Why Brains Don't Compute

by Dale Purves

This book examines what seems to be the basic challenge in neuroscience understanding how experience generated by the human brain is related to the physical world we live in. The 25 short chapters present the argument and evidence that brains address this problem on a wholly trial and error basis. The goal is to encourage neuroscientists, computer scientists, philosophers, and other interested readers to consider this concept of neural function and its implications, not least of which is the conclusion that brains don’t “compute.”

Genres: Nonfiction

184 pages, Paperback

Published May 8, 2022

About the author

Dale Purves (born March 11, 1938) is Geller Professor of Neurobiology Emeritus in the Duke Institute for Brain Sciences where he remains Research Professor with additional appointments in the department of Psychology and Brain Sciences, and the department of Philosophy at Duke University. He earned a B.A. from Yale University in 1960 and an M.D. from Harvard Medical School in 1964.

Reviews

[mesityl · Rating 2 out of 5]

What the author is trying to argue for is basically correct and consensus within the field, although he struggles to define the terms he uses and spends most of his time presenting lackluster evidence that is almost irrelevant.

The entire section on perception could be skipped (although this is also the section where he cites his own papers...). Part V on Neurophysiology was the only decent part of the book (although would probably be covered better in any good textbook) and is the clearest evidence for the point he's making.

The last two chapters of the book are spent "disproving" two other theories from neuroscience, neither of which are incompatible with his point and both of which he clearly misunderstands terribly.

[Ahmad · Rating 2 out of 5]

Doesn’t really address how brains “don’t“ compute.

[Patrick · Rating 4 out of 5]

Purves argues that the brain doesn't work like a computer, but uses a distributed and dynamic memory system to recognize patterns and make predictions. This is unlike computers which rely on explicit algorithms and rules, making them unsuited for unpredictable situations.