Computational abstraction towards a theory of the brain Brain Computation as Hierarchical Abstraction Dana H. Ballard (MIT Press, Cambridge, MA; 2015).

Who in their right mind would take on the task of trying to convey the state of current thought regarding how the human brain functions, as well as put forward a proposal for how to tie it all together, to both a nonspecialist and specialist audience, in a single volume? I cannot think of more than a few people on the planet who might have any chance at all of pulling it off without the hyperbole and sensationalism one often sees. Dana Ballard is one of those, and he does indeed pull it off. With any such attempt there are bound to be many points of contention — after all, as the author repeats throughout his book, there is still so much we do not understand about the brain, its structure, function, and behavior. Nevertheless, this is a very readable and much needed book. I can easily see it being a standard part of a reading list for graduate students starting off in neuroscience (computational or biological) or a core element of a graduate course. It would also be an excellent read for anyone seeking a snapshot of brain science. Even though this fi eld moves very quickly, adding many papers with new results and ideas daily, I feel Ballard’s book will remain a strong contribution for a long time to come. Ballard introduces early in the book the premise of his central goal, the hierarchy of abstractions that represent brain computation. He does this in three ways. First, he shows the hierarchical spatial scales present in the central nervous system (from molecules, to synapses, neurons, networks, maps, systems, and fi nally, the central nervous system as a Book review