Two members of the CNRG, Chris Eliasmith
and Charles H. Anderson,
wrote a book detailing the framework presented here. The
title is *"Neural Engineering: Computation, Representation and Dynamics
in Neurobiological Systems."*
The official MIT Press website is
here.
The book is also available at
Amazon.

Please send us an email if you have any questions or have found errors in the book (see errata below).

## Errata

### Major

- Figure 4.21 (p. 121) is printed incorrectly. The bottom response traces are missing. It should look like this.
- Section 2.2.1 reports real neuron information transmission rates as about
$3^{10}$ bits/s. This should be between about 10-700 bit/s (depending on the neuron). - In Appendix B, equation B.17, the convolution (
$*$ ) is a multiplication. This is because the equation expresses the convolution explicitly (once you remove the$*$ ). - Figure 4.3 should have values of
$\alpha=1.7$ ,$J^{bias}=1.0 nA$ , and$J_{th}=0.1 nA$ (values in the book were missing the decimal place). In addition, 4.3c is incorrect. It is approximately correct to reverse the order of the legend labels.

### Minor

- Appendix B, p. 309: the first sentence after equation B.21 should read
"where
$\langle \rangle_A$ is the convolution with the window to emphasize that this*is*providing an approximation..." The word 'is' is missing. - There's an errant bracket in the subscript for equation B.6.
- Appendix B: B.6 should be
$x(t;A)$ , not$x(t;p)$ . B.11 Should have a factor of$2 \pi$ in front. B.12 Should have no factors in front. B.16$t \alpha$ should be$t^\alpha$ , the brackets should be squared,$(t-t^\alpha)^2$ , and the expression should be normalized, i.e., multiplied by$1 / \Delta_t \sqrt(2 \pi)$ . - In chapter 9, eqns 9.9 and 9.10 should have
$p(y|x)$ in place of$p(x|y)$ (p. 278).