Connectionism holds that intelligence emerges from learned statistical regularities and distributed representations in interconnected neural units, rather than from explicitly programmed symbolic rules. section 7.6
Its central dispute with GOFAI concerns compositionality: critics argue that learned patterns must explain structured thought, while connectionists argue that symbolic-like competence can emerge from lower-level soft constraints. section 7.6
From philosophical wager to benchmark
The Malik–Hinton wager shifted attention from claims about neural networks to measured performance. PASCAL and ImageNet provided shared tests, while AlexNet supplied a concrete breakthrough for convolutional networks. section 7.6 section 7.6
