What is necessary to build an interface between a human brain and a machine? Knowledge from several areas of expertise (neuroscience, signal processing, and machine learning) put together: a recipe for a brain-computer interface (BCI). It is a new area with growing importance. A basic book with a thorough vision of the subject, including its applicability to real-world situations, would be appreciated. This is such a book. It brings a comprehensive (although superficial) set of disciplines necessary to understand a BCI--not only what it is, but also how to build it. The reader is instructed on basic neuroscience, how to record brain signals, and how to stimulate the brain in order to produce these signals. Signal processing tools are also presented, as well as machine learning techniques.

Types of BCIs, invasive, semi-invasive, and noninvasive, are discussed. A very up-to-date and extremely interesting section about applications illustrates the noble medical and the not-so-noble nonmedical uses of BCIs. It also presents experiences in the author’s lab. Rao recalls movies (Avatar, Surrogates, and Johnny Mnemonic) and comics (Spider-Man) to illustrate science fiction that is about to become true through BCIs. He also mentions a theatrical performance in New York that makes use of a BCI. The book closes with an essential discussion about ethics and legal issues. The risks of invasive and noninvasive BCIs are considered, and very important questions are raised. In addition, exercises are included at the end of each chapter, and there is an appendix about useful mathematical issues.

I like the book, the way it is written, but I found some mistakes in the figures and dimensions presented. In addition, it suffers from a lack of references and problems with figures, especially the colored ones that are smaller than their black-and-white counterparts. Perhaps the reason is the cost, although illustrations are so important for a book like this. Some figures are too small to note the relevant details. Regarding an important concept presented, the author mentions that the connectionist supervised backpropagation algorithm was discovered in the 1980s, but this is certainly arguable. Neither of these problems make the book uninteresting, and I am sure that they will be considered in a next edition.

Since the book has already been published, a second edition should include a glossary of the many acronyms used in the text. That would be very helpful. Sometimes an acronym appears without explanation, and then suddenly it appears in the next pages with an explanation. A glossary would avoid this problem: only in its first appearance would it be explained; the reader could access the glossary for the other occurrences.

In short, I recommend this book for beginners interested in BCIs. It is a very good way to begin to understand this enthralling field.

More reviews about this item: Amazon