Microcircuitry of forward and feedback connections within rat visual cortex

Feature selectivity of corticocortical feedback along the primate dorsal visual pathway

Anatomical studies have revealed a prominent role for feedback projections in the primate visual cortex. Theoretical models suggest that these projections support important brain functions such as attention, prediction, and learning. However, these models make different predictions about the relationship between feedback connectivity and neuronal stimulus selectivity. We have therefore performed simultaneous recordings in different regions of the primate dorsal visual pathway. Specifically, we recorded neural activity from the medial superior temporal (MST) area, and one of its main feedback targets, the middle temporal (MT) area. We estimated functional connectivity from correlations in the single-neuron spike trains and performed electrical microstimulation in MST to determine its causal influence on MT. Both methods revealed that inhibitory feedback occurred more commonly when the source and target neurons had very different stimulus preferences. At the same time, the strength of feedback suppression was greater for neurons with similar preferences. Excitatory feedback projections, in contrast, showed no consistent relationship with stimulus preferences. These results suggest that corticocortical feedback could play a role in shaping sensory responses according to behavioral or environmental context.NEW & NOTEWORTHY Here, we show that corticocortical feedback influences are often determined by the selectivity of the individual neurons. A common motif is the occurrence of inhibitory feedback among neurons with very different stimulus preferences. This results in strong suppression of responses in area MT when MST is electrically stimulated. Interestingly, this feedback shows a complex interaction with ongoing visual stimulation, being powerfully suppressive when visual inputs are strong, yet excitatory when visual inputs are weak.

Neuroscientific insights about computer vision models: a concise review

The development of biologically-inspired computational models has been the focus of study ever since the artificial neuron was introduced by McCulloch and Pitts in 1943. However, a scrutiny of literature reveals that most attempts to replicate the highly efficient and complex biological visual system have been futile or have met with limited success. The recent state-of the-art computer vision models, such as pre-trained deep neural networks and vision transformers, may not be biologically inspired per se. Nevertheless, certain aspects of biological vision are still found embedded, knowingly or unknowingly, in the architecture and functioning of these models. This paper explores several principles related to visual neuroscience and the biological visual pathway that resonate, in some manner, in the architectural design and functioning of contemporary computer vision models. The findings of this survey can provide useful insights for building futuristic bio-inspired computer vision models. The survey is conducted from a historical perspective, tracing the biological connections of computer vision models starting with the basic artificial neuron to modern technologies such as deep convolutional neural network (CNN) and spiking neural networks (SNN). One spotlight of the survey is a discussion on biologically plausible neural networks and bio-inspired unsupervised learning mechanisms adapted for computer vision tasks in recent times.

Visual experience reduces the spatial redundancy between cortical feedback inputs and primary visual cortex neurons

The role of experience in the organization of cortical feedback (FB) remains unknown. We measured the effects of manipulating visual experience on the retinotopic specificity of supragranular and infragranular projections from the lateromedial (LM) visual area to layer (L)1 of the mouse primary visual cortex (V1). LM inputs were, on average, retinotopically matched with V1 neurons in normally and dark-reared mice, but visual exposure reduced the fraction of spatially overlapping inputs to V1. FB inputs from L5 conveyed more surround information to V1 than those from L2/3. The organization of LM inputs from L5 depended on their orientation preference and was disrupted by dark rearing. These observations were recapitulated by a model where visual experience minimizes receptive field overlap between LM inputs and V1 neurons. Our results provide a mechanism for the dependency of surround modulations on visual experience and suggest how expected interarea coactivation patterns are learned in cortical circuits.

A mean-field model of glutamate and GABA synaptic dynamics for functional MRS

Advances in functional magnetic resonance spectroscopy (fMRS) have enabled the quantification of activity-dependent changes in neurotransmitter concentrations in vivo. However, the physiological basis of the large changes in GABA and glutamate observed by fMRS (>10%) over short time scales of less than a minute remain unclear as such changes cannot be accounted for by known synthesis or degradation metabolic pathways. Instead, it has been hypothesized that fMRS detects shifts in neurotransmitter concentrations as they cycle from presynaptic vesicles, where they are largely invisible, to extracellular and cytosolic pools, where they are detectable. The present paper uses a computational modelling approach to demonstrate the viability of this hypothesis. A new mean-field model of the neural mechanisms generating the fMRS signal in a cortical voxel is derived. The proposed macroscopic mean-field model is based on a microscopic description of the neurotransmitter dynamics at the level of the synapse. Specifically, GABA and glutamate are assumed to cycle between three metabolic pools: packaged in the vesicles; active in the synaptic cleft; and undergoing recycling and repackaging in the astrocytic or neuronal cytosol. Computational simulations from the model are used to generate predicted changes in GABA and glutamate concentrations in response to different types of stimuli including pain, vision, and electric current stimulation. The predicted changes in the extracellular and cytosolic pools corresponded to those reported in empirical fMRS data. Furthermore, the model predicts a selective control mechanism of the GABA/glutamate relationship, whereby inhibitory stimulation reduces both neurotransmitters, whereas excitatory stimulation increases glutamate and decreases GABA. The proposed model bridges between neural dynamics and fMRS and provides a mechanistic account for the activity-dependent changes in the glutamate and GABA fMRS signals. Lastly, these results indicate that echo-time may be an important timing parameter that can be leveraged to maximise fMRS experimental outcomes.

From hallucinations to synaesthesia: A circular inference account of unimodal and multimodal erroneous percepts in clinical and drug-induced psychosis

Psychedelics distort perception and induce visual and multimodal hallucinations as well as synaesthesia. This is in contradiction with the high prevalence of distressing voices in schizophrenia. Here we introduce a unifying account of unimodal and multimodal erroneous percepts based on circular inference. We show that amplification of top-down predictions (descending loops) leads to an excessive reliance on priors and aberrant levels of integration of the sensory representations, resulting in crossmodal percepts and stronger illusions. By contrast, amplification of bottom-up information (ascending loops) results in overinterpretation of unreliable sensory inputs and high levels of segregation between sensory modalities, bringing about unimodal hallucinations and reduced vulnerability to illusions. We delineate a canonical microcircuit in which layer-specific inhibition controls the propagation of information across hierarchical levels: inhibitory interneurons in the deep layers exert control over priors, removing descending loops. Conversely, inhibition in the supragranular layers counterbalances the effects of the ascending loops. Overall, we put forward a multiscale and transnosographic account of erroneous percepts with important theoretical, conceptual and clinical implications.

The role of feedback projections in feature tuning and neuronal excitability in the early primate visual system

A general assumption in visual neuroscience is that basic receptive field properties such as orientation and direction selectivity are constructed within intrinsic neuronal circuits and feedforward projections. In addition, it is assumed that general neuronal excitability and responsiveness in early visual areas is to a great extent independent of feedback input originating in areas higher in the stream. Here, we review the contribution of feedback projections from MT, V4 and pulvinar to the receptive field properties of V2 neurons in the anesthetized and paralyzed monkey. Importantly, our results contradict both of these assumptions. We separately inactivated each of these three brain regions using GABA pressure injections, while simultaneously recording V2 single unit activity before and hours after inactivation. Recordings and GABA injections were carried out in topographically corresponding regions of the visual field. We outline the changes in V2 activity, responsiveness and receptive field properties for early, mid and late post-injection phases. Immediately after injection, V2 activity is globally suppressed. Subsequently, there is an increase in stimulus-driven relative to spontaneous neuronal activity, which improves the signal-to-noise coding for the oriented moving bars. Notably, V2 tuning properties change substantially relative to its pre-injection selectivity profile. The resulting increase or decrease in selectivity could not be readily predicted based on the selectivity profile of the inactivated site. Finally, V2 activity rebounds before returning to it pre-injection profile Our results show that feedback projections profoundly impact neuronal circuits in early visual areas, and may have been heretofore largely underestimated in their physiological role.

Cortical and Subcortical Circuits for Cross-Modal Plasticity Induced by Loss of Vision

Cortical areas are highly interconnected both via cortical and subcortical pathways, and primary sensory cortices are not isolated from this general structure. In primary sensory cortical areas, these pre-existing functional connections serve to provide contextual information for sensory processing and can mediate adaptation when a sensory modality is lost. Cross-modal plasticity in broad terms refers to widespread plasticity across the brain in response to losing a sensory modality, and largely involves two distinct changes: cross-modal recruitment and compensatory plasticity. The former involves recruitment of the deprived sensory area, which includes the deprived primary sensory cortex, for processing the remaining senses. Compensatory plasticity refers to plasticity in the remaining sensory areas, including the spared primary sensory cortices, to enhance the processing of its own sensory inputs. Here, we will summarize potential cellular plasticity mechanisms involved in cross-modal recruitment and compensatory plasticity, and review cortical and subcortical circuits to the primary sensory cortices which can mediate cross-modal plasticity upon loss of vision.

Untangling the cortico-thalamo-cortical loop: cellular pieces of a knotty circuit puzzle

Functions of the neocortex depend on its bidirectional communication with the thalamus, via cortico-thalamo-cortical (CTC) loops. Recent work dissecting the synaptic connectivity in these loops is generating a clearer picture of their cellular organization. Here, we review findings across sensory, motor and cognitive areas, focusing on patterns of cell type-specific synaptic connections between the major types of cortical and thalamic neurons. We outline simple and complex CTC loops, and note features of these loops that appear to be general versus specialized. CTC loops are tightly interlinked with local cortical and corticocortical (CC) circuits, forming extended chains of loops that are probably critical for communication across hierarchically organized cerebral networks. Such CTC-CC loop chains appear to constitute a modular unit of organization, serving as scaffolding for area-specific structural and functional modifications. Inhibitory neurons and circuits are embedded throughout CTC loops, shaping the flow of excitation. We consider recent findings in the context of established CTC and CC circuit models, and highlight current efforts to pinpoint cell type-specific mechanisms in CTC loops involved in consciousness and perception. As pieces of the connectivity puzzle fall increasingly into place, this knowledge can guide further efforts to understand structure-function relationships in CTC loops.

Top-Down Inference in the Auditory System: Potential Roles for Corticofugal Projections

It has become widely accepted that humans use contextual information to infer the meaning of ambiguous acoustic signals. In speech, for example, high-level semantic, syntactic, or lexical information shape our understanding of a phoneme buried in noise. Most current theories to explain this phenomenon rely on hierarchical predictive coding models involving a set of Bayesian priors emanating from high-level brain regions (e.g., prefrontal cortex) that are used to influence processing at lower-levels of the cortical sensory hierarchy (e.g., auditory cortex). As such, virtually all proposed models to explain top-down facilitation are focused on intracortical connections, and consequently, subcortical nuclei have scarcely been discussed in this context. However, subcortical auditory nuclei receive massive, heterogeneous, and cascading descending projections at every level of the sensory hierarchy, and activation of these systems has been shown to improve speech recognition. It is not yet clear whether or how top-down modulation to resolve ambiguous sounds calls upon these corticofugal projections. Here, we review the literature on top-down modulation in the auditory system, primarily focused on humans and cortical imaging/recording methods, and attempt to relate these findings to a growing animal literature, which has primarily been focused on corticofugal projections. We argue that corticofugal pathways contain the requisite circuitry to implement predictive coding mechanisms to facilitate perception of complex sounds and that top-down modulation at early (i.e., subcortical) stages of processing complement modulation at later (i.e., cortical) stages of processing. Finally, we suggest experimental approaches for future studies on this topic.

Activity in Lateral Visual Areas Contributes to Surround Suppression in Awake Mouse V1

Neuronal response to sensory stimuli depends on the context. The response in primary visual cortex (V1), for instance, is reduced when a stimulus is surrounded by a similar stimulus [1-3]. The source of this surround suppression is partially known. In mouse, local horizontal integration by somatostatin-expressing interneurons contributes to surround suppression [4]. In primates, however, surround suppression arises too quickly to come from local horizontal integration alone, and myelinated axons from higher visual areas, where cells have larger receptive fields, are thought to provide additional surround suppression [5, 6]. Silencing higher visual areas indeed decreased surround suppression in the awake primate by increasing responses to large stimuli [7, 8], although not under anesthesia [9, 10]. In smaller mammals, like mice, fast surround suppression could be possible without feedback. Recent studies revealed a small reduction in V1 responses when silencing higher areas [11, 12] but have not investigated surround suppression. To determine whether higher visual areas contribute to V1 surround suppression, even when this is not necessary for fast processing, we inhibited the areas lateral to V1, particularly the lateromedial area (LM), a possible homolog of primate V2 [13], while recording in V1 of awake and anesthetized mice. We found that part of the surround suppression depends on activity from lateral visual areas in the awake, but not anesthetized, mouse. Inhibiting the lateral visual areas specifically increased responses in V1 to large stimuli. We present a model explaining how excitatory feedback to V1 can have these suppressive effects for large stimuli.

Posterior thalamic nucleus axon terminals have different structure and functional impact in the motor and somatosensory vibrissal cortices

Rodents extract information about nearby objects from the movement of their whiskers through dynamic computations that are carried out by a network of forebrain structures that includes the thalamus and the primary sensory (S1BF) and motor (M1wk) whisker cortices. The posterior nucleus (Po), a higher order thalamic nucleus, is a key hub of this network, receiving cortical and brainstem sensory inputs and innervating both motor and sensory whisker-related cortical areas. In a recent study in rats, we showed that Po inputs differently impact sensory processing in S1BF and M1wk. Here, in C57BL/6 mice, we measured Po synaptic bouton layer distribution and size, compared cortical unit response latencies to "in vivo" Po activation, and pharmacologically examined the glutamatergic receptor mechanisms involved. We found that, in S1BF, a large majority (56%) of Po axon varicosities are located in layer (L)5a and only 12% in L2-L4, whereas in M1wk this proportion is inverted to 18% and 55%, respectively. Light and electron microscopic measurements showed that Po synaptic boutons in M1wk layers 3-4 are significantly larger (~ 50%) than those in S1BF L5a. Electrical Po stimulation elicits different area-specific response patterns. In S1BF, responses show weak or no facilitation, and involve both ionotropic and metabotropic glutamate receptors, whereas in M1wk, unit responses exhibit facilitation to repetitive stimulation and involve ionotropic NMDA glutamate receptors. Because of the different laminar distribution of axon terminals, synaptic bouton size and receptor mechanisms, the impact of Po signals on M1wk and S1BF, although simultaneous, is likely to be markedly different.

Experience-Dependent Development of Feature-Selective Synchronization in the Primary Visual Cortex

Early visual experience is essential for the maturation of visual functions in which the primary visual cortex plays crucial roles. The extraction of visual features based on response selectivity of individual neurons, a fundamental process in the cortex, is basically established by eye opening in rodents, suggesting that visual experience is required for the development of neural functions other than feature extraction. Here, we show that synchronized firing, which is important for visual information processing, occurs selectively in adjacent neurons sharing similar orientation or spatial frequency preferences in layers 2-4 (upper layer) of rat visual cortex. This feature-selective spike synchrony was rudimentary when the eyes opened and became prominent during the first few weeks after eye opening only in the presence of pattern vision. In contrast, synchronization in layers 5-6 (lower layer) was almost independent of orientation similarity and more weakly dependent on spatial frequency similarity compared with upper layer synchrony. Lower layer synchronization was strengthened during development after eye opening independently of visual experience as a whole. However, the feature selectivity of synchronization was regulated by visual inputs, whereas the inputs without contours were sufficient for this regulation. Therefore, we speculate that feature-selective synchronization in the upper layer may convey detailed information on visual objects to the higher-order cortex, whereas weakly feature-selective synchronization in the lower layer may covey rather rough visual information to the subcortical areas or higher-order cortex. A major role of visual experience may be to establish the specific neural circuits underlying highly feature-selective synchronization.SIGNIFICANCE STATEMENT The neuronal mechanisms underlying experience-dependent improvement of visual functions still remain unresolved. In this study, we investigated whether early visual experience contributes to the development of synchronized neural firing in the primary visual cortex, which plays important roles in visual information processing. We found that synchronized firing depends more remarkably on the similarity of preferred visual stimuli in the upper than lower layer neurons. Pattern vision during development was required for the establishment of spike synchrony in the upper but not the lower layer. These findings provide a new view regarding the role of sensory experience in the functional development of the cortex and the differences in the modes of information processing in the upper and lower cortical layers.

Multi-scale account of the network structure of macaque visual cortex

Cortical network structure has been extensively characterized at the level of local circuits and in terms of long-range connectivity, but seldom in a manner that integrates both of these scales. Furthermore, while the connectivity of cortex is known to be related to its architecture, this knowledge has not been used to derive a comprehensive cortical connectivity map. In this study, we integrate data on cortical architecture and axonal tracing data into a consistent multi-scale framework of the structure of one hemisphere of macaque vision-related cortex. The connectivity model predicts the connection probability between any two neurons based on their types and locations within areas and layers. Our analysis reveals regularities of cortical structure. We confirm that cortical thickness decays with cell density. A gradual reduction in neuron density together with the relative constancy of the volume density of synapses across cortical areas yields denser connectivity in visual areas more remote from sensory inputs and of lower structural differentiation. Further, we find a systematic relation between laminar patterns on source and target sides of cortical projections, extending previous findings from combined anterograde and retrograde tracing experiments. Going beyond the classical schemes, we statistically assign synapses to target neurons based on anatomical reconstructions, which suggests that layer 4 neurons receive substantial feedback input. Our derived connectivity exhibits a community structure that corresponds more closely with known functional groupings than previous connectivity maps and identifies layer-specific directional differences in cortico-cortical pathways. The resulting network can form the basis for studies relating structure to neural dynamics in mammalian cortex at multiple scales.

Silencing "Top-Down" Cortical Signals Affects Spike-Responses of Neurons in Cat's "Intermediate" Visual Cortex

We examined the effects of reversible inactivation of a higher-order, pattern/form-processing, postero-temporal visual (PTV) cortex on the background activities and spike-responses of single neurons in the ipsilateral cytoarchitectonic area 19 (putative area V3) of anesthetized domestic cats. Very occasionally (2/28), silencing recurrent “feedback” signals from PTV, resulted in significant and reversible reduction in background activity of area 19 neurons. By contrast, in large proportions of area 19 neurons, PTV inactivation resulted in: (i) significant reversible changes in the peak magnitude of their responses to visual stimuli (35.5%; 10/28); (ii) substantial reversible changes in direction selectivity indices (DSIs; 43%; 12/28); and (iii) reversible, upward shifts in preferred stimulus velocities (37%; 7/19). Substantial (≥20°) shifts in preferred orientation and/or substantial (≥20°) changes in width of orientation-tuning curves of area 19 neurons were however less common (26.5%; 4/15). In a series of experiments conducted earlier, inactivation of PTV also induced upward shifts in the preferred velocities of the ipsilateral cytoarchitectonic area 17 (V1) neurons responding optimally at low velocities. These upward shifts in preferred velocities of areas 19 and 17 neurons were often accompanied by substantial increases in DSIs. Thus, in both the primary visual cortex and the “intermediate” visual cortex (area 19), feedback from PTV plays a modulatory role in relation to stimulus velocity preferences and/or direction selectivity, that is, the properties which are usually believed to be determined by the inputs from the dorsal thalamus and/or feedforward inputs from the primary visual cortices. The apparent specialization of area 19 for processing information about stationary/slowly moving visual stimuli is at least partially determined, by the feedback from the higher-order pattern-processing visual area. Overall, the recurrent signals from the higher-order, pattern/form-processing visual cortex appear to play an important role in determining the magnitude of spike-responses and some “motion-related” receptive field properties of a substantial proportion of neurons in the intermediate form-processing visual area—area 19.

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

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

Recruitment of inhibition and excitation across mouse visual cortex depends on the hierarchy of interconnecting areas

Diverse features of sensory stimuli are selectively processed in distinct brain areas. The relative recruitment of inhibitory and excitatory neurons within an area controls the gain of neurons for appropriate stimulus coding. We examined how such a balance of inhibition and excitation is differentially recruited across multiple levels of a cortical hierarchy by mapping the locations and strengths of synaptic inputs to pyramidal and parvalbumin (PV)-expressing neurons in feedforward and feedback pathways interconnecting primary (V1) and two higher visual areas. While interareal excitation was stronger in PV than in pyramidal neurons in all layer 2/3 pathways, we observed a gradual scaling down of the inhibition/excitation ratio from the most feedforward to the most feedback pathway. Our results indicate that interareal gain control depends on the hierarchical position of the source and the target, the direction of information flow through the network, and the laminar location of target neurons.

DOI:http://dx.doi.org/10.7554/eLife.19332.001

The visual cortex is the part of the brain responsible for the conscious sense of vision. It is made up of multiple connected areas, and each area has a different expertise for analyzing images. The areas exchange information about the outside world via connections between cells called neurons. Communication between the areas works like a hierarchy with deeper, more connected areas in the brain extracting more complex information from a visual scene.

Communication in the cortex requires repeated stimulation or “excitation” of pathways of neurons; this risks damage or loss of sensitivity. But all of the communication in the hierarchy is excitatory, meaning that a signal from one area activates other areas in the visual cortex. So, how does the brain avoid becoming over-stimulated? The answer is that connections between the areas of the visual cortex also contact inhibitory neurons that suppress brain activity. However, it is not clear how the level of inhibition in different areas of the visual cortex is fine-tuned to avoid over-stimulation while maintaining accurate perception of vision.

D’Souza et al. now report how three distinct areas of the mouse visual cortex communicate to process visual signals. The approach involved making particular pathways of neurons sensitive to light, such that they could be activated separately with a laser. Next, D’Souza et al. measured the activity of both inhibitory and excitatory neurons that link the different brain areas. The experiments showed that the inhibitory neurons are more strongly activated in the areas of the brain that are further up the hierarchy. This indicates that our ability to make sense of more complex features of visual signals requires higher levels of inhibitory control. The next step is to examine how the brain activates and controls inhibitory neurons, and how this depends on the situation an animal is in and the task it is performing.

DOI:http://dx.doi.org/10.7554/eLife.19332.002

Competitive Dynamics in MSTd: A Mechanism for Robust Heading Perception Based on Optic Flow

Human heading perception based on optic flow is not only accurate, it is also remarkably robust and stable. These qualities are especially apparent when observers move through environments containing other moving objects, which introduce optic flow that is inconsistent with observer self-motion and therefore uninformative about heading direction. Moving objects may also occupy large portions of the visual field and occlude regions of the background optic flow that are most informative about heading perception. The fact that heading perception is biased by no more than a few degrees under such conditions attests to the robustness of the visual system and warrants further investigation. The aim of the present study was to investigate whether recurrent, competitive dynamics among MSTd neurons that serve to reduce uncertainty about heading over time offer a plausible mechanism for capturing the robustness of human heading perception. Simulations of existing heading models that do not contain competitive dynamics yield heading estimates that are far more erratic and unstable than human judgments. We present a dynamical model of primate visual areas V1, MT, and MSTd based on that of Layton, Mingolla, and Browning that is similar to the other models, except that the model includes recurrent interactions among model MSTd neurons. Competitive dynamics stabilize the model's heading estimate over time, even when a moving object crosses the future path. Soft winner-take-all dynamics enhance units that code a heading direction consistent with the time history and suppress responses to transient changes to the optic flow field. Our findings support recurrent competitive temporal dynamics as a crucial mechanism underlying the robustness and stability of perception of heading.

Shape perception enhances perceived contrast: evidence for excitatory predictive feedback?

Predictive coding theory suggests that predictable responses are “explained away” (i.e., reduced) by feedback. Experimental evidence for feedback inhibition, however, is inconsistent: most neuroimaging studies show reduced activity by predictive feedback, while neurophysiology indicates that most inter-areal cortical feedback is excitatory and targets excitatory neurons. In this study, we asked subjects to judge the luminance of two gray disks containing stimulus outlines: one enabling predictive feedback (a 3D-shape) and one impeding it (random-lines). These outlines were comparable to those used in past neuroimaging studies. All 14 subjects consistently perceived the disk with a 3D-shape stimulus brighter; thus, predictive feedback enhanced perceived contrast. Since early visual cortex activity at the population level has been shown to have a monotonic relationship with subjective contrast perception, we speculate that the perceived contrast enhancement could reflect an increase in neuronal activity. In other words, predictive feedback may have had an excitatory influence on neuronal responses. Control experiments ruled out attention bias, local feature differences and response bias as alternate explanations.

A Top-Down Cortical Circuit for Accurate Sensory Perception

A fundamental issue in cortical processing of sensory information is whether top-down control circuits from higher brain areas to primary sensory areas not only modulate but actively engage in perception. Here, we report the identification of a neural circuit for top-down control in the mouse somatosensory system. The circuit consisted of a long-range reciprocal projection between M2 secondary motor cortex and S1 primary somatosensory cortex. In vivo physiological recordings revealed that sensory stimulation induced sequential S1 to M2 followed by M2 to S1 neural activity. The top-down projection from M2 to S1 initiated dendritic spikes and persistent firing of S1 layer 5 (L5) neurons. Optogenetic inhibition of M2 input to S1 decreased L5 firing and the accurate perception of tactile surfaces. These findings demonstrate that recurrent input to sensory areas is essential for accurate perception and provide a physiological model for one type of top-down control circuit.

Physiology of layer 5 pyramidal neurons in mouse primary visual cortex: coincidence detection through bursting

L5 pyramidal neurons are the only neocortical cell type with dendrites reaching all six layers of cortex, casting them as one of the main integrators in the cortical column. What is the nature and mode of computation performed in mouse primary visual cortex (V1) given the physiology of L5 pyramidal neurons? First, we experimentally establish active properties of the dendrites of L5 pyramidal neurons of mouse V1 using patch-clamp recordings. Using a detailed multi-compartmental model, we show this physiological setup to be well suited for coincidence detection between basal and apical tuft inputs by controlling the frequency of spike output. We further show how direct inhibition of calcium channels in the dendrites modulates such coincidence detection. To establish the singe-cell computation that this biophysics supports, we show that the combination of frequency-modulation of somatic output by tuft input and (simulated) calcium-channel blockage functionally acts as a composite sigmoidal function. Finally, we explore how this computation provides a mechanism whereby dendritic spiking contributes to orientation tuning in pyramidal neurons.

Neurons in the brain have elaborate dendritic morphologies, hosting a variety of nonlinear channels that give way to single cell computation. In this study, we perform patch clamp recordings in the apical dendrites to establish the spatial distribution of nonlinear channels and the signals they support in the dendrites of layer 5 pyramidal neurons of the mouse primary visual cortex. Using this data, we create a detailed single cell model and simulate synaptic input. We then summarize the results of the simulations using a simple abstracted model, that ultimately describes the computation layer 5 pyramidal neurons perform on synaptic input. We find that this computation is a form of nonlinear frequency-modulation that works in a dendritic-spike dependent manner. Finally, we show how this computation allows dendritic spikes to contribute to the orientation tuning of pyramidal neurons in the visual cortex.

Very-long-range disynaptic V1 connections through layer 6 pyramidal neurons revealed by transneuronal tracing with rabies virus

Neurons in primary visual cortex (V1) integrate across the representation of the visual field through networks of long-range projecting pyramidal neurons. These projections, which originate from within V1 and through feedback from higher visual areas, are likely to play a key role in such visual processes as low contrast facilitation and extraclassical surround suppression. The extent of the visual field representation covered by feedback is generally much larger than that covered through monosynaptic horizontal connections within V1, and, although it may be possible that multisynaptic horizontal connections across V1 could also lead to more widespread spatial integration, nothing is known regarding such circuits. In this study, we used injections of the CVS-11 strain of rabies virus to examine disynaptic long-range horizontal connections within macaque monkey V1. Injections were made around the representation of 5° eccentricity in the lower visual field. Along the opercular surface of V1, we found that the majority of connected neurons extended up to 8 mm in most layers, consistent with twice the typically reported distances of monosynaptic connections. In addition, mainly in layer 6, a steady presence of connected neurons within V1 was observed up to 16 mm away. A relatively high percentage of these connected neurons had large-diameter somata characteristic of Meynert cells, which are known to project as far as 8 mm individually. Several neurons, predominantly in layer 6, were also found deep within the calcarine sulcus, reaching as far as 20° of eccentricity, based on estimates, and extending well into the upper visual field representation. Thus, our anatomical results provide evidence for a wide-ranging disynaptic circuit within V1, mediated largely through layer 6, that accounts for integration across a large region of the visual field.

Modulation of orientation-selective neurons by motion: when additive, when multiplicative?

The recurrent interaction among orientation-selective neurons in the primary visual cortex (V1) is suited to enhance contours in a noisy visual scene. Motion is known to have a strong pop-up effect in perceiving contours, but how motion-sensitive neurons in V1 support contour detection remains vastly elusive. Here we suggest how the various types of motion-sensitive neurons observed in V1 should be wired together in a micro-circuitry to optimally extract contours in the visual scene. Motion-sensitive neurons can be selective about the direction of motion occurring at some spot or respond equally to all directions (pandirectional). We show that, in the light of figure-ground segregation, direction-selective motion neurons should additively modulate the corresponding orientation-selective neurons with preferred orientation orthogonal to the motion direction. In turn, to maximally enhance contours, pandirectional motion neurons should multiplicatively modulate all orientation-selective neurons with co-localized receptive fields. This multiplicative modulation amplifies the local V1-circuitry among co-aligned orientation-selective neurons for detecting elongated contours. We suggest that the additive modulation by direction-specific motion neurons is achieved through synaptic projections to the somatic region, and the multiplicative modulation by pandirectional motion neurons through projections to the apical region of orientation-specific pyramidal neurons. For the purpose of contour detection, the V1-intrinsic integration of motion information is advantageous over a downstream integration as it exploits the recurrent V1-circuitry designed for that task.

Distinct balance of excitation and inhibition in an interareal feedforward and feedback circuit of mouse visual cortex

Mouse visual cortex is subdivided into multiple distinct, hierarchically organized areas that are interconnected through feedforward (FF) and feedback (FB) pathways. The principal synaptic targets of FF and FB axons that reciprocally interconnect primary visual cortex (V1) with the higher lateromedial extrastriate area (LM) are pyramidal cells (Pyr) and parvalbumin (PV)-expressing GABAergic interneurons. Recordings in slices of mouse visual cortex have shown that layer 2/3 Pyr cells receive excitatory monosynaptic FF and FB inputs, which are opposed by disynaptic inhibition. Most notably, inhibition is stronger in the FF than FB pathway, suggesting pathway-specific organization of feedforward inhibition (FFI). To explore the hypothesis that this difference is due to diverse pathway-specific strengths of the inputs to PV neurons we have performed subcellular Channelrhodopsin-2-assisted circuit mapping in slices of mouse visual cortex. Whole-cell patch-clamp recordings were obtained from retrobead-labeled FF(V1→LM)- and FB(LM→V1)-projecting Pyr cells, as well as from tdTomato-expressing PV neurons. The results show that the FF(V1→LM) pathway provides on average 3.7-fold stronger depolarizing input to layer 2/3 inhibitory PV neurons than to neighboring excitatory Pyr cells. In the FB(LM→V1) pathway, depolarizing inputs to layer 2/3 PV neurons and Pyr cells were balanced. Balanced inputs were also found in the FF(V1→LM) pathway to layer 5 PV neurons and Pyr cells, whereas FB(LM→V1) inputs to layer 5 were biased toward Pyr cells. The findings indicate that FFI in FF(V1→LM) and FB(LM→V1) circuits are organized in a pathway- and lamina-specific fashion.

Superficially projecting principal neurons in layer V of medial entorhinal cortex in the rat receive excitatory retrosplenial input

Principal cells in layer V of the medial entorhinal cortex (MEC) have a nodal position in the cortical-hippocampal network. They are the main recipients of hippocampal output and receive inputs from several cortical areas, including a prominent one from the retrosplenial cortex (RSC), likely targeting basal dendrites of layer V neurons. The latter project to extrahippocampal structures but also relay information to the superficial layers of MEC, closing the hippocampal-entorhinal loop. In the rat, we electrophysiologically and morphologically characterized RSC input into MEC and conclude that RSC provides an excitatory input to layer V pyramidal cells. Ultrastructural analyses of anterogradely labeled RSC projections showed that RSC axons in layer V of MEC form predominantly asymmetrical, likely excitatory, synapses on dendritic spines (90%) or shafts (8%), with 2% symmetrical, likely inhibitory, synapses on shafts and spines. The overall excitatory nature of the RSC input was confirmed by an optogenetic approach. Patterned laser stimulation of channelrhodopsin-expressing presynaptic RSC axons evoked exclusively EPSPs in recorded postsynaptic layer V cells. All responding layer V pyramidal cells had an axon extending toward the white matter. Half of these neurons also sent an axon to superficial layers. Confocal imaging of RSC synapses onto MEC layer V neurons shown to project superficially by way of retrogradely labeling from superficial layers confirmed that proximal dendrites of superficially projecting cells are among the targets of inputs from RSC. The excitatory RSC input thus interacts with both entorhinal-cortical and entorhinal-hippocampal circuits.

Tracing inputs to inhibitory or excitatory neurons of mouse and cat visual cortex with a targeted rabies virus

BACKGROUND

Cortical inhibition plays a critical role in controlling and modulating cortical excitation, and a more detailed understanding of the neuronal circuits contributing to each will provide more insight into their roles in complex cortical computations. Traditional neuronal tracers lack a means for easily distinguishing between circuits of inhibitory and excitatory neurons. To overcome this limitation, we have developed a technique for retrogradely labeling inputs to local clusters of inhibitory or excitatory neurons, but not both, using neurotropic adenoassociated and lentiviral vectors, cell-type-specific promoters, and a modified rabies virus.

RESULTS

Applied to primary visual cortex (V1) in mouse, the cell-type-specific tracing technique labeled thousands of presynaptically connected neurons and revealed that the dominant source of input to inhibitory and excitatory neurons is local in origin. Neurons in other visual areas are also labeled; the percentage of these intercortical inputs to excitatory neurons is somewhat higher (~20%) than to inhibitory neurons (<10%), suggesting that intercortical connections have less direct control over inhibition. The inputs to inhibitory neurons were also traced in cat V1, and when aligned with the orientation preference map revealed for the first time that long-range inputs to inhibitory neurons are well tuned to orientation.

CONCLUSIONS

These novel findings for inhibitory and excitatory circuits in the visual cortex demonstrate the efficacy of our new technique and its ability to work across species, including larger-brained mammals such as the cat. This paves the way for a better understanding of the roles of specific cell types in higher-order perceptual and cognitive processes.

The cell-type specific cortical microcircuit: relating structure and activity in a full-scale spiking network model

In the past decade, the cell-type specific connectivity and activity of local cortical networks have been characterized experimentally to some detail. In parallel, modeling has been established as a tool to relate network structure to activity dynamics. While available comprehensive connectivity maps ( Thomson, West, et al. 2002; Binzegger et al. 2004) have been used in various computational studies, prominent features of the simulated activity such as the spontaneous firing rates do not match the experimental findings. Here, we analyze the properties of these maps to compile an integrated connectivity map, which additionally incorporates insights on the specific selection of target types. Based on this integrated map, we build a full-scale spiking network model of the local cortical microcircuit. The simulated spontaneous activity is asynchronous irregular and cell-type specific firing rates are in agreement with in vivo recordings in awake animals, including the low rate of layer 2/3 excitatory cells. The interplay of excitation and inhibition captures the flow of activity through cortical layers after transient thalamic stimulation. In conclusion, the integration of a large body of the available connectivity data enables us to expose the dynamical consequences of the cortical microcircuitry.

Canonical microcircuits for predictive coding

This Perspective considers the influential notion of a canonical (cortical) microcircuit in light of recent theories about neuronal processing. Specifically, we conciliate quantitative studies of microcircuitry and the functional logic of neuronal computations. We revisit the established idea that message passing among hierarchical cortical areas implements a form of Bayesian inference-paying careful attention to the implications for intrinsic connections among neuronal populations. By deriving canonical forms for these computations, one can associate specific neuronal populations with specific computational roles. This analysis discloses a remarkable correspondence between the microcircuitry of the cortical column and the connectivity implied by predictive coding. Furthermore, it provides some intuitive insights into the functional asymmetries between feedforward and feedback connections and the characteristic frequencies over which they operate.

Communication and wiring in the cortical connectome

In cerebral cortex, the huge mass of axonal wiring that carries information between near and distant neurons is thought to provide the neural substrate for cognitive and perceptual function. The goal of mapping the connectivity of cortical axons at different spatial scales, the cortical connectome, is to trace the paths of information flow in cerebral cortex. To appreciate the relationship between the connectome and cortical function, we need to discover the nature and purpose of the wiring principles underlying cortical connectivity. A popular explanation has been that axonal length is strictly minimized both within and between cortical regions. In contrast, we have hypothesized the existence of a multi-scale principle of cortical wiring where to optimize communication there is a trade-off between spatial (construction) and temporal (routing) costs. Here, using recent evidence concerning cortical spatial networks we critically evaluate this hypothesis at neuron, local circuit, and pathway scales. We report three main conclusions. First, the axonal and dendritic arbor morphology of single neocortical neurons may be governed by a similar wiring principle, one that balances the conservation of cellular material and conduction delay. Second, the same principle may be observed for fiber tracts connecting cortical regions. Third, the absence of sufficient local circuit data currently prohibits any meaningful assessment of the hypothesis at this scale of cortical organization. To avoid neglecting neuron and microcircuit levels of cortical organization, the connectome framework should incorporate more morphological description. In addition, structural analyses of temporal cost for cortical circuits should take account of both axonal conduction and neuronal integration delays, which appear mostly of the same order of magnitude. We conclude the hypothesized trade-off between spatial and temporal costs may potentially offer a powerful explanation for cortical wiring patterns.

Cortical GABAergic interneurons in cross-modal plasticity following early blindness

Early loss of a given sensory input in mammals causes anatomical and functional modifications in the brain via a process called cross-modal plasticity. In the past four decades, several animal models have illuminated our understanding of the biological substrates involved in cross-modal plasticity. Progressively, studies are now starting to emphasise on cell-specific mechanisms that may be responsible for this intermodal sensory plasticity. Inhibitory interneurons expressing γ-aminobutyric acid (GABA) play an important role in maintaining the appropriate dynamic range of cortical excitation, in critical periods of developmental plasticity, in receptive field refinement, and in treatment of sensory information reaching the cerebral cortex. The diverse interneuron population is very sensitive to sensory experience during development. GABAergic neurons are therefore well suited to act as a gate for mediating cross-modal plasticity. This paper attempts to highlight the links between early sensory deprivation, cortical GABAergic interneuron alterations, and cross-modal plasticity, discuss its implications, and further provide insights for future research in the field.

Different glutamate receptors convey feedforward and recurrent processing in macaque V1

Neurons in the primary visual cortex (V1) receive feedforward input from the thalamus, which shapes receptive-field properties. They additionally receive recurrent inputs via horizontal connections within V1 and feedback from higher visual areas that are thought to be important for conscious visual perception. Here, we investigated what roles different glutamate receptors play in conveying feedforward and recurrent inputs in macaque V1. As a measure of recurrent processing, we used figure-ground modulation (FGM), the increased activity of neurons representing figures compared with background, which depends on feedback from higher areas. We found that feedforward-driven activity was strongly reduced by the AMPA receptor antagonist 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), whereas this drug had no effect on FGM. In contrast, blockers of the NMDA receptor reduced FGM, whereas their effect on visually driven activity varied with the subunit specificity of the drug. The NMDA receptor blocker 2-amino-5-phosphonovalerate (APV) caused a slight reduction of the visual response, whereas ifenprodil, which targets NMDA receptors containing the NMDA receptor NR2B subunit, increased the visual response. These findings demonstrate that glutamate receptors contribute differently to feedforward and recurrent processing in V1 and suggest ways to selectively disrupt recurrent processing so that its role in visual perception can be elucidated.

GABA inactivation of visual area MT modifies the responsiveness and direction selectivity of V2 neurons in Cebus monkeys

We investigated the contribution of the projections from area MT to the receptive field properties of cells in visual area V2 in anesthetized and paralyzed Cebus apella monkeys. We recorded extracellular single-unit activity using tungsten microelectrodes in three monkeys before and after pressure injection of a 0.25-mol/l GABA solution. The visual stimulus consisted of a single bar moving in one of eight directions. In total, 72 V2 neurons were studied in 18 sessions of GABA injection into area MT. A group of 22 neurons was investigated over a shorter period of time ranging from 15 to 60 min, during which the activity did not return to baseline levels. The remaining 50 neurons were studied over a period of at least 2 h, and no statistical difference was observed in the neuronal response before and long after GABA inactivation. The effects on these 50 neurons consisted of an early (1-20 min) significant general decrease in excitability with changes in either orientation or direction selectivity. The differential decrease in excitability resulted in an intermediate improvement (20-40 min) of the signal-to-noise ratio for the stimulus-driven activity. The inactivation depended on the quantity of GABA injected into area MT and persisted for a period of 2 h. The GABA inactivation in area MT produced inhibition of most cells (72%) and a significant change of direction tuning in the majority (56%) of V2 neurons. Both increases and also decreases in the direction tuning of V2 neurons were observed. These feedback projections are capable of modulating not only the levels of spontaneous and driven activity of V2 neurons but also the V2 receptive field properties, such as direction selectivity.

GABA inactivation of area V4 changes receptive-field properties of V2 neurons in Cebus monkeys

To investigate the contribution of feedback circuits from area V4 to the receptive-field properties of V2 neurons, we used tungsten microelectrodes to record extracellular single units in these visual areas, before and after pressure injections of a solution of 0.25 mol/L of GABA in two anesthetized and paralyzed Cebus apella monkeys. The visual stimulus consisted of a single bar moving in one of eight directions. Using a device made of four stainless steel pipettes and one central tungsten electrode, we inactivated, with different amounts of GABA, topographically corresponding areas of V4, while studying V2 neurons. We studied a total of 36 V2 neurons during six sessions of GABA injections into area V4. GABA inactivation of visual area V4 produced a general decrease in the excitability of the neurons, which included a decrease in spontaneous and driven activities, followed by changes in direction selectivity. The changes in selectivity were toward an increase in directional selectivity and decrease in orientation selectivity. Thus, feedback connections arising from V4, an area of the ventral steams of visual information processing, are capable of not only modulating the spontaneous and driven activity of V2 neurons, but also of modifying V2 receptive field properties, such as its direction and/or orientation selectivity.

Diffusive feedback influences on hierarchical information processing

In visual information processing, feedforward projection from primary to secondary visual cortex (V1-to-V2) is essential for integrating combinations of oriented bars in order to extract angular information embedded within contours that represent the shape of objects. For feedback (V2-to-V1) projection, two distinct types of pathways have been observed: clustered projection and diffused projection. The former innervates V1 domains with a preferred orientation similar to that of V2 cells of origin. In contrast, the latter innervates without such orientation specificity. V2 cells send their axons to V1 domains with both similar and dissimilar orientation preferences. It is speculated that the clustered feedback projection has a role in contour integration. The role of the diffused feedback projection, however, remains to be seen. We simulated a minimal, functional V1-V2 neural network model. The diffused feedback projection contributed to achieving ongoing-spontaneous subthreshold membrane oscillations in V1 cells, thereby reducing the reaction time of V1 cells to a pair of bars that represents specific angular information. Interestingly, the feedback influence took place even before V2 responses, which might stem largely from ongoing-spontaneous signaling from V2. We suggest that the diffusive feedback influence from V2 could act early in V1 responses and accelerate their reaction speed to sensory stimulation in order to rapidly extract angular information.

Characteristics of synaptic connections between rodent primary somatosensory and motor cortices

The reciprocal connections between primary motor (M1) and primary somatosensory cortices (S1) are hypothesized to play a crucial role in the ability to update motor plans in response to changes in the sensory periphery. These interactions provide M1 with information about the sensory environment that in turn signals S1 with anticipatory knowledge of ongoing motor plans. In order to examine the synaptic basis of sensorimotor feedforward (S1-M1) and feedback (M1-S1) connections directly, we utilized whole-cell recordings in slices that preserve these reciprocal sensorimotor connections. Our findings indicate that these regions are connected via direct monosynaptic connections in both directions. Larger magnitude responses were observed in the feedforward direction (S1-M1), while the feedback (M1-S1) responses occurred at shorter latencies. The morphology as well as the intrinsic firing properties of the neurons in these pathways indicates that both excitatory and inhibitory neurons are targeted. Differences in synaptic physiology suggest that there exist specializations within the sensorimotor pathway that may allow for the rapid updating of sensory-motor processing within the cortex in response to changes in the sensory periphery.

The functional roles of feedback projections in the visual system

Neurons in the nervous system make connections with ascending feedforward projections and descending feedback projections, as well as projections from neural structures at the identical hierarchical level. These neurons form extremely complicated neural networks and pathways. Compared with the role of the feedforward projection, much less is known concerning the functional roles of the feedback projection. Visual cortex is a good model for studying functional roles of cortical feedback projections which involve many high functions, such as attention, searching and cognition. The present review mainly focused on the functional roles of feedback projections in the visual system.

Indirect pathway between the primary auditory and visual cortices through layer V pyramidal neurons in V2L in mouse and the effects of bilateral enucleation

Visual cortical areas are activated by auditory stimuli in blind mice. Direct heteromodal cortical connections have been shown between the primary auditory cortex (A1) and primary visual cortex (V1), and between A1 and secondary visual cortex (V2). Auditory afferents to V2 terminate in close proximity to neurons that project to V1, and potentially constitute an effective indirect pathway between A1 and V1. In this study, we injected a retrograde adenoviral vector that expresses enhanced green fluorescent protein under a synapsin promotor in V1 and biotinylated dextran amine as an anterograde tracer in A1 to determine: (i) whether A1 axon terminals establish synaptic contacts onto the lateral part of V2 (V2L) neurons that project to V1; and (ii) if this indirect cortical pathway is altered by a neonatal enucleation in mice. Complete dendritic arbors of layer V pyramidal neurons were reconstructed in 3D, and putative contacts between pre-synaptic auditory inputs and postsynaptic visual neurons were analysed using a laser-scanning confocal microscope. Putative synaptic contacts were classified as high-confidence and low-confidence contacts, and charted onto dendritic trees. As all reconstructed layer V pyramidal neurons received auditory inputs by these criteria, we conclude that V2L acts as an important relay between A1 and V1. Auditory inputs are preferentially located onto lower branch order dendrites in enucleated mice. Also, V2L neurons are subject to morphological reorganizations in both apical and basal dendrites after the loss of vision. The A1-V2L-V1 pathway could be involved in multisensory processing and contribute to the auditory activation of the occipital cortex in the blind rodent.

Fast and robust image segmentation by small-world neural oscillator networks

Inspired by the temporal correlation theory of brain functions, researchers have presented a number of neural oscillator networks to implement visual scene segmentation problems. Recently, it is shown that many biological neural networks are typical small-world networks. In this paper, we propose and investigate two small-world models derived from the well-known LEGION (locally excitatory and globally inhibitory oscillator network) model. To form a small-world network, we add a proper proportion of unidirectional shortcuts (random long-range connections) to the original LEGION model. With local connections and shortcuts, the neural oscillators can not only communicate with neighbors but also exchange phase information with remote partners. Model 1 introduces excitatory shortcuts to enhance the synchronization within an oscillator group representing the same object. Model 2 goes further to replace the global inhibitor with a sparse set of inhibitory shortcuts. Simulation results indicate that the proposed small-world models could achieve synchronization faster than the original LEGION model and are more likely to bind disconnected image regions belonging together. In addition, we argue that these two models are more biologically plausible.

A network of spiking neurons that can represent interval timing: mean field analysis

Despite the vital importance of our ability to accurately process and encode temporal information, the underlying neural mechanisms are largely unknown. We have previously described a theoretical framework that explains how temporal representations, similar to those reported in the visual cortex, can form in locally recurrent cortical networks as a function of reward modulated synaptic plasticity. This framework allows networks of both linear and spiking neurons to learn the temporal interval between a stimulus and paired reward signal presented during training. Here we use a mean field approach to analyze the dynamics of non-linear stochastic spiking neurons in a network trained to encode specific time intervals. This analysis explains how recurrent excitatory feedback allows a network structure to encode temporal representations.

Dynamics of extraclassical surround modulation in three types of V1 neurons

Visual stimuli outside of the classical receptive field (CRF) can influence the response of neurons in primary visual cortex (V1). While recording single units in cat, we presented drifting sinusoidal gratings in circular apertures of different sizes to investigate this extraclassical surround modulation over time. For the full 2-s stimulus time course, three types of neurons were found: 1) 68% of the cells were “suppressive,” 2) 25% were “plateau” cells that showed response saturation with no suppression, and 3) the remaining 6% of cells were “facilitative.” Analysis of the response dynamics revealed that at response onset, activity of one-half of facilitative cells, 70% of plateau cells, and all suppressive cells is suppressed by the surround. However, over the next 20–30 ms, surround modulation changes to stronger suppression for suppressive cells, substantial facilitation for facilitative cells, and weak facilitation for plateau cells. For all three cell types, these modulatory effects then stabilize between 100 and 200 ms from stimulus onset. Thus our findings illustrate two stages of surround modulation. Early modulation is mainly suppressive regardless of cell type and, because of rapid onset, may rely on feedforward mechanisms. Surround modulation that evolves later in time is not always suppressive, depending on cell type, and may be generated through different combinations of cortical circuits. Additional analysis of modulation throughout the cortical column suggests the possibility that the larger excitatory fields of facilitative cells, primarily found in infragranular layers, may contribute to the second stage of suppression through intracolumnar circuitry.

Evaluation of inputs to rat primary auditory cortex from the suprageniculate nucleus and extrastriate visual cortex

Evidence indicates that visual stimuli influence cells in the primary auditory cortex. To evaluate potential sources of this visual input and how they enter into the circuitry of the auditory cortex, we examined axonal terminations in the primary auditory cortex from nonprimary extrastriate visual cortex (V2M, V2L) and from the multimodal thalamic suprageniculate nucleus (SG). Gross biocytin/biotinylated dextran amine (BDA) injections into the SG or extrastriate cortex labeled inputs terminating primarily in superficial and deep layers. SG projects primarily to layers I, V, and VI while V2M and V2L project primarily to layers I and VI, with V2L also targeting layers II/III. Layer I inputs differ in that SG terminals are concentrated superficially, V2L are deeper, and V2M are equally distributed throughout. Individual axonal reconstructions document that single axons can 1) innervate multiple layers; 2) run considerable distances in layer I; and 3) run preferentially in the dorsoventral direction similar to isofrequency axes. At the electron microscopic level, SG and V2M terminals 1) are the same size regardless of layer; 2) are non-gamma-aminobutyric acid (GABA)ergic; 3) are smaller than ventral medial geniculate terminals synapsing in layer IV; 4) make asymmetric synapses onto dendrites/spines that 5) are non-GABAergic and 6) are slightly larger in layer I. Thus, both areas provide a substantial feedback-like input with differences that may indicate potentially different roles.

Influence of 'feedback' signals on spatial integration in receptive fields of cat area 17 neurons

'Feedback' signals from mammalian extrastriate visual cortices are reported to exert primarily an excitatory influence on the classical receptive field (CRF) of neurons in the primary visual cortex (V1). However, given the much larger CRFs of neurons in extrastriate visual cortices it is not yet understood how feedback signals influence the spatial integration of visual signals by V1 neurons. To investigate this, we reversibly inactivated one of the 'form-processing' extrastriate visual cortices, the postero-temporal visual (PTV) cortex, and examined changes in responses of V1 neurons to drifting grating patches up to 28 degrees in diameter. We found that during inactivation of PTV cortex the magnitude of the responses to CRF-confined stimuli and that to large stimuli inducing maximum suppression (i.e. minimum responses) was significantly reduced, while the spatial extent of the CRF remained largely unaffected. As a result, the relative strength of the surround suppression increased marginally. This effect was apparent in both simple and complex cells. It was also strong and consistent in cells located in supragranular and infragranular layers. For those cells exhibiting some relief from surround suppression or 'counter-suppression' when large stimuli patches were applied, the effect on counter-suppression was heterogeneous. Overall, the relative integrated responses to the 28 degrees grating patches were also decreased when PTV cortex was inactivated. Thus, a substantial reduction in the CRF response and the largely unaffected spatial extent of the CRF as well as a weak surround effect observed in the present study are consistent with a multiplicative scaling effect.

Perceptron learning rule derived from spike-frequency adaptation and spike-time-dependent plasticity

It is widely believed that sensory and motor processing in the brain is based on simple computational primitives rooted in cellular and synaptic physiology. However, many gaps remain in our understanding of the connections between neural computations and biophysical properties of neurons. Here, we show that synaptic spike-time-dependent plasticity (STDP) combined with spike-frequency adaptation (SFA) in a single neuron together approximate the well-known perceptron learning rule. Our calculations and integrate-and-fire simulations reveal that delayed inputs to a neuron endowed with STDP and SFA precisely instruct neural responses to earlier arriving inputs. We demonstrate this mechanism on a developmental example of auditory map formation guided by visual inputs, as observed in the external nucleus of the inferior colliculus (ICX) of barn owls. The interplay of SFA and STDP in model ICX neurons precisely transfers the tuning curve from the visual modality onto the auditory modality, demonstrating a useful computation for multimodal and sensory-guided processing.

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.

The synaptic connections between cortical areas V1 and V2 in macaque monkey

The primary visual cortex (V1) and V2 together form approximately 24% of the total neocortex of the macaque monkey and have each other as their major partners. The major target of the V1 projection to V2 is layer 4, where it forms clusters of boutons, which form asymmetric (excitatory) synapses mainly with dendritic spines (75%). The remainder form synapses with dendritic shafts. The synapses found on spines were often more complex, perforated postsynaptic densities than those found on dendritic shafts. The reciprocal projection from V2 to V1 targeted layers 1, 2/3, and 5 and was formed of axons of different morphologies. One axon type, originating from superficial layer pyramidal cells, had a morphology resembling those of local pyramidal cell collaterals. These axons arborized in layers 1, 2/3, and 5 of V1. Another type of axon, arborizing in layer 1, was slender (0.3 microm), unbranched, unmyelinated, and uniformly covered with boutons terminaux and formed asymmetric synapses mainly with slender spines. Yet a third type of axon also confined to layer 1, was thick (>1 microm), branched, heavily myelinated, and formed separate small clusters of large ( approximately 1 microm) en passant multisynaptic boutons that formed asymmetric synapses mainly with large flat spines. These data show the existence of a reciprocal excitatory loop between V1 and V2 that is formed by different axonal types, each with preferred layers of termination.

Learning reward timing in cortex through reward dependent expression of synaptic plasticity

The ability to represent time is an essential component of cognition but its neural basis is unknown. Although extensively studied both behaviorally and electrophysiologically, a general theoretical framework describing the elementary neural mechanisms used by the brain to learn temporal representations is lacking. It is commonly believed that the underlying cellular mechanisms reside in high order cortical regions but recent studies show sustained neural activity in primary sensory cortices that can represent the timing of expected reward. Here, we show that local cortical networks can learn temporal representations through a simple framework predicated on reward dependent expression of synaptic plasticity. We assert that temporal representations are stored in the lateral synaptic connections between neurons and demonstrate that reward-modulated plasticity is sufficient to learn these representations. We implement our model numerically to explain reward-time learning in the primary visual cortex (V1), demonstrate experimental support, and suggest additional experimentally verifiable predictions.

Reconciling predictive coding and biased competition models of cortical function

A simple variation of the standard biased competition model is shown, via some trivial mathematical manipulations, to be identical to predictive coding. Specifically, it is shown that a particular implementation of the biased competition model, in which nodes compete via inhibition that targets the inputs to a cortical region, is mathematically equivalent to the linear predictive coding model. This observation demonstrates that these two important and influential rival theories of cortical function are minor variations on the same underlying mathematical model.

fMRI reveals that non-local processing in ventral retinotopic cortex underlies perceptual grouping by temporal synchrony

When spatially separated objects appear and disappear in a synchronous manner, they perceptually group into a single global object that itself appears and disappears. We employed functional magnetic resonance imaging (fMRI) to identify brain regions involved in this type of perceptual grouping. Subjects viewed four chromatically-defined disks (one per visual quadrant) that flashed on and off. We contrasted %BOLD signal changes between blocks of synchronously flashing disks (Grouping) with blocks of asynchronously flashing disks (no-Grouping).

RESULTS

A region of interest analysis revealed %BOLD signal change in the Grouping condition was significantly greater than in the no-Grouping condition within retinotopic areas V2, V3, and V4v. Within a single quadrant of the visual field, the spatio-temporal information present in the image was identical across the two stimulus conditions. As such, the two conditions could not be distinguished from each other on the basis of the rate or pattern of flashing within a single visual quadrant. The observed results must therefore arise through nonlocal interactions between or within these retinotopic areas, or arise from outside these retinotopic areas. Furthermore, when V2 and V3 were split into ventral and dorsal sub-ROIs, ventral retinotopic areas V2v and V3v preferentially differentiated between the two conditions whereas the corresponding dorsal areas V2d and V3d did not. In contrast, within hMT+, %BOLD signal was significantly greater in the no-Grouping condition.

CONCLUSION

Nonlocal processing within, between, or to ventral retinotopic cortex at least as early as V2v, and including V3v, and V4v, underlies perceptual grouping via temporal synchrony.