The Five Breakthroughs of Intelligence — Book review of Max Bennett’s A Brief History of Intelligence — Evidence, analogies, and limits.

Max Bennett’s A Brief History of Intelligence argues that human intelligence accumulated through five evolutionary breakthroughs: (1) steering/valence in early bilaterians (approach/avoid), (2) reinforcement learning in vertebrates (dopamine‑driven trial‑and‑error), (3) simulation in mammals (cortical world‑models, hippocampal replay), (4) mentalizing in primates (theory of mind and social strategy), and (5) language in humans (symbols and compositionality). Each layer added new capability without replacing earlier ones. The book uses these layers as a functional blueprint for what today’s AI still lacks.

The strongest sections ground the basal ganglia ↔ RL analogy and the cortex ↔ predictive generative model view with clear evidence and experiments; the hippocampus ↔ episodic memory + replay mapping is equally persuasive. The thalamus ↔ attention/routing and specific circuitry for theory of mind are presented as promising but more speculative. The narrative is integrative and practical, connecting brain mechanisms to AI design choices and to the limits of current systems (e.g., LLMs’ weak world models and shallow social cognition).