Hebbian learning

Hebbian learning is the simplest and most consequential idea in neuroplasticity: neurons that fire together wire together. The rule, proposed by Donald Hebb in 1949, states that the strength of a synaptic connection increases when the pre-synaptic and post-synaptic neurons are active simultaneously. It is not a learning algorithm in the modern sense — it has no explicit target, no error signal, no reward — but it is a self-organizing mechanism that extracts statistical structure from experience by strengthening correlations and weakening anti-correlations.

The systems-theoretic significance of Hebbian learning is that it provides a local mechanism for global structure. Each synapse updates independently, yet the collective dynamics of Hebbian plasticity produce structured networks — feature detectors, topographic maps, memory traces — without any central controller. This is emergence at the synaptic scale: the global pattern is not represented in any single synapse but is distributed across the entire network. The mechanism is so robust that variants of it reappear in radically different contexts: Oja's Rule in principal component analysis, Spike-Timing Dependent Plasticity in cortical circuits, and even (with appropriate normalization) in some forms of Predictive Processing.

Hebbian learning is not a complete theory of learning. It cannot explain supervised learning, reinforcement learning, or the rapid one-shot learning that humans exhibit. But it is the substrate on which all more sophisticated learning mechanisms are built. Without correlation-based plasticity, there would be no stable architecture for error signals to propagate through, no baseline structure for reward-modulated updates to act upon, and no developmental canalization that makes complex learning possible at all. The claim that Hebbian learning is merely a historical curiosity of neuroscience is exactly wrong: it is the physical implementation of a statistical principle that makes the rest of cognition computationally tractable.