Plasticity Rules Research Articles

A spiking neural program for sensorimotor control during foraging in flying insects

Foraging is a vital behavioral task for living organisms. Behavioral strategies and abstract mathematical models thereof have been described in detail for various species. To explore the link between underlying neural circuits and computational principles, we present how a biologically detailed neural circuit model of the insect mushroom body implements sensory processing, learning, and motor control. We focus on cast and surge strategies employed by flying insects when foraging within turbulent odor plumes. Using a spike-based plasticity rule, the model rapidly learns to associate individual olfactory sensory cues paired with food in a classical conditioning paradigm. We show that, without retraining, the system dynamically recalls memories to detect relevant cues in complex sensory scenes. Accumulation of this sensory evidence on short time scales generates cast-and-surge motor commands. Our generic systems approach predicts that population sparseness facilitates learning, while temporal sparseness is required for dynamic memory recall and precise behavioral control. Our work successfully combines biological computational principles with spike-based machine learning. It shows how knowledge transfer from static to arbitrary complex dynamic conditions can be achieved by foraging insects and may serve as inspiration for agent-based machine learning.

From statistical regularities in multisensory inputs to peripersonal space representation and body ownership: Insights from a neural network model.

Peripersonal space (PPS), the interface between the self and the environment, is represented by a network of multisensory neurons with visual (or auditory) receptive fields anchored to specific body parts, and tactile receptive fields covering the same body parts. Neurophysiological and behavioural features of hand PPS representation have been previously modelled through a neural network constituted by one multisensory population integrating tactile inputs with visual/auditory external stimuli. Reference frame transformations were not explicitly modelled, as stimuli were encoded in pre‐computed hand‐centred coordinates. Here we present a novel model, aiming to overcome this limitation by including a proprioceptive population encoding hand position. We confirmed behaviourally the plausibility of the proposed architecture, showing that visuo‐proprioceptive information is integrated to enhance tactile processing on the hand. Moreover, the network's connectivity was spontaneously tuned through a Hebbian‐like mechanism, under two minimal assumptions. First, the plasticity rule was designed to learn the statistical regularities of visual, proprioceptive and tactile inputs. Second, such statistical regularities were simply those imposed by the body structure. The network learned to integrate proprioceptive and visual stimuli, and to compute their hand‐centred coordinates to predict tactile stimulation. Through the same mechanism, the network reproduced behavioural correlates of manipulations implicated in subjective body ownership: the invisible and the rubber hand illusion. We thus propose that PPS representation and body ownership may emerge through a unified neurocomputational process; the integration of multisensory information consistently with a model of the body in the environment, learned from the natural statistics of sensory inputs.

Exploring Neuromodulation for Dynamic Learning.

A continual learning system requires the ability to dynamically adapt and generalize to new tasks with access to only a few samples. In the central nervous system, across species, it is observed that continual and dynamic behavior in learning is an active result of a mechanism known as neuromodulation. Therefore, in this work, neuromodulatory plasticity is embedded with dynamic learning architectures as a first step toward realizing power and area efficient few shot learning systems. An inbuilt modulatory unit regulates learning based on the context and internal state of the system. This renders the system an ability to self modify its weights. In one of the proposed architectures, ModNet, a modulatory layer is introduced in a random projection framework. ModNet's learning capabilities are enhanced by integrating attention along with compartmentalized plasticity mechanisms. Moreover, to explore modulatory mechanisms in conjunction with backpropagation in deeper networks, a modulatory trace learning rule is introduced. The proposed learning rule, uses a time dependent trace to modify the synaptic connections as a function of ongoing states and activations. The trace itself is updated via simple plasticity rules thus reducing the demand on resources. The proposed ModNet and learning rules demonstrate the ability to learn from few samples, train quickly, and perform few-shot image classification in a computationally efficient manner. The simple ModNet and the compartmentalized ModNet architecture learn benchmark image classification tasks in just 2 epochs. The network with modulatory trace achieves an average accuracy of 98.8%±1.16 on the omniglot dataset for five-way one-shot image classification task while requiring 20x fewer trainable parameters in comparison to other state of the art models.

Halide Perovskite Quantum Dots Photosensitized‐Amorphous Oxide Transistors for Multimodal Synapses

AbstractDeployment of novel artificial synapses serves as the crucial unit for building neuromorphic hardware to drive data‐intensive applications. Emulation of complex neural behavior through conventional Si‐based devices requires a large number of elements which increases fabrication complexity and brings challenges of connectivity. Hence, there is a need to investigate alternative material systems and device architectures for emulating richer neural behavior comprising of lesser elements. Herein, a thin‐film transistor‐like synaptic device using all‐inorganic cesium lead bromide (CsPbBr3) perovskite quantum dots (QDs) and amorphous indium gallium zinc oxide semiconductor active material is explored for brain‐inspired computing. The incorporation of CsPbBr3 QDs as a photosensitizer aids in realizing light‐dependent synaptic memory. Furthermore, type II heterostructure can serve as a basis for electro‐optical programming. The proposed artificial synapse demonstrates a materials combination that can decouple optical absorption and charge transport property and provides freedom to tune the spectral region. Harnessing the advantages of novel materials, the devices obey spike‐timing‐dependent plasticity rules, inculcate associative learning and linear nonvolatile blind updates. This architecture paves way for efficient building of neuromorphic hardware elements with facile tunability and tailorable plasticity.

Formability prediction using bifurcation criteria and GTN damage model

In this paper, four plastic instability criteria, which are based on the bifurcation theory, are coupled with the GTN damage model for the prediction of diffuse and localized necking. General bifurcation (GB) criterion and limit-point bifurcation (LPB) criterion are used to predict diffuse necking, while loss of ellipticity (LOE) criterion and loss of strong ellipticity (LOSE) criterion are used to predict localized necking. The resulting constitutive equations and instability criteria are implemented into the finite element code ABAQUS/Standard. The constitutive equations are formulated within the framework of large deformations and fully three-dimensional approach. Since the developed numerical tools have intended applications mainly for thin sheet metals; therefore, the plane-stress conditions are considered within the instability criteria. The present contribution focuses on the effect of destabilizing mechanisms, due to non-associative plasticity and non-normal plastic flow rule, on the prediction of forming limit diagrams (FLDs). Theoretical classification of the bifurcation criteria, in terms of their order of prediction of critical necking strains, is first presented. Then, several variants of the GTN model are combined with the bifurcation criteria for the prediction of FLDs for fictitious materials. It is shown that the hierarchical prediction order of the different instability criteria is consistent with the theoretical classification, for all the considered variants of the GTN model. More specifically, it is shown that the GB criterion provides a lower bound to all bifurcation criteria, in terms of necking prediction, while the LOE criterion represents an upper bound.

A spike-timing-dependent plasticity rule for dendritic spines

The structural organization of excitatory inputs supporting spike-timing-dependent plasticity (STDP) remains unknown. We performed a spine STDP protocol using two-photon (2P) glutamate uncaging (pre) paired with postsynaptic spikes (post) in layer 5 pyramidal neurons from juvenile mice. Here we report that pre-post pairings that trigger timing-dependent LTP (t-LTP) produce shrinkage of the activated spine neck and increase in synaptic strength; and post-pre pairings that trigger timing-dependent LTD (t-LTD) decrease synaptic strength without affecting spine shape. Furthermore, the induction of t-LTP with 2P glutamate uncaging in clustered spines (<5 μm apart) enhances LTP through a NMDA receptor-mediated spine calcium accumulation and actin polymerization-dependent neck shrinkage, whereas t-LTD was dependent on NMDA receptors and disrupted by the activation of clustered spines but recovered when separated by >40 μm. These results indicate that synaptic cooperativity disrupts t-LTD and extends the temporal window for the induction of t-LTP, leading to STDP only encompassing LTP.

Emergence of Stable Synaptic Clusters on Dendrites Through Synaptic Rewiring.

The connectivity structure of neuronal networks in cortex is highly dynamic. This ongoing cortical rewiring is assumed to serve important functions for learning and memory. We analyze in this article a model for the self-organization of synaptic inputs onto dendritic branches of pyramidal cells. The model combines a generic stochastic rewiring principle with a simple synaptic plasticity rule that depends on local dendritic activity. In computer simulations, we find that this synaptic rewiring model leads to synaptic clustering, that is, temporally correlated inputs become locally clustered on dendritic branches. This empirical finding is backed up by a theoretical analysis which shows that rewiring in our model favors network configurations with synaptic clustering. We propose that synaptic clustering plays an important role in the organization of computation and memory in cortical circuits: we find that synaptic clustering through the proposed rewiring mechanism can serve as a mechanism to protect memories from subsequent modifications on a medium time scale. Rewiring of synaptic connections onto specific dendritic branches may thus counteract the general problem of catastrophic forgetting in neural networks.

Extracellular GABA waves regulate coincidence detection in excitatory circuits.

Key points Rapid changes in neuronal network activity trigger widespread waves of extracellular GABA in hippocampal neuropil.Elevations of extracellular GABA narrow the coincidence detection window for excitatory inputs to CA1 pyramidal cells.GABA transporters control the effect of extracellular GABA on coincidence detection.Small changes in the kinetics of dendritic excitatory currents amplify when reaching the soma. Coincidence detection of excitatory inputs by principal neurons underpins the rules of signal integration and Hebbian plasticity in the brain. In the hippocampal circuitry, detection fidelity is thought to depend on the GABAergic synaptic input through a feedforward inhibitory circuit also involving the hyperpolarisation‐activated Ih current. However, afferent connections often bypass feedforward circuitry, suggesting that a different GABAergic mechanism might control coincidence detection in such cases. To test whether fluctuations in the extracellular GABA concentration [GABA] could play a regulatory role here, we use a GABA 'sniffer' patch in acute hippocampal slices of the rat and document strong dependence of [GABA] on network activity. We find that blocking GABAergic signalling strongly widens the coincidence detection window of direct excitatory inputs to pyramidal cells whereas increasing [GABA] through GABA uptake blockade shortens it. The underlying mechanism involves membrane‐shunting tonic GABAA receptor current; it does not have to rely on Ih but depends strongly on the neuronal GABA transporter GAT‐1. We use dendrite‐soma dual patch‐clamp recordings to show that the strong effect of membrane shunting on coincidence detection relies on nonlinear amplification of changes in the decay of dendritic synaptic currents when they reach the soma. Our results suggest that, by dynamically regulating extracellular GABA, brain network activity can optimise signal integration rules in local excitatory circuits.

SBSNN: Stochastic-Bits Enabled Binary Spiking Neural Network With On-Chip Learning for Energy Efficient Neuromorphic Computing at the Edge

In this work, we propose stochastic Binary Spiking Neural Network (sBSNN) composed of stochastic spiking neurons and binary synapses (stochastic only during training) that computes probabilistically with one-bit precision for power-efficient and memory-compressed neuromorphic computing. We present an energy-efficient implementation of the proposed sBSNN using ‘stochastic bit’ as the core computational primitive to realize the stochastic neurons and synapses, which are fabricated in 90nm CMOS process, to achieve efficient on-chip training and inference for image recognition tasks. The measured data shows that the ‘stochastic bit’ can be programmed to mimic spiking neurons, and stochastic Spike Timing Dependent Plasticity (or sSTDP) rule for training the binary synaptic weights without expensive random number generators. Our results indicate that the proposed sBSNN realization offers possibility of up to $32\times $ neuronal and synaptic memory compression compared to full precision (32-bit) SNN and energy efficiency of 89.49 TOPS/Watt for two-layer fully-connected SNN.

Efficient Resistive Switching and Spike Rate Dependent Plasticity in a New CuCrO2 Memristor for Plausible Neuromorphic Systems

In this article, we introduce a new class of p-type transparent conductive oxide (TCO) CuCrO2 (150 nm) heterogeneously integrated onto fluorine doped tin oxide (FTO)/glass for forming-free memristor-based neuromorphic applications. The fabricated Al/CuCrO2/FTO memristors demonstrate a reliable bipolar resistive switching with an ON/ OFF ratio of 1000. The retention of the device was found to be steady even beyond 106 s, which demonstrates its nonvolatility. The current–voltage ( ${I}$ – ${V}$ ) characteristics were fit to evaluate its transport properties and a band diagram was projected to have a better insight of the device operational principles. To validate the experimental observations, a new model has been developed, and the simulated ${I}$ – ${V}$ behavior was analogous to the experimental one. Efforts were then devoted to observe long-term potentiation (LTP) and long-term depression (LTD) utilizing identical but opposite pulses to evaluate the device’s efficacy for synaptic applications. The synaptic behavior was well controlled by the pulse (pulse amplitude and width) variations. The conductance change was found to be symmetric and then saturated, which reflects the popular biological Hebbian rules. Finally, a long-term synaptic modulation has been implemented by establishing the spike rate dependent plasticity (SRDP) rule, which is a part of spiking neural networks and advantageous to mimic the brain’s capability at low power. All the obtained experimental results were systematically corroborated by neural network simulation. Overall, our approach provides a new road map toward the development of TCO-based alternative memristors, which can be employed to mimic the synaptic plasticity for energy-efficient bioinspired neuromorphic systems and non-von Neumann computer architectures.

Nanoscale Room-Temperature Multilayer Skyrmionic Synapse for Deep Spiking Neural Networks

Magnetic skyrmions have attracted considerable interest, especially after their recent experimental demonstration at room temperature in multilayers. The robustness, nanoscale size and non-volatility of skyrmions have triggered a substantial amount of research on skyrmion-based low-power, ultra-dense nanocomputing and neuromorphic systems such as artificial synapses. Room-temperature operation is required to integrate skyrmionic synapses in practical future devices. Here, we numerically propose a nanoscale skyrmionic synapse composed of magnetic multilayers that enables room-temperature device operation tailored for optimal synaptic resolution. We demonstrate that when embedding such multilayer skyrmionic synapses in a simple spiking neural network (SNN) with unsupervised learning via the spike-timing-dependent plasticity rule, we can achieve only a 78% classification accuracy in the MNIST handwritten data set under realistic conditions. We propose that this performance can be significantly improved to about 98.61% by using a deep SNN with supervised learning. Our results illustrate that the proposed skyrmionic synapse can be a potential candidate for future energy-efficient neuromorphic edge computing.

Signs of memory in a plastic frustrated Kuramoto model of neurons

We study the effects brought about by adding frustration to the plastic Kuramoto model of neurons; i.e. we simulate for the first time the plastic Kuramoto–Sakaguchi model, in which the nonlocal coupling matrix for the phases is determined from the spike-timing-dependent plasticity rule. We find that frustration leads to the birhythmicity of the synchronous states. The multistability in the dynamics, with frustration as its crucial element, results in hysteresis, even when the learning rule is symmetric. Fine structure in the hysteresis loop reflects the complexity of the underlying energy landscape. A pacemaker is used to probe this landscape. As the neurons spontaneously self-organise, memory is formed by the configuration of synaptic strengths embodied in such synchronous states. We thus have a simple model of synaptic memory.

Synaptic Plasticity Dynamics for Deep Continuous Local Learning (DECOLLE).

A growing body of work underlines striking similarities between biological neural networks and recurrent, binary neural networks. A relatively smaller body of work, however, addresses the similarities between learning dynamics employed in deep artificial neural networks and synaptic plasticity in spiking neural networks. The challenge preventing this is largely caused by the discrepancy between the dynamical properties of synaptic plasticity and the requirements for gradient backpropagation. Learning algorithms that approximate gradient backpropagation using local error functions can overcome this challenge. Here, we introduce Deep Continuous Local Learning (DECOLLE), a spiking neural network equipped with local error functions for online learning with no memory overhead for computing gradients. DECOLLE is capable of learning deep spatio temporal representations from spikes relying solely on local information, making it compatible with neurobiology and neuromorphic hardware. Synaptic plasticity rules are derived systematically from user-defined cost functions and neural dynamics by leveraging existing autodifferentiation methods of machine learning frameworks. We benchmark our approach on the event-based neuromorphic dataset N-MNIST and DvsGesture, on which DECOLLE performs comparably to the state-of-the-art. DECOLLE networks provide continuously learning machines that are relevant to biology and supportive of event-based, low-power computer vision architectures matching the accuracies of conventional computers on tasks where temporal precision and speed are essential.

Rapid Recalibration of Peri-Personal Space: Psychophysical, Electrophysiological, and Neural Network Modeling Evidence.

Interactions between individuals and the environment occur within the peri-personal space (PPS). The encoding of this space plastically adapts to bodily constraints and stimuli features. However, these remapping effects have not been demonstrated on an adaptive time-scale, trial-to-trial. Here, we test this idea first via a visuo-tactile reaction time (RT) paradigm in augmented reality where participants are asked to respond as fast as possible to touch, as visual objects approach them. Results demonstrate that RTs to touch are facilitated as a function of visual proximity, and the sigmoidal function describing this facilitation shifts closer to the body if the immediately precedent trial had indexed a smaller visuo-tactile disparity. Next, we derive the electroencephalographic correlates of PPS and demonstrate that this multisensory measure is equally shaped by recent sensory history. Finally, we demonstrate that a validated neural network model of PPS is able to account for the present results via a simple Hebbian plasticity rule. The present findings suggest that PPS encoding remaps on a very rapid time-scale and, more generally, that it is sensitive to sensory history, a key feature for any process contextualizing subsequent incoming sensory information (e.g., a Bayesian prior).

T cell receptor repertoire features display universal rules for selection and plasticity in the functional CD4 T cell subsets

Abstract Helper T cell choice of functional subset may depend on the mode and strength of T cell receptor interaction with peptide-MHC complex on antigen presenting cells. However, the extent, complexity, and homogeneity of this influence of TCR structure on the selection into a functional subset, as well as further stability of acquired clonal programs remain poorly studied. Here we present the first analysis of deep sequenced T-cell receptor repertoires of the eight effector CD4 T cell subsets (Th1, Th2, Th2a, Th17, Th1-17, Th22, Treg, and Tfh) from peripheral blood of 5 healthy donors, revealing unexpectedly prominent, multivariate subset-specific differences that are highly reproducible across unrelated donors. Similar analysis performed for the sorted naive CD4 T cell subsets showed that specific TCR features of Tregs (short and strongly interacting CDR3 region) are rooted in thymic selection, while differences between other subsets accumulate later upon T cell functional engagement. We next provide the first deep analysis of the plasticity of human effector CD4 subsets ex vivo. We demonstrate high plasticity within Th17 and Th22 functional subsets and prominent clonal exchange of both Th17 and Th22 with Th2. Th1-17 subset shares clones with Th1 but not with the Th17 subset. Finally, we investigate relative publicity of effector CD4 subset repertoires across donors. We reveal high publicity of Treg and Tfh, and high privacy of Th22 and Th2a repertoires, which corresponds to the number of added N-nucleotides for these subsets, suggesting their earlier fetal and evolutionary origin. Altogether, we provide the first detailed picture of distinct repertoire features, plasticity, and publicity of helper CD4 T cell subsets.

Flipping the Switch: Homeostatic Tuning of Chandelier Synapses Follows Developmental Changes in GABA Polarity

Homeostatic plasticity rules are well described for glutamatergic synapses but less clear for GABAergic synapses. In this issue of Neuron, Pan-Vazquez etal. (2020) leverage the precise targeting of chandelier-to-pyramidal cell connectivity to understand how homeostatic plasticity regulates GABAergic synapses, showing that these synapses maintain homeostatic rules even as they flip from exciting to inhibiting pyramidal cells.

Heterosynaptic Plasticity Determines the Set Point for Cortical Excitatory-Inhibitory Balance

Excitation in neural circuits must be carefully controlled by inhibition to regulate information processing and network excitability. During development, cortical inhibitory and excitatory inputs are initially mismatched but become co-tuned or balanced with experience. However, little is known about how excitatory-inhibitory balance is defined at most synapses or about the mechanisms for establishing or maintaining this balance at specific set points. Here we show how coordinated long-term plasticity calibrates populations of excitatory-inhibitory inputs onto mouse auditory cortical pyramidal neurons. Pairing pre- and postsynaptic activity induced plasticity at paired inputs and different forms of heterosynaptic plasticity at the strongest unpaired synapses, which required minutes of activity and dendritic Ca2+ signaling to be computed. Theoretical analyses demonstrated how the relative rate of heterosynaptic plasticity could normalize and stabilize synaptic strengths to achieve any possible excitatory-inhibitory correlation. Thus, excitatory-inhibitory balance is dynamic and cell specific, determined by distinct plasticity rules across multiple excitatory and inhibitory synapses.

Impact of functional synapse clusters on neuronal response selectivity

Clustering of functionally similar synapses in dendrites is thought to affect neuronal input-output transformation by triggering local nonlinearities. However, neither the in vivo impact of synaptic clusters on somatic membrane potential (sVm), nor the rules of cluster formation are elucidated. We develop a computational approach to measure the effect of functional synaptic clusters on sVm response of biophysical model CA1 and L2/3 pyramidal neurons to in vivo-like inputs. We demonstrate that small synaptic clusters appearing with random connectivity do not influence sVm. With structured connectivity, ~10–20 synapses/cluster are optimal for clustering-based tuning via state-dependent mechanisms, but larger selectivity is achieved by 2-fold potentiation of the same synapses. We further show that without nonlinear amplification of the effect of random clusters, action potential-based, global plasticity rules cannot generate functional clustering. Our results suggest that clusters likely form via local synaptic interactions, and have to be moderately large to impact sVm responses.

Simulation of bicrystal deformation including grain boundary effects: Atomistic computations and crystal plasticity finite element analysis

Grain boundaries play a pivotal role in dictating the deformation behavior of crystalline materials. Modeling their effects within the framework of physically based crystal plasticity approaches demands a multiscale description of the underlying phenomena. This work puts-forth a two-scale atomistic to crystal plasticity approach for determining the deformation behavior of a ductile face-centered cubic material. The approach uses atomistic computations to quantify the activation energies for nucleation of partial dislocations from a grain boundary under tensile loading. To this end, embedded-atom method based atomistic simulations involving nudged elastic band method are utilized to compute stress-dependent activation parameters of the grain boundary. The extracted parameters are then used as input to the flow rule of crystal plasticity at higher length scale, which is based on transition state theory. At this scale, the grain boundaries are explicitly accounted for by assigning a finite thickness and are differentiated from their bulk counterparts by prescribing distinct flow parameters extracted from atomistic simulations. The predictive capabilities of the proposed methodology are then assessed by performing numerical simulations on uniaxial tensile behavior of Ni bicrystal and validating them from the published experiments and computations available in the literature. This research paradigm can be pushed forward by incorporating more complex grain boundary behaviors at higher length scales while preserving the richness provided by atomic scales.

Concurrent Thalamostriatal and Corticostriatal Spike-Timing-Dependent Plasticity and Heterosynaptic Interactions Shape Striatal Plasticity Map.

The striatum integrates inputs from the cortex and thalamus, which display concomitant or sequential activity. The striatum assists in forming memory, with acquisition of the behavioral repertoire being associated with corticostriatal (CS) plasticity. The literature has mainly focused on that CS plasticity, and little remains known about thalamostriatal (TS) plasticity rules or CS and TS plasticity interactions. We undertook here the study of these plasticity rules. We found bidirectional Hebbian and anti-Hebbian spike-timing-dependent plasticity (STDP) at the thalamic and cortical inputs, respectively, which were driving concurrent changes at the striatal synapses. Moreover, TS- and CS-STDP induced heterosynaptic plasticity. We developed a calcium-based mathematical model of the coupled TS and CS plasticity, and simulations predict complex changes in the CS and TS plasticity maps depending on the precise cortex-thalamus-striatum engram. These predictions were experimentally validated using triplet-based STDP stimulations, which revealed the significant remodeling of the CS-STDP map upon TS activity, which is notably the induction of the LTD areas in the CS-STDP for specific timing regimes. TS-STDP exerts a greater influence on CS plasticity than CS-STDP on TS plasticity. These findings highlight the major impact of precise timing in cortical and thalamic activity for the memory engram of striatal synapses.