Hebbian plasticity

Hebbian plasticity is the foundational learning rule of neural systems, first articulated by Canadian psychologist Donald Hebb in 1949: "When an axon of cell A is near enough to excite a cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells such that A's efficiency, as one of the cells firing B, is increased." This principle — often summarized as "cells that fire together wire together" — provides a local, unsupervised mechanism by which neural circuits can encode correlations in their inputs without external supervision.

Hebbian plasticity is not a single mechanism but a family of related processes. Classical Hebbian learning strengthens synapses when pre- and postsynaptic activity is correlated, but variants include anti-Hebbian learning (weakening correlated synapses), BCM theory (which incorporates a sliding threshold for potentiation vs depression based on average postsynaptic activity), and spike-timing-dependent plasticity (STDP), where the precise temporal order of pre- and postsynaptic spikes determines the direction and magnitude of change. These variants transform the simple correlation rule into a rich temporal code that can encode causality, not just co-occurrence.

The theoretical significance of Hebbian plasticity extends far beyond neuroscience. It is the ancestor of unsupervised learning algorithms in machine learning, including Oja's rule and principal component analysis networks. It demonstrates that global structure can emerge from local rules without centralized instruction — a principle that recurs in collective construction, stigmergy, and self-organization.

Hebbian plasticity is often celebrated as the mechanism of learning, but it is equally the mechanism of confinement. A Hebbian system learns its inputs so thoroughly that it becomes unable to learn anything that contradicts them. This is the neurological basis of confirmation bias: a network that has wired itself to represent a correlation will resist rewiring to represent a counterexample. The same principle that makes Hebbian learning powerful makes it conservative. Every learning rule is also a forgetting rule, and Hebbian plasticity forgets dissent.