Biology subjects

Dorkenwald, S.

Publications and source records attributed to Dorkenwald, S..

At least 19 recordsLinked to original sources

bioRxiv2026

Cell type-resolved connectomic reconstruction of the spinal dorsal horn reveals sensory neuron synaptic organization and modality-specific inhibition

Touch, temperature, chemical, and noxious stimuli are detected by more than a dozen peripheral sensory neuron subtypes that together mediate somatosensation. Somatosensory neurons form synapses in the spinal cord dorsal horn, the first site of central somatosensory processing, to engage the downstream circuitry that drives both spinal reflexes and ascending pathways for perception. Yet how sensory neuron synapses are organized into the spinal cord circuitry remains poorly understood. Here we present two connectomic tools for studying somatosensory circuits of the spinal cord dorsal horn: dSC1, a large-scale, synapse-resolved electron microscopy (EM) dataset of the mouse dorsal horn, and SPINE (SPinal cord Integrated NEuron classifier), a deep-learning pipeline that identifies major sensory neuron subtypes directly from EM images. Applying SPINE to the dSC1 dataset, we found marked heterogeneity in synaptic architecture across physiologically distinct sensory neuron subtypes: low-threshold mechanoreceptors form substantially more synapses and receive far more GABAergic presynaptic inhibition than high-threshold mechanoreceptors, thermoreceptors and nociceptors. Cell-type-resolved circuit reconstruction further identified five populations of spinal inhibitory neurons (IhNs) that form synapses onto sensory neuron axons, each preferentially driven by and presynaptically inhibiting a single sensory subtype - a homotypic, modality-specific inhibitory motif. Most of these IhNs also feedforwardly inhibit other sensory pathways - a heterotypic, cross modality motif. Together, dSC1 and SPINE enable cell-type-resolved spinal cord circuit reconstruction and reveal a dual-mode logic of sensory-evoked inhibition that shapes modality-specific signaling at the earliest stage of somatosensory processing.

neuroscience↗

bioRxiv2026

Connectome-based biophysical modeling of a figure-ground discrimination circuit

Figure-ground discrimination (FGD) via relative motion represents a highly conserved visual computation essential for breaking camouflage. Over forty years ago, the Reichardt-Poggio-Hausen (RPH) model proposed a circuit architecture to explain this phenomenon, relying on an inhibitory wide-field pool cell to aggregate global motion and subsequently inhibit elementary motion detectors. However, a biological implementation of the RPH model has remained elusive. Using Drosophila connectomics data and biophysically accurate modeling, we reveal a biological circuit implementing this canonical architecture. Strikingly, we find that while the dendrites of the elemental motion detectors, T4a, are tuned to motion, their axon terminals compute relative motion. This secondary computation is achieved via presynaptic shunting inhibition from the ventral Centrifugal Horizontal (vCH) cell, which acts as a global motion integrator that normalizes the signal carried by the T4 axon terminals during whole-field optic flow. We extend our model to include the motion-opponency and the binocular communication pathways and show that FGD depends on the T4a-vCH interactions. We further show that specific circuit adjustments are sufficient for our model to make predictions consistent with several physiological and behavioral experiments across different fly species. Ultimately, by uniting connectomic connectivity with morphologically accurate biophysics, our approach demonstrates how detailed, synapse-resolution models of the brain can yield fine-grained biological predictions.

neuroscience↗

bioRxiv2026

Organizational principles governing synapse types in a whole-brain connectome

Neurons are classically viewed as polarized elements in which dendrites receive synaptic input and axons provide output. However, many neurons contain intermingled presynaptic and postsynaptic sites, enabling non-canonical forms of connectivity beyond axo-dendritic interactions. How such mixed polarity shapes connectivity across an entire nervous system remains unknown. Here, using the complete Drosophila connectome of an adult brain (approximately 140,000 neurons and >80 million synapses), we systematically mapped synapse types across neuronal compartments, cell types, and brain regions. We found that mixed polarity is widespread yet highly structured across neuronal populations. Approximately one third of synapses between polarized neurons are non-canonical, including axo-axonic, dendrodendritic, and dendro-axonic interactions. Remarkably, synapse type can be predicted from a simple axis of neuronal complexity derived from morphology and connectivity. Low-complexity neurons, which dominate early visual circuits, preferentially interact through dendrites, whereas larger and more highly connected neurons enriched in central brain regions preferentially interact through axons. Mixed polarity also strongly promotes reciprocal connectivity, which is dominated by non-canonical motifs. Finally, we identified postsynaptic terminals on presynaptic boutons as a widespread anatomical motif associated with local reciprocal and non-canonical interactions. Together, these findings reveal organizational principles linking neuronal architecture, synapse type, and circuit connectivity across a complete brain.

neuroscience↗

bioRxiv2026

Learning using switching synaptic plasticity rules

Synapses differ in molecular and ultrastructural state, but the computational role of this heterogeneity remains unclear and computational models do not account for it. Local Hebbian-like plasticity is well supported experimentally, whereas learning complex behaviors most likely requires non-local credit-assignment signals. Recent cortical electron microscopy data suggest that strong and weak synapses occupy distinct structural states, with large synapses often containing a spine apparatus that can alter calcium dynamics and plasticity. Here, we assigned different computational roles to different morphological synaptic types: we asked how networks with state-dependent plasticity operate and whether the mechanisms and structural features they develop are consistent with biological observations. We built a plasticity-switching recurrent neural network (psRNN) in which synaptic state is modeled by recurrent-weight thresholds. Synapses in the low-weight state use a local Hebbian-like rule, whereas synapses in the high-weight state receive an idealized non-local credit-assignment signal implemented with backpropagation through time. psRNNs learned cognitive tasks with temporal and working-memory demands in fewer trials than fully backpropagation-trained RNNs, despite routing BP-based updates to fewer recurrent weights. The advantage arose from three interacting mechanisms: 1) Hebbian-like updates exposed the credit-assignment component to several nearby parameter configurations, 2) the network dynamically built a task-relevant recurrent initialization, and 3) synapses switched into the credit-assignment subset before Hebbian growth destabilized learning. Structurally, trained psRNNs produced more anti-symmetric weights, lower effective rank, and stronger feedforward structure than control RNNs. Our results shed light on the computational role of heterogeneous synapses, which may regulate distinct types of learning signals, and we generate falsifiable predictions for connectomic analyses of recurrent cortical circuits.

neuroscience↗

bioRxiv2026

A quantitative census of millions of postsynaptic structures in a large electron microscopy volume of mouse visual cortex

Neurons display remarkable sub-cellular specificity in their synaptic targeting, which varies by cell type--for example, excitatory neurons prefer to target the spines of other excitatory cells. Modern electron microscopy connectomes enable the study of this sub-cellular specificity and its context in a circuit at unprecedented scale and resolution. However, this scale has also made it challenging to create accurate and efficient methods for classifying and segmenting fine cell components (including spines) across entire volumes. Here, we present a cost-efficient computational pipeline for classifying postsynaptic targets and segmenting structures such as spines. Our method relies only on having a mesh representation of a neuron and avoids processing image or segmentation data directly. Instead, we leverage tools from geometry processing to create features capturing the local geometry of a neurons surface. We couple this core technique with computational and storage optimizations, enabling reliable deployment over hundreds of thousands of neurons for a few hundred dollars in cloud compute cost. We then show that a simple classifier trained on the MICrONS mouse visual cortex dataset can use these mesh-based features to accurately classify synapses as targeting somas, dendritic shafts, or spines (weighted F1 score 0.961). Using this pipeline, we create a map of the postsynaptic structures at over 207.3 million synapses in MICrONS. We present an overview of this census of postsynaptic targeting, finding expected patterns (e.g., excitatory neurons preferentially targeting excitatory spines) as well as unexpected exceptions (e.g., Layer 5 near-projecting and Layer 6 corticothalamic cells often connecting to excitatory neuron shafts). We also demonstrate that these tools can be used to detect spines receiving multiple synaptic inputs, revealing surprising variability in their frequency across cells even within a cell type. We make our postsynaptic target predictions available for study, as well as the code for the computational pipeline and cloud deployment. Beyond MICrONS, we find that the model generalizes well to the H01 connectome without retraining (weighted F1 score 0.949), indicating that these tools will be useful in future connectomics reconstructions. More generally, our work demonstrates that representations derived from neuronal meshes can be a scalable and generalizable primitive for describing morphologies.

neuroscience↗

bioRxiv2025

New Synapse Detection in the Whole-Brain Connectome of Drosophila

The FlyWire Drosophila brain connectome1 is a graph of roughly 140K neurons and >50 million synaptic connections2,3 reconstructed from the FAFB EM dataset4. Challenges in synapse detection were identified for neurons with features such as dark cytosols, axo-axonic synapses, and complex polyadic synapses, due to limitations in ground truth data for these cells and the inherent complexity of these synapse types. To address these issues, we trained new neural networks using iteratively generated ground truth annotations and detected synapses across the entire FAFB dataset, producing what we refer to here as the Princeton synapses. These synapses were evaluated in both control regions, such as subareas of the mushroom body calyx and lateral horn, which were also chosen by Buhmann et al.3 for evaluation, as well as challenging regions, including Johnstons Organ neurons (JONs), photoreceptors, and other cell types. The new model shows significant improvements, achieving up to a 0.23 F-score increase in challenging areas, while maintaining performance in control regions. Princeton synapses also show an 8-9% improvement in neuron clustering within cell types and better left/right symmetry scores, especially for photoreceptors. Additionally, neuron type membership can be predicted from connectivity patterns alone with weighted F-scores of 0.93 for Princeton synapses versus 0.91 for Buhmann synapses. The updated Princeton synapses are now accessible via Codex (codex.flywire.ai).

neuroscience↗

bioRxiv2025

Sexually-dimorphic neurons in the Drosophila whole-brain connectome

Sexual dimorphisms in the brain arise from differences in cell number, morphology, and connectivity, and underlie sex-specific behaviors. In Drosophila, these differences are largely governed by the transcription factors Fruitless (Fru) and Doublesex (Dsx), expressed in neurons implicated in reproductive and social behaviors. However, how these neurons are integrated into the connectome to drive different behaviors in different sexes remains unclear. Here, we identify 98 putative Fru- or Dsx-expressing cell types ([~]1700 neurons) in the female whole-brain connectome by matching electron microscopy reconstructions to light microscopy images. We then analyze their connectivity within the whole-brain network. Fru/Dsx neurons are strongly interconnected, but are not part of a private network. We measure distances to the sensory periphery to examine multisensory interactions, and map connections to descending neurons that drive behavior. We further cluster Fru/Dsx neurons based on whole-brain connectivity, and uncover functional groupings. We analyze a subset of strongly connected Fru/Dsx neurons to make functional predictions, and test some of these with experiments. Together, this work provides a brain-wide framework for understanding the organization of sexually dimorphic circuits and establishes a resource for dissecting the neural basis of social and sexual behaviors.

neuroscience↗

bioRxiv2024

Comparative connectomics of the descending and ascending neurons of the Drosophila nervous system: stereotypy and sexual dimorphism

In most complex nervous systems there is a clear anatomical separation between the nerve cord, which contains most of the final motor outputs necessary for behaviour, and the brain. In insects, the neck connective is both a physical and information bottleneck connecting the brain and the ventral nerve cord (VNC, spinal cord analogue) and comprises diverse populations of descending (DN), ascending (AN) and sensory ascending neurons, which are crucial for sensorimotor signalling and control. Integrating three separate EM datasets, we now provide a complete connectomic description of the ascending and descending neurons of the female nervous system of Drosophila and compare them with neurons of the male nerve cord. Proofread neuronal reconstructions have been matched across hemispheres, datasets and sexes. Crucially, we have also matched 51% of DN cell types to light level data defining specific driver lines as well as classifying all ascending populations. We use these results to reveal the general architecture, tracts, neuropil innervation and connectivity of neck connective neurons. We observe connected chains of descending and ascending neurons spanning the neck, which may subserve motor sequences. We provide a complete description of sexually dimorphic DN and AN populations, with detailed analysis of circuits implicated in sex-related behaviours, including female ovipositor extrusion (DNp13), male courtship (DNa12/aSP22) and song production (AN hemilineage 08B). Our work represents the first EM-level circuit analyses spanning the entire central nervous system of an adult animal.

neuroscience↗

bioRxiv2023

The Synaptic Architecture of Layer 5 Thick Tufted Excitatory Neurons in the Visual Cortex of Mice

The neocortex is one of the most critical structures that makes us human, and it is involved in a variety of cognitive functions from perception to sensory integration and motor control. Composed of repeated modules, or microcircuits, the neocortex relies on distinct cell types as its fundamental building blocks. Despite significant progress in characterizing these cell types 1-5, an understanding of the complete synaptic partners associated with individual excitatory cell types remain elusive. Here, we investigate the connectivity of arguably the most well recognized and studied excitatory neuron in the neocortex: the thick tufted layer 5 pyramidal cell 6-10 also known as extra telencephalic (ET) 11 neurons. Although the synaptic interactions of ET neurons have been extensively explored, a comprehensive characterization of their local connectivity remains lacking. To address this knowledge gap, we leveraged a 1 mm3 electron microscopic (EM) dataset. We found that ET neurons primarily establish connections with inhibitory cells in their immediate vicinity. However, when they extend their axons to other cortical regions, they tend to connect more with excitatory cells. We also find that the inhibitory cells targeted by ET neurons are a specific group of cell types, and they preferentially inhibit ET cells. Finally, we observed that the most common excitatory targets of ET neurons are layer 5 IT neurons and layer 6 pyramidal cells, whereas synapses with other ET neurons are not as common. These findings challenge current views of the connectivity of ET neurons and suggest a circuit design that involves local competition among ET neurons and collaboration with other types of excitatory cells. Our results also highlight a specific circuit pattern where a subclass of excitatory cells forms a network with specific inhibitory cell types, offering a framework for exploring the connectivity of other types of excitatory cells.

neuroscience↗

bioRxiv2023

Network Statistics of the Whole-Brain Connectome of Drosophila

Brains comprise complex networks of neurons and connections. Network analysis applied to the wiring diagrams of brains can offer insights into how brains support computations and regulate information flow. The completion of the first whole-brain connectome of an adult Drosophila, the largest connectome to date, containing 130,000 neurons and millions of connections, offers an unprecedented opportunity to analyze its network properties and topological features. To gain insights into local connectivity, we computed the prevalence of two- and three-node network motifs, examined their strengths and neurotransmitter compositions, and compared these topological metrics with wiring diagrams of other animals. We discovered that the network of the fly brain displays rich club organization, with a large population (30% percent of the connectome) of highly connected neurons. We identified subsets of rich club neurons that may serve as integrators or broadcasters of signals. Finally, we examined subnetworks based on 78 anatomically defined brain regions or neuropils. These data products are shared within the FlyWire Codex and will serve as a foundation for models and experiments exploring the relationship between neural activity and anatomical structure.

neuroscience↗

bioRxiv2023

CAVE: Connectome Annotation Versioning Engine

Advances in Electron Microscopy, image segmentation and computational infrastructure have given rise to large-scale and richly annotated connectomic datasets which are increasingly shared across communities. To enable collaboration, users need to be able to concurrently create new annotations and correct errors in the automated segmentation by proofreading. In large datasets, every proofreading edit relabels cell identities of millions of voxels and thousands of annotations like synapses. For analysis, users require immediate and reproducible access to this constantly changing and expanding data landscape. Here, we present the Connectome Annotation Versioning Engine (CAVE), a computational infrastructure for immediate and reproducible connectome analysis in up-to petascale datasets ([~]1mm3) while proofreading and annotating is ongoing. For segmentation, CAVE provides a distributed proofreading infrastructure for continuous versioning of large reconstructions. Annotations in CAVE are defined by locations such that they can be quickly assigned to the underlying segment which enables fast analysis queries of CAVEs data for arbitrary time points. CAVE supports schematized, extensible annotations, so that researchers can readily design novel annotation types. CAVE is already used for many connectomics datasets, including the largest datasets available to date.

neuroscience↗

bioRxiv2023

Neuronal wiring diagram of an adult brain

Connections between neurons can be mapped by acquiring and analyzing electron microscopic (EM) brain images. In recent years, this approach has been applied to chunks of brains to reconstruct local connectivity maps that are highly informative, yet inadequate for understanding brain function more globally. Here, we present the first neuronal wiring diagram of a whole adult brain, containing 5x107 chemical synapses between [~]130,000 neurons reconstructed from a female Drosophila melanogaster. The resource also incorporates annotations of cell classes and types, nerves, hemilineages, and predictions of neurotransmitter identities. Data products are available by download, programmatic access, and interactive browsing and made interoperable with other fly data resources. We show how to derive a projectome, a map of projections between regions, from the connectome. We demonstrate the tracing of synaptic pathways and the analysis of information flow from inputs (sensory and ascending neurons) to outputs (motor, endocrine, and descending neurons), across both hemispheres, and between the central brain and the optic lobes. Tracing from a subset of photoreceptors all the way to descending motor pathways illustrates how structure can uncover putative circuit mechanisms underlying sensorimotor behaviors. The technologies and open ecosystem of the FlyWire Consortium set the stage for future large-scale connectome projects in other species.

neuroscience↗

bioRxiv2023

A consensus cell type atlas from multiple connectomes reveals principles of circuit stereotypy and variation

The fruit fly Drosophila melanogaster combines surprisingly sophisticated behaviour with a highly tractable nervous system. A large part of the flys success as a model organism in modern neuroscience stems from the concentration of collaboratively generated molecular genetic and digital resources. As presented in our FlyWire companion paper1, this now includes the first full brain connectome of an adult animal. Here we report the systematic and hierarchical annotation of this [~]130,000-neuron connectome including neuronal classes, cell types and developmental units (hemilineages). This enables any researcher to navigate this huge dataset and find systems and neurons of interest, linked to the literature through the Virtual Fly Brain database2. Crucially, this resource includes 4,552 cell types. 3,094 are rigorous consensus validations of cell types previously proposed in the "hemibrain" connectome3. In addition, we propose 1,458 new cell types, arising mostly from the fact that the FlyWire connectome spans the whole brain, whereas the hemibrain derives from a subvolume. Comparison of FlyWire and the hemibrain showed that cell type counts and strong connections were largely stable, but connection weights were surprisingly variable within and across animals. Further analysis defined simple heuristics for connectome interpretation: connections stronger than 10 unitary synapses or providing >1% of the input to a target cell are highly conserved. Some cell types showed increased variability across connectomes: the most common cell type in the mushroom body, required for learning and memory, is almost twice as numerous in FlyWire as the hemibrain. We find evidence for functional homeostasis through adjustments of the absolute amount of excitatory input while maintaining the excitation-inhibition ratio. Finally, and surprisingly, about one third of the cell types proposed in the hemibrain connectome could not yet be reliably identified in the FlyWire connectome. We therefore suggest that cell types should be defined to be robust to inter-individual variation, namely as groups of cells that are quantitatively more similar to cells in a different brain than to any other cell in the same brain. Joint analysis of the FlyWire and hemibrain connectomes demonstrates the viability and utility of this new definition. Our work defines a consensus cell type atlas for the fly brain and provides both an intellectual framework and open source toolchain for brain-scale comparative connectomics.

neuroscience↗

bioRxiv2023

Synaptic architecture of leg and wing motor control networks in Drosophila

Animal movement is controlled by motor neurons (MNs), which project out of the central nervous system to activate muscles. MN activity is coordinated by complex premotor networks that allow individual muscles to contribute to many different behaviors. Here, we use connectomics to analyze the wiring logic of premotor circuits controlling the Drosophila leg and wing. We find that both premotor networks cluster into modules that link MNs innervating muscles with related functions. Within most leg motor modules, the synaptic weights of each premotor neuron are proportional to the size of their target MNs, establishing a circuit basis for hierarchical MN recruitment. In contrast, wing premotor networks lack proportional synaptic connectivity, which may allow wing steering muscles to be recruited with different relative timing. By comparing the architecture of distinct limb motor control systems within the same animal, we identify common principles of premotor network organization and specializations that reflect the unique biomechanical constraints and evolutionary origins of leg and wing motor control.

neuroscience↗

bioRxiv2023

A leaky integrate-and-fire computational model based on the connectome of the entire adult Drosophila brain reveals insights into sensorimotor processing

The forthcoming assembly of the adult Drosophila melanogaster central brain connectome, containing over 125,000 neurons and 50 million synaptic connections, provides a template for examining sensory processing throughout the brain. Here, we create a leaky integrate-and-fire computational model of the entire Drosophila brain, based on neural connectivity and neurotransmitter identity, to study circuit properties of feeding and grooming behaviors. We show that activation of sugar-sensing or water-sensing gustatory neurons in the computational model accurately predicts neurons that respond to tastes and are required for feeding initiation. Computational activation of neurons in the feeding region of the Drosophila brain predicts those that elicit motor neuron firing, a testable hypothesis that we validate by optogenetic activation and behavioral studies. Moreover, computational activation of different classes of gustatory neurons makes accurate predictions of how multiple taste modalities interact, providing circuit-level insight into aversive and appetitive taste processing. Our computational model predicts that the sugar and water pathways form a partially shared appetitive feeding initiation pathway, which our calcium imaging and behavioral experiments confirm. Additionally, we applied this model to mechanosensory circuits and found that computational activation of mechanosensory neurons predicts activation of a small set of neurons comprising the antennal grooming circuit that do not overlap with gustatory circuits, and accurately describes the circuit response upon activation of different mechanosensory subtypes. Our results demonstrate that modeling brain circuits purely from connectivity and predicted neurotransmitter identity generates experimentally testable hypotheses and can accurately describe complete sensorimotor transformations.

neuroscience↗

bioRxiv2023

Integrating EM and Patch-seq data: Synaptic connectivity and target specificity of predicted Sst transcriptomic types

Neural circuit function is shaped both by the cell types that comprise the circuit and the connections between those cell types1. Neural cell types have previously been defined by morphology 2, 3, electrophysiology 4, 5, transcriptomic expression 6-8, connectivity 9-13, or even a combination of such modalities 14-16. More recently, the Patch-seq technique has enabled the characterization of morphology (M), electrophysiology (E), and transcriptomic (T) properties from individual cells 17-20. Using this technique, these properties were integrated to define 28, inhibitory multimodal, MET-types in mouse primary visual cortex 21. It is unknown how these MET-types connect within the broader cortical circuitry however. Here we show that we can predict the MET-type identity of inhibitory cells within a large-scale electron microscopy (EM) dataset and these MET-types have distinct ultrastructural features and synapse connectivity patterns. We found that EM Martinotti cells, a well defined morphological cell type 22, 23 known to be Somatostatin positive (Sst+) 24, 25, were successfully predicted to belong to Sst+ MET-types. Each identified MET-type had distinct axon myelination patterns and synapsed onto specific excitatory targets. Our results demonstrate that morphological features can be used to link cell type identities across imaging modalities, which enables further comparison of connectivity in relation to transcriptomic or electrophysiological properties. Furthermore, our results show that MET-types have distinct connectivity patterns, supporting the use of MET-types and connectivity to meaningfully define cell types.

neuroscience↗

bioRxiv2023

NEURD: A mesh decomposition framework forautomated proofreading and morphologicalanalysis of neuronal EM reconstructions

We are now in the era of millimeter-scale electron microscopy (EM) volumes collected at nanometer resolution (Shapson-Coe et al., 2021; Consortium et al., 2021). Dense reconstruction of cellular compartments in these EM volumes has been enabled by recent advances in Machine Learning (ML) (Lee et al., 2017; Wu et al., 2021; Lu et al., 2021; Macrina et al., 2021). Automated segmentation methods produce exceptionally accurate reconstructions of cells, but post-hoc proofreading is still required to generate large connectomes free of merge and split errors. The elaborate 3-D meshes of neurons in these volumes contain detailed morphological information at multiple scales, from the diameter, shape, and branching patterns of axons and dendrites, down to the fine-scale structure of dendritic spines. However, extracting these features can require substantial effort to piece together existing tools into custom workflows. Building on existing open-source software for mesh manipulation, here we present "NEURD", a software package that decomposes meshed neurons into compact and extensively-annotated graph representations. With these feature-rich graphs, we automate a variety of tasks such as state of the art automated proofreading of merge errors, cell classification, spine detection, axon-dendritic proximities, and other annotations. These features enable many downstream analyses of neural morphology and connectivity, making these massive and complex datasets more accessible to neuroscience researchers focused on a variety of scientific questions.

neuroscience↗

bioRxiv2023

Functional connectomics reveals general wiring rule in mouse visual cortex

Understanding the relationship between circuit connectivity and function is crucial for uncovering how the brain implements computation. In the mouse primary visual cortex (V1), excitatory neurons with similar response properties are more likely to be synaptically connected, but previous studies have been limited to within V1, leaving much unknown about broader connectivity rules. In this study, we leverage the millimeter-scale MICrONS dataset to analyze synaptic connectivity and functional properties of individual neurons across cortical layers and areas. Our results reveal that neurons with similar responses are preferentially connected both within and across layers and areas -- including feedback connections -- suggesting the universality of the like-to-like connectivity across the visual hierarchy. Using a validated digital twin model, we separated neuronal tuning into feature (what neurons respond to) and spatial (receptive field location) components. We found that only the feature component predicts fine-scale synaptic connections, beyond what could be explained by the physical proximity of axons and dendrites. We also found a higher-order rule where postsynaptic neuron cohorts downstream of individual presynaptic cells show greater functional similarity than predicted by a pairwise like-to-like rule. Notably, recurrent neural networks (RNNs) trained on a simple classification task develop connectivity patterns mirroring both pairwise and higher-order rules, with magnitude similar to those in the MICrONS data. Lesion studies in these RNNs reveal that disrupting like-to-like connections has a significantly greater impact on performance compared to lesions of random connections. These findings suggest that these connectivity principles may play a functional role in sensory processing and learning, highlighting shared principles between biological and artificial systems.

neuroscience↗