Frontiers in Computational Neuroscience

A mean field model (MFM) is a mesoscopic description of neuronal population dynamics that can reduce the complexity of neural microcircuits into equations preserving key functional properties. The generation of a MFM is a complex mathematical process that starts with the incorporation of single neuron input/output relationships and local connectivity. Once neuron electroresponsiveness and synaptic properties are defined, in principle, the process can be automatized. Here we develop a tool for automatic MFM derivation from biophysically grounded spiking networks (Auto-MFM) by performing micro-to-mesoscale parameter remapping, estimating input/output relationships specific for different neuronal populations (i.e., transfer functions), and optimizing transfer function parameters. Auto-MFM was tested using a spiking cerebellar circuit as a generative model. The cerebellar MFM derived with Auto-MFM accurately reproduced cerebellar population dynamics of the corresponding spiking network, matching mean and time-varying firing rates across a wide range of stimulation patterns. Auto-MFM allowed us to model and explore physiological and pathological circuit variants; indeed, it was used to map ataxia-related structural connectivity alterations of the cerebellar network, in which Purkinje cells with simplified dendritic structure altered the cerebellar connectivity. Furthermore, Auto-MFM was used to create a library of cerebellar MFMs by sweeping the level of the excitatory conductance at mossy fiber - granule cell synapse, which is altered in several neuropathologies. Auto-MFM is thus proving a flexible and powerful tool to generate region-specific MFMs of healthy and pathological brain networks to be embedded in brain digital models.

Cerebellar neural circuit dynamics rely on a rich repertoire of synaptic and excitability mechanisms, which are thought to determine network computation in physiological and pathological conditions. In this work, we develop and validate a biologically-grounded spiking neural network of the cerebellar cortex, embedding key mechanisms of cellular excitability and synaptic transmission, and assess their impact on signal processing. Neuronal input-output functions, short-term synaptic plasticity, receptor-specific kinetics, and NMDA channel voltage-dependent gating were calibrated against detailed multicompartmental models through automatic tuning procedures. Incorporating these realistic biological properties allowed the network model to simulate key features observed in recordings from acute cerebellar slices. The neuronal discharge and local field potentials elicited by mossy fiber stimulation faithfully reproduced the natural patterns with millisecond precision. Then, selective receptor switch-off revealed the contribution of NMDA, GABA, and AMPA receptors to the frequency-dependent input-output function of the granular layer and Purkinje cells, linking previous empirical findings to specific synaptic mechanisms. This model combines high computational performance with biological realism and offers a computationally efficient framework to investigate neurophysiological phenomena and the neural correlates of behavior in large-scale long-lasting simulations, such as those needed to address the neural underpinnings of learning and of cerebellar pathologies.

Categorisation is often treated as a form of compression: a high-dimensional stimulus space is reduced to a smaller set of behaviourally or cognitively useful classes. However, compression alone does not determine whether a category map is useful. The present manuscript develops an information-theoretic framework for evaluating categorisation in terms of both category complexity and target-relevant information preservation. Across a set of synthetic demonstrations, alternative category maps over the same stimulus space are shown to preserve different target variables, including identity, action, nuisance, and hierarchical category structure. The framework is then extended to learned visual representations by analysing layer-derived category maps from a pretrained ResNet-50 network applied to CIFAR-10 images. Two scenarios are compared: a clean-only object run and a pooled nuisance run containing clean, blurred, pixelated, and noise-perturbed images. The results show that category maps can have substantial entropy while preserving information about a variable that is not aligned with the specified target, and that the value of a categorisation depends on the target variable to be preserved. The manuscript argues that categorisation should therefore be evaluated not only by compression or separability, but by the information retained about a specified cognitive, behavioural, or computational target.

Equilibrium propagation (EP) is a biologically plausible alternative to backpropagation that has demonstrated competitive performance across a range of machine learning tasks. Recent work has extended EP to spiking neural networks (SNNs), leveraging leaky integrate-and-fire (LIF) neurons and spike-based plasticity rules to improve biological realism while maintaining strong performance. In this work, we propose an EP-based SNN framework that combines LIF neural dynamics with a predictive learning rule, replacing conventional spike-timing-dependent plasticity (STDP) with a learning rule more directly aligned with predictive coding principles. We evaluate the proposed model on multiple image classification benchmarks, including MNIST, KMNIST, and Fashion-MNIST, and compare its performance with a BP-trained LIF SNN baseline. Our results show that the proposed EP-based LIF model (EP+LIF) achieves competitive accuracy across datasets, with performance approaching that of the BP-trained counterpart (BP+LIF) while relying on a biologically motivated local learning rule. In addition, analysis of hidden-layer spiking activity reveals that EP+LIF produces more persistent hidden-state activity, whereas BP+LIF yields sparser spiking representations. These results demonstrate that predictive learning can support effective EP-based training in LIF spiking networks, while also highlighting differences in activity patterns that motivate future work on activity regulation and sparse spiking dynamics.

Understanding how animals produce a versatile locomotor repertoire requires unraveling the interplay between higher centers, decentralized locomotor circuits, and sensory feedback. However, the principles governing their integration remain elusive. We investigated amphibious centipedes through stepwise neural lesions and neuromechanical modeling. Behavioral experiments revealed that while decentralized circuits autonomously generate coordination, the brain and subesophageal ganglion provide situational flexibility, such as modulating trunk undulation and initiating leg folding. Integrating these findings, our model demonstrated how higher centers selectively inhibit or release lower circuit dynamics. Simulations verified that varying only a few descending control parameters reproduces transitions between slow walking, fast walking, and swimming. This work may capture the essence of the locomotor circuitry that harnesses decentralized self-organization to coordinate the bodys large degrees of freedom.

Memory consolidation is the process by which temporary, malleable memories are transformed into more stable, longer-lasting forms. On a coarse anatomical scale, consolidation redistributes memories in the brain, but it remains poorly understood how these changes manifest themselves on the finer, cellular scale of neuronal engrams and how they relate to the cognitive level. In this study, we developed a phenomenological model of engram dynamics under systems consolidation. The model describes consolidation as a brain-wide phenomenon, where memories deterministically follow a trajectory through a space of patterns distributed among brain regions. It captures a broad range of features of memory consolidation, including selective consolidation, semantization, and power-law forgetting. In the model, consolidation is accompanied by population-level changes in neuronal representations that resemble the widely observed phenomenon of representational drift. When only a subset of neurons is observed, the deterministic dynamics of the model can appear stochastic, and a readout of task features deteriorates over time even when a stable readout exists for the full system. Our model offers a dynamical systems perspective on memory consolidation as a distributed process, moving beyond the classic region-centered view, and provides a functional interpretation of drift as a means of redistributing engrams for improved memory retention.

AO_SCPLOWBSTRACTC_SCPLOWOne of the most compelling ideas for bridging neuroscience and artificial neural networks is the establishment of a framework based on three main components: network architecture, optimization mechanism, and loss (or objective) function to be minimized. While the first two components have been extensively explored, the definition of a loss or objective function in neuroscience has been addressed less thoroughly, often from perspectives such as predictive coding. In this work, we propose an elementary loss function grounded in the comparison of neuronal responses to two signals: an external one, used for learning, and an internal one, reflecting the acquired knowledge. The loss function is thus simply the basic difference between the two, which, in terms of logical signals, corresponds to a well-known non-linearly separable function: the XOR function. We illustrate with a computational example how a binarized image recognition algorithm can be straightforwardly implemented in an autoencoder, and we show how a neuronal motif organized around an inhibitory neuron could implement such XOR operation and provide a feedback signal that makes optimization possible.

Human object recognition is robust to challenging conditions, such as when ones view of an object is fragmented due to an occluding foreground object. In comparison, deep neural networks (DNNs) are typically more susceptible to occlusion, suggesting that human vision relies on distinct mechanisms. Here, we investigated the role of visual diet in the emergence of these mechanisms by asking whether human-like robustness might arise in DNNs when trained with image datasets that better reflect the properties of occlusion in natural vision. We trained convolutional and transformer DNNs to classify clear images only, images augmented with artificial occluders (i.e., geometric shapes) or natural occluders (objects segmented from photographs). We then evaluated DNN occlusion robustness and compared their performance profiles with 30 human participants. We found that DNNs trained with artificial occluders remained vulnerable to natural occlusion and exhibited less human-like performance than those trained with natural occlusion. Our findings suggest that human robustness to visual occlusion arises from learning to disentangle natural objects from each other rather than simply learning to recognize objects from partial views. They also imply that commonly used forms of artificial occlusion are unsuitable for the evaluation or promotion of robustness to real-world occlusion in DNNs.

Concept cells are neurons in the medial temporal lobe that represent an abstract concept, such as a familiar person, in a context-independent way. They are activated by heterogeneous and potentially multi-modal sensory input related to the concept, such as viewing a photo of the person, reading the persons name, or hearing the persons voice. Learning the concept implies connecting the cortical cells that are triggered when such features are perceived with the cells of the concept, resulting in an associative memory engram. Traditional models explain the formation of such an engram with Hebbian learning through synaptic plasticity, the strengthening of existing synapses in response to co-stimulation. However, the low edge density of the brain suggests that direct connections in the form of preexisting synapses between these cells are relatively unlikely, rendering such constructs inefficient and volatile. Instead, it is plausible that more persistent concept engrams rely on structural plasticity, involving the creation of synapses de novo. Yet, it has still been unclear how neurons can project their axons across such a considerable distance, finding a target they do not know in advance. In this paper, we simulate the formation of such structural engrams in the connectomes of healthy subjects, offering a model hypothesis of how such engrams can be formed on a structural level. Based on our model, we further demonstrate how activating a concept can trigger related concepts through overlapping sensory associations, in a process we call a percept-concept loop.

The awake brain is known to display asynchronous (AS) states during periods of attention and arousal, but the responsiveness properties of such states remain unclear. Here, we investigate this question using computational models of spiking networks of excitatory and inhibitory neurons, mimicking recurrently-connected networks in layer 2/3 of the cerebral cortex. The networks can generate a continuum of AS states, but with different responsiveness characteristics. By using a mean-field model to infer the dynamic properties of the system, we find that there are two families of AS states, which we call "underdamped" (UD) and "overdamped" (OD). Responsiveness is maximised at the transition between OD and UD states, which identifies a "working point" that may present advantageous computational properties.

Hippocampus is known to replay activity patterns to recall and process memories, which is often related to Hopfield-type attractor dynamics. Another line of theoretical studies suggests that hippocampal replay prioritizes replay of experiences to accelerate value learning for efficient decision making. It is unknown how hippocampal attractor dynamics perform prioritized memory sampling, and more broadly, how we can consistently relate dynamical (bottom-up) and functional (top-down) theories of hippocampal replay. In this paper, we propose an extended Hopfield-type attractor network model with momentum, kinetic energy, and conservation of the total energy, which is called momentum Hopfield model. We show that our model can be interpreted as CA3-CA1 network model with intrinsic oscillation, and such network model reproduces hippocampal replay in 1-D and 2-D spatial structures. We also prove that our model functionally works as Markov-chain Monte Carlo sampling in which recall frequencies of memory patterns can be arbitrarily biased. Using this property, we implemented prioritized experience replay using our model, which actually accelerated reinforcement learning for spatial navigation. Our model explains how dynamics of hippocampal circuits realize efficient memory sampling, providing a theoretical link between dynamics and functions of hippocampal replay.

Recurrent neural networks (RNNs) have proven to be highly successful in emulating human-like cognitive functions such as working memory. In recent years, RNNs are evolving to incorporate more biophysical realism to produce more plausible predictions on how cognitive tasks are solved in real neural circuits. However, there are major challenges in constructing and training networks with the complex and nonlinear properties of real neurons. A major component of the success of RNNs is that they share the same mathematical base as deep neural networks, permitting highly efficient optimization of model parameters using standard deep learning techniques. To do so, they use abstract representations of neurons which fail to capture the impact of cell-level biophysical and morphologic properties that may benefit network-level function. Expanding task-trained RNNs with biophysical properties such as dendrites and active ionic currents poses substantial challenges, as it moves these models away from the validated training regimes known to be highly effective for RNNs. To address this gap, we developed a biophysically detailed reservoir computing (BRC) framework with the goal of extracting mechanistic insights from biophysical neural models, and propose that these insights can be used to guide model choices that will work for specific categories of cognitive tasks. The BRC network was constructed with synaptically coupled excitatory and inhibitory cells, in which the excitatory cells include multicompartment biophysically active dendrites; motivated by empirical studies suggesting dendrites have desirable computational benefits (e.g. pattern classification and coincidence detection). We trained the BRC network to do a simplified working memory task where it had to maintain the representation of an extrinsic "cue" input. We studied the impact of extrinsic input time constants (fast AMPA vs slow NMDA) and location (dendrite vs soma) on the ability of a network to solve the task. Our results revealed that cue inputs through NMDA receptors are particularly efficient for solving the working memory task. Further, the properties of NMDA receptors are uniquely suited for cue inputs delivered at the dendrite, as networks trained with dendritic AMPA cue inputs failed to solve the task. Detailed examination of the cell and network dynamics that solve the task reveals distinct local network configurations and computing principles for the different types of extrinsic input. Overall, much like the body of mechanistic insights that have underpinned the success of training RNNs, this study lays the groundwork for applying the BRC framework to train biophysically detailed neural models to solve complex human-like cognitive tasks.

Convolutional neural networks (CNNs) have become essential models for predicting neural activity and behavior in visual tasks. However, their ability to capture higher-level cognitive functions, such as numerosity discrimination, remains debated. Numerosity, the ability to perceive and estimate the number of items in a visual scene, is often proposed to rely on specialized number-detector units within CNNs, analogous to number-selective neurons observed in the brain. In this study, we use CORnet, a biologically inspired CNN architecture inspired by the organization of the primate visual system. To address a limitation of classical Representational Similarity Analysis (RSA)--its assumption that all units contribute equally--we apply pruning, a feature selection approach that identifies the units most relevant for explaining behavioral similarity structure. Our results show that number-detector units are not critical for population-level representations of numerosity, challenging their proposed role in previous studies.

The impact of cellular activities on large-scale brain dynamics is thought to determine brain functioning and disease, yet the causal relationships of neural mechanisms across scales remain unclear. Recently, the cerebellum has been reported to affect whole-brain dynamics during sensorimotor integration. To disclose the underlying mechanisms, we have developed a multiscale digital brain co-simulator, in which a spiking neural network of the olivo-cerebellar microcircuit is embedded in a mouse virtual brain and wired with other nodes using an atlas-based long-range connectome. Parameters and bi-directional interfaces between the spiking olivo-cerebellar network and other rate-coded modules were tuned to match experimental data of primary sensory and motor cortex (M1 and S1) power spectral densities and neuronal spiking rates. Then, the role of the cerebellar circuitry on sensorimotor integration was analyzed by lesioning critical circuit connections in silico. Simulations showed that spike processing within the cerebellar circuit is key to explaining the gamma-band coherence between M1 and S1 during sensorimotor integration. These results provide a mechanistic explanation of how the cerebellum promotes the formation of sensorimotor contingencies in relevant cortical modules as the basis of its critical role in sensorimotor prediction. On a broader perspective, this modelling approach opens new perspectives for the multiscale investigation of brain physiological and pathological states in relation to specific cellular and microcircuit properties.

The learning dynamics of biological brains and artificial neural networks are of interest to both neuroscience and machine learning. A key difference between them is that neural networks are often trained from a randomly initialized state whereas each brain is the product of generations of evolutionary optimization, yielding innate structures that enable few-shot learning and inbuilt reflexes. Artificial neural networks, by contrast, require non-ethological quantities of training data to attain comparable performance. To investigate the effect of evolutionary optimization on the learning dynamics of neural networks, we combined algorithms simulating natural selection and online learning to produce a method for evolutionarily conditioning artificial neural networks, and applied it to both reinforcement and supervised learning contexts. We found the evolutionary conditioning algorithm, by itself, performs comparably to an unoptimized baseline. However, evolutionarily conditioned networks show signs of unique and latent learning dynamics, and can be rapidly fine-tuned to optimal performance. These results suggest evolution constitutes an inductive bias that tunes neural systems to enable rapid learning.

Generalization is a fundamental criterion for evaluating learning effectiveness, a domain where biological intelligence excels yet artificial intelligence continues to face challenges. In biological learning and memory, the well-documented spacing effect shows that appropriately spaced intervals between learning trials can significantly improve behavioral performance. While multiple theories have been proposed to explain its underlying mechanisms, one compelling hypothesis is that spaced training promotes integration of input and innate variations, thereby enhancing generalization to novel but related scenarios. Here we examine this hypothesis by introducing a bio-inspired spacing effect into artificial neural networks, integrating input and innate variations across spaced intervals at the neuronal, synaptic, and network levels. These spaced ensemble strategies yield significant performance gains across various benchmark datasets and network architectures. Biological experiments on Drosophila further validate the complementary effect of appropriate variations and spaced intervals in improving generalization, which together reveal a convergent computational principle shared by biological learning and machine learning.

Animals and humans use multiple behavioral strategies to perform tasks. However, neural implementations of multiple strategies remain elusive, as some studies propose distinct pathways, while others observe overlapping brain regions associated with strategies. We propose a hybrid deep reinforcement learning (H-DRL) method, in which one network model implements model-free and inference-based behaviors through synaptic plasticity and recurrent activity. H-DRL uses a single updating rule and switches the strategy according to task demands without an explicit arbitrator. H-DRL reproduced mixed strategies of humans in a two-step task. In the mouse perceptual decision-making task, H-DRL adapted the recurrent dynamics with rich learning when the task condition required inference-based behavior, while adopting model-free behavior with lazy learning for a simple condition. The activity of H-DRL units showed condition-dependent maintenance of previous events, consistent with orbitofrontal cortical activity in mice. Our model provides a unified view that one cortical network automatically determines strategies in use depending on task conditions.

Electrophysiological data, which serve as a biological signal that bridges neural activity and behavioral tasks, provide an innovative approach to neuroscience research. In this study, we constructed a dataset that contains over 2000 neurons across 117 days recorded in 20 mice containing 28,573 trials. Data for 5 mice were collected from the Secondary Motor Cortex (M2) region 8 mice was derived from the Ventrolateral Striatum (VLS) and 7 mice were from Substantia Nigra pars Reticulata (SNR). We induced licking behavior in head-fixed mice by periodically delivering water through a spout while simultaneously recording spiking activity from three brain regions and behavior related electrical signals. This dataset ensures precise temporal alignment between neural activity and behavioral events, offering a robust foundation for investigating neural encoding mechanisms and simulation of neural activities. This dataset establishes a precise spike-to-event mapping, which enables high decoding accuracy using Multilayer Perceptron (MLP) and Support Vector Machine (SVM). It can serve as a high-quality benchmark for developing encoding and decoding algorithms in neural networks, particularly Spiking Neural Networks (SNNs).

Tracking dynamic moving objects in the external world is ethologically important for many organisms. Recent experiments have examined neural dynamics supporting such behaviors by employing visually-guided pursuit in freely moving rodents, yet computational principles underlying this cognitive process are not well understood. To address this, we developed a recurrent neural network model for examining the predictive behaviors and computations that emerge during pursuit. We demonstrate that the model generates internal predictions of the targets future locations, with anticipatory behaviors increasing with exposure to stereotyped trajectories of the target. These internal predictions can be used by the model to pursue a target in a complex environment, and the models emergent strategy is aligned with behavior when tested in rodents. In investigating the computations that underlie the models ability to perform predictive pursuit, we found units sensitive to the position of the target relative to the artificial agent, a representation analogous to egocentric target neurons observed in animals performing pursuit tasks. Ablating these units significantly reduced model performance, establishing a causal role of this functional response type in efficient pursuit. Given the complexity of the task and agent behavior, we hypothesized that RNN models may use high-dimensional neural codes to support predictive pursuit. To test this, we trained models of varying rank and found that anticipatory behavior emerged only when the rank was sufficiently high, despite strong pursuit performance in lower rank models. All RNNs encoded the egocentric location of the target, whereas allocentric self and target locations emerged only in high-dimensional networks. Overall, our results suggest that, unlike commonly studied vision, motor, or memory tasks, predictive pursuit emerges in high-dimensional networks with sufficient resources.

Rhythmic neural activity underlies essential biological functions such as locomotion, breathing, and feeding. Computational models are widely used to study how such rhythms emerge from interactions between neuron-level and circuit-level dynamics. Intrinsically bursting neurons are key components of many central pattern generators (CPGs), yet existing models span a tradeoff between biological realism and practical usability. Biophysical models involve many parameters that are difficult to tune, whereas abstract models often integrate poorly into neural circuit simulations. We propose a simplified model of intrinsically bursting neurons derived from a reduced non-spiking biophysical formulation. The model integrates readily into neural circuits while enabling direct and independent control of bursting characteristics, including duration, amplitude, and shape. We show that the model reproduces single-unit biophysical responses to diverse stimuli as well as circuit-level activity patterns from crustacean and mammalian CPGs. This model provides a practical tool for studying rhythm generation in neural circuits.