Inhibitory stabilization of the cortical network underlies visual surround suppression

Feature-based attention in early vision for the modulation of figure-ground segregation

We investigated psychophysically whether feature-based attention modulates the perception of figure-ground (F-G) segregation and, based on the results, we investigated computationally the neural mechanisms underlying attention modulation. In the psychophysical experiments, the attention of participants was drawn to a specific motion direction and they were then asked to judge the side of figure in an ambiguous figure with surfaces consisting of distinct motion directions. The results of these experiments showed that the surface consisting of the attended direction of motion was more frequently observed as figure, with a degree comparable to that of spatial attention (Wagatsuma et al., 2008). These experiments also showed that perception was dependent on the distribution of feature contrast, specifically the motion direction differences. These results led us to hypothesize that feature-based attention functions in a framework similar to that of spatial attention. We proposed a V1-V2 model in which feature-based attention modulates the contrast of low-level feature in V1, and this modulation of contrast changes directly the surround modulation of border-ownership-selective cells in V2; thus, perception of F-G is biased. The model exhibited good agreement with human perception in the magnitude of attention modulation and its invariance among stimuli. These results indicate that early-level features that are modified by feature-based attention alter subsequent processing along afferent pathway, and that such modification could even change the perception of object.

Different orientation tuning of near- and far-surround suppression in macaque primary visual cortex mirrors their tuning in human perception

In primary visual cortex (V1), neuronal responses to stimuli inside the receptive field (RF) are usually suppressed by stimuli in the RF surround. This suppression is orientation specific. Similarly, in human vision surround stimuli can suppress perceived contrast of a central stimulus in an orientation-dependent manner. The surround consists of two regions likely generated by different circuits: a near-surround generated predominantly by geniculocortical and intra-V1 horizontal connections, and a far-surround generated exclusively by interareal feedback. Using stimuli confined to the near- or far-surround of V1 neurons, and similar stimuli in human psychophysics, we find that near-surround suppression is more sharply orientation tuned than far-surround suppression in both macaque V1 and human perception. These results point to a similarity between surround suppression in macaque V1 and human vision, and suggest that feedback circuits are less orientation biased than horizontal circuits. We find the sharpest tuning of near-surround suppression in V1 layers (3, 4B, 4Cα) with patterned and orientation-specific horizontal connections. Sharpest tuning of far-surround suppression occurs in layer 4B, suggesting greater orientation specificity of feedback to this layer. Different orientation tuning of near- and far-surround suppression may reflect a statistical bias in natural images, whereby nearby edges have higher probability than distant edges of being co-oriented and belonging to the same contour. Surround suppression would, thus, increase the coding efficiency of frequently co-occurring contours and the saliency of less frequent ones. Such saliency increase can help detect small orientation differences in nearby edges (for contour completion), but large orientation differences in distant edges (for directing saccades/attention).

No consistent relationship between gamma power and peak frequency in macaque primary visual cortex

Neural activity in the gamma frequency range ("gamma") is elevated during active cognitive states. Gamma has been proposed to play an important role in cortical function, although this is debated. Understanding what function gamma might fulfill requires a better understanding of its properties and the mechanisms that generate it. Gamma is characterized by its spectral power and peak frequency, and variations in both parameters have been associated with changes in behavioral performance. Modeling studies suggest these properties are co-modulated, but this has not been established. To test the relationship between these properties, we measured local field potentials (LFPs) and neuronal spiking responses in primary visual cortex of anesthetized monkeys, for drifting sinusoidal gratings of different sizes, contrasts, orientations and masked with different levels of noise. We find that there is no fixed relationship between LFP gamma power and peak frequency, and neither is related to the strength of spiking activity. We propose a simple model that can account for the complex stimulus dependence we observe, and suggest that separate mechanisms determine gamma power and peak frequency.

Strength of gamma rhythm depends on normalization

Manipulating a divisive normalization mechanism independently of attention in monkeys suggests that gamma power reflects excitation-inhibition interactions rather than plays a functional role in attentional processing.

Neuronal assemblies often exhibit stimulus-induced rhythmic activity in the gamma range (30–80 Hz), whose magnitude depends on the attentional load. This has led to the suggestion that gamma rhythms form dynamic communication channels across cortical areas processing the features of behaviorally relevant stimuli. Recently, attention has been linked to a normalization mechanism, in which the response of a neuron is suppressed (normalized) by the overall activity of a large pool of neighboring neurons. In this model, attention increases the excitatory drive received by the neuron, which in turn also increases the strength of normalization, thereby changing the balance of excitation and inhibition. Recent studies have shown that gamma power also depends on such excitatory–inhibitory interactions. Could modulation in gamma power during an attention task be a reflection of the changes in the underlying excitation–inhibition interactions? By manipulating the normalization strength independent of attentional load in macaque monkeys, we show that gamma power increases with increasing normalization, even when the attentional load is fixed. Further, manipulations of attention that increase normalization increase gamma power, even when they decrease the firing rate. Thus, gamma rhythms could be a reflection of changes in the relative strengths of excitation and normalization rather than playing a functional role in communication or control.

Brain signals often show a stimulus-induced rhythm in the “gamma” band (30–80 Hz) whose magnitude depends on attentional load, leading to suggestions that gamma rhythm plays a functional role in routing signals across cortical areas. However, gamma power also depends on simple stimulus features such as size or contrast, which suggests that gamma could arise from basic cortical processes involving excitation–inhibition interactions. One such process is divisive normalization, a mechanism that suppresses the response of a neuron by the overall activity of a large pool of neighboring neurons. Recent studies have shown that attention increases the strength of both excitation and normalization. We hypothesized that the increase in gamma power in an attention task is due to the effect of attention on excitation and normalization. By manipulating the normalization strength independent of attentional load in macaque monkeys, we show that gamma power increases with increasing normalization, even when attentional load is held fixed. Thus, gamma rhythms could be a reflection of changes in the relative strengths of excitation and normalization rather than playing a functional role in communication or control.

High-resolution fMRI reveals laminar differences in neurovascular coupling between positive and negative BOLD responses

The six cortical layers have distinct anatomical and physiological properties, like different energy use and different feedforward and feedback connectivity. It is not known if and how layer-specific neural processes are reflected in the fMRI signal. To address this question we used high-resolution fMRI to measure BOLD, CBV, and CBF responses to stimuli that elicit positive and negative BOLD signals in macaque primary visual cortex. We found that regions with positive BOLD responses had parallel increases in CBV and CBF, whereas areas with negative BOLD responses showed a decrease in CBF but an increase in CBV. For positive BOLD responses, CBF and CBV increased in the center of the cortex, but for negative BOLD responses, CBF decreased superficially while CBV increased in the center. Our findings suggest different mechanisms for neurovascular coupling for BOLD increases and decreases, as well as laminar differences in neurovascular coupling.

Inhibition dominates sensory responses in the awake cortex

The activity of the cerebral cortex is thought to depend on the precise relationship between synaptic excitation and inhibition. In the visual cortex, in particular, intracellular measurements have related response selectivity to coordinated increases in excitation and inhibition. These measurements, however, have all been made during anaesthesia, which strongly influences cortical state and therefore sensory processing. The synaptic activity that is evoked by visual stimulation during wakefulness is unknown. Here we measured visually evoked responses--and the underlying synaptic conductances--in the visual cortex of anaesthetized and awake mice. Under anaesthesia, responses could be elicited from a large region of visual space and were prolonged. During wakefulness, responses were more spatially selective and much briefer. Whole-cell patch-clamp recordings of synaptic conductances showed a difference in synaptic inhibition between the two conditions. Under anaesthesia, inhibition tracked excitation in amplitude and spatial selectivity. By contrast, during wakefulness, inhibition was much stronger than excitation and had extremely broad spatial selectivity. We conclude that during wakefulness, cortical responses to visual stimulation are dominated by synaptic inhibition, restricting the spatial spread and temporal persistence of neural activity. These results provide a direct glimpse of synaptic mechanisms that control sensory responses in the awake cortex.

Dynamics of local input normalization result from balanced short- and long-range intracortical interactions in area V1

To efficiently drive many behaviors, sensory systems have to integrate the activity of large neuronal populations within a limited time window. These populations need to rapidly achieve a robust representation of the input image, probably through canonical computations such as divisive normalization. However, little is known about the dynamics of the corticocortical interactions implementing these rapid and robust computations. Here, we measured the real-time activity of a large neuronal population in V1 using voltage-sensitive dye imaging in behaving monkeys. We found that contrast gain of the population increases over time with a time constant of ~30 ms and propagates laterally over the cortical surface. This dynamic is well accounted for by a divisive normalization achieved through a recurrent network that transiently increases in size after response onset with a slow swelling speed of 0.007-0.014 m/s, suggesting a polysynaptic intracortical origin. In the presence of a surround, this normalization pool is gradually balanced by lateral inputs propagating from distant cortical locations. This results in a centripetal propagation of surround suppression at a speed of 0.1-0.3 m/s, congruent with horizontal intracortical axons speed. We propose that a simple generalized normalization scheme can account for both the dynamical contrast response function through recurrent polysynaptic intracortical loops and for the surround suppression through long-range monosynaptic horizontal spread. Our results demonstrate that V1 achieves a rapid and robust context-dependent input normalization through a timely push-pull between local and lateral networks. We suggest that divisive normalization, a fundamental canonical computation, should be considered as a dynamic process.

Sharper orientation tuning of the extraclassical suppressive-surround due to a neuron's location in the V1 orientation map emerges late in time

Neuronal responses in primary visual cortex (V1) can be suppressed by a stimulus presented to the extraclassical surround, and such interactions are thought to be critical for figure ground segregation and form perception. While surround suppression likely originates from both feedforward afferents and multiple cortical circuits, it is unclear what role each circuit plays in the surround's orientation tuning. To investigate this we recorded from single units in V1 of anesthetized cat and analyzed the orientation tuning of the suppressive-surround over time. In addition, based on orientation maps derived through optical imaging prior to recording, neurons were classified as being located in domains or pinwheels. For both types of neurons, shortly after response onset (10 ms) the suppressive-surround is broadly tuned to orientation, but this is followed by a steep improvement in tuning over the next ∼30 ms. While the tuning of the pinwheel cells plateaus at this point, tuning is enhanced further for domain cells, especially those located superficially in the cortex, reaching a peak at 80 ms from response onset. This relatively slow evolution of the orientation tuning of the suppressive surround suggests that fast-arriving feedforward circuits (10 ms) likely only provide broadly tuned suppression, but that feedback from higher visual areas which is likely to arrive over the next 30 ms and can cover both the receptive field center and the extraclassical surround contributes to the initial steep rise in tuning for both cell types. Moreover, we speculate that the even later enhancement in tuning for domain neurons could mean the involvement of inputs from relatively long-range lateral connections, which not only propagate slowly but also link like-oriented domains corresponding to the receptive field of only the extraclassical surround.

A neural circuit for spatial summation in visual cortex

The response of cortical neurons to a sensory stimulus is modulated by the context. In the visual cortex, for example, stimulation of a pyramidal cell's receptive-field surround can attenuate the cell's response to a stimulus in the centre of its receptive field, a phenomenon called surround suppression. Whether cortical circuits contribute to surround suppression or whether the phenomenon is entirely relayed from earlier stages of visual processing is debated. Here we show that, in contrast to pyramidal cells, the response of somatostatin-expressing inhibitory neurons (SOMs) in the superficial layers of the mouse visual cortex increases with stimulation of the receptive-field surround. This difference results from the preferential excitation of SOMs by horizontal cortical axons. By perturbing the activity of SOMs, we show that these neurons contribute to pyramidal cells' surround suppression. These results establish a cortical circuit for surround suppression and attribute a particular function to a genetically defined type of inhibitory neuron.

Dynamics of spiking neurons: between homogeneity and synchrony

Randomly connected networks of neurons driven by Poisson inputs are often assumed to produce "homogeneous" dynamics, characterized by largely independent firing and approximable by diffusion processes. At the same time, it is well known that such networks can fire synchronously. Between these two much studied scenarios lies a vastly complex dynamical landscape that is relatively unexplored. In this paper, we discuss a phenomenon which commonly manifests in these intermediate regimes, namely brief spurts of spiking activity which we call multiple firing events (MFE). These events do not depend on structured network architecture nor on structured input; they are an emergent property of the system. We came upon them in an earlier modeling paper, in which we discovered, through a careful benchmarking process, that MFEs are the single most important dynamical mechanism behind many of the V1 phenomena we were able to replicate. In this paper we explain in a simpler setting how MFEs come about, as well as their potential dynamic consequences. Although the mechanism underlying MFEs cannot easily be captured by current population dynamics models, this phenomena should not be ignored during analysis; there is a growing body of evidence that such collaborative activity may be a key towards unlocking the possible functional properties of many neuronal networks.

Traveling waves in visual cortex

Electrode recordings and imaging studies have revealed that localized visual stimuli elicit waves of activity that travel across primary visual cortex. Traveling waves are present also during spontaneous activity, but they can be greatly reduced by widespread and intensive visual stimulation. In this Review, we summarize the evidence in favor of these traveling waves. We suggest that their substrate may lie in long-range horizontal connections and that their functional role may involve the integration of information over large regions of space.

Adult visual cortical plasticity

The visual cortex has the capacity for experience-dependent change, or cortical plasticity, that is retained throughout life. Plasticity is invoked for encoding information during perceptual learning, by internally representing the regularities of the visual environment, which is useful for facilitating intermediate-level vision--contour integration and surface segmentation. The same mechanisms have adaptive value for functional recovery after CNS damage, such as that associated with stroke or neurodegenerative disease. A common feature to plasticity in primary visual cortex (V1) is an association field that links contour elements across the visual field. The circuitry underlying the association field includes a plexus of long-range horizontal connections formed by cortical pyramidal cells. These connections undergo rapid and exuberant sprouting and pruning in response to removal of sensory input, which can account for the topographic reorganization following retinal lesions. Similar alterations in cortical circuitry may be involved in perceptual learning, and the changes observed in V1 may be representative of how learned information is encoded throughout the cerebral cortex.

Mechanisms of neuronal computation in mammalian visual cortex

Orientation selectivity in the primary visual cortex (V1) is a receptive field property that is at once simple enough to make it amenable to experimental and theoretical approaches and yet complex enough to represent a significant transformation in the representation of the visual image. As a result, V1 has become an area of choice for studying cortical computation and its underlying mechanisms. Here we consider the receptive field properties of the simple cells in cat V1--the cells that receive direct input from thalamic relay cells--and explore how these properties, many of which are highly nonlinear, arise. We have found that many receptive field properties of V1 simple cells fall directly out of Hubel and Wiesel's feedforward model when the model incorporates realistic neuronal and synaptic mechanisms, including threshold, synaptic depression, response variability, and the membrane time constant.

Frequency specific spatial interactions in human electrocorticography: V1 alpha oscillations reflect surround suppression

Electrical brain signals are often decomposed into frequency ranges that are implicated in different functions. Using subdural electrocorticography (ECoG, intracranial EEG) and functional magnetic resonance imaging (fMRI), we measured frequency spectra and BOLD responses in primary visual cortex (V1) and intraparietal sulcus (IPS). In V1 and IPS, 30-120 Hz (gamma, broadband) oscillations allowed population receptive field (pRF) reconstruction comparable to fMRI estimates. Lower frequencies, however, responded very differently in V1 and IPS. In V1, broadband activity extends down to 3 Hz. In the 4-7 Hz (theta) and 18-30 Hz (beta) ranges broadband activity increases power during stimulation within the pRF. However, V1 9-12 Hz (alpha) frequency oscillations showed a different time course. The broadband power here is exceeded by a frequency-specific power increase during stimulation of the area outside the pRF. As such, V1 alpha oscillations reflected surround suppression of the pRF, much like negative fMRI responses. They were consequently highly localized, depending on stimulus and pRF position, and independent between nearby electrodes. In IPS, all 3-25 Hz oscillations were strongest during baseline recording and correlated between nearby electrodes, consistent with large-scale disengagement. These findings demonstrate V1 alpha oscillations result from locally active functional processes and relate these alpha oscillations to negative fMRI signals. They highlight that similar oscillations in different areas reflect processes with different functional roles. However, both of these roles of alpha seem to reflect suppression of spiking activity.

Fast recruitment of recurrent inhibition in the cat visual cortex

Neurons of the same column in L4 of the cat visual cortex are likely to share the same sensory input from the same region of the visual field. Using visually-guided patch clamp recordings we investigated the biophysical properties of the synapses of neighboring layer 4 neurons. We recorded synaptic connections between all types of excitatory and inhibitory neurons in L4. The E–E, E–I, and I–E connections had moderate CVs and failure rates. However, E–I connections had larger amplitudes, faster rise-times, and shorter latencies. Identification of the sites of putative synaptic contacts together with compartmental simulations on 3D reconstructed cells, suggested that E–I synapses tended to be located on proximal dendritic branches, which would explain their larger EPSP amplitudes and faster kinetics. Excitatory and inhibitory synapses were located at the same distance on distal dendrites of excitatory neurons. We hypothesize that this co-localization and the fast recruitment of local inhibition provides an efficient means of modulating excitation in a precisely timed way.

Relationship between orientation sensitivity and spatiotemporal receptive field structures of neurons in the cat lateral geniculate nucleus

Although it is thought that orientation selectivity first emerges in the primary visual cortex, several studies have reported that neurons in the cat lateral geniculate nucleus (LGN) are sensitive to stimulus orientation, especially for high spatial frequency (SF) stimuli. To understand how this orientation sensitivity emerges, we investigated the spatiotemporal structures of linear receptive fields (RFs) of LGN neurons. Orientation tunings at several SFs were measured using sinusoidal drifting grating stimuli. Fine spatiotemporal structures of the linear RFs were measured using a reverse correlation technique and two-dimensional dynamic Gaussian white noise stimuli. A non-linear response modulation function was estimated by comparing measured responses with responses predicted from a linear RF structure. Although we found that a population of LGN neurons exhibited significantly elongated linear RF centers and that the angles of the long axes corresponded well to the preferred orientations, the orientation tunings predicted from the linear RFs were significantly broader than those measured. These results suggest that orientation-tuned non-linear response modulation induced by stimulation outside the classical RF contributes to the sharp orientation tuning seen in LGN neurons.

Surround suppression and sparse coding in visual and barrel cortices

During natural vision the entire retina is stimulated. Likewise, during natural tactile behaviors, spatially extensive regions of the somatosensory surface are co-activated. The large spatial extent of naturalistic stimulation means that surround suppression, a phenomenon whose neural mechanisms remain a matter of debate, must arise during natural behavior. To identify common neural motifs that might instantiate surround suppression across modalities, we review models of surround suppression and compare the evidence supporting the competing ideas that surround suppression has either cortical or sub-cortical origins in visual and barrel cortex. In the visual system there is general agreement lateral inhibitory mechanisms contribute to surround suppression, but little direct experimental evidence that intracortical inhibition plays a major role. Two intracellular recording studies of V1, one using naturalistic stimuli (Haider et al., 2010), the other sinusoidal gratings (Ozeki et al., 2009), sought to identify the causes of reduced activity in V1 with increasing stimulus size, a hallmark of surround suppression. The former attributed this effect to increased inhibition, the latter to largely balanced withdrawal of excitation and inhibition. In rodent primary somatosensory barrel cortex, multi-whisker responses are generally weaker than single whisker responses, suggesting multi-whisker stimulation engages similar surround suppressive mechanisms. The origins of suppression in S1 remain elusive: studies have implicated brainstem lateral/internuclear interactions and both thalamic and cortical inhibition. Although the anatomical organization and instantiation of surround suppression in the visual and somatosensory systems differ, we consider the idea that one common function of surround suppression, in both modalities, is to remove the statistical redundancies associated with natural stimuli by increasing the sparseness or selectivity of sensory responses.

Location-dependent effects of inhibition on local spiking in pyramidal neuron dendrites

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.

Cortical Surround Interactions and Perceptual Salience via Natural Scene Statistics

Spatial context in images induces perceptual phenomena associated with salience and modulates the responses of neurons in primary visual cortex (V1). However, the computational and ecological principles underlying contextual effects are incompletely understood. We introduce a model of natural images that includes grouping and segmentation of neighboring features based on their joint statistics, and we interpret the firing rates of V1 neurons as performing optimal recognition in this model. We show that this leads to a substantial generalization of divisive normalization, a computation that is ubiquitous in many neural areas and systems. A main novelty in our model is that the influence of the context on a target stimulus is determined by their degree of statistical dependence. We optimized the parameters of the model on natural image patches, and then simulated neural and perceptual responses on stimuli used in classical experiments. The model reproduces some rich and complex response patterns observed in V1, such as the contrast dependence, orientation tuning and spatial asymmetry of surround suppression, while also allowing for surround facilitation under conditions of weak stimulation. It also mimics the perceptual salience produced by simple displays, and leads to readily testable predictions. Our results provide a principled account of orientation-based contextual modulation in early vision and its sensitivity to the homogeneity and spatial arrangement of inputs, and lends statistical support to the theory that V1 computes visual salience.

Strong recurrent networks compute the orientation tuning of surround modulation in the primate primary visual cortex

In macaque primary visual cortex (V1), neuronal responses to stimuli inside the receptive field (RF) are modulated by stimuli in the RF surround. This modulation is orientation specific. Previous studies suggested that, for some cells, this specificity may not be fixed but changes with the stimulus orientation presented to the RF. We demonstrate, in recording studies, that this tuning behavior is instead highly prevalent in V1 and, in theoretical work, that it arises only if V1 operates in a regime of strong local recurrence. Strongest surround suppression occurs when the stimuli in the RF and the surround are iso-oriented, and strongest facilitation when the stimuli are cross-oriented. This is the case even when the RF is suboptimally activated by a stimulus of nonpreferred orientation but only if this stimulus can activate the cell when presented alone. This tuning behavior emerges from the interaction of lateral inhibition (via the surround pathways), which is tuned to the preferred orientation of the RF, with weakly tuned, but strong, local recurrent connections, causing maximal withdrawal of recurrent excitation at the feedforward input orientation. Thus, horizontal and feedback modulation of strong recurrent circuits allows the tuning of contextual effects to change with changing feedforward inputs.

Increased surround modulation of perceived contrast in the elderly

PURPOSE

The appearance of a stimulus depends on its background, with high-contrast backgrounds resulting in lower perceived contrast. Increased perceptual surround suppression effects have been reported in the elderly. Our experiments tested whether enhanced surround suppression in the elderly arises because of age-dependent differences in brightness induction mechanisms that are sensitive to phase information at the border of the central stimulus.

METHODS

Fifteen younger (18 to 33 years) and 18 older (61 to 84 years) adults participated. Using a method of constant stimuli, perceived contrast was measured for a sine wave grating with and without an annular surround. Annuli were either in-phase with the central grating (suppresses the perceived central contrast) or out-of-phase (typically enhances perceived central contrast). The experiment was repeated using stimuli where the contrast was reduced for younger observers to approximately match the effective contrast available to older adults.

RESULTS

With the surround present, the older group matched the contrast of the central target to an average lower contrast than younger adults [F(1,31) = 17.4, p < 0.001]. The magnitude of contrast suppression differences between older and younger observers was invariant of annulus grating phase [F(1,31) = 0.036, p = 0.85] and was of similar magnitude when the stimuli were approximately matched between groups for differences in contrast detection [F(1,31) = 0.06, p = 0.81].

CONCLUSIONS

Normal ageing increases perceived contrast surround suppression, irrespective of information at the stimulus border between center and surround. Conditions that result in perceived contrast enhancement on average in younger adults result in contrast suppression in the elderly. Our findings suggest that age-related differences are likely in the appearance of objects in natural environments where background contrast varies.

Normalization as a canonical neural computation

There is increasing evidence that the brain relies on a set of canonical neural computations, repeating them across brain regions and modalities to apply similar operations to different problems. A promising candidate for such a computation is normalization, in which the responses of neurons are divided by a common factor that typically includes the summed activity of a pool of neurons. Normalization was developed to explain responses in the primary visual cortex and is now thought to operate throughout the visual system, and in many other sensory modalities and brain regions. Normalization may underlie operations such as the representation of odours, the modulatory effects of visual attention, the encoding of value and the integration of multisensory information. Its presence in such a diversity of neural systems in multiple species, from invertebrates to mammals, suggests that it serves as a canonical neural computation.

Orientation selectivity of synaptic input to neurons in mouse and cat primary visual cortex

Primary visual cortex (V1) is the site at which orientation selectivity emerges in mammals: visual thalamus afferents to V1 respond equally to all stimulus orientations, whereas their target V1 neurons respond selectively to stimulus orientation. The emergence of orientation selectivity in V1 has long served as a model for investigating cortical computation. Recent evidence for orientation selectivity in mouse V1 opens cortical computation to dissection by genetic and imaging tools, but also raises two essential questions: (1) How does orientation selectivity in mouse V1 neurons compare with that in previously described species? (2) What is the synaptic basis for orientation selectivity in mouse V1? A comparison of orientation selectivity in mouse and in cat, where such measures have traditionally been made, reveals that orientation selectivity in mouse V1 is weaker than in cat V1, but that spike threshold plays a similar role in narrowing selectivity between membrane potential and spike rate. To uncover the synaptic basis for orientation selectivity, we made whole-cell recordings in vivo from mouse V1 neurons, comparing neuronal input selectivity-based on membrane potential, synaptic excitation, and synaptic inhibition-to output selectivity based on spiking. We found that a neuron's excitatory and inhibitory inputs are selective for the same stimulus orientations as is its membrane potential response, and that inhibitory selectivity is not broader than excitatory selectivity. Inhibition has different dynamics than excitation, adapting more rapidly. In neurons with temporally modulated responses, the timing of excitation and inhibition was different in mice and cats.

Local model for contextual modulation in the cerebral cortex

A neural response to a sensory stimulus in cerebral cortex is modulated when other stimuli are presented simultaneously. The other stimuli can modulate responses even when they do not drive the neural output alone, indicating a non-linear summation of synaptic activity. The mechanisms of the nonlinearity have remained unclear. Here, I explore a model which considers both network and intracellular processes, and which can account for various types of contextual modulation. The processes include synaptic sensitivity function, determination of inhibition strength, dendritic decay of membrane voltage, and summation of excitatory and inhibitory membrane voltages. First, the model assumes that excitatory and inhibitory units have the same input sensitivity function, which is more broadly tuned than the output tuning function. Second, a central property of the model is that inhibition is a fraction of excitation, determined by covariance between the input and the sensitivity function. With proper fraction, a model neuron sums apparently decorrelated input, regardless of correlations in the original input. Third, the model assumes that synaptic input lands anisotropically on the dendrites, which together with passive dendritic decay cause exponential decay in summation along the input space. This explains the difference between input sensitivity function and output tuning function, and thus accounts for the division between driving classical and modulating extra-classical receptive fields. The model simulations replicate single-cell area summation function, far surround facilitation, and a shift in tuning function due to contextual stimulation. The model is very general, and should be applicable to various interactions between cortical representations.

The mechanisms for compression and reflection of cortical waves

Waves are common in cortical networks and may be important for carrying information about a stimulus from one local circuit to another. In a recent study of visually evoked waves in rat cortex, compression and reflection of waves are observed as the activation passes from visual areas V1 to V2. The authors of this study apply bicuculline (BMI) and demonstrate that the reflection disappears. They conclude that inhibition plays a major role in compression and reflection. We present several models for propagating waves in heterogeneous media and show that the velocity and thus compression depends weakly on inhibition. We propose that the main site of action of BMI with respect to wave propagation is on the threshold for firing which we suggest is related to action on potassium channels. We combine numerical and analytic methods to explore both compression and reflection in an excitable system with synaptic coupling.

Phase sensitivities, excitatory summation fields, and silent suppressive receptive fields of single neurons in the parastriate cortex of the cat

We have recorded single-neuron activity from cytoarchitectonic area 18 of anesthetized (0.4–0.7% isoflurane in 65% N2O-35% O2 gaseous mixture) domestic cats. Neurons were identified as simple or complex on the basis of the ratios between the phase-variant (F1) component and the mean firing rate (F0) of spike responses to optimized (orientation, direction, spatial and temporal frequencies, size) high-contrast, luminance-modulated, sine-wave drifting gratings (simple: F1/F0 spike-response ratios > 1; complex: F1/F0 spike-response ratios < 1). The predominance (∼80%) of simple cells among the neurons recorded from the principal thalamorecipient layers supports the idea that most simple cells in area 18 might constitute a putative early stage in the visual information processing. Apart from the “spike-generating” regions (the classical receptive fields, CRFs), the receptive fields of three-quarters of area 18 neurons contain silent, extraclassical suppressive regions (ECRFs). The spatial extent of summation areas of excitatory responses was negatively correlated with the strength of the ECRF-induced suppression of spike responses. Lowering the stimulus contrast resulted in an expansion of the summation areas of excitatory responses accompanied by a reduction in the strength of the ECRF-induced suppression. The spatial and temporal frequency and orientation tunings of the ECRFs were much broader than those of the CRFs. Hence, the ECRFs of area 18 neurons appear to be largely “inherited” from their dorsal thalamic inputs. In most area 18 cells, costimulation of CRFs and ECRFs resulted in significant increases in F1/F0 spike-response ratios, and thus there was a contextually modulated functional continuum between the simple and complex cells.

A model for the origin and properties of flicker-induced geometric phosphenes

We present a model for flicker phosphenes, the spontaneous appearance of geometric patterns in the visual field when a subject is exposed to diffuse flickering light. We suggest that the phenomenon results from interaction of cortical lateral inhibition with resonant periodic stimuli. We find that the best temporal frequency for eliciting phosphenes is a multiple of intrinsic (damped) oscillatory rhythms in the cortex. We show how both the quantitative and qualitative aspects of the patterns change with frequency of stimulation and provide an explanation for these differences. We use Floquet theory combined with the theory of pattern formation to derive the parameter regimes where the phosphenes occur. We use symmetric bifurcation theory to show why low frequency flicker should produce hexagonal patterns while high frequency produces pinwheels, targets, and spirals.

Renewal theory of coupled neuronal pools: stable states and slow trajectories

A theory is provided to analyze the dynamics of delay-coupled pools of spiking neurons based on stability analysis of stationary firing. Transitions between stable and unstable regimes can be predicted by bifurcation analysis of the underlying integral dynamics. Close to the bifurcation point the network exhibits slowly changing activities and allows for slow collective phenomena like continuous attractors.

The many forms and functions of long term plasticity at GABAergic synapses

On February 12th 1973, Bliss and Lomo submitted their findings on activity-dependent plasticity of glutamatergic synapses. After this groundbreaking discovery, long-term potentiation (LTP) and depression (LTD) gained center stage in the study of learning, memory, and experience-dependent refinement of neural circuits. While LTP and LTD are extensively studied and their relevance to brain function is widely accepted, new experimental and theoretical work recently demonstrates that brain development and function relies on additional forms of plasticity, some of which occur at nonglutamatergic synapses. The strength of GABAergic synapses is modulated by activity, and new functions for inhibitory synaptic plasticity are emerging. Together with excitatory neurons, inhibitory neurons shape the excitability and dynamic range of neural circuits. Thus, the understanding of inhibitory synaptic plasticity is crucial to fully comprehend the physiology of brain circuits. Here, I will review recent findings about plasticity at GABAergic synapses and discuss how it may contribute to circuit function.

Differential connectivity and response dynamics of excitatory and inhibitory neurons in visual cortex

Neuronal responses during sensory processing are influenced by both the organization of intracortical connections and the statistical features of sensory stimuli. How these intrinsic and extrinsic factors govern the activity of excitatory and inhibitory populations is unclear. Using two-photon calcium imaging in vivo and intracellular recordings in vitro, we investigated the dependencies between synaptic connectivity, feature selectivity and network activity in pyramidal cells and fast-spiking parvalbumin-expressing (PV) interneurons in mouse visual cortex. In pyramidal cell populations, patterns of neuronal correlations were largely stimulus-dependent, indicating that their responses were not strongly dominated by functionally biased recurrent connectivity. By contrast, visual stimulation only weakly modified co-activation patterns of fast-spiking PV cells, consistent with the observation that these broadly tuned interneurons received very dense and strong synaptic input from nearby pyramidal cells with diverse feature selectivities. Therefore, feedforward and recurrent network influences determine the activity of excitatory and inhibitory ensembles in fundamentally different ways.

Biologically inspired EM image alignment and neural reconstruction

MOTIVATION

Three-dimensional reconstruction of consecutive serial-section transmission electron microscopy (ssTEM) images of neural tissue currently requires many hours of manual tracing and annotation. Several computational techniques have already been applied to ssTEM images to facilitate 3D reconstruction and ease this burden.

RESULTS

Here, we present an alternative computational approach for ssTEM image analysis. We have used biologically inspired receptive fields as a basis for a ridge detection algorithm to identify cell membranes, synaptic contacts and mitochondria. Detected line segments are used to improve alignment between consecutive images and we have joined small segments of membrane into cell surfaces using a dynamic programming algorithm similar to the Needleman-Wunsch and Smith-Waterman DNA sequence alignment procedures. A shortest path-based approach has been used to close edges and achieve image segmentation. Partial reconstructions were automatically generated and used as a basis for semi-automatic reconstruction of neural tissue. The accuracy of partial reconstructions was evaluated and 96% of membrane could be identified at the cost of 13% false positive detections.

AVAILABILITY

An open-source reference implementation is available in the Supplementary information.

CONTACT

seymour.kb@ed.ac.uk; douglas.armstrong@ed.ac.uk

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

GABAA inhibition controls response gain in visual cortex

GABA(A) inhibition is thought to play multiple roles in sensory cortex, such as controlling responsiveness and sensitivity, sharpening selectivity, and mediating competitive interactions. To test these proposals, we recorded in cat primary visual cortex (V1) after local iontophoresis of gabazine, the selective GABA(A) antagonist. Gabazine increased responsiveness by as much as 300%. It slightly decreased selectivity for stimulus orientation and direction, often by raising responses to all orientations. Strikingly, gabazine affected neither contrast sensitivity nor cross-orientation suppression, the competition seen when stimuli of different orientation are superimposed. These results were captured by a simple model in which GABA(A) inhibition has the same selectivity as excitation and keeps responses to unwanted stimuli below threshold. We conclude that GABA(A) inhibition in V1 helps enhance stimulus selectivity but is not responsible for competition among superimposed stimuli. It controls the sensitivity of V1 neurons by adjusting their response gain, without affecting their input gain.

Cholinergic control of cortical network interactions enables feedback-mediated attentional modulation

Attention increases our ability to detect behaviorally relevant stimuli. At the neuronal level this is supported by increased firing rates of neurons representing the attended object. In primary visual cortex an attention-mediated activity increase depends on the presence of the neuromodulator acetylcholine. Using a spiking network model of visual cortex we have investigated how acetylcholine interacts with biased feedback to enable attentional processing. Although acetylcholine affects cortical processing in a multitude of manners, we restricted our analysis to four of its main established actions. These were (i) a reduction in firing rate adaptation by reduction in M-currents (muscarinic), (ii) an increase in thalamocortical synaptic efficacy by nicotinic presynaptic receptors, (iii) a reduction in lateral interactions by muscarinic presynaptic receptors, and (iv) an increase in inhibitory drive by muscarinic receptors located on inhibitory interneurons. We found that acetylcholine contributes to feedback-mediated attentional modulation, mostly by reducing intracortical interactions and also to some extent by increasing the inhibitory drive. These findings help explain why acetylcholine is necessary for top-down-driven attentional modulation, and suggest a close interdependence of cholinergic and feedback drive in mediating cognitive function.

Somatosensory response properties of excitatory and inhibitory neurons in rat motor cortex

In sensory cortical networks, peripheral inputs differentially activate excitatory and inhibitory neurons. Inhibitory neurons typically have larger responses and broader receptive field tuning compared with excitatory neurons. These differences are thought to underlie the powerful feedforward inhibition that occurs in response to sensory input. In the motor cortex, as in the somatosensory cortex, cutaneous and proprioceptive somatosensory inputs, generated before and during movement, strongly and dynamically modulate the activity of motor neurons involved in a movement and ultimately shape cortical command. Human studies suggest that somatosensory inputs modulate motor cortical activity in a center excitation, surround inhibition manner such that input from the activated muscle excites motor cortical neurons that project to it, whereas somatosensory input from nearby, nonactivated muscles inhibit these neurons. A key prediction of this hypothesis is that inhibitory and excitatory motor cortical neurons respond differently to somatosensory inputs. We tested this prediction with the use of multisite extracellular recordings in anesthetized rats. We found that fast-spiking (presumably inhibitory) neurons respond to tactile and proprioceptive inputs at shorter latencies and larger response magnitudes compared with regular-spiking (presumably excitatory) neurons. In contrast, we found no differences in the receptive field size of these neuronal populations. Strikingly, all fast-spiking neuron pairs analyzed with cross-correlation analysis displayed common excitation, which was significantly more prevalent than common excitation for regular-spiking neuron pairs. These findings suggest that somatosensory inputs preferentially evoke feedforward inhibition in the motor cortex. We suggest that this provides a mechanism for dynamic selection of motor cortical modules during voluntary movements.

Membrane voltage fluctuations reduce spike frequency adaptation and preserve output gain in CA1 pyramidal neurons in a high-conductance state

Modulating the gain of the input-output function of neurons is critical for processing of stimuli and network dynamics. Previous gain control mechanisms have suggested that voltage fluctuations play a key role in determining neuronal gain in vivo. Here we show that, under increased membrane conductance, voltage fluctuations restore Na(+) current and reduce spike frequency adaptation in rat hippocampal CA1 pyramidal neurons in vitro. As a consequence, membrane voltage fluctuations produce a leftward shift in the frequency-current relationship without a change in gain, relative to an increase in conductance alone. Furthermore, we show that these changes have important implications for the integration of inhibitory inputs. Due to the ability to restore Na(+) current, hyperpolarizing membrane voltage fluctuations mediated by GABA(A)-like inputs can increase firing rate in a high-conductance state. Finally, our data show that the effects on gain and synaptic integration are mediated by voltage fluctuations within a physiologically relevant range of frequencies (10-40 Hz).

Neurometabolic coupling differs for suppression within and beyond the classical receptive field in visual cortex

Neurons in visual cortex exhibit two major types of stimulus elicited suppression. One, cross-orientation suppression, occurs within the classical receptive field (CRF) when an orthogonal grating is superposed on one at optimal orientation. The second, surround suppression, occurs when the size of an optimally oriented grating extends beyond the CRF. Previous proposals suggest that intracortical inhibition is responsible for surround suppression whereas feedforward processes may underlie cross-orientation suppression. To gain more insight concerning these types of suppression, we have included measurements of metabolic function in addition to neural responses. We made co-localized measurements of multiple unit neural activity and tissue oxygen concentrations in the striate cortex of anaesthetized cats while using visual stimuli to activate the two kinds of suppression. Results show that the amplitude of the initial negative oxygen response increases with stimulus size but neural responses decrease as size extends beyond the CRF. This shows that oxygen consumption increases with stimulus size regardless of reduced neural response. On the other hand, amplitudes of both the initial negative oxygen component and the neural responses are simultaneously attenuated by the orthogonal mask in cross-orientation suppression. These different neurometabolic response patterns are consistent with suggestions that the two types of suppressive processes arise from different neural mechanisms.

Surround suppression by high spatial frequency stimuli in the cat primary visual cortex

Surround suppression is a phenomenon whereby stimulation of the extraclassical receptive field suppressively modulates the visual responses of neurons in the primary visual cortex (V1) (also known as area 17). It is known that surround suppression tunes to spatial frequencies (SFs) that are much lower and broader than the frequencies to which the classical receptive field tunes. In this study, we tested the effects of varying SFs on surround suppression by using a circular sinusoidal grating patch that covered both the classical receptive field and the extraclassical receptive field. Using area-summation tuning curves, we found high-SF-tuned surround suppression in the cat V1. This high-SF-tuned surround suppression causes the SF tuning to shift to low SF for large stimuli. By simulating a model neuron lacking a suppressive surround mechanism, we confirmed that these preferred SF shifts do not occur in the absence of surround suppression. We surmise that the high-SF-tuned suppression, which shifts the preferred SF according to size, functionally contributes to the scale-invariant processing of visual images in V1.

Modulation of firing rate by background synaptic noise statistics in rat visual cortical neurons

It has been shown previously that background synaptic noise modulates the response gain of neocortical neurons. However, the role of the statistical properties of the noise in modulating firing rate is not known. Here, the dependence of firing rate on the statistical properties of the excitatory to inhibitory balance (EI) in cortical pyramidal neurons was studied. Excitatory glutamatergic and inhibitory GABAergic synaptic conductances were simulated as two stochastic processes and injected into individual neurons in vitro through use of the dynamic-clamp system. Response gain was significantly modulated as a function of the statistical interactions between excitatory and inhibitory synaptic conductances. Firing rates were compared for noisy synaptic conductance steps by varying either the EI correlation or the relative delay between correlated E and I. When inhibitory synaptic conductances exhibited a short temporal delay (5 ms) relative to correlated excitatory synaptic conductances, the response gain was increased compared with noise with no temporal delay but with an equivalent degree of correlation. The dependence of neuronal firing rate on the EI delay of the noisy background synaptic conductance suggests that individual excitatory pyramidal neurons are sensitive to the EI balance of the synaptic conductance. Therefore the statistical EI interactions encoded within the synaptic subthreshold membrane fluctuations are able to modulate neuronal firing properties.

Attention reduces stimulus-driven gamma frequency oscillations and spike field coherence in V1

Rhythmic activity of neuronal ensembles has been proposed to play an important role in cognitive functions such as attention, perception, and memory. Here we investigate whether rhythmic activity in V1 of the macaque monkey (macaca mulatta) is affected by top-down visual attention. We measured the local field potential (LFP) and V1 spiking activity while monkeys performed an attention-demanding detection task. We show that gamma oscillations were strongly modulated by the stimulus and by attention. Stimuli that engaged inhibitory mechanisms induced the largest gamma LFP oscillations and the largest spike field coherence. Directing attention toward a visual stimulus at the receptive field of the recorded neurons decreased LFP gamma power and gamma spike field coherence. This decrease could reflect an attention-mediated reduction of surround inhibition. Changes in synchrony in V1 would thus be a byproduct of reduced inhibitory drive, rather than a mechanism that directly aids perceptual processing.

Local non-linear interactions in the visual cortex may reflect global decorrelation

The classical receptive field in the primary visual cortex have been successfully explained by sparse activation of relatively independent units, whose tuning properties reflect the statistical dependencies in the natural environment. Robust surround modulation, emerging from stimulation beyond the classical receptive field, has been associated with increase of lifetime sparseness in the V1, but the system-wide modulation of response strength have currently no theoretical explanation. We measured fMRI responses from human visual cortex and quantified the contextual modulation with a decorrelation coefficient (d), derived from a subtractive normalization model. All active cortical areas demonstrated local non-linear summation of responses, which were in line with hypothesis of global decorrelation of voxels responses. In addition, we found sensitivity to surrounding stimulus structure across the ventral stream, and large-scale sensitivity to the number of simultaneous objects. Response sparseness across voxel population increased consistently with larger stimuli. These data suggest that contextual modulation for a stimulus event reflect optimization of the code and perhaps increase in energy efficiency throughout the ventral stream hierarchy. Our model provides a novel prediction that average suppression of response amplitude for simultaneous stimuli across the cortical network is a monotonic function of similarity of response strengths in the network when the stimuli are presented alone.

GABA concentration is reduced in visual cortex in schizophrenia and correlates with orientation-specific surround suppression

The neural mechanisms underlying cognitive deficits in schizophrenia remain essentially unknown. The GABA hypothesis proposes that reduced neuronal GABA concentration and neurotransmission results in cognitive impairments in schizophrenia. However, few in vivo studies have directly examined this hypothesis. We used magnetic resonance spectroscopy (MRS) at high field to measure visual cortical GABA levels in 13 subjects with schizophrenia and 13 demographically matched healthy control subjects. We found that the schizophrenia group had an approximately 10% reduction in GABA concentration. We further tested the GABA hypothesis by examining the relationship between visual cortical GABA levels and orientation-specific surround suppression (OSSS), a behavioral measure of visual inhibition thought to be dependent on GABAergic synaptic transmission. Previous work has shown that subjects with schizophrenia exhibit reduced OSSS of contrast discrimination (Yoon et al., 2009). For subjects with both MRS and OSSS data (n = 16), we found a highly significant positive correlation (r = 0.76) between these variables. GABA concentration was not correlated with overall contrast discrimination performance for stimuli without a surround (r = -0.10). These results suggest that a neocortical GABA deficit in subjects with schizophrenia leads to impaired cortical inhibition and that GABAergic synaptic transmission in visual cortex plays a critical role in OSSS.

Synaptic and network mechanisms of sparse and reliable visual cortical activity during nonclassical receptive field stimulation

During natural vision, the entire visual field is stimulated by images rich in spatiotemporal structure. Although many visual system studies restrict stimuli to the classical receptive field (CRF), it is known that costimulation of the CRF and the surrounding nonclassical receptive field (nCRF) increases neuronal response sparseness. The cellular and network mechanisms underlying increased response sparseness remain largely unexplored. Here we show that combined CRF + nCRF stimulation increases the sparseness, reliability, and precision of spiking and membrane potential responses in classical regular spiking (RS(C)) pyramidal neurons of cat primary visual cortex. Conversely, fast-spiking interneurons exhibit increased activity and decreased selectivity during CRF + nCRF stimulation. The increased sparseness and reliability of RS(C) neuron spiking is associated with increased inhibitory barrages and narrower visually evoked synaptic potentials. Our experimental observations were replicated with a simple computational model, suggesting that network interactions among neuronal subtypes ultimately sharpen recurrent excitation, producing specific and reliable visual responses.

Function of inhibition in visual cortical processing

Although sensory processing in V1 has been extensively characterized, the role of GABAergic inhibition is still not well understood. Advances in molecular biology have now removed significant barriers to the direct investigation of inhibitory processes in vivo. Recent studies have provided important insights into the influence of GABAergic inhibition on cortical processing at both the single cell level, where inhibition helps to shape cortical receptive fields, and at the network level, where inhibition is critical for generating cortical oscillations and setting network state.

Contrast-dependence of surround suppression in Macaque V1: experimental testing of a recurrent network model

Neuronal responses in primary visual cortex (V1) to optimally oriented high-contrast stimuli in the receptive field (RF) center are suppressed by stimuli in the RF surround, but can be facilitated when the RF center is stimulated at low contrast. The neural circuits and mechanisms for surround modulation are still unknown. We previously proposed that topdown feedback connections mediate suppression from the "far" surround, while "near' surround suppression is mediated primarily by horizontal connections. We implemented this idea in a recurrent network model of V1. A model assumption needed to account for the contrast-dependent sign of surround modulation is a response asymmetry between excitation and inhibition; accordingly, inhibition, but not excitation, is silent for weak visual inputs to the RF center, and surround stimulation can evoke facilitation. A prediction stemming from this same assumption is that surround suppression is weaker for low than for high contrast stimuli in the RF center. Previous studies are inconsistent with this prediction. Using single unit recordings in macaque V1, we confirm this model's prediction. Model simulations demonstrate that our results can be reconciled with those from previous studies. We also performed a systematic comparison of the experimentally measured surround suppression strength with predictions of the model operated in different parameter regimes. We find that the original model, with strong horizontal and no feedback excitation of local inhibitory neurons, can only partially account quantitatively for the experimentally measured suppression. Strong direct feedback excitation of V1 inhibitory neurons is necessary to account for the experimentally measured surround suppression strength.

Balanced amplification: a new mechanism of selective amplification of neural activity patterns

In cerebral cortex, ongoing activity absent a stimulus can resemble stimulus-driven activity in size and structure. In particular, spontaneous activity in cat primary visual cortex (V1) has structure significantly correlated with evoked responses to oriented stimuli. This suggests that, from unstructured input, cortical circuits selectively amplify specific activity patterns. Current understanding of selective amplification involves elongation of a neural assembly's lifetime by mutual excitation among its neurons. We introduce a new mechanism for selective amplification without elongation of lifetime: "balanced amplification." Strong balanced amplification arises when feedback inhibition stabilizes strong recurrent excitation, a pattern likely to be typical of cortex. Thus, balanced amplification should ubiquitously contribute to cortical activity. Balanced amplification depends on the fact that individual neurons project only excitatory or only inhibitory synapses. This leads to a hidden feedforward connectivity between activity patterns. We show in a detailed biophysical model that this can explain the cat V1 observations.