The fossil record shows a four-fold increase from the australopithecine brain to that of Homo sapiens, but modern comparative neuroanatomy indicates a variety of relative changes from the chimp to the human brain: decreases (olfactory, visual, motor cortex), small increases (auditory cortex) and large increases (prefrontal, polymodal association cortex)(1). Posterior polymodal association cortex evolved prior to the expansion of the frontal lobe(2), leading Oppenheimer(3) to ask: What were the perceptions that drove the emergence of brain regions where visual, tactile and auditory sensory pathways first converge at the neocortical level(4). The empirical evidence indicates that, during the first 2 million years of expansion of the primate brain, our ancestors were involved in making unhafted, simple stone tools by repeatedly striking one stone with another. The evolutionary puzzle is why the new visuomotor coordination and new conceptual insights necessary for stone-tool making did not lead immediately to more complex tools? The answer lies in the difficulty of triadic associations(5). Unlike dyadic associations, that can be rapidly learned in unsupervised neural networks, triadic associations are inherently complex. The mechanisms of trimodal (truly multimodal, as distinct from cross-modal) perception, three-way associations, conditional associations and triadic cognition are explained using neural network simulations(6).