Dimensionality reduction for large-scale neural recordings

Stabilizing brain-computer interfaces through alignment of latent dynamics

Intracortical brain-computer interfaces (iBCIs) restore motor function to people with paralysis by translating brain activity into control signals for external devices. In current iBCIs, instabilities at the neural interface result in a degradation of decoding performance, which necessitates frequent supervised recalibration using new labeled data. One potential solution is to use the latent manifold structure that underlies neural population activity to facilitate a stable mapping between brain activity and behavior. Recent efforts using unsupervised approaches have improved iBCI stability using this principle; however, existing methods treat each time step as an independent sample and do not account for latent dynamics. Dynamics have been used to enable high-performance prediction of movement intention, and may also help improve stabilization. Here, we present a platform for Nonlinear Manifold Alignment with Dynamics (NoMAD), which stabilizes decoding using recurrent neural network models of dynamics. NoMAD uses unsupervised distribution alignment to update the mapping of nonstationary neural data to a consistent set of neural dynamics, thereby providing stable input to the decoder. In applications to data from monkey motor cortex collected during motor tasks, NoMAD enables accurate behavioral decoding with unparalleled stability over weeks- to months-long timescales without any supervised recalibration.

The effects of the post-delay epochs on working memory error reduction

Accurate retrieval of the maintained information is crucial for working memory. This process primarily occurs during post-delay epochs, when subjects receive cues and generate responses. However, the computational and neural mechanisms that underlie these post-delay epochs to support robust memory remain poorly understood. To address this, we trained recurrent neural networks (RNNs) on a color delayed-response task, where certain colors (referred to as common colors) were more frequently presented for memorization. We found that the trained RNNs reduced memory errors for common colors by decoding a broader range of neural states into these colors through the post-delay epochs. This decoding process was driven by convergent neural dynamics and a non-dynamic, biased readout process during the post-delay epochs. Our findings highlight the importance of post-delay epochs in working memory and suggest that neural systems adapt to environmental statistics by using multiple mechanisms across task epochs.

Capturing the emergent dynamical structure in biophysical neural models

Complex neural systems can display structured emergent dynamics. Capturing this structure remains a significant scientific challenge. Using information theory, we apply Dynamical Independence (DI) to uncover the emergent dynamical structure in a minimal 5-node biophysical neural model, shaped by the interplay of two key aspects of brain organisation: integration and segregation. In our study, functional integration within the biophysical neural model is modulated by a global coupling parameter, while functional segregation is influenced by adding dynamical noise, which counteracts global coupling. Leveraging transfer entropy, DI defines a dimensionally-reduced macroscopic variable (e.g., a coarse-graining) as emergent to the extent that it behaves as an independent dynamical process, distinct from the micro-level dynamics. Dynamical dependence (a departure from dynamical independence) is measured by minimising the transfer entropy from microlevel variables to macroscopic variables across spatial scales. Our results indicate that the degree of emergence of macroscopic variables is relatively minimised at balanced points of integration and segregation and maximised at the extremes. Additionally, our method identifies to which degree the macroscopic dynamics are localised across microlevel nodes, thereby elucidating the emergent dynamical structure through the relationship between microscopic and macroscopic processes. We find that deviation from a balanced point between integration and segregation results in a less localised, more distributed emergent dynamical structure as identified by DI. This finding suggests that a balance of functional integration and segregation is associated with lower levels of emergence (higher dynamical dependence), which may be crucial for sustaining coherent, localised emergent macroscopic dynamical structures. This work also provides a complete computational implementation for the identification of emergent neural dynamics that could be applied both in silico and in vivo.

Toward Principles of Brain Network Organization and Function

The brain is immensely complex, with diverse components and dynamic interactions building upon one another to orchestrate a wide range of behaviors. Understanding patterns of these complex interactions and how they are coordinated to support collective neural function is critical for parsing human and animal behavior, treating mental illness, and developing artificial intelligence. Rapid experimental advances in imaging, recording, and perturbing neural systems across various species now provide opportunities to distill underlying principles of brain organization and function. Here, we take stock of recent progress and review methods used in the statistical analysis of brain networks, drawing from fields of statistical physics, network theory, and information theory. Our discussion is organized by scale, starting with models of individual neurons and extending to large-scale networks mapped across brain regions. We then examine organizing principles and constraints that shape the biological structure and function of neural circuits. We conclude with an overview of several critical frontiers, including expanding current models, fostering tighter feedback between theory and experiment, and leveraging perturbative approaches to understand neural systems. Alongside these efforts, we highlight the importance of contextualizing their contributions by linking them to formal accounts of explanation and causation.

Magnetoencephalography Dimensionality Reduction Informed by Dynamic Brain States

Complex spontaneous brain dynamics mirror the large number of interactions taking place among regions, supporting higher functions. Such complexity is manifested in the interregional dependencies among signals derived from different brain areas, as observed utilising neuroimaging techniques, like magnetoencephalography. The dynamics of this data produce numerous subsets of active regions at any moment as they evolve. Notably, converging evidence shows that these states can be understood in terms of transient coordinated events that spread across the brain over multiple spatial and temporal scales. Those can be used as a proxy of the 'effectiveness' of the dynamics, as they become stereotyped or disorganised in neurological diseases. However, given the high-dimensional nature of the data, representing them has been challenging thus far. Dimensionality reduction techniques are typically deployed to describe complex interdependencies and improve their interpretability. However, many dimensionality reduction techniques lose information about the sequence of configurations that took place. Here, we leverage a newly described algorithm, potential of heat-diffusion for affinity-based transition embedding (PHATE), specifically designed to preserve the dynamics of the system in the low-dimensional embedding space. We analysed source-reconstructed resting-state magnetoencephalography from 18 healthy subjects to represent the dynamics of the configuration in low-dimensional space. After reduction with PHATE, unsupervised clustering via K-means is applied to identify distinct clusters. The topography of the states is described, and the dynamics are represented as a transition matrix. All the results have been checked against null models, providing a parsimonious account of the large-scale, fast, aperiodic dynamics during resting-state. The study applies the PHATE algorithm to source-reconstructed magnetoencephalography (MEG) data, reducing dimensionality while preserving large-scale neural dynamics. Results reveal distinct configurations, or 'states', of brain activity, identified via unsupervised clustering. Their transitions are characterised by a transition matrix. This method offers a simplified yet rich view of complex brain interactions, opening new perspectives on large-scale brain dynamics in health and disease.

Visual Statistical Learning Alters Low-Dimensional Cortical Architecture

Our brains are in a constant state of generating predictions, implicitly extracting environmental regularities to support later cognition and behavior, a process known as statistical learning (SL). While prior work investigating the neural basis of SL has focused on the activity of single brain regions in isolation, much less is known about how distributed brain areas coordinate their activity to support such learning. Using fMRI and a classic visual SL task, we investigated changes in whole-brain functional architecture as human female and male participants implicitly learned to associate pairs of images, and later, when predictions generated from learning were violated. By projecting individuals' patterns of cortical and subcortical functional connectivity onto a low-dimensional manifold space, we found that SL was associated with changes along a single neural dimension describing covariance across the visual-parietal and perirhinal cortex (PRC). During learning, we found regions within the visual cortex expanded along this dimension, reflecting their decreased communication with other networks, whereas regions within the dorsal attention network (DAN) contracted, reflecting their increased connectivity with higher-order cortex. Notably, when SL was interrupted, we found the PRC and entorhinal cortex, which did not initially show learning-related effects, now contracted along this dimension, reflecting their increased connectivity with the default mode and DAN, and decreased covariance with visual cortex. While prior research has linked SL to either broad cortical or medial temporal lobe changes, our findings suggest an integrative view, whereby cortical regions reorganize during association formation, while medial temporal lobe regions respond to their violation.

Analysis of High-Dimensional Coordination in Human Movement Using Variance Spectrum Scaling and Intrinsic Dimensionality

Full-body movement involving multi-segmental coordination has been essential to our evolution as a species, but its study has been focused mostly on the analysis of one-dimensional data. The field is poised for a change by the availability of high-density recording and data sharing. New ideas are needed to revive classical theoretical questions such as the organization of the highly redundant biomechanical degrees of freedom and the optimal distribution of variability for efficiency and adaptiveness. In movement science, there are popular methods that up-dimensionalize: they start with one or a few recorded dimensions and make inferences about the properties of a higher-dimensional system. The opposite problem, dimensionality reduction, arises when making inferences about the properties of a low-dimensional manifold embedded inside a large number of kinematic degrees of freedom. We present an approach to quantify the smoothness and degree to which the kinematic manifold of full-body movement is distributed among embedding dimensions. The principal components of embedding dimensions are rank-ordered by variance. The power law scaling exponent of this variance spectrum is a function of the smoothness and dimensionality of the embedded manifold. It defines a threshold value below which the manifold becomes non-differentiable. We verified this approach by showing that the Kuramoto model obeys the threshold when approaching global synchronization. Next, we tested whether the scaling exponent was sensitive to participants' gait impairment in a full-body motion capture dataset containing short gait trials. Variance scaling was highest in healthy individuals, followed by osteoarthritis patients after hip replacement, and lastly, the same patients before surgery. Interestingly, in the same order of groups, the intrinsic dimensionality increased but the fractal dimension decreased, suggesting a more compact but complex manifold in the healthy group. Thinking about manifold dimensionality and smoothness could inform classic problems in movement science and the exploration of the biomechanics of full-body action.

A deep learning strategy to identify cell types across species from high-density extracellular recordings

High-density probes allow electrophysiological recordings from many neurons simultaneously across entire brain circuits but fail to reveal cell type. Here, we develop a strategy to identify cell types from extracellular recordings in awake animals and reveal the computational roles of neurons with distinct functional, molecular, and anatomical properties. We combine optogenetics and pharmacology using the cerebellum as a testbed to generate a curated ground-truth library of electrophysiological properties for Purkinje cells, molecular layer interneurons, Golgi cells, and mossy fibers. We train a semi-supervised deep learning classifier that predicts cell types with greater than 95% accuracy based on the waveform, discharge statistics, and layer of the recorded neuron. The classifier's predictions agree with expert classification on recordings using different probes, in different laboratories, from functionally distinct cerebellar regions, and across species. Our classifier extends the power of modern dynamical systems analyses by revealing the unique contributions of simultaneously recorded cell types during behavior.

A mechanism for the emergence of low-dimensional structures in brain dynamics

Recent neuroimaging advancements have led to datasets characterized by an overwhelming number of features. Different dimensionality reduction techniques have been employed to uncover low-dimensional manifold representations underlying cognitive functions, while maintaining the fundamental characteristics of the data. These range from linear algorithms to more intricate non-linear methods for manifold extraction. However, the mechanisms responsible for the emergence of these simplified architectures remain a topic of debate. Motivated by concepts from dynamical systems theory, such as averaging and time-scale separation, our study introduces a novel mechanism for the collapse of high dimension brain dynamics onto lower dimensional manifolds. In our framework, fast neuronal activity oscillations average out over time, leading to the resulting dynamics approximating task-related processes occurring at slower time scales. This leads to the emergence of low-dimensional solutions as complex dynamics collapse into slow invariant manifolds. We test this assumption via neural simulations using a simplified model and then enhance the complexity of our simulations by incorporating a large-scale brain network model to mimic realistic neuroimaging signals. We observe in the different cases the convergence of fast oscillatory fluctuations of neuronal activity across time scales that correspond to simulated behavioral configurations. Specifically, by employing various dimensionality reduction techniques and manifold extraction schemes, we observe the reduction of high-dimensional dynamics onto lower-dimensional spaces, revealing emergent low-dimensional solutions. Our findings shed light on the role of frequency and time-scale separation in neuronal activity, proposing and testing a novel theoretical framework for understanding the inner mechanisms governing low-dimensional pattern formation in brain dynamics.

Assistive sensory-motor perturbations influence learned neural representations

Task errors are used to learn and refine motor skills. We investigated how task assistance influences learned neural representations using Brain-Computer Interfaces (BCIs), which map neural activity into movement via a decoder. We analyzed motor cortex activity as monkeys practiced BCI with a decoder that adapted to improve or maintain performance over days. Over time, task-relevant information became concentrated in fewer neurons, unlike with fixed decoders. At the population level, task information also became largely confined to a few neural modes that accounted for an unexpectedly small fraction of the population variance. A neural network model suggests the adaptive decoders directly contribute to forming these more compact neural representations. Our findings show that assistive decoders manipulate error information used for long-term learning computations like credit assignment, which informs our understanding of motor learning and has implications for designing real-world BCIs.

Surround modulation is predominantly orientation-unspecific in macaque V1

Surround modulation is a fundamental property of V1 neurons, playing critical roles in stimulus integration and segregation. It is believed to be orientation-specific, as neurons' responses at preferred orientations are suppressed more by iso-oriented surrounds than by cross-oriented surrounds. Here, we investigated an alternative hypothesis that surround modulation is primarily orientation-unspecific, in that the observed "orientation-specific" surround effects actually reflect overall gain changes that affect neurons tuned to all orientations. We employed two-photon calcium imaging to compare V1 population orientation tuning functions under iso- and cross-surround modulation in awake, fixating macaques. While confirming "orientation-specific" surround suppression in individual neurons, our analysis of the population orientation tuning functions revealed that iso-surrounds induce a general orientation-unspecific suppression across all orientation-tuned neurons, plus weak orientation-specific suppression to neurons tuned to the center stimulus orientation. Furthermore, cross-surrounds mainly reduce orientation-unspecific suppression by scaling up responses of all orientation-tuned neurons. These findings suggest a model of population gain control where surround stimuli mostly scale the responses of the neuronal population. Further population coding analyses supported this conclusion, demonstrating that surround suppression leads to degraded target orientation information at least partially due to a reduced number of contributing neurons with diverse orientation preferences.

Flexible control of motor units: is the multidimensionality of motor unit manifolds a sufficient condition?

Understanding flexibility in the neural control of movement requires identifying the distribution of common inputs to the motor units. In this study, we identified large samples of motor units from two lower limb muscles: the vastus lateralis (VL; up to 60 motor units per participant) and the gastrocnemius medialis (GM; up to 67 motor units per participant). First, we applied a linear dimensionality reduction method to assess the dimensionality of the manifolds underlying the motor unit activity. We subsequently investigated the flexibility in motor unit control under two conditions: sinusoidal contractions with torque feedback, and online control with visual feedback on motor unit firing rates. Overall, we found that the activity of GM motor units was effectively captured by a single latent factor defining a unidimensional manifold, whereas the VL motor units were better represented by three latent factors defining a multidimensional manifold. Despite this difference in dimensionality, the recruitment of motor units in the two muscles exhibited similarly low levels of flexibility. Using a spiking network model, we tested the hypothesis that dimensionality derived from factorization does not solely represent descending cortical commands but is also influenced by spinal circuitry. We demonstrated that a heterogeneous distribution of inputs to motor units, or specific configurations of recurrent inhibitory circuits, could produce a multidimensional manifold. This study clarifies an important debated issue, demonstrating that while motor unit firings of a non-compartmentalized muscle can lie in a multidimensional manifold, the CNS may still have limited capacity for flexible control of these units. KEY POINTS: To generate movement, the CNS distributes both excitatory and inhibitory inputs to the motor units. The level of flexibility in the neural control of these motor units remains a topic of debate with significant implications for identifying the smallest unit of movement control. By combining experimental data and in silico models, we demonstrated that the activity of a large sample of motor units from a single muscle can be represented by a multidimensional linear manifold; however, these units show very limited flexibility in their recruitment. The dimensionality of the linear manifold may not directly reflect the dimensionality of descending inputs but could instead relate to the organization of local spinal circuits.

Uncertainties in signal recovery from heterogeneous and convoluted time series with principal component analysis

Principal component analysis (PCA) is one of the most used tools for extracting low-dimensional representations of data, in particular for time series. Performances are known to strongly depend on the quality (amount of noise) and the quantity of data. We here investigate the impact of heterogeneities, often present in real data, on the reconstruction of low-dimensional trajectories and of their associated modes. We focus in particular on the effects of sample-to-sample fluctuations and of component-dependent temporal convolution and noise in the measurements. We derive analytical predictions for the error on the reconstructed trajectory and the confusion between the modes using the replica method in a high-dimensional setting, in which the number and the dimension of the data are comparable. We find in particular that sample-to-sample variability is deleterious for the reconstruction of the signal trajectory, but beneficial for the inference of the modes, and that the fluctuations in the temporal convolution kernels prevent perfect recovery of the latent modes even for very weak measurement noise. Our predictions are corroborated by simulations with synthetic data for a variety of control parameters.

Geometry Linked to Untangling Efficiency Reveals Structure and Computation in Neural Populations

From an eagle spotting a fish in shimmering water to a scientist extracting patterns from noisy data, many cognitive tasks require untangling overlapping signals. Neural circuits achieve this by transforming complex sensory inputs into distinct, separable representations that guide behavior. Data-visualization techniques convey the geometry of these transformations, and decoding approaches quantify performance efficiency. However, we lack a framework for linking these two key aspects. Here we address this gap by introducing a data-driven analysis framework, which we call Geometry Linked to Untangling Efficiency (GLUE) with manifold capacity theory, that links changes in the geometrical properties of neural activity patterns to representational untangling at the computational level. We applied GLUE to over seven neuroscience datasets-spanning multiple organisms, tasks, and recording techniques-and found that task-relevant representations untangle in many domains, including along the cortical hierarchy, through learning, and over the course of intrinsic neural dynamics. Furthermore, GLUE can characterize the underlying geometric mechanisms of representational untangling, and explain how it facilitates efficient and robust computation. Beyond neuroscience, GLUE provides a powerful framework for quantifying information organization in data-intensive fields such as structural genomics and interpretable AI, where analyzing high-dimensional representations remains a fundamental challenge.

Material Damage to Multielectrode Arrays after Electrolytic Lesioning is in the Noise

1The quality of stable long-term recordings from chronically implanted electrode arrays is essential for experimental neuroscience and brain-computer interfaces. This work uses scanning electron microscopy (SEM) to image and analyze eight 96-channel Utah arrays previously implanted in motor cortical regions of four subjects (subject H = 2242 days implanted, F = 1875, U = 2680, C = 594), providing important contributions to a growing body of long-term implant research leveraging this imaging technology. Four of these arrays have been used in electrolytic lesioning experiments (H = 10 lesions, F = 1, U = 4, C = 1), a novel electrolytic perturbation technique using small direct currents. In addition to surveying physical damage, such as biological debris and material deterioration, this work also analyzes whether electrolytic lesioning created damage beyond what is typical for these arrays. Each electrode was scored in six damage categories, identified from the literature: abnormal debris, metal coating cracks, silicon tip breakage, parylene C delamination, parylene C cracks, and shank fracture. This analysis confirms previous results that observed damage on explanted arrays is more severe on the outer-edge electrodes versus inner electrodes. These findings also indicate that are no statistically significant differences between the damage observed on normal electrodes versus electrodes used for electrolytic lesioning. This work provides evidence that electrolytic lesioning does not significantly affect the quality of chronically implanted electrode arrays and can be a useful tool in understanding perturbations to neural systems. Finally, this work also includes the largest collection of single-electrode SEM images for previously implanted multielectrode Utah arrays, spanning eleven different intact arrays and one broken array. As the clinical relevance of chronically implanted electrodes with single-neuron resolution continues to grow, these images may be used to provide the foundation for a larger public database and inform further electrode design and analyses.

Mapping the Psychosis Spectrum - Imaging Neurosubtypes from Multi-Scale Functional Network Connectivity

This study aims to identify Psychosis Imaging Neurosubtypes (PINs)-homogeneous subgroups of individuals with psychosis characterized by distinct neurobiology derived from imaging features. Specifically, we utilized resting-state fMRI data from 2103 B-SNIP 1&2 participants (1127 with psychosis, 350 relatives, 626 controls) to compute subject-specific multiscale functional network connectivity (msFNC). We then derived a low-dimensional neurobiological subspace, termed Latent Network Connectivity (LNC), which captured system-wide interconnected multiscale information across three components (cognitive-related, typical, psychosis-related). Projections of psychosis participants' msFNC onto this subspace revealed three PINs through unsupervised learning, each with distinct cognitive, clinical, and connectivity profiles, spanning all DSM diagnoses (Schizophrenia, Bipolar, Schizoaffective). PIN-1, the most cognitively impaired, showed Cerebellar-Subcortical and Visual-Sensorimotor hypoconnectivity, alongside Visual-Subcortical hyperconnectivity. Most cognitively preserved PIN-2 showed Visual-Subcortical, Subcortical-Sensorimotor, and Subcortical-Higher Cognition hypoconnectivity. PIN-3 exhibited intermediate cognitive function, showing Cerebellar-Subcortical hypoconnectivity alongside Cerebellar-Sensorimotor and Subcortical-Sensorimotor hyperconnectivity. Notably, 55% of relatives aligned with the same neurosubtype as their affected family members-a significantly higher rate than random chance (p-valueRelatives-to-PIN-1 < 0.001, p-valueRelatives-to-PIN-2 < 0.05, p-valueRelatives-to-PIN-3 < 0.001) compared to a non-significant 37% DSM-based classification, supporting a biological basis of these neurosubtypes. Cognitive performance reliably aligns with distinct brain connectivity patterns, which are also evident in relatives, supporting their construct validity. Our PINs differed from original B-SNIP Biotypes, which were determined from electrophysiological, cognitive, and oculomotor data. These findings underscore the limitations of DSM-based classifications in capturing the biological complexity of psychotic disorders and highlight the potential of imaging-based neurosubtypes to enhance our understanding of the psychosis spectrum.

Random compressed coding with neurons

Classical models of efficient coding in neurons assume simple mean responses-"tuning curves"- such as bell-shaped or monotonic functions of a stimulus feature. Real neurons, however, can be more complex: grid cells, for example, exhibit periodic responses that impart the neural population code with high accuracy. But do highly accurate codes require fine-tuning of the response properties? We address this question with the use of a simple model: a population of neurons with random, spatially extended, and irregular tuning curves. Irregularity enhances the local resolution of the code but gives rise to catastrophic, global errors. For optimal smoothness of the tuning curves, when local and global errors balance out, the neural population compresses information about a continuous stimulus into a low-dimensional representation, and the resulting distributed code achieves exponential accuracy. An analysis of recordings from monkey motor cortex points to such "compressed efficient coding." Efficient codes do not require a finely tuned design-they emerge robustly from irregularity or randomness.

Predicting Future Development of Stress-Induced Anhedonia From Cortical Dynamics and Facial Expression

The current state of mental health treatment for individuals diagnosed with major depressive disorder leaves billions of individuals with first-line therapies that are ineffective or burdened with undesirable side effects. One major obstacle is that distinct pathologies may currently be diagnosed as the same disease and prescribed the same treatments. The key to developing antidepressants with ubiquitous efficacy is to first identify a strategy to differentiate between heterogeneous conditions. Major depression is characterized by hallmark features such as anhedonia and a loss of motivation1,2, and it has been recognized that even among inbred mice raised under identical housing conditions, we observe heterogeneity in their susceptibility and resilience to stress3. Anhedonia, a condition identified in multiple neuropsychiatric disorders, is described as the inability to experience pleasure and is linked to anomalous medial prefrontal cortex (mPFC) activity4. The mPFC is responsible for higher order functions5-8, such as valence encoding; however, it remains unknown how mPFC valence-specific neuronal population activity is affected during anhedonic conditions. To test this, we implemented the unpredictable chronic mild stress (CMS) protocol9-11 in mice and examined hedonic behaviors following stress and ketamine treatment. We used unsupervised clustering to delineate individual variability in hedonic behavior in response to stress. We then performed in vivo 2-photon calcium imaging to longitudinally track mPFC valence-specific neuronal population dynamics during a Pavlovian discrimination task. Chronic mild stress mice exhibited a blunted effect in the ratio of mPFC neural population responses to rewards relative to punishments after stress that rebounds following ketamine treatment. Also, a linear classifier revealed that we can decode susceptibility to chronic mild stress based on mPFC valence-encoding properties prior to stress-exposure and behavioral expression of susceptibility. Lastly, we utilized markerless pose tracking computer vision tools to predict whether a mouse would become resilient or susceptible based on facial expressions during a Pavlovian discrimination task. These results indicate that mPFC valence encoding properties and behavior are predictive of anhedonic states. Altogether, these experiments point to the need for increased granularity in the measurement of both behavior and neural activity, as these factors can predict the predisposition to stress-induced anhedonia.

Slow cortical dynamics generate context processing and novelty detection

The cortex amplifies responses to novel stimuli while suppressing redundant ones. Novelty detection is necessary to efficiently process sensory information and build predictive models of the environment, and it is also altered in schizophrenia. To investigate the circuit mechanisms underlying novelty detection, we used an auditory "oddball" paradigm and two-photon calcium imaging to measure responses to simple and complex stimuli across mouse auditory cortex. Stimulus statistics and complexity generated specific responses across auditory areas. Neuronal ensembles reliably encoded auditory features and temporal context. Interestingly, stimulus-evoked population responses were particularly long lasting, reflecting stimulus history and affecting future responses. These slow cortical dynamics encoded stimulus temporal context and generated stronger responses to novel stimuli. Recurrent neural network models trained on the oddball task also exhibited slow network dynamics and recapitulated the biological data. We conclude that the slow dynamics of recurrent cortical networks underlie processing and novelty detection.

Contractions in human cerebellar-cortical manifold structure underlie motor reinforcement learning

How the brain learns new motor commands through reinforcement involves distributed neural circuits beyond known frontal-striatal pathways, yet a comprehensive understanding of this broader neural architecture remains elusive. Here, using human functional MRI (N = 46, 27 females) and manifold learning techniques, we identified a low-dimensional neural space that captured the dynamic changes in whole-brain functional organization during a reward-based trajectory learning task. By quantifying participants' learning rates through an Actor-Critic model, we discovered that periods of accelerated learning were characterized by significant manifold contractions across multiple brain regions, including areas of limbic and hippocampal cortex, as well as the cerebellum. This contraction reflected enhanced network integration, with notably stronger connectivity between several of these regions and the sensorimotor cerebellum correlating with higher learning rates. These findings challenge the traditional view of the cerebellum as solely involved in error-based learning, supporting the emerging view that it coordinates with other brain regions during reinforcement learning.Significance Statement This study reveals how distributed brain systems, including the cerebellum and hippocampus, alter their functional connectivity to support motor learning through reinforcement. Using advanced manifold learning techniques on functional MRI data, we examined changes in regional connectivity during reward-based learning and their relationship to learning rate. For several brain regions, we found that periods of heightened learning were associated with increased cerebellar connectivity, suggesting a key role for the cerebellum in reward-based motor learning. These findings challenge the traditional view of the cerebellum as solely involved in supervised (error-based) learning and add to a growing rodent literature supporting a role for cerebellar circuits in reward-driven learning.

Less is more: selection from a small set of options improves BCI velocity control

Objective.Decoding algorithms used in invasive brain-computer interfaces (iBCIs) typically convert neural activity into continuously varying velocity commands. We hypothesized that putting constraints on which decoded velocity commands are permissible could improve user performance. To test this hypothesis, we designed the discrete direction selection (DDS) decoder, which uses neural activity to select among a small menu of preset cursor velocities.Approach. We tested DDS in a closed-loop cursor control task against many common continuous velocity decoders in both a human-operated real-time iBCI simulator (the jaBCI) and in a monkey using an iBCI. In the jaBCI, we compared performance across four visits by each of 48 naïve, able-bodied human subjects using either DDS, direct regression with assist (an affine map from neural activity to cursor velocity, DR-A), ReFIT, or the velocity Kalman Filter (vKF). In a follow up study to verify the jaBCI results, we compared a monkey's performance using an iBCI with either DDS or the Wiener filter decoder (a direct regression decoder that includes time history, WF).Main Result. In the jaBCI, DDS substantially outperformed all other decoders with 93% mean targets hit per visit compared to DR-A, ReFIT, and vKF with 56%, 39%, and 26% mean targets hit, respectively. With the iBCI, the monkey achieved a 61% success rate with DDS and a 37% success rate with WF.Significance. Discretizing the decoded velocity with DDS effectively traded high resolution velocity commands for less tortuous and lower noise trajectories, highlighting the potential benefits of discretization in simplifying online BCI control.

Pathway-like Activation of 3D Neuronal Constructs with an Optical Interface

Three-dimensional neuronal organoids, spheroids, and tissue mimics are increasingly used to model cognitive processes in vitro. These 3D constructs are also used to model the effects of neurological and psychiatric disorders and to perform computational tasks. The brain's complex network of neurons is activated via feedforward sensory pathways. Therefore, an interface to 3D constructs that models sensory pathway-like inputs is desirable. In this work, an optical interface for 3D neuronal constructs was developed. Dendrites and axons extended by cortical neurons within the 3D constructs were guided into microchannel-confined bundles. These neurite bundles were then optogenetically stimulated, and evoked responses were evaluated by calcium imaging. Optical stimulation was designed to deliver distinct input patterns to the network in the 3D construct, mimicking sensory pathway inputs to cortical areas in the intact brain. Responses of the network to the stimulation possessed features of neuronal population code, including separability by input pattern and mixed selectivity of individual neurons. This work represents the first demonstration of a pathway-like activation of networks in 3D constructs. Another innovation of this work is the development of an all-optical interface to 3D neuronal constructs, which does not require the use of expensive microelectrode arrays. This interface may enable the use of 3D neuronal constructs for investigations into cortical information processing. It may also enable studies into the effects of neurodegenerative or psychiatric disorders on cortical computation.

The geometry of the neural state space of decisions

How do populations of neurons collectively encode and process information during cognitive tasks? We analyze high-yield population recordings from the macaque lateral intraparietal area (LIP) during a reaction-time random-dot-motion direction-discrimination task. We find that the trajectories of neural population activity patterns during single decisions lie within a curved two-dimensional manifold. The reaction time of trajectories systematically varies along one dimension, such that slow and fast decisions trace distinct activity patterns. Trajectories transition from a deliberation stage, in which they are noisy and remain similar between the choices, to a commitment stage, in which they are far less noisy and diverge sharply for the different choices. The deliberation phase is pronounced for slower decisions and gradually diminishes as reaction time decreases. A mechanistic circuit model provides an explanation for the observed properties, and suggests the transition between stages represents a transition from more sensory-driven to more circuit-driven dynamics. It yields two striking predictions we verify in the data. First, whether neurons are more choice selective for slow or fast trials varies systematically with the retinotopic location of their response fields. Second, the slower the trial, the more saccades undershoot the choice target. The results highlight the roles of distributed and dynamic activity patterns and intrinsic circuit dynamics in the population implementation of a cognitive task.

Facial gestures are enacted via a cortical hierarchy of dynamic and stable codes

Successful communication requires the generation and perception of a shared set of signals. Facial gestures are one fundamental set of communicative behaviors in primates, generated through the dynamic arrangement of dozens of fine muscles. While much progress has been made uncovering the neural mechanisms of face perception, little is known about those controlling facial gesture production. Commensurate with the importance of facial gestures in daily social life, anatomical work has shown that facial muscles are under direct control from multiple cortical regions, including primary and premotor in lateral frontal cortex, and cingulate in medial frontal cortex. Furthermore, neuropsychological evidence from focal lesion patients has suggested that lateral cortex controls voluntary movements, and medial emotional expressions. Here we show that lateral and medial cortical face motor regions encode both types of gestures. They do so through unique temporal activity patterns, distinguishable well-prior to movement onset. During gesture production, cortical regions encoded facial kinematics in a context-dependent manner. Our results show how cortical regions projecting in parallel downstream, but each situated at a different level of a posterior-anterior hierarchy form a continuum of gesture coding from dynamic to temporally stable, in order to produce context-related, coherent motor outputs during social communication.

Transformations in prefrontal ensemble activity underlying rapid threat avoidance learning

To survive, animals must rapidly learn to avoid aversive outcomes by predicting threats and taking preemptive actions to avoid them. Often, this involves identifying locations that are safe in the context of specific, impending threats and remaining in those locations until the threat passes. Thus, animals quickly learn how threat-predicting cues alter the implications of entering or leaving a safe location. The prelimbic subregion (PL) of the medial prefrontal cortex (mPFC) integrates learned associations to influence threat avoidance strategies.1,2,3,4,5,6,7,8,9,10,11,12 These processes become dysfunctional in mood and anxiety disorders, which are characterized by excessive avoidance.13,14 Prior work largely focused on the role of PL activity in avoidance behaviors that are fully established,12,15,16,17 leaving the prefrontal mechanisms driving avoidance learning poorly understood. To determine when and how learning-related changes emerge, we recorded PL neural activity using miniscope Ca2+ imaging18,19 as mice rapidly learned to avoid a cued threat by accessing a safe location. Early in learning, PL population dynamics accurately predicted trial outcomes and tracked individual learning rates. Once behavioral performance stabilized, neurons that encoded avoidance behaviors or risky exploration were strongly modulated by the conditioned tone. Our findings reveal that, during avoidance learning, the PL rapidly generates novel representations of whether mice will take avoidance or exploratory actions during an impending threat. We reveal the sequence of transformations that unfold in the PL and how they relate to individual learning rates.

Neural Correlation Integrated Adaptive Point Process Filtering on Population Spike Trains

Brain encodes information through neural spiking activities that modulate external environmental stimuli and underlying internal states. Population of neurons coordinate through functional connectivity to plan movement trajectories and accurately activate neuromuscular activities. Motor Brain-machine interface (BMI) is a platform to study the relationship between behaviors and neural ensemble activities. In BMI, point process filters model directly on spike timings to extract underlying states such as motion intents from observed multi-neuron spike trains. However, these methods assume the encoded information from individual neurons is conditionally independent, which leads to less precise estimation. It is necessary to incorporate functional neural connectivity into a point process filter to improve the state estimation. In this paper, we propose a neural correlation integrated adaptive point process filter (CIPPF) that can incorporate the information from functional neural connectivity from population spike trains in a recursive Bayesian framework. Functional neural connectivity information is approximated by an artificial neural network to provide extra updating information for the posterior estimation. Gaussian approximation is applied on the probability distribution to obtain a closed-form solution. Our proposed method is validated on both simulation and real data collected from the rat two-lever discrimination task. Due to the simultaneous modeling of functional neural connectivity and single neuronal tuning properties, the proposed method shows better decoding performance. This suggests the possibility to improve BMI performance by processing the coordinated neural population activities.

Motor somatotopy impacts imagery strategy success in human intracortical brain-computer interfaces

Objective:The notion of a somatotopically organized motor cortex, with movements of different body parts being controlled by spatially distinct areas of cortex, is well known. However, recent studies have challenged this notion and suggested a more distributed representation of movement control. This shift in perspective has significant implications, particularly when considering the implantation location of electrode arrays for intracortical brain-computer interfaces (iBCIs). We sought to evaluate whether the location of neural recordings from the precentral gyrus, and thus the underlying somatotopy, has any impact on the imagery strategies that can enable successful iBCI control.Approach:Three individuals with a spinal cord injury were enrolled in an ongoing clinical trial of an iBCI. Participants had two intracortical microelectrode arrays implanted in the arm and/or hand areas of the precentral gyrus based on presurgical functional imaging. Neural data were recorded while participants attempted to perform movements of the hand, wrist, elbow, and shoulder.Main results:We found that electrode arrays that were located more medially recorded significantly more activity during attempted proximal arm movements (elbow, shoulder) than did lateral arrays, which captured more activity related to attempted distal arm movements (hand, wrist). We also evaluated the relative contribution from the two arrays implanted in each participant to decoding accuracy during calibration of an iBCI decoder for translation and grasping tasks. For both task types, imagery strategy (e.g. reaching vs wrist movements) had a significant impact on the relative contributions of each array to decoding. Overall, we found some evidence of broad tuning to arm and hand movements; however, there was a clear bias in the amount of information accessible about each movement type in spatially distinct areas of cortex.Significance:These results demonstrate that classical concepts of somatotopy can have real consequences for iBCI use, and highlight the importance of considering somatotopy when planning iBCI implantation.

Analyzing the brain's dynamic response to targeted stimulation using generative modeling

Generative models of brain activity have been instrumental in testing hypothesized mechanisms underlying brain dynamics against experimental datasets. Beyond capturing the key mechanisms underlying spontaneous brain dynamics, these models hold an exciting potential for understanding the mechanisms underlying the dynamics evoked by targeted brain stimulation techniques. This paper delves into this emerging application, using concepts from dynamical systems theory to argue that the stimulus-evoked dynamics in such experiments may be shaped by new types of mechanisms distinct from those that dominate spontaneous dynamics. We review and discuss (a) the targeted experimental techniques across spatial scales that can both perturb the brain to novel states and resolve its relaxation trajectory back to spontaneous dynamics and (b) how we can understand these dynamics in terms of mechanisms using physiological, phenomenological, and data-driven models. A tight integration of targeted stimulation experiments with generative quantitative modeling provides an important opportunity to uncover novel mechanisms of brain dynamics that are difficult to detect in spontaneous settings.

Hierarchical emergence of opponent coding in auditory belt cortex

We recorded from neurons in primary auditory cortex (A1) and middle-lateral belt area (ML) while rhesus macaques either discriminated amplitude-modulated noise (AM) from unmodulated noise or passively heard the same stimuli. We used several post hoc pooling models to investigate the ability of auditory cortex to leverage population coding for AM detection. We find that pooled-response AM detection is better in the active condition than the passive condition, and better using rate-based coding than synchrony-based coding. Neurons can be segregated into two classes based on whether they increase (INC) or decrease (DEC) their firing rate in response to increasing modulation depth. In these samples, A1 had relatively fewer DEC neurons (26%) than ML (45%). When responses were pooled without segregating these classes, AM detection using rate-based coding was much better in A1 than in ML, but when pooling only INC neurons, AM detection in ML approached that found in A1. Pooling only DEC neurons resulted in impaired AM detection in both areas. To investigate the role of DEC neurons, we devised two pooling methods that opposed DEC and INC neurons-a direct subtractive method and a two-pool push-pull opponent method. Only the push-pull opponent method resulted in superior AM detection relative to indiscriminate pooling. In the active condition, the opponent method was superior to pooling only INC neurons during the late portion of the response in ML. These results suggest that the increasing prevalence of the DEC response type in ML can be leveraged by appropriate methods to improve AM detection.NEW & NOTEWORTHY We used several post hoc pooling models to investigate the ability of primate auditory cortex to leverage population coding in the detection of amplitude-modulated sounds. When cells are indiscriminately pooled, primary auditory cortex (A1) detects amplitude-modulated sounds better than middle-lateral belt (ML). When cells that decrease firing rate with increasing modulation depth are excluded, or used in a push-pull opponent fashion, detection is similar in the two areas, and macaque behavior can be approximated using reasonably sized pools.

Latent circuit inference from heterogeneous neural responses during cognitive tasks

Higher cortical areas carry a wide range of sensory, cognitive and motor signals mixed in heterogeneous responses of single neurons tuned to multiple task variables. Dimensionality reduction methods that rely on correlations between neural activity and task variables leave unknown how heterogeneous responses arise from connectivity to drive behavior. We develop the latent circuit model, a dimensionality reduction approach in which task variables interact via low-dimensional recurrent connectivity to produce behavioral output. We apply the latent circuit inference to recurrent neural networks trained to perform a context-dependent decision-making task and find a suppression mechanism in which contextual representations inhibit irrelevant sensory responses. We validate this mechanism by confirming the behavioral effects of patterned connectivity perturbations predicted by the latent circuit model. We find similar suppression of irrelevant sensory responses in the prefrontal cortex of monkeys performing the same task. We show that incorporating causal interactions among task variables is critical for identifying behaviorally relevant computations from neural response data.

Uncovering Dynamical Equations of Stochastic Decision Models Using Data-Driven SINDy Algorithm

Decision formation in perceptual decision making involves sensory evidence accumulation instantiated by the temporal integration of an internal decision variable toward some decision criterion or threshold, as described by sequential sampling theoretical models. The decision variable can be represented in the form of experimentally observable neural activities. Hence, elucidating the appropriate theoretical model becomes crucial to understanding the mechanisms underlying perceptual decision formation. Existing computational methods are limited to either fitting of choice behavioral data or linear model estimation from neural activity data. In this work, we made use of sparse identification of nonlinear dynamics (SINDy), a data-driven approach, to elucidate the deterministic linear and nonlinear components of often-used stochastic decision models within reaction time task paradigms. Based on the simulated decision variable activities of the models and assuming the noise coefficient term is known beforehand, SINDy, enhanced with approaches using multiple trials, could readily estimate the deterministic terms in the dynamical equations, choice accuracy, and decision time of the models across a range of signal-to-noise ratio values. In particular, SINDy performed the best using the more memory-intensive multi-trial approach while trial-averaging of parameters performed more moderately. The single-trial approach, although expectedly not performing as well, may be useful for real-time modeling. Taken together, our work offers alternative approaches for SINDy to uncover the dynamics in perceptual decision making and, more generally, for first-passage time problems.

Excitation-Inhibition Balance Controls Information Encoding in Neural Populations

Understanding how the complex connectivity structure of the brain shapes its information-processing capabilities is a long-standing question. By focusing on a paradigmatic architecture, we study how the neural activity of excitatory and inhibitory populations encodes information on external signals. We show that at long times information is maximized at the edge of stability, where inhibition balances excitation, both in linear and nonlinear regimes. In the presence of multiple external signals, this maximum corresponds to the entropy of the input dynamics. By analyzing the case of a prolonged stimulus, we find that stronger inhibition is instead needed to maximize the instantaneous sensitivity, revealing an intrinsic tradeoff between short-time responses and long-time accuracy. In agreement with recent experimental findings, our results pave the way for a deeper information-theoretic understanding of how the balance between excitation and inhibition controls optimal information-processing in neural populations.

Visual hallucinations in Parkinson's disease: spotlight on central cholinergic dysfunction

Visual hallucinations are a common non-motor feature of Parkinson's disease and have been associated with accelerated cognitive decline, increased mortality and early institutionalization. Despite their prevalence and negative impact on patient outcomes, the repertoire of treatments aimed at addressing this troubling symptom is limited. Over the past two decades, significant contributions have been made in uncovering the pathological and functional mechanisms of visual hallucinations, bringing us closer to the development of a comprehensive neurobiological framework. Convergent evidence now suggests that degeneration within the central cholinergic system may play a significant role in the genesis and progression of visual hallucinations. Here, we outline how cholinergic dysfunction may serve as a potential unifying neurobiological substrate underlying the multifactorial and dynamic nature of visual hallucinations. Drawing upon previous theoretical models, we explore the impact that alterations in cholinergic neurotransmission has on the core cognitive processes pertinent to abnormal perceptual experiences. We conclude by highlighting that a deeper understanding of cholinergic neurobiology and individual pathophysiology may help to improve established and emerging treatment strategies for the management of visual hallucinations and psychotic symptoms in Parkinson's disease.

Sensorimotor environment but not task rule reconfigures population dynamics in rhesus monkey posterior parietal cortex

Primates excel at mapping sensory inputs flexibly onto motor outcomes. We asked if the neural dynamics to support context-sensitive sensorimotor mapping generalizes or differs between different behavioral contexts that demand such flexibility. We compared reaching under mirror-reversed vision, a case of adaptation to a modified sensorimotor environment (SE), with anti reaching, a case of applying an abstract task rule (TR). While neural dynamics in monkey posterior parietal cortex show shifted initial states and non-aligned low-dimensional neural subspaces in the SE task, remapping is achieved in overlapping subspaces in the TR task. A recurrent neural network model demonstrates how output constraints mimicking SE and TR tasks are sufficient to generate the two fundamentally different neural computational dynamics. We conclude that sensorimotor remapping to implement an abstract task rule happens within the existing repertoire of neural dynamics, while compensation of perturbed sensory feedback requires exploration of independent neural dynamics in parietal cortex.

The role of oscillations in grid cells' toroidal topology

Persistent homology applied to the activity of grid cells in the Medial Entorhinal Cortex suggests that this activity lies on a toroidal manifold. By analyzing real data and a simple model, we show that neural oscillations play a key role in the appearance of this toroidal topology. To quantitatively monitor how changes in spike trains influence the topology of the data, we first define a robust measure for the degree of toroidality of a dataset. Using this measure, we find that small perturbations ( ~ 100 ms) of spike times have little influence on both the toroidality and the hexagonality of the ratemaps. Jittering spikes by  ~ 100-500 ms, however, destroys the toroidal topology, while still having little impact on grid scores. These critical jittering time scales fall in the range of the periods of oscillations between the theta and eta bands. We thus hypothesized that these oscillatory modulations of neuronal spiking play a key role in the appearance and robustness of toroidal topology and the hexagonal spatial selectivity is not sufficient. We confirmed this hypothesis using a simple model for the activity of grid cells, consisting of an ensemble of independent rate-modulated Poisson processes. When these rates were modulated by oscillations, the network behaved similarly to the real data in exhibiting toroidal topology, even when the position of the fields were perturbed. In the absence of oscillations, this similarity was substantially lower. Furthermore, we find that the experimentally recorded spike trains indeed exhibit temporal modulations at the eta and theta bands, and that the ratio of the power in the eta band to that of the theta band, [Formula: see text], correlates with the critical jittering time at which the toroidal topology disappears.

Anesthesia alters complexity of spontaneous and stimulus-related neuronal firing patterns in rat visual cortex

Complexity of neuronal firing patterns may serve as an indicator of sensory information processing across different states of consciousness. Recent studies have shown that spontaneous changes in brain states can occur during general anesthesia, which may influence neuronal complexity and the state of consciousness. In this study, we investigated how the firing patterns of cortical neurons, both at rest and during visual stimulation, are affected by spontaneously changing brain states under varying levels of anesthesia. Extracellular unit activity was measured in the primary visual cortex of unrestrained rats as the inhaled concentration of desflurane was incrementally reduced to 6%, 4%, 2%, and 0%. Using dimensionality reduction and density-based clustering on individual unit activities, we identified five distinct population states, which underwent dynamic transitions independent of the anesthetic level during both resting and stimulus conditions. One population state that occurred mainly in deep anesthesia exhibited a paradoxically increased number of active neurons and asynchronous spiking, suggesting a spontaneous reversal towards an awake-like condition. However, this was contradicted by the observation of low neuronal complexity in both spontaneous and stimulus-related spike activity, which more closely aligns with unconsciousness. Our findings reveal that transient neuronal states with distinct spiking patterns can emerge in visual cortex at constant anesthetic concentrations. The reduced complexity in states associated with deep anesthesia likely indicates a disruption of conscious sensory information processing.

Intracellular Lactate Dynamics in Drosophila Glutamatergic Neurons

Rates of lactate production and consumption reflect the metabolic state of many cell types, including neurons. Here, we investigate the effects of nutrient deprivation on lactate dynamics in Drosophila glutamatergic neurons by leveraging the limiting effects of the diffusion barrier surrounding cells in culture. We found that neurons constitutively consume lactate when availability of trehalose, the glucose disaccharide preferred by insects, is limited by the diffusion barrier. Acute mechanical disruption of the barrier reduced this reliance on lactate. Through kinetic modeling and experimental validation, we demonstrate that neuronal lactate consumption rates correlate inversely with their mitochondrial density. Further, we found that lactate levels in neurons exhibited temporal correlations that allowed prediction of cytosolic lactate dynamics after the disruption of the diffusion barrier from pre-perturbation lactate fluctuations. Collectively, our findings reveal the influence of diffusion barriers on neuronal metabolic preferences, and demonstrate the existence of temporal correlations between lactate dynamics under conditions of nutrient deprivation and those evoked by the subsequent restoration of nutrient availability.

Cannabinoid Modulation of Central Amygdala Population Dynamics During Threat Investigation

Cannabinoids modulate innate avoidance, threat-reactivity, and stress adaptations via modulation amygdala-associated circuits; however, the mechanisms by which cannabinoids modulate amygdala representation of threat-related behavior are not known. We show that cannabinoid administration increases the activity of central amygdala (CeA) somatostatin neurons (SOM) and alters basal network dynamics in a manner supporting generation of antagonistic sub-ensembles within the SOM population. Moreover, diverging neuronal population trajectory dynamics and enhanced antagonistic sub-ensemble representation of threat-related behaviors, and enhanced threat-related location representation, were also observed. Lastly, cannabinoid administration increased the proportion of SOM neurons exhibiting multidimensional representation of threat-related behaviors and behavior-location conjunction. While cannabinoid receptor activation ex vivo suppressed excitatory inputs to SOM neurons, our data suggest preferential suppression of local GABA release subserves cannabinoid activation of CeA SOM neurons. These data provide insight into how cannabinoid-mediated presynaptic suppression transforms postsynaptic population dynamics and reveal cellular mechanisms by which cannabinoids could affect threat-reactivity.

Investigating the intrinsic top-down dynamics of deep generative models

Hierarchical generative models can produce data samples based on the statistical structure of their training distribution. This capability can be linked to current theories in computational neuroscience, which propose that spontaneous brain activity at rest is the manifestation of top-down dynamics of generative models detached from action-perception cycles. A popular class of hierarchical generative models is that of Deep Belief Networks (DBNs), which are energy-based deep learning architectures that can learn multiple levels of representations in a completely unsupervised way exploiting Hebbian-like learning mechanisms. In this work, we study the generative dynamics of a recent extension of the DBN, the iterative DBN (iDBN), which more faithfully simulates neurocognitive development by jointly tuning the connection weights across all layers of the hierarchy. We characterize the number of states visited during top-down sampling and investigate whether the heterogeneity of visited attractors could be increased by initiating the generation process from biased hidden states. To this end, we train iDBN models on well-known datasets containing handwritten digits and pictures of human faces, and show that the ability to generate diverse data prototypes can be enhanced by initializing top-down sampling from "chimera states", which represent high-level features combining multiple abstract representations of the sensory data. Although the models are not always able to transition between all potential target states within a single-generation trajectory, the iDBN shows richer top-down dynamics in comparison to a shallow generative model (a single-layer Restricted Bolzamann Machine). We further show that the generated samples can be used to support continual learning through generative replay mechanisms. Our findings suggest that the top-down dynamics of hierarchical generative models is significantly influenced by the shape of the energy function, which depends both on the depth of the processing architecture and on the statistical structure of the sensory data.

Heterogeneous and higher-order cortical connectivity undergirds efficient, robust, and reliable neural codes

We hypothesized that the heterogeneous architecture of biological neural networks provides a substrate to regulate the well-known tradeoff between robustness and efficiency, thereby allowing different subpopulations of the same network to optimize for different objectives. To distinguish between subpopulations, we developed a metric based on the mathematical theory of simplicial complexes that captures the complexity of their connectivity by contrasting its higher-order structure to a random control and confirmed its relevance in several openly available connectomes. Using a biologically detailed cortical model and an electron microscopic dataset, we showed that subpopulations with low simplicial complexity exhibit efficient activity. Conversely, subpopulations of high simplicial complexity play a supporting role in boosting the reliability of the network as a whole, softening the robustness-efficiency tradeoff. Crucially, we found that both types of subpopulations can and do coexist within a single connectome in biological neural networks, due to the heterogeneity of their connectivity.

Modeling Latent Neural Dynamics with Gaussian Process Switching Linear Dynamical Systems

Understanding how the collective activity of neural populations relates to computation and ultimately behavior is a key goal in neuroscience. To this end, statistical methods which describe high-dimensional neural time series in terms of low-dimensional latent dynamics have played a fundamental role in characterizing neural systems. Yet, what constitutes a successful method involves two opposing criteria: (1) methods should be expressive enough to capture complex nonlinear dynamics, and (2) they should maintain a notion of interpretability often only warranted by simpler linear models. In this paper, we develop an approach that balances these two objectives: the Gaussian Process Switching Linear Dynamical System (gpSLDS). Our method builds on previous work modeling the latent state evolution via a stochastic differential equation whose nonlinear dynamics are described by a Gaussian process (GP-SDEs). We propose a novel kernel function which enforces smoothly interpolated locally linear dynamics, and therefore expresses flexible - yet interpretable - dynamics akin to those of recurrent switching linear dynamical systems (rSLDS). Our approach resolves key limitations of the rSLDS such as artifactual oscillations in dynamics near discrete state boundaries, while also providing posterior uncertainty estimates of the dynamics. To fit our models, we leverage a modified learning objective which improves the estimation accuracy of kernel hyperparameters compared to previous GP-SDE fitting approaches. We apply our method to synthetic data and data recorded in two neuroscience experiments and demonstrate favorable performance in comparison to the rSLDS.

Progressively shifting patterns of co-modulation among premotor cortex neurons carry dynamically similar signals during action execution and observation

Many neurons in the premotor cortex show firing rate modulation whether the subject performs an action or observes another individual performing a similar action. Although such "mirror neurons" have been thought to have highly congruent discharge during execution and observation, many if not most actually show non-congruent activity. Studies of neuronal populations active during both execution and observation have shown that the most prevalent patterns of co-modulation-captured as neural trajectories-pass through subspaces which are shared in part, but in part are visited exclusively during either execution or observation. These studies focused on reaching movements for which low-dimensional neural trajectories exhibit comparatively simple dynamical motifs. But the neural dynamics of hand movements are more complex. We developed a novel approach to examine prevalent patterns of co-modulation during execution and observation of a task that involved reaching, grasping, and manipulation. Rather than following neural trajectories in subspaces that contain their entire time course, we identified time series of instantaneous subspaces, calculated principal angles among them, sampled trajectory segments at the times of selected behavioral events, and projected those segments into the time series of instantaneous subspaces. We found that instantaneous neural subspaces most often remained distinct during execution versus observation. Nevertheless, latent dynamics during execution and observation could be partially aligned with canonical correlation, indicating some similarity of the relationships among neural representations of different movements relative to one another during execution and observation. We also found that during action execution, mirror neurons showed consistent patterns of co-modulation both within and between sessions, but other non-mirror neurons that were modulated only during action execution and not during observation showed considerable variability of co-modulation.

Dynamic tracking of objects in the macaque dorsomedial frontal cortex

A central tenet of cognitive neuroscience is that humans build an internal model of the external world and use mental simulation of the model to perform physical inferences. Decades of human experiments have shown that behaviors in many physical reasoning tasks are consistent with predictions from the mental simulation theory. However, evidence for the defining feature of mental simulation - that neural population dynamics reflect simulations of physical states in the environment - is limited. We test the mental simulation hypothesis by combining a naturalistic ball-interception task, large-scale electrophysiology in non-human primates, and recurrent neural network modeling. We find that neurons in the monkeys' dorsomedial frontal cortex (DMFC) represent task-relevant information about the ball position in a multiplexed fashion. At a population level, the activity pattern in DMFC comprises a low-dimensional neural embedding that tracks the ball both when it is visible and invisible, serving as a neural substrate for mental simulation. A systematic comparison of different classes of task-optimized RNN models with the DMFC data provides further evidence supporting the mental simulation hypothesis. Our findings provide evidence that neural dynamics in the frontal cortex are consistent with internal simulation of external states in the environment.

Biologically inspired heterogeneous learning for accurate, efficient and low-latency neural network

The pursuit of artificial neural networks that mirror the accuracy, efficiency and low latency of biological neural networks remains a cornerstone of artificial intelligence (AI) research. Here, we incorporated recent neuroscientific findings of self-inhibiting autapse and neuron heterogeneity for innovating a spiking neural network (SNN) with enhanced learning and memorizing capacities. A bi-level programming paradigm was formulated to respectively learn neuron-level biophysical variables and network-level synapse weights for nested heterogeneous learning. We successfully demonstrated that our biologically inspired neuron model could reproduce neural statistics at both individual and group levels, contributing to the effective decoding of brain-computer interface data. Furthermore, the heterogeneous SNN showed higher accuracy (1%-10% improvement), superior efficiency (maximal 17.83-fold reduction in energy) and lower latency (maximal 5-fold improvement) in performing several AI tasks. For the first time, we benchmarked SNN for conducting cell type identification from scRNA-seq data. The proposed model correctly identified very rare cell types associated with severe brain diseases where typical SNNs failed.

Rastermap: a discovery method for neural population recordings

Neurophysiology has long progressed through exploratory experiments and chance discoveries. Anecdotes abound of researchers listening to spikes in real time and noticing patterns of activity related to ongoing stimuli or behaviors. With the advent of large-scale recordings, such close observation of data has become difficult. To find patterns in large-scale neural data, we developed 'Rastermap', a visualization method that displays neurons as a raster plot after sorting them along a one-dimensional axis based on their activity patterns. We benchmarked Rastermap on realistic simulations and then used it to explore recordings of tens of thousands of neurons from mouse cortex during spontaneous, stimulus-evoked and task-evoked epochs. We also applied Rastermap to whole-brain zebrafish recordings; to wide-field imaging data; to electrophysiological recordings in rat hippocampus, monkey frontal cortex and various cortical and subcortical regions in mice; and to artificial neural networks. Finally, we illustrate high-dimensional scenarios where Rastermap and similar algorithms cannot be used effectively.

A combinatorial neural code for long-term motor memory

Motor skill repertoire can be stably retained over long periods, but the neural mechanism that underlies stable memory storage remains poorly understood1-8. Moreover, it is unknown how existing motor memories are maintained as new motor skills are continuously acquired. Here we tracked neural representation of learned actions throughout a significant portion of the lifespan of a mouse and show that learned actions are stably retained in combination with context, which protects existing memories from erasure during new motor learning. We established a continual learning paradigm in which mice learned to perform directional licking in different task contexts while we tracked motor cortex activity for up to six months using two-photon imaging. Within the same task context, activity driving directional licking was stable over time with little representational drift. When learning new task contexts, new preparatory activity emerged to drive the same licking actions. Learning created parallel new motor memories instead of modifying existing representations. Re-learning to make the same actions in the previous task context re-activated the previous preparatory activity, even months later. Continual learning of new task contexts kept creating new preparatory activity patterns. Context-specific memories, as we observed in the motor system, may provide a solution for stable memory storage throughout continual learning.

Deep learning models reveal the link between dynamic brain connectivity patterns and states of consciousness

Decoding states of consciousness from brain activity is a central challenge in neuroscience. Dynamic functional connectivity (dFC) allows the study of short-term temporal changes in functional connectivity (FC) between distributed brain areas. By clustering dFC matrices from resting-state fMRI, we previously described "brain patterns" that underlie different functional configurations of the brain at rest. The networks associated with these patterns have been extensively analyzed. However, the overall dynamic organization and how it relates to consciousness remains unclear. We hypothesized that deep learning networks would help to model this relationship. Recent studies have used low-dimensional variational autoencoders (VAE) to learn meaningful representations that can help explaining consciousness. Here, we investigated the complexity of selecting such a generative model to study brain dynamics, and extended the available methods for latent space characterization and modeling. Therefore, our contributions are threefold. First, compared with probabilistic principal component analysis and sparse VAE, we showed that the selected low-dimensional VAE exhibits balanced performance in reconstructing dFCs and classifying brain patterns. We then explored the organization of the obtained low-dimensional dFC latent representations. We showed how these representations stratify the dynamic organization of the brain patterns as well as the experimental conditions. Finally, we proposed to delve into the proposed brain computational model. We first applied a receptive field analysis to identify preferred directions in the latent space to move from one brain pattern to another. Then, an ablation study was achieved where we virtually inactivated specific brain areas. We demonstrated the model's efficiency in summarizing consciousness-specific information encoded in key inter-areal connections, as described in the global neuronal workspace theory of consciousness. The proposed framework advocates the possibility of developing an interpretable computational brain model of interest for disorders of consciousness, paving the way for a dynamic diagnostic support tool.

Latent learning drives sleep-dependent plasticity in distinct CA1 subpopulations

Latent learning is a process that enables the brain to transform experiences into "cognitive maps," a form of implicit memory, without requiring reinforced training. To investigate its neural mechanisms, we record from hippocampal neurons in mice during latent learning of spatial maps and observe that the high-dimensional neural state space gradually transforms into a low-dimensional manifold that closely resembles the physical environment. This transformation process is associated with the neural reactivation of navigational experiences during sleep. Additionally, we identify a subset of hippocampal neurons that, rather than forming place fields in a novel environment, maintain weak spatial tuning but gradually develop correlated activity with other neurons. The elevated correlation introduces redundancy into the ensemble code, transforming the neural state space into a low-dimensional manifold that effectively links discrete place fields of place cells into a map-like structure. These results suggest a potential mechanism for latent learning of spatial maps in the hippocampus.

Predicting Future Development of Stress-Induced Anhedonia From Cortical Dynamics and Facial Expression

The current state of mental health treatment for individuals diagnosed with major depressive disorder leaves billions of individuals with first-line therapies that are ineffective or burdened with undesirable side effects. One major obstacle is that distinct pathologies may currently be diagnosed as the same disease and prescribed the same treatments. The key to developing antidepressants with ubiquitous efficacy is to first identify a strategy to differentiate between heterogeneous conditions. Major depression is characterized by hallmark features such as anhedonia and a loss of motivation1,2, and it has been recognized that even among inbred mice raised under identical housing conditions, we observe heterogeneity in their susceptibility and resilience to stress3. Anhedonia, a condition identified in multiple neuropsychiatric disorders, is described as the inability to experience pleasure and is linked to anomalous medial prefrontal cortex (mPFC) activity4. The mPFC is responsible for higher order functions5-8, such as valence encoding; however, it remains unknown how mPFC valence-specific neuronal population activity is affected during anhedonic conditions. To test this, we implemented the unpredictable chronic mild stress (CMS) protocol9-11 in mice and examined hedonic behaviors following stress and ketamine treatment. We used unsupervised clustering to delineate individual variability in hedonic behavior in response to stress. We then performed in vivo 2-photon calcium imaging to longitudinally track mPFC valence-specific neuronal population dynamics during a Pavlovian discrimination task. Chronic mild stress mice exhibited a blunted effect in the ratio of mPFC neural population responses to rewards relative to punishments after stress that rebounds following ketamine treatment. Also, a linear classifier revealed that we can decode susceptibility to chronic mild stress based on mPFC valence-encoding properties prior to stress-exposure and behavioral expression of susceptibility. Lastly, we utilized markerless pose tracking computer vision tools to predict whether a mouse would become resilient or susceptible based on facial expressions during a Pavlovian discrimination task. These results indicate that mPFC valence encoding properties and behavior are predictive of anhedonic states. Altogether, these experiments point to the need for increased granularity in the measurement of both behavior and neural activity, as these factors can predict the predisposition to stress-induced anhedonia.

Multiscale Cloud-Based Pipeline for Neuronal Electrophysiology Analysis and Visualization

Electrophysiology offers a high-resolution method for real-time measurement of neural activity. Longitudinal recordings from high-density microelectrode arrays (HD-MEAs) can be of considerable size for local storage and of substantial complexity for extracting neural features and network dynamics. Analysis is often demanding due to the need for multiple software tools with different runtime dependencies. To address these challenges, we developed an open-source cloud-based pipeline to store, analyze, and visualize neuronal electrophysiology recordings from HD-MEAs. This pipeline is dependency agnostic by utilizing cloud storage, cloud computing resources, and an Internet of Things messaging protocol. We containerized the services and algorithms to serve as scalable and flexible building blocks within the pipeline. In this paper, we applied this pipeline on two types of cultures, cortical organoids and ex vivo brain slice recordings to show that this pipeline simplifies the data analysis process and facilitates understanding neuronal activity.