Dual-mode dendritic devices enhanced neural network based on electrolyte gated transistors

As a fundamental component of biological neurons, dendrites have been proven to have crucial effects in neuronal activities. Single neurons with dendrite structures show high signal processing capability that is analogous to a multilayer perceptron, whereas oversimplified point neuron models are still prevalent in AI algorithms and neuromorphic systems and fundamentally limit their efficiency and functionality of the systems constructed. In this study, we propose a dual-mode dendritic device based on electrolyte gated transistor, which can be operated to generate both supralinear and sublinear current-voltage responses when receiving input voltage pulses. We propose and demonstrate that the dual-mode dendritic devices can be used as a dendritic processing block between weight matrices and output neurons so as to enhance the expression ability of the neural networks. A dual-mode dendrites-enhanced neural network is therefore constructed with only two trainable parameters in the second layer, thus achieving 1000× reduction in the amount of second layer parameter compared to multilayer perceptron. After training by back propagation, the network reaches 90.1% accuracy in MNIST handwritten digits classification, showing advantage of the present dual-mode dendritic devices in building highly efficient neuromorphic computing.

Diverse processing underlying frequency integration in midbrain neurons of barn owls

Emergent response properties of sensory neurons depend on circuit connectivity and somatodendritic processing. Neurons of the barn owl’s external nucleus of the inferior colliculus (ICx) display emergence of spatial selectivity. These neurons use interaural time difference (ITD) as a cue for the horizontal direction of sound sources. ITD is detected by upstream brainstem neurons with narrow frequency tuning, resulting in spatially ambiguous responses. This spatial ambiguity is resolved by ICx neurons integrating inputs over frequency, a relevant processing in sound localization across species. Previous models have predicted that ICx neurons function as point neurons that linearly integrate inputs across frequency. However, the complex dendritic trees and spines of ICx neurons raises the question of whether this prediction is accurate. Data from in vivo intracellular recordings of ICx neurons were used to address this question. Results revealed diverse frequency integration properties, where some ICx neurons showed responses consistent with the point neuron hypothesis and others with nonlinear dendritic integration. Modeling showed that varied connectivity patterns and forms of dendritic processing may underlie observed ICx neurons’ frequency integration processing. These results corroborate the ability of neurons with complex dendritic trees to implement diverse linear and nonlinear integration of synaptic inputs, of relevance for adaptive coding and learning, and supporting a fundamental mechanism in sound localization.

Electrical synapse asymmetry results from, and masks, neuronal heterogeneity

Electrical synapses couple inhibitory neurons across the brain, underlying a variety of functions that are modifiable by activity. Despite recent advances, many basic functions and contributions of electrical synapses within neural circuitry remain underappreciated. Among these is the source and impact of electrical synapse asymmetry. Using multi-compartmental models of neurons coupled through dendritic electrical synapses, we investigated intrinsic factors that contribute to synaptic asymmetry and that result in modulation of spike time between coupled cells. We show that electrical synapse location along a dendrite, input resistance, internal dendritic resistance, or directional conduction of the electrical synapse itself each alter asymmetry as measured by coupling between cell somas. Conversely, true synapse asymmetry can be masked by each of these properties. Furthermore, we show that asymmetry alters the spiking timing and latency of coupled cells by up to tens of milliseconds, depending on direction of conduction or dendritic location of the electrical synapse. These simulations illustrate that causes of asymmetry are multifactorial, may not be apparent in somatic measurements of electrical coupling, influence dendritic processing, and produce a variety of outcomes on spike timing of coupled cells. Our findings highlight aspects of electrical synapses that should be considered in experimental demonstrations of coupling, and when assembling networks containing electrical synapses.

Are Artificial Dendrites Useful in Neuro-Evolution?

The significant role of dendritic processing within neuronal networks has become increasingly clear. This letter explores the effects of including a simple dendrite-inspired mechanism into neuro-evolution. The phenomenon of separate dendrite activation thresholds on connections is allowed to emerge under an evolutionary process. It is shown how such processing can be positively selected for, particularly for connections between the hidden and output layers, and increases performance.

Cerebellar Golgi cell models predict dendritic processing and mechanisms of synaptic plasticity

The Golgi cells are the main inhibitory interneurons of the cerebellar granular layer. Although recent works have highlighted the complexity of their dendritic organization and synaptic inputs, the mechanisms through which these neurons integrate complex input patterns remained unknown. Here we have used 8 detailed morphological reconstructions to develop multicompartmental models of Golgi cells, in which Na, Ca, and K channels were distributed along dendrites, soma, axonal initial segment and axon. The models faithfully reproduced a rich pattern of electrophysiological and pharmacological properties and predicted the operating mechanisms of these neurons. Basal dendrites turned out to be more tightly electrically coupled to the axon initial segment than apical dendrites. During synaptic transmission, parallel fibers caused slow Ca-dependent depolarizations in apical dendrites that boosted the axon initial segment encoder and Na-spike backpropagation into basal dendrites, while inhibitory synapses effectively shunted backpropagating currents. This oriented dendritic processing set up a coincidence detector controlling voltage-dependent NMDA receptor unblock in basal dendrites, which, by regulating local calcium influx, may provide the basis for spike-timing dependent plasticity anticipated by theory.

Cerebellar Golgi cell models predict dendritic processing and mechanisms of synaptic plasticity

The Golgi cells are the main inhibitory interneurons of the cerebellar granular layer. Although recent works have highlighted the complexity of their dendritic organization and synaptic inputs, the mechanisms through which these neurons integrate complex input patterns remained unknown. Here we have used 8 detailed morphological reconstructions to develop multicompartmental models of Golgi cells, in which Na, Ca, and K channels were distributed along dendrites, soma, axonal initial segment and axon. The models faithfully reproduced a rich pattern of electrophysiological and pharmacological properties and predicted the operating mechanisms of these neurons. Basal dendrites turned out to be more tightly electrically coupled to the axon initial segment than apical dendrites. During synaptic transmission, parallel fibers caused slow Ca-dependent depolarizations in apical dendrites that boosted the axon initial segment encoder and Na-spike backpropagation into basal dendrites, while inhibitory synapses effectively shunted backpropagating currents. This oriented dendritic processing set up a coincidence detector controlling voltage-dependent NMDA receptor unblock in basal dendrites, which, by regulating local calcium influx, may provide the basis for spike-timing dependent plasticity anticipated by theory.Author SummaryThe Golgi cells are the main inhibitory interneurons of the cerebellum granular layer and play a fundamental role in controlling cerebellar processing. However, it was unclear how spikes are processed in the dendrites by specific sets of ionic channels and how they might contribute to integrate synaptic inputs and plasticity. Here we have developed detailed multicompartmental models of Golgi cells that faithfully reproduced a large set of experimental findings and revealed the nature of signal interchange between dendrites and axo-somatic compartments. A main prediction of the models is that synaptic activation of apical dendrites can effectively trigger spike generation in the axonal initial segment followed by rapid spike backpropagation into basal dendrites. Here, incoming mossy fiber inputs and backpropagating spikes regulate the voltage-dependent unblock of NMDA channels and the induction of spike timing-dependent plasticity (STDP). STDP, which was predicted by theory, may therefore be controlled by contextual information provided by parallel fibers and integrated in apical dendrites, supporting the view that spike timing is fundamental to control synaptic plasticity at the cerebellar input stage.