A modulatory effect of the feedback from higher visual areas to V1 in the mouse

Excitatory-inhibitory recurrent dynamics produce robust visual grids and stable attractors

Spatially modulated grid cells have been recently found in the rat secondary visual cortex (V2) during active navigation. However, the computational mechanism and functional significance of V2 grid cells remain unknown. To address the knowledge gap, we train a biologically inspired excitatory-inhibitory recurrent neural network to perform a two-dimensional spatial navigation task with multisensory input. We find grid-like responses in both excitatory and inhibitory RNN units, which are robust with respect to spatial cues, dimensionality of visual input, and activation function. Population responses reveal a low-dimensional, torus-like manifold and attractor. We find a link between functional grid clusters with similar receptive fields and structured excitatory-to-excitatory connections. Additionally, multistable torus-like attractors emerged with increasing sparsity in inter- and intra-subnetwork connectivity. Finally, irregular grid patterns are found in recurrent neural network (RNN) units during a visual sequence recognition task. Together, our results suggest common computational mechanisms of V2 grid cells for spatial and non-spatial tasks.

Distinct organization of two cortico-cortical feedback pathways

Neocortical feedback is critical for attention, prediction, and learning. To mechanically understand its function requires deciphering its cell-type wiring. Recent studies revealed that feedback between primary motor to primary somatosensory areas in mice is disinhibitory, targeting vasoactive intestinal peptide-expressing interneurons, in addition to pyramidal cells. It is unknown whether this circuit motif represents a general cortico-cortical feedback organizing principle. Here we show that in contrast to this wiring rule, feedback between higher-order lateromedial visual area to primary visual cortex preferentially activates somatostatin-expressing interneurons. Functionally, both feedback circuits temporally sharpen feed-forward excitation eliciting a transient increase-followed by a prolonged decrease-in pyramidal cell activity under sustained feed-forward input. However, under feed-forward transient input, the primary motor to primary somatosensory cortex feedback facilitates bursting while lateromedial area to primary visual cortex feedback increases time precision. Our findings argue for multiple cortico-cortical feedback motifs implementing different dynamic non-linear operations.

My prolonged collaboration with Ray Guillery

My active collaboration with Ray Guillery started in 1968, when he was a Full Professor at the University of Wisconsin and I was a graduate student at the University of Pennsylvania. The collaboration lasted almost 50 years with virtually no breaks. Among the ideas we proposed are that glutamatergic pathways in thalamus and cortex can be classified into drivers and modulators; that many thalamic nuclei could be classified as higher order, meaning that they receive driving input from layer 5 of cortex and participate in cortico-thalamocortical circuits; and that much of the information relayed by thalamus serves as an efference copy for motor commands initiated by cortex.

Synaptic properties of the lemniscal and paralemniscal pathways to the mouse somatosensory thalamus

Somatosensory information is thought to arrive in thalamus through two glutamatergic routes called the lemniscal and paralemniscal pathways via the ventral posterior medial (VPm) and posterior medial (POm) nuclei. Here we challenge the view that these pathways functionally represent parallel information routes. Using electrical stimulation and an optogenetic approach in brain slices from the mouse, we investigated the synaptic properties of the lemniscal and paralemniscal input to VPm and POm. Stimulation of the lemniscal pathway produced class 1, or "driver," responses in VPm relay cells, which is consistent with this being an information-bearing channel. However, stimulation of the paralemniscal pathway produced two distinct types of responses in POm relay cells: class 1 (driver) responses in 29% of the cells, and class 2, or "modulator," responses in the rest. Our data suggest that, unlike the lemniscal pathway, the paralemniscal one is not homogenous and that it is primarily modulatory. This finding requires major rethinking regarding the routes of somatosensory information to cortex and suggests that the paralemniscal route is chiefly involved in modulatory functions rather than simply being an information route parallel to the lemniscal channel.

Stream-dependent development of higher visual cortical areas

Multiple cortical areas contribute to visual processing in mice. However, the functional organization and development of higher visual areas are unclear. Here we used intrinsic signal optical imaging and two-photon calcium imaging to map visual responses in adult and developing mice. We found that visually driven activity was well correlated among higher visual areas within two distinct subnetworks resembling the dorsal and ventral visual streams. Visual response magnitude in dorsal stream areas slowly increased over the first 2 weeks of visual experience. By contrast, ventral stream areas exhibited strong responses shortly after eye opening. Neurons in a dorsal stream area showed little change in their tuning sharpness to oriented gratings while those in a ventral stream area increased stimulus selectivity and expanded their receptive fields significantly. Together, these findings provide a functional basis for grouping subnetworks of mouse visual areas and revealed stream differences in the development of receptive field properties.

Thalamus plays a central role in ongoing cortical functioning

Several challenges to current views of thalamocortical processing are offered here. Glutamatergic pathways in thalamus and cortex are divided into two distinct classes: driver and modulator. We suggest that driver inputs are the main conduits of information and that modulator inputs modify how driver inputs are processed. Different driver sources reveal two types of thalamic relays: first order relays receive subcortical driver input (for example, retinal input to the lateral geniculate nucleus), whereas higher order relays (for example, pulvinar) receive driver input from layer 5 of cortex and participate in cortico-thalamo-cortical (or transthalamic) circuits. These transthalamic circuits represent an unappreciated aspect of cortical functioning, which I discuss here. Direct corticocortical connections are often paralleled by transthalamic ones. Furthermore, driver inputs to thalamus, both first and higher order, typically arrive via branching axons, and the transthalamic branch often innervates subcortical motor centers, leading to the suggestion that these inputs to thalamus serve as efference copies.

Top-Down-Mediated Facilitation in the Visual Cortex Is Gated by Subcortical Neuromodulation

Response properties in primary sensory cortices are highly dependent on behavioral state. For example, the nucleus basalis of the forebrain plays a critical role in enhancing response properties of excitatory neurons in primary visual cortex (V1) during active exploration and learning. Given the strong reciprocal connections between hierarchically arranged cortical regions, how are increases in sensory response gain constrained to prevent runaway excitation? To explore this, we used in vivo two-photon guided cell-attached recording in conjunction with spatially restricted optogenetic photo-inhibition of higher-order visual cortex in mice. We found that the principle feedback projection to V1 originating from the lateral medial area (LM) facilitated visual responses in layer 2/3 excitatory neurons by ∼20%. This facilitation was reduced by half during basal forebrain activation due to differential response properties between LM and V1. Our results demonstrate that basal-forebrain-mediated increases in response gain are localized to V1 and are not propagated to LM and establish that subcortical modulation of visual cortex is regionally distinct.

Properties of the primary somatosensory cortex projection to the primary motor cortex in the mouse

The primary somatosensory (S1) and primary motor (M1) cortices are reciprocally connected, and their interaction has long been hypothesized to contribute to coordinated motor output. Very little is known, however, about the nature and synaptic properties of the S1 input to M1. Here we wanted to take advantage of a previously developed sensorimotor slice preparation that preserves much of the S1-to-M1 connectivity (Rocco MM, Brumberg JC. J Neurosci Methods 162: 139-147, 2007), as well as available optogenetic methodologies, in order to investigate the synaptic profile of this projection. Our data show that S1 input to pyramidal cells of M1 is highly homogeneous, possesses many features of a "driver" pathway, such as paired-pulse depression and lack of metabotropic glutamate receptor activation, and is mediated through axons that terminate in both small and large synaptic boutons. Our data suggest that S1 provides M1 with afferents that possess synaptic and anatomical characteristics ideal for the delivery of strong inputs that can "drive" postsynaptic M1 cells, thereby potentially affecting their output.

Modulatory effects of activation of metabotropic glutamate receptors on GABAergic circuits in the mouse thalamus

Metabotropic glutamate receptors (mGluRs) are widely distributed in the central nervous system and modulate the release of neurotransmitters in different ways. We have previously shown that activation of presynaptic group II mGluRs reduces the gain of GABAergic inputs in both primary visual and auditory cortices (V1 and A1). In the present study, we sought to determine whether activation of mGluRs can also affect the inhibitory inputs in thalamus. Using whole cell recordings in a mouse slice preparation, we studied two GABAergic inputs to thalamic relay cells: that of the thalamic reticular nucleus (TRN) to cells of the ventral posteromedial nucleus (VPM) and that of interneurons to cells of the lateral geniculate nucleus (LGN). We found that activation of mGluRs significantly reduced the amplitudes of inhibitory postsynaptic currents (IPSCs) evoked from TRN inputs to VPM cells, and further experiments indicated that this was due to activation of presynaptic group I and group II mGluRs. Similar results were found in the interneuronal inputs to LGN cells. Activation of presynaptic group I (type 1 but not type 5) and group II mGluRs significantly reduced the amplitudes of evoked IPSCs of the axonal inputs to relay cells, and additional experiments were consistent with previous observations that activation of type 5 mGluRs on the dendritic terminals of interneurons enhanced postsynaptic IPSCs. We concluded that group I and II mGluRs may generally reduce the amplitude of evoked GABAergic IPSCs of axonal inputs to thalamic relay cells, operating through presynaptic mechanisms, and this extends our previous findings in cortex.

Developmental decline in modulation of glutamatergic synapses in layer IV of the barrel cortex by group II metabotropic glutamate receptors

Previously, we demonstrated that group II metabotropic glutamate receptors (mGluRs) reduce glutamate release from thalamocortical synapses during early postnatal development (P7-11). To further examine the role of group II mGluRs in the modulation of somatosensory circuitry, we determined whether group II mGluRs continue to modulate thalamocortical synapses until adulthood and whether these receptors also modulate intra-cortical synapses in the barrel cortex. To address these issues, we examined the effect of the group II mGluR agonists on thalamocortical excitatory postsynaptic currents (EPSCs) and intra-barrel EPSCs in slices from animals of different ages (P7-53). We found that the depression of thalamocortical EPSCs by group II mGluRs rapidly declined after the second postnatal week. In contrast, adenosine continued to depress thalamocortical EPSCs via a presynaptic mechanism in young adult mice (P30-50). Activation of group II mGluRs also reduced intra-barrel EPSCs through a postsynaptic mechanism in young mice (P7-11). Similar to the thalamocortical synapses, the group II mGluR modulation of intra-barrel excitatory synapses declined with development. In young adult animals (P30-50), group II mGluR stimulation had little effect on intra-barrel EPSCs but did hyperpolarize the neurons. Together our results demonstrate that group II mGluRs modulate barrel cortex circuitry by presynaptic and postsynaptic mechanisms depending on the source of the synapse and that this modulation declines with development.

Visual cortical areas of the mouse: comparison of parcellation and network structure with primates

Brains have evolved to optimize sensory processing. In primates, complex cognitive tasks must be executed and evolution led to the development of large brains with many cortical areas. Rodents do not accomplish cognitive tasks of the same level of complexity as primates and remain with small brains both in relative and absolute terms. But is a small brain necessarily a simple brain? In this review, several aspects of the visual cortical networks have been compared between rodents and primates. The visual system has been used as a model to evaluate the level of complexity of the cortical circuits at the anatomical and functional levels. The evolutionary constraints are first presented in order to appreciate the rules for the development of the brain and its underlying circuits. The organization of sensory pathways, with their parallel and cross-modal circuits, is also examined. Other features of brain networks, often considered as imposing constraints on the development of underlying circuitry, are also discussed and their effect on the complexity of the mouse and primate brain are inspected. In this review, we discuss the common features of cortical circuits in mice and primates and see how these can be useful in understanding visual processing in these animals.

Adaptive learning in a compartmental model of visual cortex-how feedback enables stable category learning and refinement

The categorization of real world objects is often reflected in the similarity of their visual appearances. Such categories of objects do not necessarily form disjunct sets of objects, neither semantically nor visually. The relationship between categories can often be described in terms of a hierarchical structure. For instance, tigers and leopards build two separate mammalian categories, both of which are subcategories of the category Felidae. In the last decades, the unsupervised learning of categories of visual input stimuli has been addressed by numerous approaches in machine learning as well as in computational neuroscience. However, the question of what kind of mechanisms might be involved in the process of subcategory learning, or category refinement, remains a topic of active investigation. We propose a recurrent computational network architecture for the unsupervised learning of categorial and subcategorial visual input representations. During learning, the connection strengths of bottom-up weights from input to higher-level category representations are adapted according to the input activity distribution. In a similar manner, top-down weights learn to encode the characteristics of a specific stimulus category. Feedforward and feedback learning in combination realize an associative memory mechanism, enabling the selective top-down propagation of a category's feedback weight distribution. We suggest that the difference between the expected input encoded in the projective field of a category node and the current input pattern controls the amplification of feedforward-driven representations. Large enough differences trigger the recruitment of new representational resources and the establishment of additional (sub-) category representations. We demonstrate the temporal evolution of such learning and show how the proposed combination of an associative memory with a modulatory feedback integration successfully establishes category and subcategory representations.

NMDA spike/plateau potentials in dendrites of thalamocortical neurons

Dendritic NMDA spike/plateau potentials, first discovered in cortical pyramidal neurons, provide supralinear integration of synaptic inputs on thin and distal dendrites, thereby increasing the impact of these inputs on the soma. The more specific functional role of these potentials has been difficult to clarify, partly due to the complex circuitry of cortical neurons. Thalamocortical (TC) neurons in the dorsal lateral geniculate nucleus participate in simpler circuits. They receive their primary afferent input from retina and send their output to visual cortex. Cortex, in turn, regulates this output through massive feedback to distal dendrites of the TC neurons. The TC neurons can operate in two modes related to behavioral states: burst mode prevailing during sleep, when T-type calcium bursts largely disrupt the transfer of signals from retina to cortex, and tonic mode, which provides reliable transfer of retinal signals to cortex during wakefulness. We studied dendritic potentials in TC neurons with combined two-photon calcium imaging and whole-cell recording of responses to local dendritic glutamate iontophoresis in acute brain slices from mice. We found that NMDA spike/plateaus can be elicited locally at distal dendrites of TC neurons. We suggest that these dendritic potentials have important functions in the cortical regulation of thalamocortical transmission. NMDA spike/plateaus can induce shifts in the functional mode from burst to tonic by blockade of T-type calcium conductances. Moreover, in tonic mode, they can facilitate the transfer of retinal signals to cortex by depolarization of TC neurons.

The stimulus selectivity and connectivity of layer six principal cells reveals cortical microcircuits underlying visual processing

Sensory computations performed in the neocortex involve layer six (L6) cortico-cortical (CC) and cortico-thalamic (CT) signaling pathways. Developing an understanding of the physiological role of these circuits requires dissection of the functional specificity and connectivity of the underlying individual projection neurons. By combining whole-cell recording from identified L6 principal cells in the mouse primary visual cortex (V1) with modified rabies virus-based input mapping, we have determined the sensory response properties and upstream monosynaptic connectivity of cells mediating the CC or CT pathway. We show that CC-projecting cells encompass a broad spectrum of selectivity to stimulus orientation and are predominantly innervated by deep layer V1 neurons. In contrast, CT-projecting cells are ultrasparse firing, exquisitely tuned to orientation and direction information, and receive long-range input from higher cortical areas. This segregation in function and connectivity indicates that L6 microcircuits route specific contextual and stimulus-related information within and outside the cortical network.

•

L6 cortico-cortical neurons are broadly tuned to stimulus orientation

•

L6 cortico-thalamic neurons are sparse firing and highly tuned to stimulus orientation

•

L6 cortico-cortical neurons receive input from cells located mostly within V1

•

L6 cortico-thalamic neurons receive input from higher order cortical areas

Hierarchical representation of shapes in visual cortex-from localized features to figural shape segregation

Visual structures in the environment are segmented into image regions and those combined to a representation of surfaces and prototypical objects. Such a perceptual organization is performed by complex neural mechanisms in the visual cortex of primates. Multiple mutually connected areas in the ventral cortical pathway receive visual input and extract local form features that are subsequently grouped into increasingly complex, more meaningful image elements. Such a distributed network of processing must be capable to make accessible highly articulated changes in shape boundary as well as very subtle curvature changes that contribute to the perception of an object. We propose a recurrent computational network architecture that utilizes hierarchical distributed representations of shape features to encode surface and object boundary over different scales of resolution. Our model makes use of neural mechanisms that model the processing capabilities of early and intermediate stages in visual cortex, namely areas V1-V4 and IT. We suggest that multiple specialized component representations interact by feedforward hierarchical processing that is combined with feedback signals driven by representations generated at higher stages. Based on this, global configurational as well as local information is made available to distinguish changes in the object's contour. Once the outline of a shape has been established, contextual contour configurations are used to assign border ownership directions and thus achieve segregation of figure and ground. The model, thus, proposes how separate mechanisms contribute to distributed hierarchical cortical shape representation and combine with processes of figure-ground segregation. Our model is probed with a selection of stimuli to illustrate processing results at different processing stages. We especially highlight how modulatory feedback connections contribute to the processing of visual input at various stages in the processing hierarchy.

Treatment of amblyopia in the adult: insights from a new rodent model of visual perceptual learning

Amblyopia is the most common form of impairment of visual function affecting one eye, with a prevalence of about 1–5% of the total world population. Amblyopia usually derives from conditions of early functional imbalance between the two eyes, owing to anisometropia, strabismus, or congenital cataract, and results in a pronounced reduction of visual acuity and severe deficits in contrast sensitivity and stereopsis. It is widely accepted that, due to a lack of sufficient plasticity in the adult brain, amblyopia becomes untreatable after the closure of the critical period in the primary visual cortex. However, recent results obtained both in animal models and in clinical trials have challenged this view, unmasking a previously unsuspected potential for promoting recovery even in adulthood. In this context, non invasive procedures based on visual perceptual learning, i.e., the improvement in visual performance on a variety of simple visual tasks following practice, emerge as particularly promising to rescue discrimination abilities in adult amblyopic subjects. This review will survey recent work regarding the impact of visual perceptual learning on amblyopia, with a special focus on a new experimental model of perceptual learning in the amblyopic rat.

Modulatory effects of activation of metabotropic glutamate receptors on GABAergic circuits in the mouse cortex

Metabotropic glutamate receptors (mGluRs) have a ubiquitous distribution in the central nervous system and often serve to regulate the release of neurotransmitters. We have previously shown that activation of both presynaptic and postsynaptic mGluRs can affect the gain of glutamatergic inputs in both thalamus and cortex. In the present study, we sought to determine the effect of mGluR activation on GABAergic inputs in cortex. Using whole cell recordings in a mouse slice preparation of either primary visual or auditory cortex (V1 or A1), we tested the effects on mGluRs by applying various agonists to the slice. Two pathways were tested in each area: the GABAergic inputs in layers 2/3 activated from layer 4 and the GABAergic inputs in layer 4 activated from adjacent layer 4. In both of these pathways, we found that activation of mGluRs significantly reduced the amplitude of the evoked inhibitory postsynaptic currents. Because the effects were not blocked by the addition of GDPβS to the recording electrode, and because mGluR agonists did not affect responses to photostimulation of GABA in a low-Ca(2+) and high-Mg(2+) bathing solution, we concluded this reduction was due to activation of presynaptic mGluRs. Furthermore, using specific mGluR agonists, we found that group II mGluRs, but not group I mGluRs, were involved in these modulatory effects. Because similar results were found in both pathways in V1 and A1, a possible cortical pattern for these effects is suggested.