasive neuroimaging, with data recorded from the brains of adult volunteers under the influence of a variety of drugs. While this data provides holistic insights into the effects of psychedelics on whole-brain dynamics, the effects of psychedelics on the meso-scale dynamics of neuronal circuits remains much less explored. Here, we report the effects of the serotonergic psychedelic N,N-diproptyltryptamine (DPT) on information-processing dynamics in a sample of in vitro organotypic cultures of cortical tissue from post-natal rats. Three hours of spontaneous activity were recorded: an hour of pre-drug control, and hour of exposure to 10 μM DPT solution, and a final hour of washout, once again under control conditions. We found that DPT reversibly alters information dynamics in multiple ways: first, the DPT condition was associated with higher entropy of spontaneous firing activity and reduced the amount of time information was stored in individual neurons. Second, DPT also reduced the reversibility of neural activity, increasing the entropy produced and suggesting a drive away from equilibrium. Third, DPT altered the structure of neuronal circuits, decreasing the overall information flow coming into each neuron, but increasing the number of weak connections, creating a dynamic that combines elements of integration and disintegration. Finally, DPT decreased the higher-order statistical synergy present in sets of three neurons. Collectively, these results paint a complex picture of how psychedelics regulate information processing in meso-scale neuronal networks in cortical tissue. Implications for existing hypotheses of psychedelic action, such as the Entropic Brain Hypothesis, are discussed.

Pyramidal neurons proportionately alter the identity and survival of specific cortical interneuron subtypes.

The mammalian cerebral cortex comprises a complex neuronal network that maintains a delicate balance between excitatory neurons and inhibitory interneurons. Previous studies, including our own research, have shown that specific interneuron subtypes are closely associated with particular pyramidal neuron types, forming stereotyped local inhibitory microcircuits. However, the developmental processes that establish these precise networks are not well understood. Here we show that pyramidal neuron types are instrumental in driving the terminal differentiation and maintaining the survival of specific associated interneuron subtypes. In a wild-type cortex, the relative abundance of different interneuron subtypes aligns precisely with the pyramidal neuron types to which they synaptically target. In Fezf2 mutant cortex, characterized by the absence of layer 5 pyramidal tract neurons and an expansion of layer 6 intratelencephalic neurons, we observed a corresponding decrease in associated layer 5b interneurons and an increase in layer 6 subtypes. Interestingly, these shifts in composition are achieved through mechanisms specific to different interneuron types. While SST interneurons adjust their abundance to the change in pyramidal neuron prevalence through the regulation of programmed cell death, parvalbumin interneurons alter their identity. These findings illustrate two key strategies by which the dynamic interplay between pyramidal neurons and interneurons allows local microcircuits to be sculpted precisely. These insights underscore the precise roles of extrinsic signals from pyramidal cells in the establishment of interneuron diversity and their subsequent integration into local cortical microcircuits.

Self-organization toward 1/f noise in deep neural networks.

In biological neural networks, it has been well recognized that a healthy brain exhibits 1/f noise patterns. However, in artificial neural networks that are increasingly matching or even out-performing human cognition, this phenomenon has yet to be established. In this work, we found that similar to that of their biological counterparts, 1/f noise exists in artificial neural networks when trained on time series classification tasks. Additionally, we found that the activations of the neurons are the closest to 1/f noise when the neurons are highly utilized. Conversely, if the network is too large and many neurons are underutilized, the neuron activations deviate from 1/f noise patterns toward that of white noise.

KineWheel-DeepLabCut Automated Paw Annotation Using Alternating Stroboscopic UV and White Light Illumination.

Uncovering the relationships between neural circuits, behavior, and neural dysfunction may require rodent pose tracking. While open-source toolkits such as DeepLabCut have revolutionized markerless pose estimation using deep neural networks, the training process still requires human intervention for annotating key points of interest in video data. To further reduce human labor for neural network training, we developed a method that automatically generates annotated image datasets of rodent paw placement in a laboratory setting. It uses invisible but fluorescent markers that become temporarily visible under UV light. Through stroboscopic alternating illumination, adjacent video frames taken at 720 Hz are either UV or white light illuminated. After color filtering the UV-exposed video frames, the UV markings are identified and the paw locations are deterministically mapped. This paw information is then transferred to automatically annotate paw positions in the next white light-exposed frame that is later used for training the neural network. We demonstrate the effectiveness of our method using a KineWheel-DeepLabCut setup for the markerless tracking of the four paws of a harness-fixed mouse running on top of the transparent wheel with mirror. Our automated approach, made available open-source, achieves high-quality position annotations and significantly reduces the need for human involvement in the neural network training process, paving the way for more efficient and streamlined rodent pose tracking in neuroscience research.

Multi-timescale neural dynamics for multisensory integration.

Carrying out any everyday task, be it driving in traffic, conversing with friends or playing basketball, requires rapid selection, integration and segregation of stimuli from different sensory modalities. At present, even the most advanced artificial intelligence-based systems are unable to replicate the multisensory processes that the human brain routinely performs, but how neural circuits in the brain carry out these processes is still not well understood. In this Perspective, we discuss recent findings that shed fresh light on the oscillatory neural mechanisms thatmediate multisensory integration (MI), including power modulations, phase resetting, phase-amplitude coupling and dynamic functional connectivity. We then consider studies that also suggest multi-timescale dynamics in intrinsic ongoing neural activity and during stimulus-driven bottom-up and cognitive top-down neural network processing in the context of MI. We propose a new concept of MI that emphasizes the critical role of neural dynamics at multiple timescales within and across brain networks, enabling the simultaneous integration, segregation, hierarchical structuring and selection of information in different time windows. To highlight predictions from our multi-timescale concept of MI, real-world scenarios in which multi-timescale processes may coordinate MI in a flexible and adaptive manner are considered.