David M. Halliday

This chapter provides a comprehensive examination of the biologically plausible synaptic plasticity mechanisms fundamental to learning in Spiking Neural Networks (SNNs). Because SNNs, recognized as the third generation of neural network models, encode information through the precise temporal dynamics of discrete spikes, learning is intrinsically dependent upon the continuous modulation of synaptic weights and delays. We initially review the mechanics of chemical synaptic transmission and classical Hebbian plasticity, which are operationalized biologically through Long-Term Potentiation (LTP) and Long-Term Depression (LTD). The analysis subsequently advances to Spike-Timing Dependent Plasticity (STDP), a critical unsupervised mechanism wherein the precise relative timing of pre-synaptic and post-synaptic action potentials dictates both the magnitude and direction of synaptic weight modifications. To mitigate the inherent risk of runaway network excitation driven by these localized STDP rules, the chapter elucidates the essential regulatory role of homeostatic plasticity. Specifically, it details how activity-dependent synaptic scaling globally adjusts neuronal excitability to maintain stable, target firing rates without disrupting the relative weight distributions established by STDP. Finally, the integration of dopamine-modulated plasticity is examined, illustrating how global neuromodulatory reward signals act as reward-prediction errors to consolidate local STDP-induced changes. This mechanism bridges the gap between unsupervised Hebbian timing rules and complex, goal-oriented reinforcement learning. Collectively, these temporal and homeostatic principles establish a robust theoretical foundation for the development of more advanced learning algorithms and supervised mapping techniques in artificial networks.

Intramuscular high-frequency coherence is increased during visually guided treadmill walking as a consequence of increased supra-spinal input. The influence of walking speed on intramuscular coherence and its inter-trial reproducibility need to be established before adoption as a functional gait assessment tool in clinical settings. Here, fifteen healthy controls performed a normal and a target walking task on a treadmill at various speeds (0.3 m/s, 0.5 m/s, 0.9 m/s, and preferred) during two sessions. Intramuscular coherence was calculated between two surface EMG recordings sites of the Tibialis anterior muscle during the swing phase of walking. The results were averaged across low-frequency (5–14 Hz) and high-frequency (15–55 Hz) bands. The effect of speed, task, and time on mean coherence was assessed using three-way repeated measures ANOVA. Reliability and agreement were calculated with the intra-class correlation coefficient and Bland–Altman method, respectively. Intramuscular coherence during target walking was significantly higher than during normal walking across all walking speeds in the high-frequency band as obtained by the three-way repeated measures ANOVA. Interaction effects between task and speed were found for the low- and high-frequency bands, suggesting that task-dependent differences increase at higher walking speeds. Reliability of intramuscular coherence was moderate to excellent for most normal and target walking tasks in all frequency bands. This study confirms previous reports of increased intramuscular coherence during target walking, while providing first evidence for reproducibility and robustness of this measure as a requirement to investigate supra-spinal input.Trial registration Registry number/ClinicalTrials.gov Identifier: NCT03343132, date of registration 2017/11/17.

Key points: Gait‐related arm swing in humans supports efficient lower limb muscle activation, indicating a neural coupling between the upper and lower limbs during gait. Intermuscular coherence analyses of gait‐related electromyography from upper and lower limbs in 20 healthy participants identified significant coherence in alpha and beta/gamma bands indicating that upper and lower limbs share common subcortical and cortical drivers that coordinate the rhythmic four‐limb gait pattern. Additional directed connectivity analyses revealed that upper limb muscles drive and shape lower limb muscle activity during gait via subcortical and cortical pathways and to a lesser extent vice versa. The results provide a neural underpinning that arm swing may serve as an effective rehabilitation therapy concerning impaired gait in neurological diseases. Human gait benefits from arm swing, as it enhances efficient lower limb muscle activation in healthy participants as well as patients suffering from neurological impairment. The underlying neuronal mechanisms of such coupling between upper and lower limbs remain poorly understood. The aim of the present study was to examine this coupling by intermuscular coherence analysis during gait. Additionally, directed connectivity analysis of this coupling enabled assessment of whether gait‐related arm swing indeed drives lower limb muscles. To that end, electromyography recordings were obtained from four lower limb muscles and two upper limb muscles bilaterally, during gait, of 20 healthy participants (mean (SD) age 67 (6.8) years). Intermuscular coherence analysis revealed functional coupling between upper and lower limb muscles in the alpha and beta/gamma band during muscle specific periods of the gait cycle. These effects in the alpha and beta/gamma bands indicate involvement of subcortical and cortical sources, respectively, that commonly drive the rhythmic four‐limb gait pattern in an efficiently coordinated fashion. Directed connectivity analysis revealed that upper limb muscles drive and shape lower limb muscle activity during gait via subcortical and cortical pathways and to a lesser extent vice versa. This indicates that gait‐related arm swing reflects the recruitment of neuronal support for optimizing the cyclic movement pattern of the lower limbs. These findings thus provide a neural underpinning for arm swing to potentially serve as an effective rehabilitation therapy concerning impaired gait in neurological diseases.

Objective. Graphical networks and network metrics are widely used to understand and characterise brain networks and brain function. These methods can be applied to a range of electrophysiological data including electroencephalography, local field potential and single unit recordings. Functional networks are often constructed using pair-wise correlation between variables. The objective of this study is to demonstrate that functional networks can be more accurately estimated using partial correlation than with pair-wise correlation. Approach. We compared network metrics derived from unconditional and conditional graphical networks, obtained using coherence and multivariate partial coherence (MVPC), respectively. Graphical networks were constructed using coherence and MVPC estimates, and binary and weighted network metrics derived from these: node degree, path length, clustering coefficients and small-world index. Main results. Network metrics were applied to simulated and experimental single unit spike train data. Simulated data used a 10x10 grid of simulated cortical neurons with centre-surround connectivity. Conditional network metrics gave a more accurate representation of the known connectivity: Numbers of excitatory connections had range 3–11, unconditional binary node degree had range 6–80, conditional node degree had range 2–13. Experimental data used multi-electrode array recording with 19 single-units from left and right hippocampal brain areas in a rat model for epilepsy. Conditional network analysis showed similar trends to simulated data, with lower binary node degree and longer binary path lengths compared to unconditional networks. Significance. We conclude that conditional networks, where common dependencies are removed through partial coherence analysis, give a more accurate representation of the interactions in a graphical network model. These results have important implications for graphical network analyses of brain networks and suggest that functional networks should be derived using partial correlation, based on MVPC estimates, as opposed to the common approach of pair-wise correlation.

It is now known that astrocytes modulate the activity at the tripartite synapses where indirect signaling via the retrograde messengers, endocannabinoids, leads to a localized self-repairing capability. In this paper, a self-repairing spiking astrocyte neural network (SANN) is proposed to demonstrate a distributed self-repairing capability at the network level. The SANN uses a novel learning rule that combines the spike-timing-dependent plasticity (STDP) and Bienenstock, Cooper, and Munro (BCM) learning rules (hereafter referred to as the BSTDP rule). In this learning rule, the synaptic weight potentiation is not only driven by the temporal difference between the presynaptic and postsynaptic neuron firing times but also by the postsynaptic neuron activity. We will show in this paper that the BSTDP modulates the height of the plasticity window to establish an input-output mapping (in the learning phase) and also maintains this mapping (via self-repair) if synaptic pathways become dysfunctional. It is the functional dependence of postsynaptic neuron firing activity on the height of the plasticity window that underpins how the proposed SANN self-repairs on the fly. The SANN also uses the coupling between the tripartite synapses and ɣ-GABAergic interneurons. This interaction gives rise to a presynaptic neuron frequency filtering capability that serves to route information, represented as spike trains, to different neurons in the subsequent layers of the SANN. The proposed SANN follows a feedforward architecture with multiple interneuron pathways and astrocytes modulate synaptic activity at the hidden and output neuronal layers. The self-repairing capability will be demonstrated in a robotic obstacle avoidance application, and the simulation results will show that the SANN can maintain learned maneuvers at synaptic fault densities of up to 80% regardless of the fault locations.

Spiking Astrocyte-neuron Networks (SANNs) model the adaptive/repair feature of the human brain. They integrate astrocyte cells with spiking neurons to facilitate a distributed and fine-grained self-repair capability at the synapse level. SANNs are more complex with the addition of astrocyte cells and require longer simulation times, as they are dynamic over much longer time-scales than traditional neural networks. Therefore, dedicated FPGA accelerators offer reductions in simulation times. To support the acceleration of SANNs, the capability of fault injection to synapses and monitoring significant levels of neuron and astrocyte data for off-chip transmission to PC-based analysis, are required. This paper presents an FPGA-based monitoring platform (FMP) for injecting faults and capturing and analyzing data acquired from the SANN FPGA accelerator, Astrobyte. The FMP uses custom logic and a NIOS II based system to control fault injection and data monitoring on the FPGA. Results show accurate accelerated simulations of fault injection scenarios using FMP with speedups up to 65 times greater compared with equivalent Matlab implementations.

Much of the motor impairment associated with Parkinson’s disease is thought to arise from pathological activity in the networks formed by the basal ganglia (BG) and motor cortex. To evaluate several hypotheses proposed to explain the emergence of pathological oscillations in parkinsonism, we investigated changes to the directed connectivity in BG networks following dopamine depletion. We recorded local field potentials (LFPs) in the cortex and basal ganglia of rats rendered parkinsonian by injection of 6-hydroxydopamine (6-OHDA) and in dopamine-intact controls. We performed systematic analyses of the networks using a novel tool for estimation of directed interactions (nonparametric directionality, NPD). We used a “conditioned” version of the NPD analysis that reveals the dependence of the correlation between two signals on a third reference signal. We find evidence of the dopamine dependency of both low-beta (14–20 Hz) and high-beta/low-gamma (20–40 Hz) directed network interactions. Notably, 6-OHDA lesions were associated with enhancement of the cortical “hyperdirect” connection to the subthalamic nucleus (STN) and its feedback to the cortex and striatum. We find that pathological beta synchronization resulting from 6-OHDA lesioning is widely distributed across the network and cannot be located to any individual structure. Furthermore, we provide evidence that high-beta/gamma oscillations propagate through the striatum in a pathway that is independent of STN. Rhythms at high beta/gamma show susceptibility to conditioning that indicates a hierarchical organization compared with those at low beta. These results further inform our understanding of the substrates for pathological rhythms in salient brain networks in parkinsonism.NEW & NOTEWORTHY We present a novel analysis of electrophysiological recordings in the cortico-basal ganglia network with the aim of evaluating several hypotheses concerning the origins of abnormal brain rhythms associated with Parkinson’s disease. We present evidence for changes in the directed connections within the network following chronic dopamine depletion in rodents. These findings speak to the plausibility of a “short-circuiting” of the network that gives rise to the conditions from which pathological synchronization may arise.