Microsoft Word Kurzweil, Ray The Singularity Is Near doc

Chapter Four: Achieving the Software of Human Intelligence: 
How to Reverse Engineer the Human Brain 

---

1.
Lloyd Watts, "Visualizing Complexity in the Brain," in D. Fogel and C. Robinson, eds., 
Computational 
Intelligence: The Experts Speak
(Piscataway, N.J.: IEEE Press/Wiley, 2003), 
http://www.lloydwatts.com/wcci.pdf. 
2.
J. G. Taylor, B. Horwitz, and K. J. Friston, "The Global Brain: Imaging and Modeling," 
Neural Networks
13, 
special issue (2000): 827. 
3.
Neil A. Busis, "Neurosciences on the Internet," http://www.neuroguide.com; "Neuroscientists Have Better 
Tools on the Brain," 
Bio IT Bulletin
, http://www.bio-it.world.com/news/041503_report2345.html; "Brain 
Projects to Reap Dividends for Neurotech Firms," 
Neurotech Reports
, 
http://www.neurotechreports.com/pages/brainprojects.html. 
4.
Robert A. Freitas Jr., 
Nanomedicine
, vol. 1, 
Basic Capabilities
, section 4.8.6, "Noninvasive Neuroelectric 
Monitoring" (Georgetown, Tex.: Landes Bioscience, 1999), pp. 115–16, 
http://www.nanomedicine.com/NMI/4.8.6.htm. 
5.
Chapter 3 analyzed this issue; see the section "The Computational Capacity of the Human Brain." 
6.
Speech-recognition research and development, Kurzweil Applied Intelligence, which I founded in 1982, now 
part of ScanSoft (formerly Kurzweil Computer Products). 
7.
Lloyd Watts, U.S. Patent Application, U.S. Patent and Trademark Office, 20030095667, May 22, 2003, 
"Computation of Multi-sensor Time Delays." Abstract: "Determining a time delay between a first signal 
received at a first sensor and a second signal received at a second sensor is described. The first signal is 
analyzed to derive a plurality of first signal channels at different frequencies and the second signal is analyzed 
to derive a plurality of second signal channels at different frequencies. A first feature is detected that occurs at 
a first time in one of the first signal channels. A second feature is detected that occurs at a second time in one 
of the second signal channels. The first feature is matched with the second feature and the first time is 
compared to the second time to determine the time delay." See also Nabil H. Farhat, U.S. Patent Application 
20040073415, U.S. Patent and Trademark Office, April 15, 2004, "Dynamical Brain Model for Use in Data 
Processing Applications." 
8.
I estimate the compressed genome at about thirty to one hundred million bytes (see note 57 for chapter 2); this 
is smaller than the object code for Microsoft Word and much smaller than the source code. See Word 2003 
system requirements, October 20, 2003, http://www.microsoft.com/office/word/prodinfo/sysreq.mspx. 
9.
Wikipedia, http://en.wikipedia.org/wiki/Epigenetics. 
10.
See note 57 in chapter 2 for an analysis of the information content in the genome, which I estimate to be 30 to 
100 million bytes, therefore less than 10
9
bits. See the section "Human Memory Capacity" in chapter 3 (p. 126) 
for my analysis of the information in a human brain, estimated at 10
18
bits. 
11.
Marie Gustafsson and Christian Balkenius, "Using Semantic Web Techniques for Validation of Cognitive 
Models against Neuroscientific Data," AILS04 Workshop, SAIS/SSLS Workshop (Swedish Artificial 
Intelligence Society; Swedish Society for Learning Systems), April 15–16, 2004, Lund, Sweden, 
www.lucs.lu.se/People/Christian.Balkenius/PDF/Gustafsson.Balkenius.2004.pdf. 
12.
See discussion in chapter 3. In one useful reference, when modeling neuron by neuron, Tomaso Poggio and 
Christof Koch describe the neuron as similar to a chip with thousands of logical gates. See T. Poggio and C. 
Koch, "Synapses That Compute Motion,"