I thank Donahoe for a thoughtful and thorough review that fairly describes the book’s core arguments. Donahoe’s criticisms and comments provide an opportunity to amplify on some of our key arguments in ways that I hope will contribute to a fuller understanding of them. As Donahoe points out, Shannon’s theory of communication, from which the modern mathematical definition of information comes, is central to our argument. The one-sentence essence of our argument is that: ―The function of memory is to carry information forward in time in a computationally accessible form.‖ For this argument to have substance, it must be clear what information is; hence, that is where our book begins. Information theory assumes that the receiver of a communication (here, the brain) has, ab initio, before receiving any messages, both a representation of the set of possible messages and a probability distribution on that set. Donahoe’s objection to this assumption reveals a common misunderstanding of what it means to have ―foreknowledge of all the events that could possibly occur‖ and what it means to have a prior probability distribution over a set of possible messages denoting possible states of the world. Donahoe’s objection is that ―When a state is encountered for the first time, neither its prior existence (by definition) nor its a priori probability could be known.‖ First, consider what it means to assume that a system that is to receive messages conveying information about states of some aspect of the world has prior knowledge of those states. For the sake of concreteness, let us focus on what it means to say that an animal with color vision has a representation of possible reflectance spectra before it opens it eyes. What this means is that it has inputprocessing machinery that can generate different signals for different spectra (of equal photopic luminosity). From an evolutionary perspective, the existence of this genetically specified machinery implies (or suggests, if you like) the existence of differences in spectra in the world that the animal is likely to live in. If the sun were a monochromatic source of light, we would not expect to find visual systems with structures that enabled them to distinguish spectra. In many cave-dwelling species, the visual system has atrophied altogether. Where there are no lightconveyed messages to be received, there is no point in having machinery designed to pick up light. Where there are no spectral differences, machinery designed to
[1]
C. L. Hull.
Knowledge and purpose as habit mechanisms.
,
1930
.
[2]
N. Clayton,et al.
The rationality of animal memory: complex caching strategies of western scrub jays
,
2006
.
[3]
William Bialek,et al.
Spikes: Exploring the Neural Code
,
1996
.
[4]
Frieeland Judson Horace.
The eighth day of creation
,
1979
.
[5]
C. Gallistel,et al.
Time and Associative Learning.
,
2010,
Comparative cognition & behavior reviews.
[6]
Philipp Slusallek,et al.
Introduction to real-time ray tracing
,
2005,
SIGGRAPH Courses.
[7]
David E. Comings,et al.
The eighth day of creation.
,
1980
.
[8]
D. Knill,et al.
The Bayesian brain: the role of uncertainty in neural coding and computation
,
2004,
Trends in Neurosciences.
[9]
C. Gallistel,et al.
Risk assessment in man and mouse
,
2009,
Proceedings of the National Academy of Sciences.
[10]
B. Roche,et al.
The Behavior of Organisms?
,
1997
.
[11]
C. Gallistel,et al.
Time, rate, and conditioning.
,
2000,
Psychological review.
[12]
C. Gallistel,et al.
Temporal maps and informativeness in associative learning
,
2009,
Trends in Neurosciences.
[13]
永福 智志.
The Organization of Learning
,
2005,
Journal of Cognitive Neuroscience.