Sparse Firing Research Articles

Spiking representation learning for associative memories.

Networks of interconnected neurons communicating through spiking signals offer the bedrock of neural computations. Our brain's spiking neural networks have the computational capacity to achieve complex pattern recognition and cognitive functions effortlessly. However, solving real-world problems with artificial spiking neural networks (SNNs) has proved to be difficult for a variety of reasons. Crucially, scaling SNNs to large networks and processing large-scale real-world datasets have been challenging, especially when compared to their non-spiking deep learning counterparts. The critical operation that is needed of SNNs is the ability to learn distributed representations from data and use these representations for perceptual, cognitive and memory operations. In this work, we introduce a novel SNN that performs unsupervised representation learning and associative memory operations leveraging Hebbian synaptic and activity-dependent structural plasticity coupled with neuron-units modelled as Poisson spike generators with sparse firing (~1 Hz mean and ~100 Hz maximum firing rate). Crucially, the architecture of our model derives from the neocortical columnar organization and combines feedforward projections for learning hidden representations and recurrent projections for forming associative memories. We evaluated the model on properties relevant for attractor-based associative memories such as pattern completion, perceptual rivalry, distortion resistance, and prototype extraction.

Sparse Firing in a Hybrid Central Pattern Generator for Spinal Motor Circuits.

Central pattern generators are circuits generating rhythmic movements, such as walking. The majority of existing computational models of these circuits produce antagonistic output where all neurons within a population spike with a broad burst at about the same neuronal phase with respect to network output. However, experimental recordings reveal that many neurons within these circuits fire sparsely, sometimes as rarely as once within a cycle. Here we address the sparse neuronal firing and develop a model to replicate the behavior of individual neurons within rhythm-generating populations to increase biological plausibility and facilitate new insights into the underlying mechanisms of rhythm generation. The developed network architecture is able to produce sparse firing of individual neurons, creating a novel implementation for exploring the contribution of network architecture on rhythmic output. Furthermore, the introduction of sparse firing of individual neurons within the rhythm-generating circuits is one of the factors that allows for a broad neuronal phase representation of firing at the population level. This moves the model toward recent experimental findings of evenly distributed neuronal firing across phases among individual spinal neurons. The network is tested by methodically iterating select parameters to gain an understanding of how connectivity and the interplay of excitation and inhibition influence the output. This knowledge can be applied in future studies to implement a biologically plausible rhythm-generating circuit for testing biological hypotheses.

Synapses, predictions, and prediction errors: A neocortical computational study of MDD using the temporal memory algorithm of HTM.

Synapses and spines play a significant role in major depressive disorder (MDD) pathophysiology, recently highlighted by the rapid antidepressant effect of ketamine and psilocybin. According to the Bayesian brain and interoception perspectives, MDD is formalized as being stuck in affective states constantly predicting negative energy balance. To understand how spines and synapses relate to the predictive function of the neocortex and thus to symptoms, we used the temporal memory (TM), an unsupervised machine-learning algorithm. TM models a single neocortical layer, learns in real-time, and extracts and predicts temporal sequences. TM exhibits neocortical biological features such as sparse firing and continuous online learning using local Hebbian-learning rules. We trained a TM model on random sequences of upper-case alphabetical letters, representing sequences of affective states. To model depression, we progressively destroyed synapses in the TM model and examined how that affected the predictive capacity of the network. We found that the number of predictions decreased non-linearly. Destroying 50% of the synapses slightly reduced the number of predictions, followed by a marked drop with further destruction. However, reducing the synapses by 25% distinctly dropped the confidence in the predictions. Therefore, even though the network was making accurate predictions, the network was no longer confident about these predictions. These findings explain how interoceptive cortices could be stuck in limited affective states with high prediction error. Connecting ketamine and psilocybin's proposed mechanism of action to depression pathophysiology, the growth of new synapses would allow representing more futuristic predictions with higher confidence. To our knowledge, this is the first study to use the TM model to connect changes happening at synaptic levels to the Bayesian formulation of psychiatric symptomatology. Linking neurobiological abnormalities to symptoms will allow us to understand the mechanisms of treatments and possibly, develop new ones.

A Fast Homeostatic Inhibitory Plasticity Rule Circuit with a Memristive Synapse

Learning from the robust mechanism of the biological nervous system is critical for creating reliable neuromorphic hardware. The homeostatic inhibition plasticity rule is a robust biological mechanism to balance Hebbian plasticity and resist external environmental disturbances and local damage. It plays an essential role in maintaining the homeostatic sparse firing patterns of the nervous system. This paper imitates this mechanism and provides a fast homeostatic inhibitory plasticity rule circuit with a memristive synapse. Firstly, the design method and principle of the circuit are demonstrated. Secondly, the function of the circuit was verified in PSpice© using a commercial Knowm memristor as a synapse. The PSpice© simulation results show that the circuit can achieve a weight update curve similar to the biological homeostatic inhibitory plasticity rule, and the time scale of the circuit is improved by a factor of 1000 compared to that of the biological nervous system. Furthermore, the circuit has wide applicability due to the tunable qualities of the homeostatic learning window, scaling factor, and homeostatic factor. This study provides new opportunities for building fast and reliable neuromorphic hardware.

State-Dependent Modulation of Activity in Distinct Layer 6 Corticothalamic Neurons in Barrel Cortex of Awake Mice.

Layer 6 corticothalamic (L6 CT) neurons are in a strategic position to control sensory input to the neocortex, yet we understand very little about their functions. Apart from studying their anatomic, physiological, and synaptic properties, most recent efforts have focused on the activity-dependent influences CT cells can exert on thalamic and cortical neurons through causal optogenetic manipulations. However, few studies have attempted to study them during behavior. To address this gap, we performed juxtacellular recordings from optogenetically identified CT neurons in whisker-related primary somatosensory cortex (wS1) of awake, head-fixed mice (either sex) free to rest quietly or self-initiate bouts of whisking and locomotion. We found a rich diversity of response profiles exhibited by CT cells. Their spiking patterns were either modulated by whisking-related behavior (∼28%) or not (∼72%). Whisking-responsive neurons exhibited both increases (activated-type) and decreases in firing rates (suppressed-type) that aligned with whisking onset better than locomotion. We also encountered responsive neurons with preceding modulations in firing rate before whisking onset. Overall, whisking better explained these changes in rates than overall changes in arousal. Whisking-unresponsive CT cells were generally quiet, with many having low spontaneous firing rates (sparse-type) and others being completely silent (silent-type). Remarkably, the sparse firing CT population preferentially spiked at the state transition point when pupil diameter constricted, and the mouse entered quiet wakefulness. Thus, our results demonstrate that L6 CT cells in wS1 show diverse spiking patterns, perhaps subserving distinct functional roles related to precisely timed responses during complex behaviors and transitions between discrete waking states.SIGNIFICANCE STATEMENT Layer 6 corticothalamic neurons provide a massive input to the sensory thalamus and local connectivity within cortex, but their role in thalamocortical processing remains unclear because of difficulty accessing and isolating their activity. Although several recent optogenetic studies reveal that the net influence of corticothalamic actions, suppression versus enhancement, depends critically on the rate these neurons fire, the factors that influence their spiking are poorly understood, particularly during wakefulness. Using the well-established Ntsr1-Cre line to target this elusive population in the whisker somatosensory cortex of awake mice, we found that corticothalamic neurons show diverse state-related responses and modulations in firing rate. These results suggest separate corticothalamic populations can differentially influence thalamocortical excitability during rapid state transitions in awake, behaving animals.

Role of Interaction Delays in the Synchronization of Inhibitory Networks.

Neural oscillations provide a means for efficient and flexible communication among different brain areas. Understanding the mechanisms of the generation of brain oscillations is crucial to determine principles of communication and information transfer in the brain circuits. It is well known that the inhibitory neurons play a major role in the generation of oscillations in the gamma range, in pure inhibitory networks, or in the networks composed of excitatory and inhibitory neurons. In this study, we explore the impact of different parameters and, in particular, the delay in the transmission of the signals between the neurons, on the dynamics of inhibitory networks. We show that increasing delay in a reasonable range increases the synchrony and stabilizes the oscillations. Unstable gamma oscillations characterized by a highly variable amplitude of oscillations can be observed in an intermediate range of delays. We show that in this range of delays, other experimentally observed phenomena such as sparse firing, variable amplitude and period, and the correlation between the instantaneous amplitude and period could be observed. The results broaden our understanding of the mechanism of the generation of the gamma oscillations in the inhibitory networks, known as the ING (interneuron-gamma) mechanism.

A Long Short-Term Memory for AI Applications in Spike-based Neuromorphic Hardware

Spike-based neuromorphic hardware holds promise for more energy-efficient implementations of deep neural networks (DNNs) than standard hardware such as GPUs. But this requires us to understand how DNNs can be emulated in an event-based sparse firing regime, as otherwise the energy advantage is lost. In particular, DNNs that solve sequence processing tasks typically employ long short-term memory units that are hard to emulate with few spikes. We show that a facet of many biological neurons, slow after-hyperpolarizing currents after each spike, provides an efficient solution. After-hyperpolarizing currents can easily be implemented in neuromorphic hardware that supports multi-compartment neuron models, such as Intel’s Loihi chip. Filter approximation theory explains why after-hyperpolarizing neurons can emulate the function of long short-term memory units. This yields a highly energy-efficient approach to time-series classification. Furthermore, it provides the basis for an energy-efficient implementation of an important class of large DNNs that extract relations between words and sentences in order to answer questions about the text. Deep learning could be less energy intensive when implemented on spike-based neuromorphic chips. An approach inspired by a characteristic feature of biological neurons, the presence of slowly changing internal currents, is developed to emulate long short-term memory units in a sparse spiking regime for neuromorphic implementation.

Network Model With Reduced Metabolic Rate Predicts Spatial Synchrony of Neuronal Activity.

In a cerebral hypometabolic state, cortical neurons exhibit slow synchronous oscillatory activity with sparse firing. How such a synchronization spatially organizes as the cerebral metabolic rate decreases have not been systemically investigated. We developed a network model of leaky integrate-and-fire neurons with an additional dependency on ATP dynamics. Neurons were scattered in a 2D space, and their population activity patterns at varying ATP levels were simulated. The model predicted a decrease in firing activity as the ATP production rate was lowered. Under hypometabolic conditions, an oscillatory firing pattern, that is, an ON-OFF cycle arose through a failure of sustainable firing due to reduced excitatory positive feedback and rebound firing after the slow recovery of ATP concentration. The firing rate oscillation of distant neurons developed at first asynchronously that changed into burst suppression and global synchronization as ATP production further decreased. These changes resembled the experimental data obtained from anesthetized rats, as an example of a metabolically suppressed brain. Together, this study substantiates a novel biophysical mechanism of neuronal network synchronization under limited energy supply conditions.

Sparsification of AP firing in adult-born hippocampal granule cells via voltage-dependent α5-GABAA receptors

GABA can depolarize immature neurons close to the action potential (AP) threshold in development and adult neurogenesis. Nevertheless, GABAergic synapses effectively inhibit AP firing in newborn granule cells of the adult hippocampus as early as two weeks post-mitosis. The underlying mechanisms are largely unclear. Here, we analyze GABAergic inputs in newborn hippocampal granule cells mediated by soma-targeting parvalbumin and dendrite-targeting somatostatin interneurons. Surprisingly, both interneuron subtypes activate α5-subunit-containing GABAA receptors (α5-GABAARs) in young neurons, showing a nonlinear voltage dependence with increasing conductance around the AP threshold. By contrast, in mature cells, parvalbumin interneurons mediate linear GABAergic synaptic currents lacking α5-subunits, while somatostatin interneurons continue to target nonlinear α5-GABAARs. Computational modeling shows that the voltage-dependent amplification of α5-GABAAR opening in young neurons is crucial for inhibition of AP firing to generate balanced and sparse firing activity, even with depolarized GABA reversal potential.

Stimulus Feature-Specific Control of Layer 2/3 Subthreshold Whisker Responses by Layer 4 in the Mouse Primary Somatosensory Cortex.

In the barrel field of the rodent primary somatosensory cortex (S1bf), excitatory cells in layer 2/3 (L2/3) display sparse firing but reliable subthreshold response during whisker stimulation. Subthreshold responses encode specific features of the sensory stimulus, for example, the direction of whisker deflection. According to the canonical model for the flow of sensory information across cortical layers, activity in L2/3 is driven by layer 4 (L4). However, L2/3 cells receive excitatory inputs from other regions, raising the possibility that L4 partially drives L2/3 during whisker stimulation. To test this hypothesis, we combined patch-clamp recordings from L2/3 pyramidal neurons in S1bf with selective optogenetic inhibition of L4 during passive whisker stimulation in both anesthetized and awake head-restrained mice. We found that L4 optogenetic inhibition did not abolish the subthreshold whisker-evoked response nor it affected spontaneous membrane potential fluctuations of L2/3 neurons. However, L4 optogenetic inhibition decreased L2/3 subthreshold responses to whisker deflections in the preferred direction, and it increased L2/3 responses to stimuli in the nonpreferred direction, leading to a change in the direction tuning. Our results contribute to reveal the circuit mechanisms underlying the processing of sensory information in the rodent S1bf.

Kv4.1, a Key Ion Channel For Low Frequency Firing of Dentate Granule Cells, Is Crucial for Pattern Separation.

The dentate gyrus (DG) in the hippocampus may play key roles in remembering distinct episodes through pattern separation, which may be subserved by the sparse firing properties of granule cells (GCs) in the DG. Low intrinsic excitability is characteristic of mature GCs, but ion channel mechanisms are not fully understood. Here, we investigated ionic channel mechanisms for firing frequency regulation in hippocampal GCs using male and female mice, and identified Kv4.1 as a key player. Immunofluorescence analysis showed that Kv4.1 was preferentially expressed in the DG, and its expression level determined by Western blot analysis was higher at 8-week than 3-week-old mice, suggesting a developmental regulation of Kv4.1 expression. With respect to firing frequency, GCs are categorized into two distinctive groups: low-frequency (LF) and high-frequency (HF) firing GCs. Input resistance (Rin) of most LF-GCs is lower than 200 MΩ, suggesting that LF-GCs are fully mature GCs. Kv4.1 channel inhibition by intracellular perfusion of Kv4.1 antibody increased firing rates and gain of the input-output relationship selectively in LF-GCs with no significant effect on resting membrane potential and Rin, but had no effect in HF-GCs. Importantly, mature GCs from mice depleted of Kv4.1 transcripts in the DG showed increased firing frequency, and these mice showed an impairment in contextual discrimination task. Our findings suggest that Kv4.1 expression occurring at late stage of GC maturation is essential for low excitability of DG networks and thereby contributes to pattern separation.SIGNIFICANCE STATEMENT The sparse activity of dentate granule cells (GCs), which is essential for pattern separation, is supported by high inhibitory inputs and low intrinsic excitability of GCs. Low excitability of GCs is thought to be attributable to a high K+ conductance at resting membrane potentials, but this study identifies Kv4.1, a depolarization-activated K+ channel, as a key ion channel that regulates firing of GCs without affecting resting membrane potentials. Kv4.1 expression is developmentally regulated and Kv4.1 currents are detected only in mature GCs that show low-frequency firing, but not in less mature high-frequency firing GCs. Furthermore, mice depleted of Kv4.1 transcripts in the dentate gyrus show impaired pattern separation, suggesting that Kv4.1 is crucial for sparse coding and pattern separation.

Measuring Sharp Waves and Oscillatory Population Activity With the Genetically Encoded Calcium Indicator GCaMP6f.

GCaMP6f is among the most widely used genetically encoded calcium indicators for monitoring neuronal activity. Applications are at both the cellular and population levels. Here, we explore two important and under-explored issues. First, we have tested if GCaMP6f is sensitive enough for the detection of population activity with sparse firing, similar to the sensitivity of the local field potential (LFP). Second, we have tested if GCaMP6f is fast enough for the detection of fast network oscillations critical for the encoding and consolidation of memory. We have focused this study on the activity of the hippocampal network including sharp waves (SWs), carbachol-induced theta oscillations, and interictal-like spikes. We compare simultaneous LFP and optical GCaMP6f fluorescent recordings in Thy1-GCaMP6f mouse hippocampal slices. We observe that SWs produce a clear population GCaMP6f signal above noise with an average magnitude of 0.3% ΔF/F. This population signal is highly correlated with the LFP, albeit with a delay of 40.3 ms (SD 10.8 ms). The population GCaMP6f signal follows the LFP evoked by 20 Hz stimulation with high fidelity, while electrically evoked oscillations up to 40 Hz were detectable with reduced amplitude. GCaMP6f and LFP signals showed a large amplitude discrepancy. The amplitude of GCaMP6f fluorescence increased by a factor of 28.9 (SD 13.5) between spontaneous SWs and carbachol-induced theta bursts, while the LFP amplitude increased by a factor of 2.4 (SD 1.0). Our results suggest that GCaMP6f is a useful tool for applications commonly considered beyond the scope of genetically encoded calcium indicators. In particular, population GCaMP6f signals are sensitive enough for detecting synchronous network events with sparse firing and sub-threshold activity, as well as asynchronous events with only a nominal LFP. In addition, population GCaMP6f signals are fast enough for monitoring theta and beta oscillations (<25 Hz). Faster calcium indicators (e.g., GCaMP7) will further improve the frequency response for the detection of gamma band oscillations. The advantage of population optical over LFP recordings are that they are non-contact and free from stimulation artifacts. These features may be particularly useful for high-throughput recordings and applications sensitive to stimulus artifact, such as monitoring responses during continuous stimulation.

Diverse tuning underlies sparse activity in layer 2/3 vibrissal cortex of awake mice.

Sparse population activity is a common feature observed across cortical areas, yet the implications for sensory coding are not clear. We recorded single neuron activity in the vibrissal somatosensory cortex of awake head-fixed mice using the cell-attached technique. Unlike the anaesthetised condition, in awake mice a high-velocity, piezo-controlled whisker deflection excited only a small fraction of neurons. Manual probing of whiskers revealed that the majority of these silent neurons could be activated by specific forms of whisker-object contact. Our results suggest that sparse coding in vibrissal cortex may be due to high dimensionality of the stimulus space and narrow tuning of individual neurons. It is widely reported that superficial layers of the somatosensory cortex exhibit sparse firing. This sparseness could reflect weak feedforward sensory inputs that are not sufficient to generate action potentials in these layers. Alternatively, sparseness might reflect tuning to unknown or higher-level complex features that are not fully explored in the stimulus space. Here, we examined these hypotheses by applying a range of vibrotactile and manual vibrissal stimuli in awake, head-fixed mice while performing loose-seal cell-attached recordings from the vibrissal primary somatosensory (vS1) cortex. A high-velocity stimulus delivered by a piezo-electric actuator evoked activity in a small fraction of regular spiking supragranular neurons (23%) in the awake condition. However, a majority of the supragranular regular spiking neurons (84%) were driven by manual stimulation of whiskers. Our results suggest that most neurons in the superficial layers of vS1 cortex contribute to coding in the awake condition when neurons may encounter their preferred feature(s) during whisker-object interactions.

Firing rate models for gamma oscillations.

Gamma oscillations are readily observed in a variety of brain regions during both waking and sleeping states. Computational models of gamma oscillations typically involve simulations of large networks of synaptically coupled spiking units. These networks can exhibit strongly synchronized gamma behavior, whereby neurons fire in near synchrony on every cycle, or weakly modulated gamma behavior, corresponding to stochastic, sparse firing of the individual units on each cycle of the population gamma rhythm. These spiking models offer valuable biophysical descriptions of gamma oscillations; however, because they involve many individual neuronal units they are limited in their ability to communicate general network-level dynamics. Here we demonstrate that few-variable firing rate models with established synaptic timescales can account for both strongly synchronized and weakly modulated gamma oscillations. These models go beyond the classical formulations of rate models by including at least two dynamic variables per population: firing rate and synaptic activation. The models' flexibility to capture the broad range of gamma behavior depends directly on the timescales that represent recruitment of the excitatory and inhibitory firing rates. In particular, we find that weakly modulated gamma oscillations occur robustly when the recruitment timescale of inhibition is faster than that of excitation. We present our findings by using an extended Wilson-Cowan model and a rate model derived from a network of quadratic integrate-and-fire neurons. These biophysical rate models capture the range of weakly modulated and coherent gamma oscillations observed in spiking network models, while additionally allowing for greater tractability and systems analysis. NEW & NOTEWORTHY Here we develop simple and tractable models of gamma oscillations, a dynamic feature observed throughout much of the brain with significant correlates to behavior and cognitive performance in a variety of experimental contexts. Our models depend on only a few dynamic variables per population, but despite this they qualitatively capture features observed in previous biophysical models of gamma oscillations that involve many individual spiking units.

The Spectrum of Asynchronous Dynamics in Spiking Networks as a Model for the Diversity of Non-rhythmic Waking States in the Neocortex

SummaryThe awake cortex exhibits diverse non-rhythmic network states. However, how these states emerge and how each state impacts network function is unclear. Here, we demonstrate that model networks of spiking neurons with moderate recurrent interactions display a spectrum of non-rhythmic asynchronous dynamics based on the level of afferent excitation, from afferent input-dominated (AD) regimes, characterized by unbalanced synaptic currents and sparse firing, to recurrent input-dominated (RD) regimes, characterized by balanced synaptic currents and dense firing. The model predicted regime-specific relationships between different neural biophysical properties, which were all experimentally validated in the somatosensory cortex (S1) of awake mice. Moreover, AD regimes more precisely encoded spatiotemporal patterns of presynaptic activity, while RD regimes better encoded the strength of afferent inputs. These results provide a theoretical foundation for how recurrent neocortical circuits generate non-rhythmic waking states and how these different states modulate the processing of incoming information.

Synaptic mechanisms underlying the intense firing of neocortical layer 5B pyramidal neurons in response to cortico-cortical inputs

In the neocortex, large layer 5B pyramidal neurons implement a high-density firing code. In contrast, other subtypes of pyramidal neurons, including those in layer 2/3, are functionally characterized by their sparse firing rate. Here, we investigate the synaptic basis of this behavior by comparing the properties of the postsynaptic responses evoked by cortical inputs in layer 5B and layer 2/3 pyramidal neurons in vitro. We demonstrate that a major determinant of the larger responsiveness of layer 5B with respect to layer 2/3 pyramidal neurons is the different properties in their inhibitory postsynaptic currents (IPSCs): layer 5B pyramidal neurons have IPSCs of lower amplitude and the temporal delay between the excitatory and inhibitory synaptic components is also larger in these cells. Our data also suggest that this difference depends on the lower gain of the cortical response of layer 5 parvalbumin-positive fast-spiking (PV-FS) interneurons with respect to PV-FS cells from layer 2/3. We propose that, while superficial PV-FS interneurons are well suited to provide a powerful feed-forward inhibitory control of pyramidal neuron responses, layer 5 PV-FS interneurons are mainly engaged in a feedback inhibitory loop and only after a substantial recruitment of surrounding pyramidal cells do they respond to an external input.

Spatiotemporal Structure and Dynamics of Spontaneous Oscillatory Synchrony in the Vagal Complex.

Fundamental structure and dynamics of spontaneous neuronal activities without apparent peripheral inputs were analyzed in the vagal complex (VC), whose activities had been generally thought to be produced almost passively to peripheral cues. The analysis included the caudal nucleus of the tractus solitarius—a main gateway for viscerosensory peripheral afferents and involved dynamically and critically in cardiorespiratory brainstem networks. In the present study, a possibility of self-organized brain activity was addressed in the VC. While VC neurons exhibited sparse firing in anesthetized rats and in in vitro preparations, we identified peculiar features of the emergent electrical population activity: (1) Spontaneous neuronal activity, in most cases, comprised both respiration and cardiac cycle components. (2) Population potentials of polyphasic high amplitudes reaching several millivolts emerged in synchrony with the inspiratory phase of respiratory cycles and exhibited several other characteristic temporal dynamics. (3) The spatiotemporal dynamics of local field potentials (LFPs), recorded simultaneously over multiple sites, were characterized by a stochastic emergence of high-amplitude synchrony. By adjusting amplitude and frequency (phase) over both space and time, the traveling synchrony exhibited varied degrees of coherence and power with a fluctuating balance between mutual oscillators of respiratory and cardiac frequency ranges. Full-fledged large-scale oscillatory synchrony over a wide region of the VC emerged after achieving a maximal stable balance between the two oscillators. Distinct somatic (respiratory; ~1 Hz) and visceral (autonomic; ~5 Hz) oscillators seemed to exist and communicate co-operatively in the brainstem network. Fluctuating oscillatory coupling may reflect varied degrees of synchrony influenced by the varied amplitude and frequency of neuronal activity in the VC. Intranuclear micro-, intrabulbar meso-, and wide-ranging macro-circuits involving the VC are likely to form nested networks and strategically interact to maintain a malleable whole-body homeostasis. These two brainstem oscillators could orchestrate neuronal activities of the VC, and other neuronal groups, through a phase-phase coupling mechanism to perform specific physiological functions.

Dendritic spikes in hippocampal granule cells are necessary for long-term potentiation at the perforant path synapse.

Long-term potentiation (LTP) of synaptic responses is essential for hippocampal memory function. Perforant-path (PP) synapses on hippocampal granule cells (GCs) contribute to the formation of associative memories, which are considered the cellular correlates of memory engrams. However, the mechanisms of LTP at these synapses are not well understood. Due to sparse firing activity and the voltage attenuation in their dendrites, it remains unclear how associative LTP at distal synapses occurs. Here, we show that NMDA receptor-dependent LTP can be induced at PP-GC synapses without backpropagating action potentials (bAPs) in acute rat brain slices. Dendritic recordings reveal substantial attenuation of bAPs as well as local dendritic Na+ spike generation during PP-GC input. Inhibition of dendritic Na+ spikes impairs LTP induction at PP-GC synapse. These data suggest that dendritic spikes may constitute a key cellular mechanism for memory formation in the dentate gyrus.

Deep Learning Predicts Correlation between a Functional Signature of Higher Visual Areas and Sparse Firing of Neurons.

Visual information in the visual cortex is processed in a hierarchical manner. Recent studies show that higher visual areas, such as V2, V3, and V4, respond more vigorously to images with naturalistic higher-order statistics than to images lacking them. This property is a functional signature of higher areas, as it is much weaker or even absent in the primary visual cortex (V1). However, the mechanism underlying this signature remains elusive. We studied this problem using computational models. In several typical hierarchical visual models including the AlexNet, VggNet, and SHMAX, this signature was found to be prominent in higher layers but much weaker in lower layers. By changing both the model structure and experimental settings, we found that the signature strongly correlated with sparse firing of units in higher layers but not with any other factors, including model structure, training algorithm (supervised or unsupervised), receptive field size, and property of training stimuli. The results suggest an important role of sparse neuronal activity underlying this special feature of higher visual areas.

Combining Theory, Model, and Experiment to Explain How Intrinsic Theta Rhythms Are Generated in an In Vitro Whole Hippocampus Preparation without Oscillatory Inputs.

Scientists have observed local field potential theta rhythms (3–12 Hz) in the hippocampus for decades, but understanding the mechanisms underlying their generation is complicated by their diversity in pharmacological and frequency profiles. In addition, interactions with other brain structures and oscillatory drives to the hippocampus during distinct brain states has made it difficult to identify hippocampus-specific properties directly involved in theta generation. To overcome this, we develop cellular-based network models using a whole hippocampus in vitro preparation that spontaneously generates theta rhythms. Building on theoretical and computational analyses, we find that spike frequency adaptation and postinhibitory rebound constitute a basis for theta generation in large, minimally connected CA1 pyramidal (PYR) cell network models with fast-firing parvalbumin-positive (PV+) inhibitory cells. Sparse firing of PYR cells and large excitatory currents onto PV+ cells are present as in experiments. The particular theta frequency is more controlled by PYR-to-PV+ cell interactions rather than PV+-to-PYR cell interactions. We identify two scenarios by which theta rhythms can emerge, and they can be differentiated by the ratio of excitatory to inhibitory currents to PV+ cells, but not to PYR cells. Only one of the scenarios is consistent with data from the whole hippocampus preparation, which leads to the prediction that the connection probability from PV+ to PYR cells needs to be larger than from PYR to PV+ cells. Our models can serve as a platform on which to build and develop an understanding of in vivo theta generation.