Spikes: Exploring the Neural Code

by Fred Rieke, David Warland, Rob de Ruyter van Steveninck

What does it mean to say that a certain set of spikes is the right answer to a computational problem? In what sense does a spike train convey information about the sensory world? Spikes begins by providing precise formulations of these and related questions about the representation of sensory signals in neural spike trains. The answers to these questions are then pursued in experiments on sensory neurons. Intended for neurobiologists with an interest in mathematical analysis of neural data as well as the growing number of physicists and mathematicians interested in information processing by real nervous systems, Spikes provides a self-contained review of relevant concepts in information theory and statistical decision theory. Our perception of the world is driven by input from the sensory nerves. This input arrives encoded as sequences of identical spikes. Much of neural computation involves processing these spike trains. What does it mean to say that a certain set of spikes is the right answer to a computational problem? In what sense does a spike train convey information about the sensory world? Spikes begins by providing precise formulations of these and related questions about the representation of sensory signals in neural spike trains. The answers to these questions are then pursued in experiments on sensory neurons. The authors invite the reader to play the role of a hypothetical observer inside the brain who makes decisions based on the incoming spike trains. Rather than asking how a neuron responds to a given stimulus, the authors ask how the brain could make inferences about an unknown stimulus from a given neural response. The flavor of some problems faced by the organism is captured by analyzing the way in which the observer can make a running reconstruction of the sensory stimulus as it evolves in time. These ideas are illustrated by examples from experiments on several biological systems.

Intended for neurobiologists with an interest in mathematical analysis of neural data as well as the growing number of physicists and mathematicians interested in information processing by real nervous systems, Spikes provides a self-contained review of relevant concepts in information theory and statistical decision theory. A quantitative framework is used to pose precise questions about the structure of the neural code. These questions in turn influence both the design and analysis of experiments on sensory neurons.

Genres: Neuroscience, Science, Textbooks, Nonfiction, Biology, Brain, Computers

416 pages, Paperback

First published November 15, 1996

Reviews

[Max Pensack · Rating 3 out of 5]

Not the masterpiece of exegesis I had hoped for, but still provides a nice narrative overview of the history of modeling approaches in sensory neuroscience.

Summary:
1) Spikes are the currency of neural communication, but understanding this communication is not so simple as deciding between rate vs. timing codes.
2) Linear decoding is fairly powerful.
3) Sparse coding allows for >1 bits of information per spike.
4) This information can be used to guide behavior at near optimal levels, given physical limits.
5) Characterizing these physical limits is a challenging task, and this doesn't go away just by look at populations of neurons.

[Bria]

This is right up my alley - information theory and the brain - but sadly I had not the energy to do more than scan my eyes over most of it. I did, however, learn a very vital fact about how real science is done, viz. in studying frog sensory systems: "The coding in these afferents can be studied by recording from a single afferent fiber while shaking the entire frog."

[Jon · Rating 3 out of 5]

This book taught me a lot about neural signal encoding, and just how nerve signals work in general. However, it was not an easy read and most of the text is obviously written by an academic. There are some sections which go into excessive detail, however if you don't get hung up in those sections, the entire book does an great job of explaining many of the basic concepts taught in neural science programs such as rate encoding and the concept of the homoculus. This is not really light or easy reading though, but worth the effort if you are interested in neural science.

[J. Lucas · Rating 4 out of 5]

These guys are pretty sharp, and the discussion of Bayes' Rule as a guiding principle for neural function approximation is quite thorough. That said, they really underestimate the importance of population codes, etc. - possibly because they don't fit into their mathematical formalism compactly. A good one, though.

[mkfs · Rating 5 out of 5]

Despite being a bit dated (1999), Spikes is an excellent neuro text. Much of the information gleaned from the experiments is now common knowledge, which at first makes the book a bit uninteresting to read. The true value is in the methodology, which is explained in great depth (and in a surprisingly accessible way) by the authors.

[Philippe Roi · Rating 5 out of 5]

A scientifical travel by 'spikes' through the nerves. A fair state of the art (see p. 23 note 1) in regards to research in neurosciences to this day. But most of all, a sharp analysis of what we know today about information transmission. A reference book in the field of neurosciences.

[DJ]

info theory implications for neuroscience

[Ben · Rating 4 out of 5]

Nice overview and a lot of statistics about how neurons can interact and influence each others.