Neuron-Centric Plasticity

3 Neurons and their connections

Looking back, at least since the Hopfield networks of 1982, synapse-centric plasticity became the norm. This means that each synapse is independently capable of adjusting its weights as a form of memory and learning in the brain. The question remained which “learning rule”, which mode of updating during the electrical activations, the neural spikes, in the brain seemed realistic, and what a system built with it could accomplish. However, in the brain, there is only limited plasticity that remains local to the synapse. While spines provide a restricted form of internal signaling that allows to adjust synaptic strength on short time-scales (at the same time supporting ion channel and GPCR based plasticity), long-term adjustments require signaling in the cytoplasm and even genetic read-out in the nucleus. Automatically we have a neuron-centric model here where the state of the neuron and its internal signaling determines plasticity at the membrane — including the presynaptic boutons and the postsynaptic densities. Outside signals — e.g. signals from other neurons — influence the outcome of plasticity processes, but they are also not strictly synapse-specific, instead dendritic integration plays a role. —

What can we achieve with neuron-centric models? This question has been rarely asked. We have published a paper recently, where we obtained a large reduction in the number of synapses required for…