Feedforward, feedback and inhibitory connections in primate visual cortex

Cortical Layer-Dependent Signaling in Cognition: Three Computational Modes of the Canonical Circuit

The cerebral cortex performs computations via numerous six-layer modules. The operational dynamics of these modules were studied primarily in early sensory cortices using bottom-up computation for response selectivity as a model, which has been recently revolutionized by genetic approaches in mice. However, cognitive processes such as recall and imagery require top-down generative computation. The question of whether the layered module operates similarly in top-down generative processing as in bottom-up sensory processing has become testable by advances in the layer identification of recorded neurons in behaving monkeys. This review examines recent advances in laminar signaling in these two computations, using predictive coding computation as a common reference, and shows that each of these computations recruits distinct laminar circuits, particularly in layer 5, depending on the cognitive demands. These findings highlight many open questions, including how different interareal feedback pathways, originating from and terminating at different layers, convey distinct functional signals.

Laminar multi-contrast fMRI at 7T allows differentiation of neuronal excitation and inhibition underlying positive and negative BOLD responses

A major challenge for human neuroimaging using functional MRI is the differentiation of neuronal excitation and inhibition which may induce positive and negative BOLD responses. Here we present an innovative multi-contrast laminar functional MRI technique that offers comprehensive and quantitative imaging of neurovascular (CBF, CBV, BOLD) and metabolic (CMRO2) responses across cortical layers at 7 Tesla. This technique was first validated through a finger-tapping experiment, revealing 'double-peak' laminar activation patterns within the primary motor cortex. By employing a ring-shaped visual stimulus that elicited positive and negative BOLD responses, we further observed distinct neurovascular and metabolic responses across cortical layers and eccentricities in the primary visual cortex. This suggests potential feedback inhibition of neuronal activities in both superficial and deep cortical layers underlying the negative BOLD signals in the fovea, and also illustrates the neuronal activities in visual areas adjacent to the activated eccentricities.

Dependence of Contextual Modulation in Macaque V1 on Interlaminar Signal Flow

In visual cortex, neural correlates of subjective perception can be generated by modulation of activity from beyond the classical receptive field (CRF). In macaque V1, activity generated by nonclassical receptive field (nCRF) stimulation involves different intracortical circuitry than activity generated by CRF stimulation, suggesting that interactions between neurons across V1 layers differ under CRF and nCRF stimulus conditions. We measured border ownership modulation within large populations of V1 neurons. We found that neurons in single columns preferred the same side of objects located outside of the CRF. In addition, we found that interactions between pairs of neurons situated across feedback/horizontal and input layers differed between CRF and nCRF stimulation. Furthermore, the magnitude of border ownership modulation was predicted by greater information flow from feedback/horizontal to input layers. These results demonstrate that the flow of signals between layers covaries with the degree to which neurons integrate information from beyond the CRF.

Preferential transduction of parvalbumin-expressing cortical neurons by AAV-mDLX5/6 vectors

A major goal of modern neuroscience is to understand the functions of the varied neuronal types that comprise the mammalian brain. Toward this end, some types of neurons can be targeted and manipulated with enhancer-bearing AAV vectors. These vectors hold great promise to advance basic and translational neuroscience, but to realize this potential, their selectivity must be characterized. In this study, we investigated the selectivity of AAV vectors carrying an enhancer of the murine Dlx5 and Dlx6 genes. Vectors were injected into the visual cortex of two macaque monkeys, the frontal cortex of two others, and the somatosensory/motor cortex of three rats. Post-mortem immunostaining revealed that parvalbumin-expressing neurons were transduced efficiently in all cases but calretinin-expressing neurons were not. We speculate that this specificity is a consequence of differential activity of this DLX5/6 enhancer in adult neurons of different developmental lineages.

Microscale visualization of cellular features in adult macaque visual cortex

Expansion microscopy and light sheet imaging enable fine-scale resolution of intracellular features that comprise neural circuits. Most current techniques visualize sparsely distributed features across whole brains or densely distributed features within individual brain regions. Here, we visualize dense distributions of immunolabeled proteins across early visual cortical areas in adult macaque monkeys. This process may be combined with multiphoton or magnetic resonance imaging to produce multimodal atlases in large, gyrencephalic brains.

Internal feedback in the cortical perception-action loop enables fast and accurate behavior

Animals move smoothly and reliably in unpredictable environments. Models of sensorimotor control, drawing on control theory, have assumed that sensory information from the environment leads to actions, which then act back on the environment, creating a single, unidirectional perception-action loop. However, the sensorimotor loop contains internal delays in sensory and motor pathways, which can lead to unstable control. We show here that these delays can be compensated by internal feedback signals that flow backward, from motor toward sensory areas. This internal feedback is ubiquitous in neural sensorimotor systems, and we show how internal feedback compensates internal delays. This is accomplished by filtering out self-generated and other predictable changes so that unpredicted, actionable information can be rapidly transmitted toward action by the fastest components, effectively compressing the sensory input to more efficiently use feedforward pathways: Tracts of fast, giant neurons necessarily convey less accurate signals than tracts with many smaller neurons, but they are crucial for fast and accurate behavior. We use a mathematically tractable control model to show that internal feedback has an indispensable role in achieving state estimation, localization of function (how different parts of the cortex control different parts of the body), and attention, all of which are crucial for effective sensorimotor control. This control model can explain anatomical, physiological, and behavioral observations, including motor signals in the visual cortex, heterogeneous kinetics of sensory receptors, and the presence of giant cells in the cortex of humans as well as internal feedback patterns and unexplained heterogeneity in neural systems.

Visuomotor interactions in the mouse forebrain mediated by extrastriate cortico-cortical pathways

Introduction

The mammalian visual system can be broadly divided into two functional processing pathways: a dorsal stream supporting visually and spatially guided actions, and a ventral stream enabling object recognition. In rodents, the majority of visual signaling in the dorsal stream is transmitted to frontal motor cortices via extrastriate visual areas surrounding V1, but exactly where and to what extent V1 feeds into motor-projecting visual regions is not well known.

Methods

We employed a dual labeling strategy in male and female mice in which efferent projections from V1 were labeled anterogradely, and motor-projecting neurons in higher visual areas were labeled with retrogradely traveling adeno-associated virus (rAAV-retro) injected in M2. We characterized the labeling in both flattened and coronal sections of dorsal cortex and made high-resolution 3D reconstructions to count putative synaptic contacts in different extrastriate areas.

Results

The most pronounced colocalization V1 output and M2 input occurred in extrastriate areas AM, PM, RL and AL. Neurons in both superficial and deep layers in each project to M2, but high resolution volumetric reconstructions indicated that the majority of putative synaptic contacts from V1 onto M2-projecting neurons occurred in layer 2/3.

Discussion

These findings support the existence of a dorsal processing stream in the mouse visual system, where visual signals reach motor cortex largely via feedforward projections in anteriorly and medially located extrastriate areas.

What multiplexing means for the interpretation of functional MRI data

Despite technology advances that have enabled routine acquisition of functional MRI data with sub-millimeter resolution, the inferences that cognitive neuroscientists must make to link fMRI data to behavior are complicated. Thus, a single dataset subjected to different analyses can be interpreted in different ways. This article presents two optical analogies that can be useful for framing fMRI analyses in a way that allows for multiple interpretations of fMRI data to be valid simultaneously without undermining each other. The first is reflection: when an object is reflected in a mirrored surface, it appears as if the reflected object is sharing space with the mirrored object, but of course it is not. This analogy can be a good guide for interpreting the fMRI signal, since even at sub-millimeter resolutions the signal is determined by a mixture of local and long-range neural computations. The second is refraction. If we view an object through a multi-faceted prism or gemstone, our view will change-sometimes dramatically-depending on our viewing angle. In the same way, interpretation of fMRI data (inference of underlying neuronal activity) can and should be different depending on the analysis approach. Rather than representing a weakness of the methodology, or the superiority of one approach over the other (for example, simple regression analysis versus multi-voxel pattern analysis), this is an expected consequence of how information is multiplexed in the neural networks of the brain: multiple streams of information are simultaneously present in each location. The fact that any one analysis typically shows only one view of the data also puts some parentheses around fMRI practitioners' constant search for ground truth against which to compare their data. By holding our interpretations lightly and understanding that many interpretations of the data can all be true at the same time, we do a better job of preparing ourselves to appreciate, and eventually understand, the complexity of the brain and the behavior it produces.

Segmentation of Medical Image Using Novel Dilated Ghost Deep Learning Model

Image segmentation and computer vision are becoming more important in computer-aided design. A computer algorithm extracts image borders, colours, and textures. It also depletes resources. Technical knowledge is required to extract information about distinctive features. There is currently no medical picture segmentation or recognition software available. The proposed model has 13 layers and uses dilated convolution and max-pooling to extract small features. Ghost model deletes the duplicated features, makes the process easier, and reduces the complexity. The Convolution Neural Network (CNN) generates a feature vector map and improves the accuracy of area or bounding box proposals. Restructuring is required for healing. As a result, convolutional neural networks segment medical images. It is possible to acquire the beginning region of a segmented medical image. The proposed model gives better results as compared to the traditional models, it gives an accuracy of 96.05, Precision 98.2, and recall 95.78. The first findings are improved by thickening and categorising the image's pixels. Morphological techniques may be used to segment medical images. Experiments demonstrate that the recommended segmentation strategy is effective. This study rethinks medical image segmentation methods.

Conserved patterns of functional organization between cortex and thalamus in mice

SignificanceNeuroanatomical tracing provides just a partial picture of information flow in the brain, because excitatory synapses are not all equal. Some strongly drive postsynaptic targets to transfer information, whereas others weakly modulate their responsiveness. Here, we show conserved patterns of synaptic function across somatosensory and visual thalamocortical circuits in mice involving higher-order thalamic nuclei. These nuclei serve as hubs in transthalamic or cortico-thalamo-cortical pathways. We report that feedforward transthalamic circuits in the somatosensory and visual systems operate to efficiently transmit information, whereas feedback transthalamic circuits act to modulate their target areas. These patterns may generalize to other brain systems and show how methods of synapse physiology and molecular biology can inform the exploration of brain circuitry and information processing.

Stimulus dependence of directed information exchange between cortical layers in macaque V1

Perception and cognition require the integration of feedforward sensory information with feedback signals. Using different sized stimuli, we isolate spectral signatures of feedforward and feedback signals, and their effect on communication between layers in primary visual cortex of male macaque monkeys. Small stimuli elicited gamma frequency oscillations predominantly in the superficial layers. These Granger-causally originated in upper layer 4 and lower supragranular layers. Unexpectedly, large stimuli generated strong narrow band gamma oscillatory activity across cortical layers. They Granger-causally arose in layer 5, were conveyed through layer six to superficial layers, and violated existing models of feedback spectral signatures. Equally surprising, with large stimuli, alpha band oscillatory activity arose predominantly in granular and supragranular layers and communicated in a feedforward direction. Thus, oscillations in specific frequency bands are dynamically modulated to serve feedback and feedforward communication and are not restricted to specific cortical layers in V1.

Recurrent processing improves occluded object recognition and gives rise to perceptual hysteresis

Over the past decades, object recognition has been predominantly studied and modelled as a feedforward process. This notion was supported by the fast response times in psychophysical and neurophysiological experiments and the recent success of deep feedforward neural networks for object recognition. Recently, however, this prevalent view has shifted and recurrent connectivity in the brain is now believed to contribute significantly to object recognition — especially under challenging conditions, including the recognition of partially occluded objects. Moreover, recurrent dynamics might be the key to understanding perceptual phenomena such as perceptual hysteresis. In this work we investigate if and how artificial neural networks can benefit from recurrent connections. We systematically compare architectures comprised of bottom-up, lateral, and top-down connections. To evaluate the impact of recurrent connections for occluded object recognition, we introduce three stereoscopic occluded object datasets, which span the range from classifying partially occluded hand-written digits to recognizing three-dimensional objects. We find that recurrent architectures perform significantly better than parameter-matched feedforward models. An analysis of the hidden representation of the models suggests that occluders are progressively discounted in later time steps of processing. We demonstrate that feedback can correct the initial misclassifications over time and that the recurrent dynamics lead to perceptual hysteresis. Overall, our results emphasize the importance of recurrent feedback for object recognition in difficult situations.

Directed information exchange between cortical layers in macaque V1 and V4 and its modulation by selective attention

Achieving behavioral goals requires integration of sensory and cognitive information across cortical laminae and cortical regions. How this computation is performed remains unknown. Using local field potential recordings and spectrally resolved conditional Granger causality (cGC) analysis, we mapped visual information flow, and its attentional modulation, between cortical layers within and between macaque brain areas V1 and V4. Stimulus-induced interlaminar information flow within V1 dominated upwardly, channeling information toward supragranular corticocortical output layers. Within V4, information flow dominated from granular to supragranular layers, but interactions between supragranular and infragranular layers dominated downwardly. Low-frequency across-area communication was stronger from V4 to V1, with little layer specificity. Gamma-band communication was stronger in the feedforward V1-to-V4 direction. Attention to the receptive field of V1 decreased communication between all V1 layers, except for granular-to-supragranular layer interactions. Communication within V4, and from V1 to V4, increased with attention across all frequencies. While communication from V4 to V1 was stronger in lower-frequency bands (4 to 25 Hz), attention modulated cGCs from V4 to V1 across all investigated frequencies. Our data show that top-down cognitive processes result in reduced communication within cortical areas, increased feedforward communication across all frequency bands, and increased gamma-band feedback communication.

Neocortical Layer 1: An Elegant Solution to Top-Down and Bottom-Up Integration

Many of our daily activities, such as riding a bike to work or reading a book in a noisy cafe, and highly skilled activities, such as a professional playing a tennis match or a violin concerto, depend upon the ability of the brain to quickly make moment-to-moment adjustments to our behavior in response to the results of our actions. Particularly, they depend upon the ability of the neocortex to integrate the information provided by the sensory organs (bottom-up information) with internally generated signals such as expectations or attentional signals (top-down information). This integration occurs in pyramidal cells (PCs) and their long apical dendrite, which branches extensively into a dendritic tuft in layer 1 (L1). The outermost layer of the neocortex, L1 is highly conserved across cortical areas and species. Importantly, L1 is the predominant input layer for top-down information, relayed by a rich, dense mesh of long-range projections that provide signals to the tuft branches of the PCs. Here, we discuss recent progress in our understanding of the composition of L1 and review evidence that L1 processing contributes to functions such as sensory perception, cross-modal integration, controlling states of consciousness, attention, and learning.

Why do imagery and perception look and feel so different?

Despite the past few decades of research providing convincing evidence of the similarities in function and neural mechanisms between imagery and perception, for most of us, the experience of the two are undeniably different, why? Here, we review and discuss the differences between imagery and perception and the possible underlying causes of these differences, from function to neural mechanisms. Specifically, we discuss the directional flow of information (top-down versus bottom-up), the differences in targeted cortical layers in primary visual cortex and possible different neural mechanisms of modulation versus excitation. For the first time in history, neuroscience is beginning to shed light on this long-held mystery of why imagery and perception look and feel so different. This article is part of the theme issue 'Offline perception: voluntary and spontaneous perceptual experiences without matching external stimulation'.

Complementary Ca2+ Activity of Sensory Activated and Suppressed Layer 6 Corticothalamic Neurons Reflects Behavioral State

Layer 6 (L6) corticothalamic neurons project to thalamus, where they are thought to regulate sensory information transmission to cortex. However, the activity of these neurons during different behavioral states has not been described. Here, we imaged calcium changes in visual cortex L6 primary corticothalamic neurons with two-photon microscopy in head-fixed mice in response to passive viewing during a range of behavioral states, from locomotion to sleep. In addition to a substantial fraction of quiet neurons, we found sensory-activated and suppressed neurons, comprising two functionally distinct L6 feedback channels. Quiet neurons could be dynamically recruited to one or another functional channel, and the opposite, functional neurons could become quiet under different stimulation conditions or behavior states. The state dependence of neuronal activity was heterogeneous with respect to locomotion or level of alertness, although the average activity was largest during highest vigilance within populations of functional neurons. Interestingly, complementary activity of these distinct populations kept the overall corticothalamic feedback relatively constant during any given behavioral state. Thereby, in addition to sensory and non-sensory information, a constant activity level characteristic of behavioral state is conveyed to thalamus, where it can regulate signal transmission from the periphery to cortex.

Magnitude, time course, and specificity of rapid adaptation across mouse visual areas

Adaptation is a ubiquitous feature of sensory processing whereby recent experience shapes future responses. The mouse primary visual cortex (V1) is particularly sensitive to recent experience, where a brief stimulus can suppress subsequent responses for seconds. This rapid adaptation profoundly impacts perception, suggesting that its effects are propagated along the visual hierarchy. To understand how rapid adaptation influences sensory processing, we measured its effects at key nodes in the visual system: in V1, three higher visual areas (HVAs: lateromedial, anterolateral, and posteromedial), and the superior colliculus (SC) in awake mice of both sexes using single-unit recordings. Consistent with the feed-forward propagation of adaptation along the visual hierarchy, we find that neurons in layer 4 adapt less strongly than those in other layers of V1. Furthermore, neurons in the HVAs adapt more strongly, and recover more slowly, than those in V1. The magnitude and time course of adaptation was comparable in each of the HVAs and in the SC, suggesting that adaptation may not linearly accumulate along the feed-forward visual processing hierarchy. Despite the increase in adaptation in the HVAs compared with V1, the effects were similarly orientation specific across all areas. These data reveal that adaptation profoundly shapes cortical processing, with increasing impact at higher levels in the cortical hierarchy, and also strongly influencing computations in the SC. Thus, we find robust, brain-wide effects of rapid adaptation on sensory processing.NEW & NOTEWORTHY Rapid adaptation dynamically alters sensory signals to account for recent experience. To understand how adaptation affects sensory processing and perception, we must determine how it impacts the diverse set of cortical and subcortical areas along the hierarchy of the mouse visual system. We find that rapid adaptation strongly impacts neurons in primary visual cortex, the higher visual areas, and the colliculus, consistent with its profound effects on behavior.

Macroscopic gradients of synaptic excitation and inhibition in the neocortex

With advances in connectomics, transcriptome and neurophysiological technologies, the neuroscience of brain-wide neural circuits is poised to take off. A major challenge is to understand how a vast diversity of functions is subserved by parcellated areas of mammalian neocortex composed of repetitions of a canonical local circuit. Areas of the cerebral cortex differ from each other not only in their input-output patterns but also in their biological properties. Recent experimental and theoretical work has revealed that such variations are not random heterogeneities; rather, synaptic excitation and inhibition display systematic macroscopic gradients across the entire cortex, and they are abnormal in mental illness. Quantitative differences along these gradients can lead to qualitatively novel behaviours in non-linear neural dynamical systems, by virtue of a phenomenon mathematically described as bifurcation. The combination of macroscopic gradients and bifurcations, in tandem with biological evolution, development and plasticity, provides a generative mechanism for functional diversity among cortical areas, as a general principle of large-scale cortical organization.

Control of synaptic plasticity in deep cortical networks

Humans and many other animals have an enormous capacity to learn about sensory stimuli and to master new skills. However, many of the mechanisms that enable us to learn remain to be understood. One of the greatest challenges of systems neuroscience is to explain how synaptic connections change to support maximally adaptive behaviour. Here, we provide an overview of factors that determine the change in the strength of synapses, with a focus on synaptic plasticity in sensory cortices. We review the influence of neuromodulators and feedback connections in synaptic plasticity and suggest a specific framework in which these factors can interact to improve the functioning of the entire network.

Medical Image Segmentation Algorithm Based on Feedback Mechanism CNN

With the development of computer vision and image segmentation technology, medical image segmentation and recognition technology has become an important part of computer-aided diagnosis. The traditional image segmentation method relies on artificial means to extract and select information such as edges, colors, and textures in the image. It not only consumes considerable energy resources and people's time but also requires certain expertise to obtain useful feature information, which no longer meets the practical application requirements of medical image segmentation and recognition. As an efficient image segmentation method, convolutional neural networks (CNNs) have been widely promoted and applied in the field of medical image segmentation. However, CNNs that rely on simple feedforward methods have not met the actual needs of the rapid development of the medical field. Thus, this paper is inspired by the feedback mechanism of the human visual cortex, and an effective feedback mechanism calculation model and operation framework is proposed, and the feedback optimization problem is presented. A new feedback convolutional neural network algorithm based on neuron screening and neuron visual information recovery is constructed. So, a medical image segmentation algorithm based on a feedback mechanism convolutional neural network is proposed. The basic idea is as follows: The model for obtaining an initial region with the segmented medical image classifies the pixel block samples in the segmented image. Then, the initial results are optimized by threshold segmentation and morphological methods to obtain accurate medical image segmentation results. Experiments show that the proposed segmentation method has not only high segmentation accuracy but also extremely high adaptive segmentation ability for various medical images. The research in this paper provides a new perspective for medical image segmentation research. It is a new attempt to explore more advanced intelligent medical image segmentation methods. It also provides technical approaches and methods for further development and improvement of adaptive medical image segmentation technology.

Approximating the Architecture of Visual Cortex in a Convolutional Network

Deep convolutional neural networks (CNNs) have certain structural, mechanistic, representational, and functional parallels with primate visual cortex and also many differences. However, perhaps some of the differences can be reconciled. This study develops a cortex-like CNN architecture, via (1) a loss function that quantifies the consistency of a CNN architecture with neural data from tract tracing, cell reconstruction, and electrophysiology studies; (2) a hyperparameter-optimization approach for reducing this loss, and (3) heuristics for organizing units into convolutional-layer grids. The optimized hyperparameters are consistent with neural data. The cortex-like architecture differs from typical CNN architectures. In particular, it has longer skip connections, larger kernels and strides, and qualitatively different connection sparsity. Importantly, layers of the cortex-like network have one-to-one correspondences with cortical neuron populations. This should allow unambiguous comparison of model and brain representations in the future and, consequently, more precise measurement of progress toward more biologically realistic deep networks.

Microglia in the Primary Somatosensory Barrel Cortex Mediate Trigeminal Neuropathic Pain

Trigeminal neuropathic pain (TGN) is an attacking, abrupt, electric-shock headache involving abnormal cortical activity. The neural mechanism underlying TGN remains elusive. In this study, we explored the role of microglia in the primary somatosensory barrel cortex (S1BF), which is a critical region for TGN, of a mouse model of TGN that displayed significant pain-related behaviors. Using electrophysiological recordings, we found robust neuronal hyperactivity in glutamatergic neurons of S1BF (Glu^S1BF). Chemogenetic inhibition of Glu^S1BF neurons significantly relieved mechanical allodynia in TGN mice. In naïve mice, chemogenetic activation of Glu^S1BF neurons induced pain sensitization. In addition, we found that microglia in the S1BF (microglia^S1BF) were significantly activated, with density and morphology changes. Intraperitoneal administration of minocycline, a microglia inhibitor, attenuated pain sensitization, and decreased Glu^S1BF neuronal activity. Together, these findings demonstrate the putative importance of microglia as a key regulator in TGN through actions on Glu^S1BF neuronal adaptation.

Dynamic Computation in Visual Thalamocortical Networks

Contemporary neurodynamical frameworks, such as coordination dynamics and winnerless competition, posit that the brain approximates symbolic computation by transitioning between metastable attractive states. This article integrates these accounts with electrophysiological data suggesting that coherent, nested oscillations facilitate information representation and transmission in thalamocortical networks. We review the relationship between criticality, metastability, and representational capacity, outline existing methods for detecting metastable oscillatory patterns in neural time series data, and evaluate plausible spatiotemporal coding schemes based on phase alignment. We then survey the circuitry and the mechanisms underlying the generation of coordinated alpha and gamma rhythms in the primate visual system, with particular emphasis on the pulvinar and its role in biasing visual attention and awareness. To conclude the review, we begin to integrate this perspective with longstanding theories of consciousness and cognition.

Myelin and Modeling: Bootstrapping Cortical Microcircuits

Histological studies of myelin-stained sectioned cadaver brain and in vivo myelin-weighted magnetic resonance imaging (MRI) show that the cerebral cortex is organized into cortical areas with generally well-defined boundaries, which have consistent internal patterns of myelination. The process of myelination is largely driven by neural experience, in which the axonal passage of action potentials stimulates neighboring oligodendrocytes to perform their task. This bootstrapping process, such that the traffic of action potentials facilitates increased traffic, suggests the hypothesis that the specific pattern of myelination (myeloarchitecture) in each cortical area reveals the principal cortical microcircuits required for the function of that area. If this idea is correct, the observable sequential maturation of specific brain areas can provide evidence for models of the stages of cognitive development.

Dynamics and orientation selectivity in a cortical model of rodent V1 with excess bidirectional connections

Recent experiments have revealed fine structure in cortical microcircuitry. In particular, bidirectional connections are more prevalent than expected by chance. Whether this fine structure affects cortical dynamics and function has not yet been studied. Here we investigate the effects of excess bidirectionality in a strongly recurrent network model of rodent V1. We show that reciprocal connections have only a very weak effect on orientation selectivity. We find that excess reciprocity between inhibitory neurons slows down the dynamics and strongly increases the Fano factor, while for reciprocal connections between excitatory and inhibitory neurons it has the opposite effect. In contrast, excess bidirectionality within the excitatory population has a minor effect on the neuronal dynamics. These results can be explained by an effective delayed neuronal self-coupling which stems from the fine structure. Our work suggests that excess bidirectionality between inhibitory neurons decreases the efficiency of feature encoding in cortex by reducing the signal to noise ratio. On the other hand it implies that the experimentally observed strong reciprocity between excitatory and inhibitory neurons improves the feature encoding.

Pattern recognition with neuronal avalanche dynamics

Pattern recognition is a fundamental neuronal process which enables a cortical system to interpret visual stimuli. How the brain learns to recognize patterns is, however, an unsolved problem. The frequently employed method of back propagation excels at this task but has been found to be unbiological in many aspects. In this Rapid Communication we achieve pattern recognition tasks in a biologically, fully consistent framework. We consider a neuronal network exhibiting avalanche dynamics, as observed experimentally, and implement negative feedback signals. These are chemical signals, such as dopamine, which mediate synaptic plasticity and sculpt the network to achieve certain tasks. The system is able to distinguish horizontal and vertical lines with high accuracy, as well as to perform well at the more complicated task of handwritten digit recognition. Resulting from the learning mechanism, spatially separate activity regions emerge, as observed in the primary visual cortex using functional magnetic resonance imaging techniques. The results therefore suggest that negative feedback signals offer an explanation for the emergence of distinct activity areas in the visual cortex.

Structure and function of dual-source cholinergic modulation in early vision

Behavioral states such as arousal and attention have profound effects on sensory processing, determining how-even whether-a stimulus is perceived. This state-dependence is believed to arise, at least in part, in response to inputs from subcortical structures that release neuromodulators such as acetylcholine, often nonsynaptically. The mechanisms that underlie the interaction between these nonsynaptic signals and the more point-to-point synaptic cortical circuitry are not well understood. This review highlights the state of the field, with a focus on cholinergic action in early visual processing. Key anatomical and physiological features of both the cholinergic and the visual systems are discussed. Furthermore, presenting evidence of cholinergic modulation in visual thalamus and primary visual cortex, we explore potential functional roles of acetylcholine and its effects on the processing of visual input over the sleep-wake cycle, sensory gain control during wakefulness, and consider evidence for cholinergic support of visual attention.

Fast Inhibitory Decay Facilitates Adult-like Temporal Processing in Layer 5 of Developing Primary Auditory Cortex

The protracted maturational process of temporal processing in layer 4 (L4) of primary auditory cortex (A1) has been extensively studied. Accumulating evidences show that layer 5 (L5) receives direct thalamic inputs as well. How the temporal responses in L5 may developmentally emerge remains unclear. Using in vivo loose-patch recordings in rat A1, we found that putative pyramidal (Pyr) neurons in developing L5 exhibited adult-like stimulus-following ability but less bursting shortly after hearing onset. L5 Pyr neurons in adult A1 exhibited phase-locking similar to L4 neurons, while L5 fast-spiking (FS) neurons showed greater phase-locking at 7 and 12.5 pps. In developing L5, whole-cell recordings revealed inhibition with decay constant comparable to that in adult L5, thereby avoiding the summation of inhibition that contributed to the strong adaptation in L4. Given the targets of L5 outputs, the relatively precocious temporal processing in L5 might contribute to temporal response maturation in connected cortical and subcortical areas. Our findings were in agreement with the idea that L5 may be a "hub" for processing cortical inputs and outputs that can operate independently of L4.

Neural Encoding and Decoding with Deep Learning for Dynamic Natural Vision

Convolutional neural network (CNN) driven by image recognition has been shown to be able to explain cortical responses to static pictures at ventral-stream areas. Here, we further showed that such CNN could reliably predict and decode functional magnetic resonance imaging data from humans watching natural movies, despite its lack of any mechanism to account for temporal dynamics or feedback processing. Using separate data, encoding and decoding models were developed and evaluated for describing the bi-directional relationships between the CNN and the brain. Through the encoding models, the CNN-predicted areas covered not only the ventral stream, but also the dorsal stream, albeit to a lesser degree; single-voxel response was visualized as the specific pixel pattern that drove the response, revealing the distinct representation of individual cortical location; cortical activation was synthesized from natural images with high-throughput to map category representation, contrast, and selectivity. Through the decoding models, fMRI signals were directly decoded to estimate the feature representations in both visual and semantic spaces, for direct visual reconstruction and semantic categorization, respectively. These results corroborate, generalize, and extend previous findings, and highlight the value of using deep learning, as an all-in-one model of the visual cortex, to understand and decode natural vision.

Blindness and Human Brain Plasticity

Early blindness causes fundamental alterations of neural function across more than 25% of cortex-changes that span the gamut from metabolism to behavior and collectively represent one of the most dramatic examples of plasticity in the human brain. The goal of this review is to describe how the remarkable behavioral and neuroanatomical compensations demonstrated by blind individuals provide insights into the extent, mechanisms, and limits of human brain plasticity.

Comprehensive Morpho-Electrotonic Analysis Shows 2 Distinct Classes of L2 and L3 Pyramidal Neurons in Human Temporal Cortex

There have been few quantitative characterizations of the morphological, biophysical, and cable properties of neurons in the human neocortex. We employed feature-based statistical methods on a rare data set of 60 3D reconstructed pyramidal neurons from L2 and L3 in the human temporal cortex (HL2/L3 PCs) removed after brain surgery. Of these cells, 25 neurons were also characterized physiologically. Thirty-two morphological features were analyzed (e.g., dendritic surface area, 36 333 ± 18 157 μm²; number of basal trees, 5.55 ± 1.47; dendritic diameter, 0.76 ± 0.28 μm). Eighteen features showed a significant gradual increase with depth from the pia (e.g., dendritic length and soma radius). The other features showed weak or no correlation with depth (e.g., dendritic diameter). The basal dendritic terminals in HL2/L3 PCs are particularly elongated, enabling multiple nonlinear processing units in these dendrites. Unlike the morphological features, the active biophysical features (e.g., spike shapes and rates) and passive/cable features (e.g., somatic input resistance, 47.68 ± 15.26 MΩ, membrane time constant, 12.03 ± 1.79 ms, average dendritic cable length, 0.99 ± 0.24) were depth-independent. A novel descriptor for apical dendritic topology yielded 2 distinct classes, termed hereby as “slim-tufted” and “profuse-tufted” HL2/L3 PCs; the latter class tends to fire at higher rates. Thus, our morpho-electrotonic analysis shows 2 distinct classes of HL2/L3 PCs.

Locally induced neuronal synchrony precisely propagates to specific cortical areas without rhythm distortion

Propagation of oscillatory spike firing activity at specific frequencies plays an important role in distributed cortical networks. However, there is limited evidence for how such frequency-specific signals are induced or how the signal spectra of the propagating signals are modulated during across-layer (radial) and inter-areal (tangential) neuronal interactions. To directly evaluate the direction specificity of spectral changes in a spiking cortical network, we selectively photostimulated infragranular excitatory neurons in the rat primary visual cortex (V1) at a supra-threshold level with various frequencies, and recorded local field potentials (LFPs) at the infragranular stimulation site, the cortical surface site immediately above the stimulation site in V1, and cortical surface sites outside V1. We found a significant reduction of LFP powers during radial propagation, especially at high-frequency stimulation conditions. Moreover, low-gamma-band dominant rhythms were transiently induced during radial propagation. Contrastingly, inter-areal LFP propagation, directed to specific cortical sites, accompanied no significant signal reduction nor gamma-band power induction. We propose an anisotropic mechanism for signal processing in the spiking cortical network, in which the neuronal rhythms are locally induced/modulated along the radial direction, and then propagate without distortion via intrinsic horizontal connections for spatiotemporally precise, inter-areal communication.

Feedback determines the structure of correlated variability in primary visual cortex

The variable responses of sensory neurons tend to be weakly correlated (spike-count correlation, r_sc). This is widely thought to reflect noise in shared afferents, in which case r_sc can limit the reliability of sensory coding. However, it could also be due to feedback from higher-order brain regions. Currently, the relative contributions of these sources are unknown. We addressed this by recording from populations of V1 neurons in macaques performing different discrimination tasks involving the same visual input. We found that the structure of r_sc (the way r_sc varied with neuronal stimulus preference) changed systematically with task instruction. Therefore, even at the earliest stage in the cortical visual hierarchy, r_sc structure during task performance primarily reflects feedback dynamics. Consequently, previous proposals for how r_sc constrains sensory processing need not apply. Furthermore, we show that correlations between the activity of single neurons and choice depend on feedback engaged by the task.

Postnatal refinement of interareal feedforward projections in ferret visual cortex

We studied the postnatal refinement of feedforward (FF) projections from ferret V1 to multiple cortical targets during the period around eye opening. Our goal was to establish (a) whether the developmental refinement of FF projections parallels that of feedback (FB) cortical circuits, and (b) whether FF pathways from V1 to different target areas refine with a similar rate. We injected the tracer CTb into V1 of juvenile ferrets, and visualized the pattern of labeled axon terminals in extrastriate cortex. Bouton density of FF projections to target areas 18, 19, and 21 declined steadily from 4 to 8 weeks postnatal. However, in area Ssy this decline was delayed somewhat, not occurring until after 6 weeks. During this postnatal period, mean interbouton intervals along individual FF axons to all visual areas increased, and we observed a concomitant moderate decrease in axon density in areas 18, 21, and Ssy. These data suggest that FF circuits linking V1 to its main extrastriate targets remodel largely synchronously in the weeks following eye opening, that FF and FB cortical circuits share a broadly similar developmental timecourse, and that postnatal visual experience is critical for the refinement of both FF and FB cortical circuits.

Contextual Interactions in Grating Plaid Configurations Are Explained by Natural Image Statistics and Neural Modeling

Processing natural scenes requires the visual system to integrate local features into global object descriptions. To achieve coherent representations, the human brain uses statistical dependencies to guide weighting of local feature conjunctions. Pairwise interactions among feature detectors in early visual areas may form the early substrate of these local feature bindings. To investigate local interaction structures in visual cortex, we combined psychophysical experiments with computational modeling and natural scene analysis. We first measured contrast thresholds for 2 × 2 grating patch arrangements (plaids), which differed in spatial frequency composition (low, high, or mixed), number of grating patch co-alignments (0, 1, or 2), and inter-patch distances (1° and 2° of visual angle). Contrast thresholds for the different configurations were compared to the prediction of probability summation (PS) among detector families tuned to the four retinal positions. For 1° distance the thresholds for all configurations were larger than predicted by PS, indicating inhibitory interactions. For 2° distance, thresholds were significantly lower compared to PS when the plaids were homogeneous in spatial frequency and orientation, but not when spatial frequencies were mixed or there was at least one misalignment. Next, we constructed a neural population model with horizontal laminar structure, which reproduced the detection thresholds after adaptation of connection weights. Consistent with prior work, contextual interactions were medium-range inhibition and long-range, orientation-specific excitation. However, inclusion of orientation-specific, inhibitory interactions between populations with different spatial frequency preferences were crucial for explaining detection thresholds. Finally, for all plaid configurations we computed their likelihood of occurrence in natural images. The likelihoods turned out to be inversely related to the detection thresholds obtained at larger inter-patch distances. However, likelihoods were almost independent of inter-patch distance, implying that natural image statistics could not explain the crowding-like results at short distances. This failure of natural image statistics to resolve the patch distance modulation of plaid visibility remains a challenge to the approach.

Toward an Integration of Deep Learning and Neuroscience

Neuroscience has focused on the detailed implementation of computation, studying neural codes, dynamics and circuits. In machine learning, however, artificial neural networks tend to eschew precisely designed codes, dynamics or circuits in favor of brute force optimization of a cost function, often using simple and relatively uniform initial architectures. Two recent developments have emerged within machine learning that create an opportunity to connect these seemingly divergent perspectives. First, structured architectures are used, including dedicated systems for attention, recursion and various forms of short- and long-term memory storage. Second, cost functions and training procedures have become more complex and are varied across layers and over time. Here we think about the brain in terms of these ideas. We hypothesize that (1) the brain optimizes cost functions, (2) the cost functions are diverse and differ across brain locations and over development, and (3) optimization operates within a pre-structured architecture matched to the computational problems posed by behavior. In support of these hypotheses, we argue that a range of implementations of credit assignment through multiple layers of neurons are compatible with our current knowledge of neural circuitry, and that the brain's specialized systems can be interpreted as enabling efficient optimization for specific problem classes. Such a heterogeneously optimized system, enabled by a series of interacting cost functions, serves to make learning data-efficient and precisely targeted to the needs of the organism. We suggest directions by which neuroscience could seek to refine and test these hypotheses.

What insights can fMRI offer into the structure and function of mid-tier visual areas?

Inferring neural responses from functional magnetic resonance imaging (fMRI) data is challenging. Even if we take advantage of high-field systems to acquire data with submillimeter resolution, we are still acquiring data in which a single datum summarizes the responses of tens of thousands of neurons. Excitation and inhibition, spikes and subthreshold membrane potential modulations, local and long-range computations, and tuned and nonselective responses are mixed together in one signal. With a priori knowledge of the underlying neural population responses, careful experiment design allows us to manipulate the experiment or task design so that subpopulations are selectively modulated, and our experiments can reveal those tuning functions. However, because we want to be able to use fMRI to discover new kinds of tuning functions and selectivity, we cannot limit ourselves to experiments in which we already know what we are looking for. Broadly speaking, analyses that rely on classification of responses that are distributed across the local neural population [multi-voxel pattern analyses (MVPA)] offer the ability to discover new kinds of information representation and selectivities in neural subpopulations. There is, however, no way to determine how the information discovered with MVPA or other analyses is related to the underlying neuronal tuning functions. Therefore, we must continue to rely on behavioral, computational, and animal models to develop theories of information representation in mid-tier visual cortical areas. Once encoding models exist, fMRI can be powerful for testing these a priori models of information representation. As an aide in developing these models, an important contribution that fMRI can make to our understanding of mid-tier visual areas is derived from connectivity analyses and experiments that study information sharing between visual areas. This ability to quantify localized population average responses throughout the brain is the strength we can best leverage to discover new properties of local and long-range neural networks.

Inhibition as a Binary Switch for Excitatory Plasticity in Pyramidal Neurons

Synaptic plasticity is thought to induce memory traces in the brain that are the foundation of learning. To ensure the stability of these traces in the presence of further learning, however, a regulation of plasticity appears beneficial. Here, we take up the recent suggestion that dendritic inhibition can switch plasticity of excitatory synapses on and off by gating backpropagating action potentials (bAPs) and calcium spikes, i.e., by gating the coincidence signals required for Hebbian forms of plasticity. We analyze temporal and spatial constraints of such a gating and investigate whether it is possible to suppress bAPs without a simultaneous annihilation of the forward-directed information flow via excitatory postsynaptic potentials (EPSPs). In a computational analysis of conductance-based multi-compartmental models, we demonstrate that a robust control of bAPs and calcium spikes is possible in an all-or-none manner, enabling a binary switch of coincidence signals and plasticity. The position of inhibitory synapses on the dendritic tree determines the spatial extent of the effect and allows a pathway-specific regulation of plasticity. With appropriate timing, EPSPs can still trigger somatic action potentials, although backpropagating signals are abolished. An annihilation of bAPs requires precisely timed inhibition, while the timing constraints are less stringent for distal calcium spikes. We further show that a wide-spread motif of local circuits—feedforward inhibition—is well suited to provide the temporal precision needed for the control of bAPs. Altogether, our model provides experimentally testable predictions and demonstrates that the inhibitory switch of plasticity can be a robust and attractive mechanism, hence assigning an additional function to the inhibitory elements of neuronal microcircuits beyond modulation of excitability.

We must constantly learn in order to meet the demands of a dynamically changing environment. The basis of learning is believed to be synaptic plasticity, i.e. the potential of neuronal connections to change. Depending on context, however, it may be either useful to learn and modify connections or, alternatively, to keep an established network structure stable to maintain what has been already learned (also referred to as the plasticity-stability dilemma). The ability to switch synaptic plasticity on and off in a flexible way hence constitutes an attractive feature of neuronal processing. Here, we analyze a cellular mechanism based on the inhibition-mediated gating of coincidence signals required for Hebbian forms of excitatory synaptic plasticity. While experimental evidence in support of individual steps involved in this mechanism is accumulating, it is as of now unclear whether this mechanism can indeed operate robustly under physiologically realistic parameters of pyramidal cells, in particular, without impairing information flow in these cells altogether. Computational modeling allows us to demonstrate that this is indeed possible if inhibition is well timed (on the order of 1 ms). Moreover, we show that a specific design of the local circuit can ensure the necessary timing.

Neural representation for object recognition in inferotemporal cortex

We suggest that population representation of objects in inferotemporal cortex lie on a continuum between a purely structural, parts-based description and a purely holistic description. The intrinsic dimensionality of object representation is estimated to be around 100, perhaps with lower dimensionalities for object representations more toward the holistic end of the spectrum. Cognitive knowledge in the form of semantic information and task information feed back to inferotemporal cortex from perirhinal and prefrontal cortex respectively, providing high-level multimodal-based expectations that assist in the interpretation of object stimuli. Integration of object information across eye movements may also contribute to object recognition through a process of active vision.

Regional Specificity of GABAergic Regulation of Cross-Modal Plasticity in Mouse Visual Cortex after Unilateral Enucleation

In adult mice, monocular enucleation (ME) results in an immediate deactivation of the contralateral medial monocular visual cortex. An early restricted reactivation by open eye potentiation is followed by a late overt cross-modal reactivation by whiskers (Van Brussel et al., 2011). In adolescence (P45), extensive recovery of cortical activity after ME fails as a result of suppression or functional immaturity of the cross-modal mechanisms (Nys et al., 2014). Here, we show that dark exposure before ME in adulthood also prevents the late cross-modal reactivation component, thereby converting the outcome of long-term ME into a more P45-like response. Because dark exposure affects GABAergic synaptic transmission in binocular V1 and the plastic immunity observed at P45 is reminiscent of the refractory period for inhibitory plasticity reported by Huang et al. (2010), we molecularly examined whether GABAergic inhibition also regulates ME-induced cross-modal plasticity. Comparison of the adaptation of the medial monocular and binocular cortices to long-term ME or dark exposure or a combinatorial deprivation revealed striking differences. In the medial monocular cortex, cortical inhibition via the GABAA receptor α1 subunit restricts cross-modal plasticity in P45 mice but is relaxed in adults to allow the whisker-mediated reactivation. In line, in vivo pharmacological activation of α1 subunit-containing GABAA receptors in adult ME mice specifically reduces the cross-modal aspect of reactivation. Together with region-specific changes in glutamate acid decarboxylase (GAD) and vesicular GABA transporter expression, these findings put intracortical inhibition forward as an important regulator of the age-, experience-, and cortical region-dependent cross-modal response to unilateral visual deprivation.

SIGNIFICANCE STATEMENT

In adult mice, vision loss through one eye instantly reduces neuronal activity in the visual cortex. Strengthening of remaining eye inputs in the binocular cortex is followed by cross-modal adaptations in the monocular cortex, in which whiskers become a dominant nonvisual input source to attain extensive cortical reactivation. We show that the cross-modal component does not occur in adolescence because of increased intracortical inhibition, a phenotype that was mimicked in adult enucleated mice when treated with indiplon, a GABAA receptor α1 agonist. The cross-modal versus unimodal responses of the adult monocular and binocular cortices also mirror regional specificity in inhibitory alterations after visual deprivation. Understanding cross-modal plasticity in response to sensory loss is essential to maximize patient susceptibility to sensory prosthetics.

Learning enhances the relative impact of top-down processing in the visual cortex

Theories have proposed that, in sensory cortices, learning can enhance top-down modulation by higher brain areas while reducing bottom-up sensory drives. To address circuit mechanisms underlying this process, we examined the activity of layer 2/3 (L2/3) excitatory neurons in the mouse primary visual cortex (V1) as well as L4 excitatory neurons, the main bottom-up source, and long-range top-down projections from the retrosplenial cortex (RSC) during associative learning over days using chronic two-photon calcium imaging. During learning, L4 responses gradually weakened, whereas RSC inputs became stronger. Furthermore, L2/3 acquired a ramp-up response temporal profile, potentially encoding the timing of the associated event, which coincided with a similar change in RSC inputs. Learning also reduced the activity of somatostatin-expressing inhibitory neurons (SOM-INs) in V1 that could potentially gate top-down inputs. Finally, RSC inactivation or SOM-IN activation was sufficient to partially reverse the learning-induced changes in L2/3. Together, these results reveal a learning-dependent dynamic shift in the balance between bottom-up and top-down information streams and uncover a role of SOM-INs in controlling this process.

Visual system plasticity in mammals: the story of monocular enucleation-induced vision loss

The groundbreaking work of Hubel and Wiesel in the 1960’s on ocular dominance plasticity instigated many studies of the visual system of mammals, enriching our understanding of how the development of its structure and function depends on high quality visual input through both eyes. These studies have mainly employed lid suturing, dark rearing and eye patching applied to different species to reduce or impair visual input, and have created extensive knowledge on binocular vision. However, not all aspects and types of plasticity in the visual cortex have been covered in full detail. In that regard, a more drastic deprivation method like enucleation, leading to complete vision loss appears useful as it has more widespread effects on the afferent visual pathway and even on non-visual brain regions. One-eyed vision due to monocular enucleation (ME) profoundly affects the contralateral retinorecipient subcortical and cortical structures thereby creating a powerful means to investigate cortical plasticity phenomena in which binocular competition has no vote.In this review, we will present current knowledge about the specific application of ME as an experimental tool to study visual and cross-modal brain plasticity and compare early postnatal stages up into adulthood. The structural and physiological consequences of this type of extensive sensory loss as documented and studied in several animal species and human patients will be discussed. We will summarize how ME studies have been instrumental to our current understanding of the differentiation of sensory systems and how the structure and function of cortical circuits in mammals are shaped in response to such an extensive alteration in experience. In conclusion, we will highlight future perspectives and the clinical relevance of adding ME to the list of more longstanding deprivation models in visual system research.

Feedback stabilizes propagation of synchronous spiking in cortical neural networks

Precisely timed action potentials related to stimuli and behavior have been observed in the cerebral cortex. However, information carried by the precise spike timing has to propagate through many cortical areas, and noise could disrupt millisecond precision during the transmission. Previous studies have demonstrated that only strong stimuli that evoke a large number of spikes with small dispersion of spike times can propagate through multilayer networks without degrading the temporal precision. Here we show that feedback projections can increase the number of spikes in spike volleys without degrading their temporal precision. Feedback also increased the range of spike volleys that can propagate through multilayer networks. Our work suggests that feedback projections could be responsible for the reliable propagation of information encoded in spike times through cortex, and thus could serve as an attentional mechanism to regulate the flow of information in the cortex. Feedback projections may also participate in generating spike synchronization that is engaged in cognitive behaviors by the same mechanisms described here for spike propagation.