1. Rieke, F., D. Warland, R. de Ruyter van Steveninck, and W. Bialek. (1997) Spikes: Exploring the neural code. MIT Press.
2. Wilson, H. R. (1999). Spikes, decisions, and actions: Dynamical foundations of neuroscience. Oxford University Press.
3. Bower, J. M. and D. Beeman (eds.) (1994). The book of Genesis: Exploring realistic neural models with the GEneral NEural SImulation System. Springer-Verlag.
4. Koch, C. (1999). Biophysics of computation: Information processing in single neurons. Oxford University Press.
5. Koch, C. and I. Segev (eds.) (1989). Methods in neuronal modeling: From synapses to networks. MIT Press.
6. Arbib, M. A. (ed.) (1995). The handbook of brain theory and neural networks. MIT Press.
7. Kandel, E. R., J. H. Schwartz, and T. M. Jessell (eds.) (1991). Principles of neural science. Third Edition. Elsevier.

1. Spikes: Exploring the neural code

1.1 Overview

This book is the ninth in the computational neuroscience series put out by MIT Press and edited by T. J. Sejnowski and T. A. Poggio. The first book in the series is the Koch and Segev book we discuss later (section 5). In our view, it is one of the most important books in this series. This is because it does an excellent job of laying the foundations for understanding the temporal nature of the neural code. In particular, Spikes discusses the theory behind `decoding' the neural code of individual (or more precisely, pairs of) neurons.

Much of the book is spent analyzing the information transmission characteristics of individual neurons, with the conclusion that, in general, neurons are able to transmit between about 2 and 5 bits of information per neural spike. The book provides fairly detailed coverage of the mathematical background related to information theory and signals and systems analysis needed to understand their analysis.

Of less interest to us, another major focus of the book is on issues of hyperacuity and discriminability.

1.2 Strengths and Weaknesses

The greatest strength of this book is that is provides an excellent collection of empirical results and related theoretical discussions regarding the nature of coding in individual neurons. These ideas have been previously discussed by the authors in a number of papers, but nowhere else is the discussion as complete and thorough. We will be reviewing much of the analysis presented in this book in the section of the course concerned with temporal coding.

An additional strength is the many mathematical appendices provided for those wishing to get an in-depth look at the methods, assumptions, and tools used for generating the results discussed in the main text.

2. Spikes, decisions, and actions: Dynamical foundations of neuroscience

2.1 Overview

Spikes, decisions, and actions is an excellent introduction to low level single cell modeling. Hugh Wilson provides the necessary mathematical and neuroscientific background needed to get up to speed on the latest in single cell modeling. He begins with an insightful discussion of the integrate-and- fire model, and slowly introduces more and more biophysically realistic models. However, he doesn't simply focus on biophysical realism. Rather, he provides, presents and discusses a number of tools for analyzing the behavior of the models. As well, he discusses means of simplifying extremely complex models (which are difficult to analyze) in reasonable ways.

A central focus of the book is on nonlinear analysis applied to spiking neural models. Wilson clearly presents concepts from nonlinear systems theory and relates them to reduced models of various classes of neurons. This includes a discussion of bifurcations, limit cycles, chaos, and Lyapunov functions.

Of less interest to us, he also discusses bursting cells, central pattern generators and other synchrony-related issues currently of interest to many modelers.

2.2 Strengths and Weaknesses

The greatest strength of the book is its clear explanation of all the necessary concepts, both mathematical and neuroscientific, from nearly first principles. Anyone rusty in nonlinear systems theory, linear systems theory, and differential equations will appreciate Wilson's short, but clear discussion of these issues. As well, he introduces advanced topics in a concise way.

From our perspective, his discussion of nonlinear neurodynamics and clear exposition of the methods of reduction from high dimensional single cell models is most useful. Nowhere else have we found such a good summary of the literature in this area.

3. The book of Genesis: Exploring realistic neural models with the GEneral NEural SImulation System

3.1 Overview

GENESIS is a well-known, standard neural simulation package available for free. This book is a collection of chapters that act as tutorials in the use of the simulation package. Many of the basic modeling techniques used in computational neuroscience are covered in this book. Topics of the book include the Hodgkin-Huxley model, compartmental modeling, cable equations, voltage activated channels as well as topics specific to the simulation package.

In general, the book is written from the perspective of an experimentalist. A lot of emphasis is placed on generating models of equal complexity to those found in real neural systems. So, it is not surprising that much of the book is spent discussing single cell models, and methods for making them more realistic. Nevertheless, the simulation tool itself can be used to model networks of cells as well.

3.2 Strengths and Weaknesses

The book provides a an excellent introduction to well-known models in neuroscience. In particular, the chapters on the Hodgkin-Huxley model, compartmental modeling and cable equations provide good background, assumed by many more advanced books in the area. If you plan to do any simulations with GENESIS (the package), this book is mandatory. If you don't it is still worthwhile reading many of the earlier chapters that introduce standard modeling techniques.

From our perspective, this book provides a introduction to the concepts, vocabulary, and approach used by experimental neuroscientists to modeling single neuronal cells. This kind of background is very important to have in order to communicate effectively with more biologically oriented computational neuroscientists.

4. Biophysics of computation: Information processing in single neurons

4.1 Overview

This book from Christof Koch, a well-respected leader in the field, is a heroic effort to cover most of the topics in single-cell computational neuroscience. He provides the basic background on subjects such as cable theory, membrane equations, the Hodgkin-Huxley model, and passive dendritic trees, as well as touching on issues of nonlinear systems analysis, voltage- dependent currents, dendritic spines, active dendrites, and bursting.

For us, this book provides a perspective on the current state of the field in computational neuroscience. Although our research focusses on the systems level and this is a book more concerned with single neurons, the Biophysics of Computation provides a good primer for our research. This is so because these different levels of explanation and description are mutually dependent. Thus, we see Koch's book as an invaluable tool for determining what is important to know about the single neurons that comprise the larger systems that produce the vast variety of animal behavior.

4.2 Strengths and Weaknesses

We have yet to find a book that covers so many different topics of interest to computational neuroscientists. Unfortunately, many of the discussions are so short as to be somewhat superficial. However, the numerous citations allow the reader to pursue topics to great levels of details.

Seldom are so many of the important issues brought together in a single volume. For those wanting to know what computational neuroscientists are interested regarding single cells, this book is an unmatched resource.

5. Methods in neuronal modeling: From synapses to networks

5.1 Overview

This book is a classic in computational neuroscience. Even though it was published decades ago, many of the articles in the collection are cited by contemporary authors. The papers are generally written by some of the best known experts in the area. Unlike most of the books we've discussed so far, this one also has a significant portion of the chapters (3 out of 13) dedicated to systems level models.

The focus of this book is on the methods used in constructing and simulating realistic neural models. Thus, there are chapters dedicated to issues of implementation on parallel computers, numerical methods, and large-scale simulation construction techniques. Most of the theoretical work is in early chapters (which focus on the behavior of single neurons). Such subjects as the cable equations, nonlinear systems analysis, and compartmental modeling are addressed.

5.2 Strengths and Weaknesses

The main strength of this book is that it is a collection of now classic papers in many of the areas of computational neuroscience. Thus, it provides good background reading for a number of subjects still central to the discipline. The main weakness, from our point of view, is the lack of articles on theoretical analysis of systems level networks.

6. The handbook of brain theory and neural networks

6.1 Overview

This collection of well over 200 articles is a massive tome that provides unmatched coverage of connectionism, computational neuroscience, and artificial intelligence. It is something like an encyclopedia of brain theory. Many articles are written at an introductory level to familiarize the reader with the relevant issues, and recent insights in the particular field. Because the focus is on brain theory, most articles do not address the relevant neurobiology in any detail. That being said, there are brief introductions to neuroanatomy, single cell modeling, neural development, and plasticity. However, even these sections are more focused on analysis and simulation, than on facts about the brain.

6.2 Strengths and Weaknesses

This collection is simply unmatched in its coverage of theoretical approaches to brain-like models. Although many of the approaches are somewhat `unbiological', all are important to understanding the current state of thought in computational neuroscience. Of course, the book is advertised as being on brain theory, so the downplay of biological facts is precisely what we should expect. For those facts, we can turn to the final book, below.

7. Principles of neural science

7.1 Overview

Simply put, this text is the standard neuroscience textbook. Now in its sixth edition, the hefty 1200 page volume covers almost all aspects of neuroscience as it is currently understood. Many of the articles are written for neuroscience students, and are thus clear and seldom assume much background in neuroscience. For this reason, it is an ideal resource for those new to computational neuroscience who are coming from technical fields.

The book discusses major themes of neuroscience, large and small, and includes in-depth coverage of sensory and motor anatomy and function, the molecular basis of neuronal behavior and synaptic transmission, neural/behavioral development, and neural pathologies. If you need to know a neural fact, you can probably find it here.

7.2 Strengths and Weaknesses

The main strength of this volume is its unsurpassed value as a `quick and easy' source for neuroscientific facts. The main weakness, as we see it, is that there is seldom any attempt to systematize these facts: this is a criticism that has been leveled at neuroscience in general by some theoreticians.