Search

Spiking neural network (SNN) distinguish themselves from artificial neural network (ANN) because of their inherent temporal processing and spike-based computations, enabling a power-efficient implementation in neuromorphic hardware. In this study, we demonstrate that data processing with spiking neurons can be enhanced by co-learning the synaptic weights with two other biologically inspired neuronal features: (1) a set of parameters describing neuronal adaptation processes and (2) synaptic propagation delays. The former allows a spiking neuron to learn how to specifically react to incoming spikes based on its past. The trained adaptation parameters result in neuronal heterogeneity, which leads to a greater variety in available spike patterns and is also found in the brain. The latter enables to learn to explicitly correlate spike trains that are temporally distanced. Synaptic delays reflect the time an action potential requires to travel from one neuron to another. We show that each of the co-learned features separately leads to an improvement over the baseline SNN and that the combination of both leads to state-of-the-art SNN results on all speech recognition datasets investigated with a simple 2-hidden layer feed-forward network. Our SNN outperforms the benchmark ANN on the neuromorphic datasets (Spiking Heidelberg Digits and Spiking Speech Commands), even with fewer trainable parameters. On the 35-class Google Speech Commands dataset, our SNN also outperforms a GRU of similar size. Our study presents brain-inspired improvements in SNN that enable them to excel over an equivalent ANN of similar size on tasks with rich temporal dynamics. [ABSTRACT FROM AUTHOR]

Spiking neural network (SNN) distinguish themselves from artificial neural network (ANN) because of their inherent temporal processing and spike-based computations, enabling a power-efficient implementation in neuromorphic hardware. In this study, we demonstrate that data processing with spiking neurons can be enhanced by co-learning the synaptic weights with two other biologically inspired neuronal features: (1) a set of parameters describing neuronal adaptation processes and (2) synaptic propagation delays. The former allows a spiking neuron to learn how to specifically react to incoming spikes based on its past. The trained adaptation parameters result in neuronal heterogeneity, which leads to a greater variety in available spike patterns and is also found in the brain. The latter enables to learn to explicitly correlate spike trains that are temporally distanced. Synaptic delays reflect the time an action potential requires to travel from one neuron to another. We show that each of the co-learned features separately leads to an improvement over the baseline SNN and that the combination of both leads to state-of-the-art SNN results on all speech recognition datasets investigated with a simple 2-hidden layer feed-forward network. Our SNN outperforms the benchmark ANN on the neuromorphic datasets (Spiking Heidelberg Digits and Spiking Speech Commands), even with fewer trainable parameters. On the 35-class Google Speech Commands dataset, our SNN also outperforms a GRU of similar size. Our study presents brain-inspired improvements in SNN that enable them to excel over an equivalent ANN of similar size on tasks with rich temporal dynamics.

A model based on inhibitory coupling has been proposed to explain perceptual oscillations. This 'adapting reciprocal inhibition' model postulates that it is the strength of inhibitory coupling that determines the fate of competition between percepts. Here, we used an fMRI-based adaptation technique to reveal the influence of neighboring neuronal populations, such as reciprocal inhibition, in motion-selective hMT+/V5. If reciprocal inhibition exists in this region, the following predictions should hold: 1. stimulus-driven response would not simply decrease, as predicted by simple repetition-suppression of neuronal populations, but instead, increase due to the activity from adjacent populations; 2. perceptual decision involving competing representations, should reflect decreased reciprocal inhibition by adaptation; 3. neural activity for the competing percept should also later on increase upon adaptation. Our results confirm these three predictions, showing that a model of perceptual decision based on adapting reciprocal inhibition holds true. Finally, they also show that the net effect of the well-known repetition suppression phenomenon can be reversed by this mechanism.

The etiology of idiopathic pain conditions, such as Provoked vulvodynia (PV), is multifactorial. The efficiency of venlafaxine, serotonin-noradrenaline reuptake inhibitor (SNRIs) in modulating vulvar pain led to the hypothesis that PV might involve central mechanisms. Here, we investigate whether vulvar pain is associated with gene-expression changes in mood, stress and pain systems, including amygdala (Amg), medial prefrontal cortex (mPFC), and periaqueductal gray matter (PAG). Additionally, we examined the analgesic and anxiolytic effects of venlafaxine. We found that the development of chronic vulvar pain in an animal model of PV is associated by overexpression of genes related to neuronal-activity and neuroinflammation in the Amg, mPFC, and PAG. Additionally, changes in the expression of GABA and serotonin synthesis, and reuptake were noted in the Amg and mPFC. Unsurprisingly, anxiety-like behavior and emotional-disorder were observed in rats with chronic vulvar pain. Nevertheless, treatment with venlafaxine (37.5 mg/kg) for one month significantly improves the vulvar hypersensitivity, as well as reduces the anxiety level. More critically, the long-term gene expression adaptation in serotonin receptor and synthesis, GABA synthesis, neuroplasticity, and neuroinflammation in the Amg, mPFC, and PAG, were modulated by venlafaxine in rats with vulvar pain. Our findings suggest that vulvar hypersensitivity induced by inflammation might associated with gene expression changes in brain areas that are involved in mood, stress and pain regulation. These changes probably play a role in central sensitization, and anxiety. Strikingly, enhancing the activity of serotonin and noradrenaline via venlafaxine treatment in rats with vulvar pain induces analgesic and anxiolytic effects. • Provoked vulvodynia is an idiopathic pain condition characterized by vulvar pain. • Chronic vulvar pain (CVP) causes anxiety-like behavior and emotional-disorder in rats. • CVP associate with neuroinflammation in the mPFC, amygdala and PAG. • Venlafaxine (VLX) alleviates vulvar pain and anxiety in rats with CVP. • Gene-expression changes in 5HT, GABA synthesis and neuroinflammation modulated by VLX. [ABSTRACT FROM AUTHOR]

• fMRI-based adaptation, which has been developed as a tool to identify functional selectivity in the human brain, can also reveal the influence of neighboring neuronal populations. • Our data reveals neural evidence for a disinhibition effect as a result of the adaptation of adjacent populations, which is in line with the adapting reciprocal inhibition model. • Reciprocal inhibition can, thus, be tracked in the human brain using fMRI, adding to the understanding of human multistable perception and the neural coding of visual information. • Our results also provide a mechanism for reversal of repetition suppression effects. A model based on inhibitory coupling has been proposed to explain perceptual oscillations. This 'adapting reciprocal inhibition' model postulates that it is the strength of inhibitory coupling that determines the fate of competition between percepts. Here, we used an fMRI-based adaptation technique to reveal the influence of neighboring neuronal populations, such as reciprocal inhibition, in motion-selective hMT+/V5. If reciprocal inhibition exists in this region, the following predictions should hold: 1. stimulus-driven response would not simply decrease, as predicted by simple repetition-suppression of neuronal populations, but instead, increase due to the activity from adjacent populations; 2. perceptual decision involving competing representations, should reflect decreased reciprocal inhibition by adaptation; 3. neural activity for the competing percept should also later on increase upon adaptation. Our results confirm these three predictions, showing that a model of perceptual decision based on adapting reciprocal inhibition holds true. Finally, they also show that the net effect of the well-known repetition suppression phenomenon can be reversed by this mechanism. [ABSTRACT FROM AUTHOR]