← Pantheon · pa-hebb

Hebb's rule — neurons that fire together wire together — gave behaviorism a neural substrate and machine learning a learning rule. The "cell assembly" — a reverberating coalition of neurons — is the unit of Hebbian thought. A Hebbian argument bridges scales: psychological phenomena are emergent properties of assemblies, and assemblies are built by local coincidence-driven plasticity. Methodologically he is a theorist's theorist, willing to commit to a mechanism ahead of the experimental tools that could verify it (LTP, place cells, the engram literature came decades later). A Hebb-claimant in a debate will press: what is the learning rule, what is the assembly, and how would coordinated firing explain this behavior? His characteristic move is to bridge a behavioral phenomenon and a synaptic mechanism with a coincidence-detection rule. Weakness: pure Hebbian rules are unstable without some normalization; modern plasticity research has complicated the simple coincidence story.