Using a million cell simulation of the cerebellum: Network scaling and task generality

Abstract

Several factors combine to make it feasible to build computer simulations of the cerebellum and to test them in biologically realistic ways. These simulations can be used to help understand the computational contributions of various cerebellar components, including the relevance of the enormous number of neurons in the granule cell layer. In previous work we have used a simulation containing 12000 granule cells to develop new predictions and to account for various aspects of eyelid conditioning, a form of motor learning mediated by the cerebellum. Here we demonstrate the feasibility of scaling up this simulation to over one million granule cells using parallel graphics processing unit (GPU) technology. We observe that this increase in number of granule cells requires only twice the execution time of the smaller simulation on the GPU. We demonstrate that this simulation, like its smaller predecessor, can emulate certain basic features of conditioned eyelid responses, with a slight improvement in performance in one measure. We also use this simulation to examine the generality of the computation properties that we have derived from studying eyelid conditioning. We demonstrate that this scaled up simulation can learn a high level of performance in a classic machine learning task, the cart-pole balancing task. These results suggest that this parallel GPU technology can be used to build very large-scale simulations whose connectivity ratios match those of the real cerebellum and that these simulations can be used guide future studies on cerebellar mediated tasks and on machine learning problems.

The <em>NCAN</em> gene: schizophrenia susceptibility and cognitive dysfunction

“…In contrast, neuroimaging studies have shown that cortical cerebellar and global cerebellar volumes are significantly reduced in patients with schizophrenia compared to healthy controls 36. As the cerebellum represents at least 10% of the brain volume that contains 50% of its neurons,37 it has been established that cerebellar cortical volume reduction is implicated in cognitive dysfunction in schizophrenia 36. Moreover, the aberrant network connecting the frontal cortex and cerebellum is believed to be involved in the development of cognitive dysfunction,38 because cerebellar stimulation dynamically influences the medial frontal cortex in mice and clinically alleviates cognitive dysfunction in patients with schizophrenia 39,40.…”

Increased whole cerebellar serotonin in aged C57BL/6 mice

“…Lugaro cell activity also directly inhibits Purkinje neuron firing [33]. A single Lugaro cell makes inhibitory synaptic contact with about 150 nearby Golgi cells [31]; each Golgi cell makes inhibitory synaptic contact with about 3,000–5,000 granule cells [34]. Lugaro cell activity thus has the potential to reduce both the tonic and phasic components of Golgi cell inhibition onto large populations of granule cells.…”

Cerebellar neurodynamics during motor planning predict decision timing and outcome on single-trial level

“…Thus, in contrast to heat-evoked sensorimotor transformation which involves a hindbrain local circuit (Haesemeyer et al, 2018), the overall functional circuitry spans across the entire brain ( Figure 8), even though both studies observed strong response to heat in the telencephalon and the habenula (Figure 4). This input to the cerebellum could be further amplified by the divergent mapping of the mossy fiber input to the granular cells (Huang et al, 2013;Ito, 2006;Li et al, 2013) and then integrated at the Purkinje cell layer, which also receives excitatory input from the contralateral inferior olive (IO) through the climbing fibers (Ito, 2008). Purkinje cells modulate via inhibitory connections the output of the eurydendroid neurons which at the same time also receive direct excitatory connections from the granular cells (Ito, 2008) and feed into the ARTR, i.e.…”