Biological Learning: Synaptic Plasticity, Hebb Rule and Spike Timing Dependent Plasticity

Synonyms

Correlation-based learning; Hebb rule; Hebbian learning

Definition

The brain of humans and animals consists of a large number interconnected neurons. Learning in biological neural systems is thought to take place by changes in the connections between these neurons. Since the contact points between two neurons are called synapses, the change in the connection strength is called synaptic plasticity. The mathematical description of synaptic plasticity is called a (biological) learning rule. Most of these biological learning rules can be categorized in the context of machine learning as unsupervised learning rules, and the remaining ones as reward-based or reinforcement learning. The Hebb rule is an example of an unsupervised correlation-based learning rule formulated on the level of neuronal firing rates. Spike-timing-dependent plasticity (STDP) is an unsupervised learning rule formulated on the level of spikes. Modulation of learning rates in a Hebb rule or STDP rule by a...