Response Dynamics in an Olivocerebellar Spiking Neural Network With Non-linear Neuron Properties

Abstract

Sensorimotor signals are integrated and processed by the cerebellar circuit to predict accurate control of actions. In order to investigate how single neuron dynamics and geometrical modular connectivity affect cerebellar processing, we have built an olivocerebellar Spiking Neural Network (SNN) based on a novel simplification algorithm for single point models (Extended Generalized Leaky Integrate and Fire, EGLIF) capturing essential non-linear neuronal dynamics (e.g., pacemaking, bursting, adaptation, oscillation and resonance). EGLIF models specifically tuned for each neuron type were embedded into an olivocerebellar scaffold reproducing realistic spatial organization and physiological convergence and divergence ratios of connections. In order to emulate the circuit involved in an eye blink response to two associated stimuli, we modeled two adjacent olivocerebellar microcomplexes with a common mossy fiber input but different climbing fiber inputs (either on or off). EGLIF-SNN model simulations revealed the emergence of fundamental response properties in Purkinje cells (burst-pause) and deep nuclei cells (pause-burst) similar to those reported in vivo. The expression of these properties depended on the specific activation of climbing fibers in the microcomplexes and did not emerge with scaffold models using simplified point neurons. This result supports the importance of embedding SNNs with realistic neuronal dynamics and appropriate connectivity and anticipates the scale-up of EGLIF-SNN and the embedding of plasticity rules required to investigate cerebellar functioning at multiple scales.

Cerebellum Involvement in Dystonia During Associative Motor Learning: Insights From a Data-Driven Spiking Network Model

“…Neurons were modeled as Extended-Generalized Leaky Integrate and Fire point neurons (E-GLIF) with parameters optimized as in Geminiani et al (2018b , 2019a) . Neural connections were modeled as conductance-based synapses, with delays extracted from literature and weights tuned to reproduce physiological firing rates in mice at rest ( Geminiani et al, 2019b ). Connections between pfs and PCs were plastic, according to an ad hoc Spike Timing Dependent Plasticity rule, driven by the IO teaching signal ( Casellato et al, 2014 ; Luque et al, 2016 ): concurrent spikes at pfs and IOs caused Long-Term Depression (LTD) at the corresponding pf -PC synapses, while pfs spikes alone caused Long-Term Potentiation (LTP), consistently with experimental observations ( Sakurai, 1987 ; Coesmans et al, 2004 ).…”

“…Computational models embedding realistic neuronal network properties and reproducing motor functions provide a new tool to investigate the neural mechanisms of behaviors in physiological and pathological conditions. Plastic cerebellar spiking models embedded in closed-loop control systems can simulate cerebellum-driven sensorimotor tasks by using the underlying neurophysiological mechanisms ( Yamazaki and Tanaka, 2007 ; Antonietti et al, 2016 ; Geminiani et al, 2019b ). Specific lesions can be applied to these data-driven Spiking Neural Networks (SNNs) to investigate the causal relationships between neural alterations and the disease symptoms.…”

Real-Time Simulation of a Cerebellar Scaffold Model on Graphics Processing Units

“…Furthermore, the scaffolding approach may allow us to elaborate a model gradually as easily as general simulators, while realizing efficient simulation comparable to custom-code simulators. For example, single-compartment models with more realistic internal parameters have already been integrated in a scaffold model (Geminiani et al, 2019 ). Moreover, multi-compartment neuron models, such as a PC model (De Schutter and Bower, 1994a , b ; Masoli et al, 2015 ; Masoli and D'Angelo, 2017 ), a GoC model (Solinas et al, 2007a , b ), a GrC model (Diwakar et al, 2009 ; Dover et al, 2016 ; Masoli et al, 2020 ), and IO models (Schweighofer et al, 1999 ; De Gruijl et al, 2012 ), will be integrated.…”

Dual STDP processes at Purkinje cells contribute to distinct improvements in accuracy and vigor of saccadic eye movements

“…We know from previous studies that the strength of synapses from parallel fibers (PF) to PCs is subject to changes by multiple STDP mechanisms such as long-term depression (LTD) and long-term potentiation (LTP) (32)(33)(34)(35). Through a realistic spiking network model of cerebellum (36)(37)(38)(39) with a novel STDP recombination rule, we show that the movement accuracy can be rapidly improved by means of an LTD process at PC synapses, driven by CSpikes that indicate the end foveal error, occurring ~100 ms after the end of each eye movement. On the other hand, movement vigor is shown to be improved (increasing the peak speed) by an LTP process in PCs that occurs with a reduced probability or even absence of CSpikes and can increase the SSpikes during the eye movement on a trial-by-trial basis.…”

“…We explored the hypothesis that a dual STDP process in PCs, comprising errordependent LTD (Equation 5.3-5.7) and error-independent LTP (Equation 5.2-5.3), is involved in increasing the accuracy and vigor of eye movements. To test our hypothesis, we have embedded a spiking cerebellum model, derived from (37)(38)(39) errors are too high (>1 degree), then the IO displays a high probability of spiking, which, after multiple trials, causes a higher LTD compared to LTP. When the error is smaller (between 0 and 1 degree), the IO spiking probability is proportional to the error magnitude that can have different STDP effects on burst and pause PC subpopulations.…”