Search

Spatial attention is an essential cognitive process for visual perception, especially in complex scenes with poor luminance. Attentional modulation of neural activity has been documented across the visual cortex. However, how these changes in neural response lead to better behavioral performance on challenging tasks remain unknown. In our study, we implemented spatial attention in a deep convolutional neural network model of the ventral visual hierarchy and measured its impact on categorization performance on cluttered images with varying contrast levels. We applied attention to network units in three ways: enhancement in attended region (EAR), suppression in unattended region (SUAR) or both. When focally applied to a single convolutional layer, SUAR is more effective in boosting performance than EAR, especially in the presence of high contrast distractors. SUAR is also effective in recovering degraded performance due to low contrast targets, whereas EAR fails to recover. These results predict a novel mechanism of suppression of neural activity corresponding to the unattended parts of visual space. Intriguingly, EAR in our model achieves the same performance as SUAR when attention is diffusely applied to stacks of successive convolutional layers, irrespective of target contrast. This suggests an alternate mechanism of attention wherein enhancement of attended neural activity alone, when applied to successive cortical encoding stages, is an effective strategy for boosting performance in challenging object recognition tasks. Our results indicate that the two alternative attentional mechanisms are functionally equivalent in tackling challenging object recognition tasks.

Adaptive information processing, comprised of local computations and their efficient routing, is crucial for flexible brain function. Spatial attention is a quintessential example of this adaptive process. It is critical for recognizing and interacting with behaviorally relevant objects in a cluttered environment1. Object recognition is mediated by ensembles of computational units distributed across the ventral visual hierarchy. How the deployment of spatial attention aids these hierarchical computations is unclear. Based on pairwise correlation analysis, two key mechanisms have been proposed: First is an improvement in the efficacy of unique information directed from one encoding stage to another, suggested by evidence along the visual hierarchy2–6. Based on the theoretical results that even weak correlated variability can substantially limit the encoding capacity of a neuronal pool7, a second proposal is an improvement in the sensory information capacity of an encoding stage through a reduction in shared fluctuations8,9. However, pairwise analyses capture both unique and shared components of fluctuations, and therefore cannot disambiguate the proposed mechanisms. To test these proposals, it is crucial to estimate the attentional modulation of unique information flow across and shared information within the stages of the visual hierarchy. We investigated this in the multi-stage laminar network of visual area V4, an area strongly modulated by attention10–12. Using network-based statistical modeling, we estimated the strength of inter-layer information flow by measuring statistical dependencies that reflect how the cortical layersuniquelydrive each other’s neural activity. We quantified their modulation across attention conditions (attend-in vs. attend-away) in a change detection task and found that deployment of attention indeed strengthened unique dependencies between the input and superficial layers. Using the partial information decomposition framework13,14, we estimated modulation of shared dependencies and found that they are reduced within laminar populations, specifically the putative excitatory subpopulations. Surprisingly, we found a strengthening of unique dependencies within the laminar populations, a finding not previously predicted. Crucially, these modulation patterns were also observed across behavioral outcomes (hit vs. miss) that are thought to be mediated by endogenous state fluctuations15–17. By “decomposing” the modulation of dependency components and in combination with prior theoretical work7, our results suggest the following computational model of optimal sensory states that are attained by either task demands or endogenous fluctuations in brain state: enhanced information flowbetweenand improved information capacitywithinencoding stages.

Synaptic excitation (E) and inhibition (I) stay relatively proportional to each other over different spatiotemporal scales, orchestrating neuronal activity in the brain. This proportionality, referred to as E-I balance, is thought to be critical for neuronal functions because its disruption was observed in many neurological disorders. However, the causal evidence demonstrating its significance is scarce. Here we show that deleting Neuroligin-2 (Nlgn2), a postsynaptic adhesion molecule at inhibitory synapses, from mouse glutamatergic or GABAergic neurons reduces inhibition cell-autonomously without affecting excitation, thereby disrupting E-I balance and causing lethality. In contrast, deleting Nlgn2 constitutively or simultaneously from both glutamatergic and GABAergic neurons results in viable mice. A neural network model shows that reducing inhibition in either neuronal type is detrimental to network activity, but in both types partially re-establishes E-I balance and activity. Together, our results provide evidence for an essential role of E-I balance in brain functions and organism survival.

Identical stimuli can be perceived or go unnoticed across successive presentations, producing divergent behavioral readouts despite similarities in sensory input. We hypothesized that fluctuations in neurophysiological states in the sensory neocortex, which could alter cortical processing at the level of neural subpopulations, underlies this perceptual variability. We analyzed cortical layer-specific electrophysiological activity in visual area V4 during a cued attention task. We find that hit trials are characterized by a larger pupil diameter and lower incidence of microsaccades, indicative of a behavioral state with increased arousal and perceptual stability. Target stimuli presented at perceptual threshold evoke elevated multi-unit activity in V4 neurons in hit trials compared to miss trials, across all cortical layers. Putative excitatory and inhibitory neurons are strongly positively modulated in the input (IV) and deep (V & VI) layers of the cortex during hit trials. Excitatory neurons in the superficial cortical layers exhibit lower variability in hit trials. Deep layer neurons are less phase-locked to low frequency rhythms in hits. Hits are also characterized by greater interlaminar coherence between the superficial and deep layers in the pre-stimulus period, and a complementary pattern between the input layer and both the superficial and deep layers in the stimulus-evoked period. Taken together, these results indicate that a state of elevated levels of arousal and perceptual stability allow enhanced processing of sensory stimuli, which contributes to hits at perceptual threshold.

Contrast is a key feature of the visual scene that aids object recognition. Attention has been shown to selectively enhance the responses to low contrast stimuli in visual area V4, a critical hub that sends projections both up and down the visual hierarchy. Veridical encoding of contrast information is a key computation in early visual areas, while later stages encode higher level features that benefit from improved sensitivity to low contrast. How area V4 meets these distinct information processing demands in the attentive state is not known. We found that attentional modulation of contrast responses in area V4 is cortical layer and cell-class specific. Putative excitatory neurons in the superficial output layers that project to higher areas show enhanced boosting of low contrast information. On the other hand, putative excitatory neurons of deep output layers that project to early visual areas exhibit contrast-independent scaling. Computational modeling revealed that such layer-wise differences may result from variations in spatial integration extent of inhibitory neurons. These findings reveal that the nature of interactions between attention and contrast in V4 is highly compartmentalized, in alignment with the demands of the visual processing hierarchy.

Saccades are a ubiquitous and crucial component of our visual system, allowing for the efficient deployment of the fovea and its accompanying neural resources. Initiation of a saccade is known to cause saccadic suppression, a temporary reduction in visual sensitivity1, 2 and visual cortical firing rates3–6. While saccadic suppression has been well characterized at the level of perception and single neurons, relatively little is known about the visual cortical networks governing this phenomenon. Here we examine the effects of saccadic suppression on distinct neural subpopulations within visual area V4. We find cortical layer- and cell type-specific differences in the magnitude and timing of peri-saccadic modulation. Neurons in the input layer show changes in firing rate and inter-neuronal correlations prior to saccade onset, indicating that this layer receives information about impending saccades. Putative inhibitory interneurons in the input layer elevate their firing rate during saccades, suggesting they play a role in suppressing the activity of other cortical subpopulations. A computational model of this circuit recapitulates our empirical observations and demonstrates that an input layer-targeting pathway can initiate saccadic suppression by enhancing local inhibitory activity. Collectively, our results provide a mechanistic understanding of how eye movement signaling interacts with cortical circuitry to enforce visual stability.

SummaryBrain activity is highly regulated by GABAergic activity, which acts via GABAARs to suppress somatic spike generation as well as dendritic synaptic integration and calcium signaling. Tonic GABAergic conductances mediated by distinct receptor subtypes can also inhibit neuronal excitability and spike output, though the consequences for dendritic calcium signaling are unclear. Here, we use 2-photon calcium imaging in cortical pyramidal neurons and computational modeling to show that low affinity GABAARs containing an α5 subunit mediate a tonic hyperpolarization of the dendritic membrane potential, resulting in deinactivation of voltage-gated calcium channels and a paradoxical boosting of action potential-evoked calcium influx. We also find that GABAergic enhancement of calcium signaling modulates short-term synaptic plasticity, augmenting depolarization-induced suppression of inhibition. These results demonstrate a novel role for GABA in the control of dendritic activity and suggest a mechanism for differential modulation of electrical and biochemical signaling.

Deployment of covert attention to a spatial location can cause large decreases in low-frequency correlated variability among neurons in macaque area V4 whose receptive-fields lie at the attended location. It has been estimated that this reduction accounts for a substantial fraction of the attention-mediated improvement in sensory processing. These estimates depend on assumptions about how population signals are decoded and the conclusion that correlated variability impairs perception, is purely hypothetical. Here we test this proposal directly by optogenetically inducing low-frequency fluctuations, to see if this interferes with performance in an attention-demanding task. We find that low-frequency optical stimulation of neurons in V4 elevates correlations among pairs of neurons and impairs the animal’s ability to make fine sensory discriminations. Stimulation at higher frequencies does not impair performance, despite comparable modulation of neuronal responses. These results support the hypothesis that attention-dependent reductions in correlated variability contribute to improved perception of attended stimuli.

N-methyl-D-aspartate receptor (NMDAR) hypofunction in parvalbumin-expressing (PV+) inhibitory neurons (INs) may contribute to symptoms in patients with schizophrenia (SZ). This hypothesis was inspired by studies in humans involving NMDAR antagonists that trigger SZ symptoms. Animal models of SZ using neuropharmacology and genetic knockouts have successfully replicated some of the key observations in human subjects involving alteration of gamma band oscillations (GBO) observed in electroencephalography and magnetoencephalography signals. However, it remains to be seen if NMDAR hypofunction in PV+ neurons is fundamental to the phenotype observed in these models. In this review, we discuss some of the key computational models of GBO and their predictions in the context of NMDAR hypofunction in INs. While PV+ INs have been the main focus of SZ studies in animal models, we also discuss the implications of NMDAR hypofunction in other types of INs using computational models for GBO modulation in the visual cortex.

A signature feature of the neocortex is the dense network of horizontal connections (HCs) through which pyramidal neurons (PNs) exchange “contextual” information. In primary visual cortex (V1), HCs are thought to facilitate boundary detection, a crucial operation for object recognition, but how HCs modulate PN responses to boundary cues within their classical receptive fields (CRF) remains unknown. We began by “asking” natural images, through a structured data collection and ground truth labeling process, what function a V1 cell should use to compute boundary probability from aligned edge cues within and outside its CRF. The “answer” was an asymmetric 2-D sigmoidal function, whose nonlinear form provides the first normative account for the “multiplicative” center-flanker interactions previously reported in V1 neurons (Kapadia et al. 1995, 2000; Polat et al. 1998). Using a detailed compartmental model, we then show that this boundary-detecting classical-contextual interaction function can be computed with near perfect accuracy by NMDAR-dependent spatial synaptic interactions within PN dendrites – the site where classical and contextual inputs first converge in the cortex. In additional simulations, we show that local interneuron circuitry activated by HCs can powerfully leverage the nonlinear spatial computing capabilities of PN dendrites, providing the cortex with a highly flexible substrate for integration of classical and contextual information.Significance StatementIn addition to the driver inputs that establish their classical receptive fields, cortical pyramidal neurons (PN) receive a much larger number of “contextual” inputs from other PNs through a dense plexus of horizontal connections (HCs). However by what mechanisms, and for what behavioral purposes, HC’s modulate PN responses remains unclear. We pursued these questions in the context of object boundary detection in visual cortex, by combining an analysis of natural boundary statistics with detailed modeling PNs and local circuits. We found that nonlinear synaptic interactions in PN dendrites are ideally suited to solve the boundary detection problem. We propose that PN dendrites provide the core computing substrate through which cortical neurons modulate each other’s responses depending on context.

Understanding the anatomical and functional architecture of the brain is essential for designing neurally inspired intelligent systems. Theoretical and empirical studies suggest a role for narrowband oscillations in shaping the functional architecture of the brain through their role in coding and communication of information. Such oscillations are ubiquitous signals in the electrical activity recorded from the brain. In the cortex, oscillations detected in the gamma range (30-80 Hz) are modulated by behavioral states and sensory features in complex ways. How is this regulation achieved? Although several underlying principles for the genesis of these oscillations have been proposed, a unifying account for their regulation has remained elusive. In a network of excitatory and inhibitory neurons operating in an inhibition-stabilized regime, we show that strongly superlinear responses of inhibitory neurons facilitate bidirectional regulation of oscillation frequency and power. In such a network, the balance of drives to the excitatory and inhibitory populations determines how the power and frequency of oscillations are modulated. The model accounts for the puzzling increase in their frequency with the salience of visual stimuli, and a decrease with their size. Oscillations in our model grow stronger as the mean firing level is reduced, accounting for the size dependence of visually evoked gamma rhythms, and suggesting a role for oscillations in improving the signal-to-noise ratio (SNR) of signals in the brain. Empirically testing such predictions is still challenging, and implementing the proposed coding and communication strategies in neuromorphic systems could assist in our understanding of the biological system.

Precise spike times carry information and are important for synaptic plasticity. Synchronizing oscillations such as gamma bursts could coordinate spike times, thus regulating information transmission in the cortex. Oscillations are driven by inhibitory neurons and are modulated by sensory stimuli and behavioral states. How their power and frequency are regulated is an open question. Using a model cortical circuit, we propose a regulatory mechanism that depends on the activity balance of monosynaptic and disynaptic pathways to inhibitory neurons: Monosynaptic input causes more powerful oscillations whereas disynaptic input increases the frequency of oscillations. The balance of stimulation to the two pathways modulates the overall distribution of spikes, with stronger disynaptic stimulation (e.g., preferred stimuli inside visual receptive fields) producing high firing rates and weak oscillations; in contrast, stronger monosynaptic stimulation (e.g., suppressive contextual stimulation from outside visual receptive fields) generates low firing rates and strong oscillatory regulation of spike timing, as observed in alert cortex processing complex natural stimuli. By accounting for otherwise paradoxical experimental findings, our results demonstrate how the frequency and power of oscillations, and hence spike times, can be modulated by both sensory input and behavioral context, with powerful oscillations signifying a cortical state under inhibitory control in which spikes are sparse and spike timing is precise.

Gamma band rhythms may synchronize distributed cell assemblies to facilitate information transfer within and across brain areas, yet their underlying mechanisms remain hotly debated. Most circuit models postulate that soma-targeting parvalbumin-positive GABAergic neurons are the essential inhibitory neuron subtype necessary for gamma rhythms. Using cell-type-specific optogenetic manipulations in behaving animals, we show that dendrite-targeting somatostatin (SOM) interneurons are critical for a visually induced, context-dependent gamma rhythm in visual cortex. A computational model independently predicts that context-dependent gamma rhythms depend critically on SOM interneurons. Further in vivo experiments show that SOM neurons are required for long-distance coherence across the visual cortex. Taken together, these data establish an alternative mechanism for synchronizing distributed networks in visual cortex. By operating through dendritic and not just somatic inhibition, SOM-mediated oscillations may expand the computational power of gamma rhythms for optimizing the synthesis and storage of visual perceptions.

To deepen our understanding of object recognition, it is critical to understand the nature of transformations that occur in intermediate stages of processing in the ventral visual pathway, such as area V4. Neurons in V4 are selective to local features of global shape, such as extended contours. Previously, we found that V4 neurons selective for curved elements exhibit a high degree of spatial variation in their preference. If spatial variation in curvature selectivity was also marked by distinct temporal response patterns at different spatial locations, then it might be possible to untangle this information in subsequent processing based on temporal responses. Indeed, we find that V4 neurons whose receptive fields exhibit intricate selectivity also show variation in their temporal responses across locations. A computational model that decodes stimulus identity based on population responses benefits from using this temporal information, suggesting that it could provide a multiplexed code for spatio-temporal features.

Neuromodulatory signaling is generally considered broad in its impact across cortex. However, variations in the characteristics of cortical circuits may introduce regionally-specific responses to diffuse modulatory signals. Features such as patterns of axonal innervation, tissue tortuosity and molecular diffusion, effectiveness of degradation pathways, subcellular receptor localization, and patterns of receptor expression can lead to local modification of modulatory inputs. We propose that modulatory compartments exist in cortex and can be defined by variation in structural features of local circuits. Further, we argue that these compartments are responsible for local regulation of neuromodulatory tone. For the cholinergic system, these modulatory compartments are regions of cortical tissue within which signaling conditions for acetylcholine are relatively uniform, but between which signaling can vary profoundly. In the visual system, evidence for the existence of compartments indicates that cholinergic modulation likely differs across the visual pathway. We argue that the existence of these compartments calls for thinking about cholinergic modulation in terms of finer-grained control of local cortical circuits than is implied by the traditional view of this system as a diffuse modulator. Further, an understanding of modulatory compartments provides an opportunity to better understand and perhaps correct signal modifications that lead to pathological states.

In pursuit of the goal to understand and eventually reproduce the diverse functions of the brain, a key challenge lies in reverse engineering the peculiar biology-based “technology” that underlies the brain's remarkable ability to process and store information. The basic building block of the nervous system is the nerve cell, or “neuron,” yet after more than 100 years of neurophysiological study and 60 years of modeling, the information processing functions of individual neurons, and the parameters that allow them to engage in so many different types of computation (sensory, motor, mnemonic, executive, etc.) remain poorly understood. In this paper, we review both historical and recent findings that have led to our current understanding of the analog spatial processing capabilities of dendrites, the major input structures of neurons, with a focus on the principal cell type of the neocortex and hippocampus, the pyramidal neuron (PN). We encapsulate our current understanding of PN dendritic integration in an abstract layered model whose spatially sensitive branch-subunits compute multidimensional sigmoidal functions. Unlike the 1-D sigmoids found in conventional neural network models, multidimensional sigmoids allow the cell to implement a rich spectrum of nonlinear modulation effects directly within their dendritic trees.

Cortical computations are critically dependent on interactions between pyramidal neurons (PNs) and a menagerie of inhibitory interneuron types. A key feature distinguishing interneuron types is the spatial distribution of their synaptic contacts onto PNs, but the location-dependent effects of inhibition are mostly unknown, especially under conditions involving active dendritic responses. We studied the effect of somatic vs. dendritic inhibition on local spike generation in basal dendrites of layer 5 PNs both in neocortical slices and in simple and detailed compartmental models, with equivalent results: somatic inhibition divisively suppressed the amplitude of dendritic spikes recorded at the soma while minimally affecting dendritic spike thresholds. In contrast, distal dendritic inhibition raised dendritic spike thresholds while minimally affecting their amplitudes. On-the-path dendritic inhibition modulated both the gain and threshold of dendritic spikes depending on its distance from the spike initiation zone. Our findings suggest that cortical circuits could assign different mixtures of gain vs. threshold inhibition to different neural pathways, and thus tailor their local computations, by managing their relative activation of soma- vs. dendrite-targeting interneurons., Author Summary Establishing how inhibitory neurons shape the computing functions of neural circuits is crucial to understanding both normal function and dysfunction in the human brain. It has been known for over a century that different classes of inhibitory interneurons project to different sub-regions of the neurons they contact – some primarily target cell bodies, others the dendrites, still others the axon. It remains poorly understood, however, how these different projection patterns influence synaptic integration in the target neuron populations. By providing new data from intracellular recordings in brain slices, and a simple but powerful model of the location-dependent effects of inhibition on dendritic spike generation, our study (1) demonstrates the importance of the absolute and relative locations of excitatory and inhibitory inputs to pyramidal neurons, the principal cells of the cerebral cortex, and (2) helps to establish a more solid theoretical understanding of these complex integrative phenomena. As high resolution mapping of the cortical “connectome” becomes available in the coming years, our work will be helpful in interpreting the computing functions of cortical tissue both at the single neuron and circuit levels.

Neocortical pyramidal neurons (PNs) receive thousands of excitatory synaptic contacts on their basal dendrites. Some act as classical driver inputs while others are thought to modulate PN responses based on sensory or behavioral context, but the biophysical mechanisms that mediate classical-contextual interactions in these dendrites remain poorly understood. We hypothesized that if two excitatory pathways bias their synaptic projections towards proximal vs. distal ends of the basal branches, the very different local spike thresholds and attenuation factors for inputs near and far from the soma might provide the basis for a classical-contextual functional asymmetry. Supporting this possibility, we found both in compartmental models and electrophysiological recordings in brain slices that the responses of basal dendrites to spatially separated inputs are indeed strongly asymmetric. Distal excitation lowers the local spike threshold for more proximal inputs, while having little effect on peak responses at the soma. In contrast, proximal excitation lowers the threshold, but also substantially increases the gain of distally-driven responses. Our findings support the view that PN basal dendrites possess significant analog computing capabilities, and suggest that the diverse forms of nonlinear response modulation seen in the neocortex, including uni-modal, cross-modal, and attentional effects, could depend in part on pathway-specific biases in the spatial distribution of excitatory synaptic contacts onto PN basal dendritic arbors., Author Summary Pyramidal neurons (PNs) are the principal neurons of the cerebral cortex and therefore lie at the heart of the brain's higher sensory, motor, affective, memory, and executive functions. But how do they work? In particular, how do they manage interactions between the classical “driver” inputs that give rise to their basic response properties, and “contextual” inputs that nonlinearly modulate those responses? It is known that PNs are contacted by thousands of excitatory synapses scattered about their dendrites, but despite decades of research, the “rules” that govern how inputs at different locations in the dendritic tree combine to influence the cell's firing rate remain poorly understood. We show here that two excitatory inputs contacting the same dendrite interact in an asymmetric nonlinear way that depends on their absolute and relative locations, where the resulting spectrum of location-dependent synaptic interactions constitutes a previously unknown form of spatial analog computation. In addition to suggesting a possible substrate for classical-contextual interactions in PN dendrites, our results imply that the computing functions of cortical circuits can only be fully understood when the detailed map of synaptic connectivity – the cortical connectome – is known down to the subdendritic level.