Neural Mapping Research Articles

Intraoperative Facial Nerve Monitoring during Parotidectomy: The Current Practices and Patterns of the Korean Society of Head and Neck Surgery (KSHNS).

Objectives: This study aimed to evaluate the current practices and trends of intraoperative facial nerve (FN) monitoring (IOFNM) during parotidectomy. Methods: A questionnaire containing 33 questions collecting information on the usage, indications, settings, techniques, loss of signal (LOS) management, anesthesiologist cooperation, and perception of usefulness of IOFNM was distributed among 348 members of the Korean Society of Head and Neck Surgery (KSHNS) via a dedicated website. Results: The response rate was approximately 25.6%, and 97% of the respondents reported using IOFNM selectively or routinely during parotidectomy. IOFNM usage decreased as the surgeon's level of experience increased (p = 0.089), from 100% in those with less than 5 years of experience to 75% in those with 20 or more years. Approximately 95% of respondents reported that the initial event threshold for electromyography activity used was 50-149 μV. Moreover, 52.4% of respondents performed neural mapping of the FN before visual identification. Initial management of LOS in visually intact FNs included checking the IOFNM system (75.3%), confirmation of muscle relaxant dosage (75.3%), and facial twitch identification (58.8%). Further management included proceeding with surgery regardless of persistent LOS (81.2%) and steroid administration sometimes or all of the time (72.9%). Overall, 98.8% of respondents found IOFNM beneficial for safe execution of parotidectomy. Conclusions: The majority of KSHNS surgeons used IOFNM during parotidectomy, although the clinical implementation of the procedure and LOS management varied between practitioners. This could be attributed to the lack of standardized protocols for IOFNM, emphasizing the need for the development of evidence-based consensus guidelines for all institutions.

Transparent MXene Microelectrode Arrays for Multimodal Mapping of Neural Dynamics.

Transparent microelectrode arrays have proven useful in neural sensing, offering a clear interface for monitoring brain activity without compromising high spatial and temporal resolution. The current landscape of transparent electrode technology faces challenges in developing durable, highly transparent electrodes while maintaining low interface impedance and prioritizing scalable processing and fabrication methods. To address these limitations, we introduce artifact-resistant transparent MXene microelectrode arrays optimized for high spatiotemporal resolution recording of neural activity. With 60% transmittance at 550 nm, these arrays enable simultaneous imaging and electrophysiology for multimodal neural mapping. Electrochemical characterization shows low impedance of 563 ± 99 kΩ at 1 kHz and a charge storage capacity of 58 mC cm⁻² without chemical doping. In vivo experiments in rodent models demonstrate the transparent arrays' functionality and performance. In a rodent model of chemically-induced epileptiform activity, we tracked ictal wavefronts via calcium imaging while simultaneously recording seizure onset. In the rat barrel cortex, we recorded multi-unit activity across cortical depths, showing the feasibility of recording high-frequency electrophysiological activity. The transparency and optical absorption properties of Ti₃C₂Tx MXene microelectrodes enable high-quality recordings and simultaneous light-based stimulation and imaging without contamination from light-induced artifacts.

Lévy noise-induced coherence resonance in neural maps

In this contribution, we investigate coherence resonance in map-based neural models in the presence of Lévy noise. The impact of different parameters of Lévy noise on the oscillatory activity is analyzed for several neural maps, including the Rulkov map, the Chialvo map, and the Courbage–Nekorkin–Vdovin map. We demonstrate that the coherence of noise-excited oscillations decreases as the stability index decreases. On the other hand, changing the skewness parameter can both reduce and, conversely, enhance coherence. We also explore and establish the peculiarities of coherence resonance in the neural maps under study depending on the Lévy noise parameters.

Training Neural Mapping Schemes for Satellite Altimetry With Simulation Data

AbstractSatellite altimetry combined with data assimilation and optimal interpolation schemes have deeply renewed our ability to monitor sea surface dynamics. Recently, deep learning schemes have emerged as appealing solutions to address space‐time interpolation problems. However, the training of state‐of‐the‐art neural schemes on real‐world case‐studies is hindered by the sparse space‐time coverage of the sea surface of real altimetry data set. Here, we introduce an innovative approach that leverages state‐of‐the‐art ocean models to train simulation‐based neural schemes for the mapping of sea surface height and demonstrate their performance on real altimetry data sets. We analyze further how the ocean simulation data set used during the training phase impacts this performance. This experimental analysis covers both the resolution from eddy‐present configurations to eddy‐rich ones, forced simulations versus reanalyzes using data assimilation and tide‐free versus tide‐resolving simulations. Our benchmarking framework focuses on a Gulf Stream region for a realistic 5‐altimeter constellation using NEMO ocean simulations and 4DVarNet mapping schemes. All simulation‐based 4DVarNets outperform the operational observation‐driven and reanalysis products, namely DUACS and GLORYS. The more realistic the ocean simulation data set used during the training phase, the better the mapping. The best 4DVarNet mapping was trained from an eddy‐rich and tide‐free simulation data sets. It improves the resolved longitudinal scale from 151 km for DUACS and 241 km for GLORYS to 98 km and reduces the root mean square error by 23% and 61%. These results open research avenues for new synergies between ocean modeling and ocean observation using learning‐based approaches.

Exploring Contextual Representation and Multi-modality for End-to-end Autonomous Driving

Learning contextual and spatial environmental representations enhances autonomous vehicle’s hazard anticipation and decision-making in complex scenarios. Recent perception systems enhance spatial understanding with sensor fusion but often lack global environmental context. Humans, when driving, naturally employ neural maps that integrate various factors such as historical data, situational subtleties, and behavioral predictions of other road users to form a rich contextual understanding of their surroundings. This neural map-based comprehension is integral to making informed decisions on the road. In contrast, even with their significant advancements, autonomous systems have yet to fully harness this depth of human-like contextual understanding. Motivated by this, our work draws inspiration from human driving patterns and seeks to formalize the sensor fusion approach within an end-to-end autonomous driving framework. We introduce a framework that integrates three cameras (left, right, and center) to emulate the human field of view, coupled with top-down bird-eye-view semantic data to enhance contextual representation. The sensor data is fused and encoded using a self-attention mechanism, leading to an auto-regressive waypoint prediction module. We treat feature representation as a sequential problem, employing a vision transformer to distill the contextual interplay between sensor modalities. The efficacy of the proposed method is experimentally evaluated in both open and closed-loop settings. Our method achieves displacement error by 0.67m in open-loop settings, surpassing current methods by 6.9% on the nuScenes dataset. In closed-loop evaluations on CARLA’s Town05 Long and Longest6 benchmarks, the proposed method enhances driving performance, route completion, and reduces infractions.

Digital twinning of cardiac electrophysiology for congenital heart disease.

In recent years, blending mechanistic knowledge with machine learning has had a major impact in digital healthcare. In this work, we introduce a computational pipeline to build certified digital replicas of cardiac electrophysiology in paediatric patients with congenital heart disease. We construct the patient-specific geometry by means of semi-automatic segmentation and meshing tools. We generate a dataset of electrophysiology simulations covering cell-to-organ level model parameters and using rigorous mathematical models based on differential equations. We previously proposed Branched Latent Neural Maps (BLNMs) as an accurate and efficient means to recapitulate complex physical processes in a neural network. Here, we employ BLNMs to encode the parametrized temporal dynamics of in silico 12-lead electrocardiograms (ECGs). BLNMs act as a geometry-specific surrogate model of cardiac function for fast and robust parameter estimation to match clinical ECGs in paediatric patients. Identifiability and trustworthiness of calibrated model parameters are assessed by sensitivity analysis and uncertainty quantification.

Visual boundary cues suffice to anchor place and grid cells in virtual reality

The hippocampal formation contains neurons responsive to an animal's current location and orientation, which together provide the organism with a neural map of space.1,2,3 Spatially tuned neurons rely on external landmark cues and internally generated movement information to estimate position.4,5 An important class of landmark cue are the boundaries delimiting an environment, which can define place cell field position6,7 and stabilize grid cell firing.8 However, the precise nature of the sensory information used to detect boundaries remains unknown. We used 2-dimensional virtual reality (VR)9 to show that visual cues from elevated walls surrounding the environment are both sufficient and necessary to stabilize place and grid cell responses in VR, when only visual and self-motion cues are available. By contrast, flat boundaries formed by the edges of a textured floor did not stabilize place and grid cells, indicating only specific forms of visual boundary stabilize hippocampal spatial firing. Unstable grid cells retain internally coherent, hexagonally arranged firing fields, but these fields "drift" with respect to the virtual environment over periods >5 s. Optic flow from a virtual floor does not slow drift dynamics, emphasizing the importance of boundary-related visual information. Surprisingly, place fields are more stable close to boundaries even with floor and wall cues removed, suggesting invisible boundaries are inferred using the motion of a discrete, separate cue (a beacon signaling reward location). Subsets of place cells show allocentric directional tuning toward the beacon, with strength of tuning correlating with place field stability when boundaries are removed.

Linguistic modulation of the neural encoding of phonemes.

Speech comprehension entails the neural mapping of the acoustic speech signal onto learned linguistic units. This acousto-linguistic transformation is bi-directional, whereby higher-level linguistic processes (e.g. semantics) modulate the acoustic analysis of individual linguistic units. Here, we investigated the cortical topography and linguistic modulation of the most fundamental linguistic unit, the phoneme. We presented natural speech and "phoneme quilts" (pseudo-randomly shuffled phonemes) in either a familiar (English) or unfamiliar (Korean) language to native English speakers while recording functional magnetic resonance imaging. This allowed us to dissociate the contribution of acoustic vs. linguistic processes toward phoneme analysis. We show that (i) the acoustic analysis of phonemes is modulated by linguistic analysis and (ii) that for this modulation, both of acoustic and phonetic information need to be incorporated. These results suggest that the linguistic modulation of cortical sensitivity to phoneme classes minimizes prediction error during natural speech perception, thereby aiding speech comprehension in challenging listening situations.

Commentaries on “Reconsidering the path for neural and physiological methods in consumer psychology”

AbstractThe initial version of the article by Clithero, Karmarkar, Nave, and Plassmann (Journal of Consumer Psychology, 2024) was critiqued by open comments from a small group of scholars. Their suggestions encouraged the authors to clarify challenging relationships between brain processes and emotions, beliefs, and actions. The revision expanded fMRI and EEG to include measures of vision, facial expression, breathing, heart rhythms, and blood chemistry. The paper provides multiple avenues of joint work between neurological and psychological scholars. The comments below reflect different reactions to the final article. Wes Hutchinson acknowledges that neuroscience insights complement cognitive measures that generate explicit measures of thought, emotion, or preferences, but he warns that repeated measures over time are problematic for both types of measurement, and the inherent complexity of brain–behavior relationships is often underestimated. With both orientations, understanding the functioning of human behavior is akin to making sense of an orchestra, where the interactive blending of different instruments and musicians reflects a complex activity that generates sounds, emotions, and stories. Both consumer neuroscientists and psychologists need to broaden their paradigmatic approaches with bodily measures and advanced psychological procedures to overcome challenges to joint progress. Martin Reiman asserts that despite difficulties with measures that have different levels of abstraction or velocity, research has provided remarkable associations between brain activity and consumer behavior. Effective studies merging brain and behavior can effectively proceed with studies that differ in two dimensions: first, by altering the number of variables, and second, by shifting whether the scientific paradigm is inductive or deductive. In its simple form, the Excavation path explores brain activity when a person is exposed to specific statements or emotions. In its most challenging form, Integrative Studies generate predictions from theories that test the convergent validity of divergent measures and leverage skills from different researchers. Studies reflecting high levels on one dimension but low levels on the other can also provide fruitful research opportunities. Brian Knutson, like Reimann, counters the idea that consumer psychology has not lived up to its promises. He references studies showing that activity from very specific areas of the brain reliably predicts choices better than explicit ratings or choices. Such research generates deductions from increasingly precise neural maps that enable confirmation of theory. That said, he acknowledges that consumer neuroscience is not able to identify a brain button that would alter choice through manipulated neurostimulation. However, since human brains are similar across people, the depth of neural insights that are consistent across a small sample of 40 respondents may generate greater insights than conventional marketing research with 2000 respondents. The cost of neuroscience will further decrease with gains in reliability, validity, and generalizability, particularly if augmented with bodily measures. He acknowledges that the theoretical side has developed more slowly than applications, particularly applications that are supported by sponsoring organizations more satisfied with local insights than general models.

MIPS-Fusion: Multi-Implicit-Submaps for Scalable and Robust Online Neural RGB-D Reconstruction

We introduce MIPS-Fusion, a robust and scalable online RGB-D reconstruction method based on a novel neural implicit representation - multi-implicit-submap. Different from existing neural RGB-D reconstruction methods lacking either flexibility with a single neural map or scalability due to extra storage of feature grids, we propose a pure neural representation tackling both difficulties with a divide-and-conquer design. In our method, neural submaps are incrementally allocated alongside the scanning trajectory and efficiently learned with local neural bundle adjustments. The submaps can be refined individually in a back-end optimization and optimized jointly to realize submap-level loop closure. Meanwhile, we propose a hybrid tracking approach combining randomized and gradient-based pose optimizations. For the first time, randomized optimization is made possible in neural tracking with several key designs to the learning process, enabling efficient and robust tracking even under fast camera motions. The extensive evaluation demonstrates that our method attains higher reconstruction quality than the state of the arts for large-scale scenes and under fast camera motions.

Temporal hierarchy of observed goal-directed actions

During social interactions, we continuously integrate current and previous information over varying timescales to infer other people's action intentions. Motor cognition theories argue for a hierarchical organization of goal-directed actions based on temporal scales. Accordingly, transient motor primitives are represented at lower levels of the hierarchy, a combination of primitives building motor sequences at subordinate levels, and more stable overarching action goals at superordinate levels. A neural topography of hierarchal timescales for information accumulation was previously shown in the visual and auditory domains. However, whether such a temporal hierarchy can also account for observed goal-directed action representations in motor pathways remains to be determined. Thus, the current study examined the neural architecture underlying the processing of observed goal-directed actions using inter-subject correlation (ISC) of fMRI activity. Observers (n = 24) viewed sequential hand movements presented in their intact order or piecewise scrambled at three timescales pertaining to goal-directed action evolution (Primitives: ± 1.5 s, Sub-Goals: ± 4 s, and High-Goals: ± 10 s). The results revealed differential intrinsic temporal capacities for integrating goal-directed action information across brain areas engaged in action observation. Longer timescales (> ± 10 s) were found in the posterior parietal and dorsal premotor compared to the ventral premotor (± 4 s) and anterior parietal (± 1.5 s) cortex. Moreover, our results revealed a hemispheric bias with more extended timescales in the right MT+, primary somatosensory, and early visual cortices compared to their homotopic regions in the left hemisphere. Our findings corroborate a hierarchical neural mapping of observed actions based on temporal scales of goals and provide further support for a ubiquitous time-dependent neural organization of information processing across multiple modalities.

City-scale continual neural semantic mapping with three-layer sampling and panoptic representation

Neural implicit representations are drawing much attention from the robotics community recently, as they are expressive, continuous and compact. However, city-scale continual implicit dense mapping based on sparse LiDAR input is still an under-explored challenge. To this end, we successfully build a city-scale continual neural mapping system with a panoptic representation that consists of environment-level and instance-level modeling. Given a stream of sparse LiDAR point cloud, it maintains a model that maps 3D coordinates to signed distance field (SDF) values. To address the difficulty of representing geometric information at different levels in city-scale space, we propose a tailored three-layer sampling strategy to sample the global, local and near-surface domains dynamically. Meanwhile, to realize high fidelity mapping of instance under incomplete observation, category-specific prior is introduced to better model the geometric details. We evaluate on the public SemanticKITTI [1] dataset and demonstrate the significance of the newly proposed three-layer sampling strategy and panoptic representation, using both quantitative and qualitative results. Codes and model will be publicly available.

Branched Latent Neural Maps

We introduce Branched Latent Neural Maps (BLNMs) to learn finite dimensional input–output maps encoding complex physical processes. A BLNM is defined by a simple and compact feedforward partially-connected neural network that structurally disentangles inputs with different intrinsic roles, such as the time variable from model parameters of a differential equation, while transferring them into a generic field of interest. BLNMs leverage latent outputs to enhance the learned dynamics and break the curse of dimensionality by showing excellent in-distribution generalization properties with small training datasets and short training times on a single processor. Indeed, their in-distribution generalization error remains comparable regardless of the adopted discretization during the testing phase. Moreover, the partial connections, in place of a fully-connected structure, significantly reduce the number of tunable parameters. We show the capabilities of BLNMs in a challenging test case involving biophysically detailed electrophysiology simulations in a biventricular cardiac model of a pediatric patient with hypoplastic left heart syndrome. The model includes a 1D Purkinje network for fast conduction and a 3D heart-torso geometry. Specifically, we trained BLNMs on 150 in silico generated 12-lead electrocardiograms (ECGs) while spanning 7 model parameters, covering cell-scale, organ-level and electrical dyssynchrony. Although the 12-lead ECGs manifest very fast dynamics with sharp gradients, after automatic hyperparameter tuning the optimal BLNM, trained in less than 3 h on a single CPU, retains just 7 hidden layers and 19 neurons per layer. The resulting mean square error is on the order of 10−4 on an independent test dataset comprised of 50 additional electrophysiology simulations. In the online phase, the BLNM allows for 5000x faster real-time simulations of cardiac electrophysiology on a single core standard computer and can be employed to solve inverse problems via global optimization in a few seconds of computational time. This paper provides a novel computational tool to build reliable and efficient reduced-order models for digital twinning in engineering applications. The Julia implementation is publicly available under MIT License at https://github.com/StanfordCBCL/BLNM.jl.

Accurate Implicit Neural Mapping With More Compact Representation in Large-Scale Scenes Using Ranging Data

Accurate Implicit Neural Mapping With More Compact Representation in Large-Scale Scenes Using Ranging Data

There is life outside the gait lab: Effectiveness of a self-organising neural map for recognising 24/7 activities of daily living

There is life outside the gait lab: Effectiveness of a self-organising neural map for recognising 24/7 activities of daily living

Invited Session IV: Studies of the visual cortex with sub-millimeter resolution: Three-photon imaging reveals the neural basis of fMRI across cortical layers.

V1 is ideally suited to study the spatial organization of neurovascular coupling at the level of synapses, neurons, individual blood vessels and laminar-resolution fMRI. This is because at least in layers 1 and 2/3 of V1, the functional micro-architecture for neurons, synapses and blood vessels has been determined using 2-photon imaging (Ohki et al 2005 Nature; Kara and Boyd 2009 Nature; O'Herron et al 2016 Nature). Hence, feature selectivity, e.g., orientation and direction selectivity of spiking, synaptic and hemodynamic activity in layer 2/3 is known. However, the micro-architecture of layer 4 neural activity (spiking and synaptic) along with individual blood vessel responses is unknown because conventional 2-photon imaging cannot access deeper cortical layers. The organizing principles of neural maps and the selectivity of hemodynamic responses is of paramount importance for laminar processing because the thalamic inputs arriving into layer 4 are untuned. 3-photon imaging triples the imaging depth compared to 2-photon imaging. Using this optical technique and high-resolution fMRI, we have determined the extent to which different types of neural (spiking, synaptic) and vascular signals (blood flow from individual vessels and fMRI voxels) are coupled across cortical layers in the cat V1. Our data show systematic changes in selectivity of hemodynamic signals across cortical layers that have clear underpinnings in neural circuitry and the propagation of hemodynamic signals.

The Intrinsic Similarity of Topological Structure in Biological Neural Networks.

Most previous studies mainly have focused on the analysis of structural properties of individual neuronal networks from C. elegans. In recent years, an increasing number of synapse-level neural maps, also known as biological neural networks, have been reconstructed. However, it is not clear whether there are intrinsic similarities of structural properties of biological neural networks from different brain compartments or species. To explore this issue, we collected nine connectomes at synaptic resolution including C. elegans, and analyzed their structural properties. We found that these biological neural networks possess small-world properties and modules. Excluding the Drosophila larval visual system, these networks have rich clubs. The distributions of synaptic connection strength for these networks can be fitted by the truncated pow-law distributions. Additionally, compared with the power-law model, a log-normal distribution is a better model to fit the complementary cumulative distribution function (CCDF) of degree for these neuronal networks. Moreover, we also observed that these neural networks belong to the same superfamily based on the significance profile (SP) of small subgraphs in the network. Taken together, these findings suggest that biological neural networks share intrinsic similarities in their topological structure, revealing some principles underlying the formation of biological neural networks within and across species.

Deep learning for full-field ultrasonic characterization

This study takes advantage of recent advances in machine learning to establish a physics-based data analytic platform for distributed reconstruction of mechanical properties in layered components from full waveform data. In this vein, two logics, namely the direct inversion and physics-informed neural networks (PINNs), are explored. The direct inversion entails three steps: (i) spectral denoising and differentiation of the full-field data, (ii) building appropriate neural maps to approximate the profile of unknown physical and regularization parameters on their respective domains, and (iii) simultaneous training of the neural networks by minimizing the Tikhonov-regularized PDE loss using data from (i). PINNs furnish efficient surrogate models of complex systems with predictive capabilities via multitask learning where the field variables are modeled by neural maps endowed with (scaler or distributed) auxiliary parameters such as physical unknowns and loss function weights. PINNs are then trained by minimizing a measure of data misfit subject to the underlying physical laws as constraints. In this study, to facilitate learning from ultrasonic data, the PINNs loss adopts (a) wavenumber-dependent Sobolev norms to compute the data misfit, and (b) non-adaptive weights in a specific scaling framework to naturally balance the loss objectives by leveraging the form of PDEs germane to elastic-wave propagation. Both paradigms are examined via synthetic and laboratory test data. In the latter case, the reconstructions are performed at multiple frequencies and the results are verified by a set of complementary experiments highlighting the importance of verification and validation in data-driven modeling.

Variant anatomy of non-recurrent laryngeal nerve: when and how should it be taught in surgical residency?

While the performance of a thyroidectomy is generally associated with a low risk of injury to the recurrent laryngeal nerve (RLN), the presence of a non-recurrent nerve (NRLN) increases the risk of this complication. Generally, the intraoperative detection via visual appreciation of variant anatomy of the RLN has been regarded as poor, possibly due to a lack of knowledge of both the normal and aberrant anatomy of the RLN. Articles for the review were searched through PubMed using the search terms and their combinations: "non-recurrent laryngeal nerve," "thyroidectomy," "injury," "palsy," "variant anatomy," and "residency," from January 1, 2000, to December 2022. Papers considered for the review were the articles published in English, with additional classic and articles of surgical importance retrieved from the reference list of papers. Only papers relevant to the scope of the review were considered for this review. The NRLN has been found to be associated with concurrent vascular abnormalities, such as the presence of an aberrant right subclavian artery (ARSA) or an arteria lusoria originating from the aortic arch. However, it seems that both the normal as well as aberrant anatomy of the RLN is currently not emphasized enough during postgraduate surgical training. With the increased use of intraoperative neuromonitoring (IONM), detection of NRLN has become possible through appropriate neural mapping during thyroid surgery, besides other pointers such as visualization during surgery, computerised tomography, and duplex ultrasound scans to visualize the variant vascular anatomy. There is also a possible role for cadaveric courses, either during medical school or in a post-graduate setting-adapted to the student's level to teach the variant anatomy. With the development of newer techniques such as artificial intelligence, there are potential new options for teaching and training anatomy in the near future. Adequate knowledge of the normal and aberrant anatomy of the RLN remains essential for the best outcomes in thyroid surgery, even in the era of the IONM. Moving forward, the knowledge of (aberrant) anatomy should be made an integral part of the core competencies of both medical students and surgical trainees. It is imperative that leaders of the different field work closely together to combine their knowledge towards providing their trainees with the best possible training options.

The development of bi-directionally coupled self-organizing neurovascular networks captures orientation-selective neural and hemodynamic cortical responses.

Networks of neurons are the primary substrate of information processing. Conversely, blood vessels in the brain are generally viewed to have physiological functions unrelated to information processing, such as the timely supply of oxygen, and other nutrients to the neural tissue. However, recent studies have shown that cerebral microvessels, like neurons, exhibit tuned responses to sensory stimuli. Tuned neural responses to sensory stimuli may be enhanced with experience-dependent Hebbian plasticity and other forms of learning. Hence, it is possible that the microvascular network might also be subject to some form of competitive learning rules during early postnatal development such that its fine-scale structure becomes optimized for metabolic delivery to a given neural micro-architecture. To explore the possibility of adaptive lateral interactions and tuned responses in cerebral microvessels, we modelled the cortical neurovascular network by interconnecting two laterally connected self-organizing networks. The afferent and lateral connections of the neural and vascular networks were defined by trainable weights. By varying the topology of lateral connectivity in the vascular network layer, we observed that the partial correspondence of feature selectivity between neural and hemodynamic responses could be explained by lateral coupling across local blood vessels such that the central domain receives an excitatory drive of more blood flow and a distal surrounding region where blood flow is reduced. Critically, our simulations suggest a new role for feedback from the vascular to the neural network because the radius of vascular perfusion determines whether the cortical neural map develops into a clustered vs. salt-and-pepper organization.