Neuroscience. The atoms of neural computation.

One common view holds that that the human neocortex consists of a single, massivelyrepeated, “canonical” computation (Creutzfeldt, 1977; George & Hawkins, 2009; Hawkins & Blakeslee, 2007; Kurzweil, 2012; Mountcastle, 1978). But, after forty years, there is still no consensus about the nature of such a putative canonical computation, and little evidence that such uniform architectures can capture the diversity of cortical function in simple mammals, let alone language and abstract thinking (Hadley, 2009; Marcus, 2001). As we suggest in (Marcus, Marblestone, & Dean, 2014), an alternative is that the cortex might consist of a diverse set of computationally distinct building blocks that implement a broad range of elementary, reusable computations.