Page 446 - Deep Learning

Notes to Pages 184–192 429

represented by a dozen rules. For complex strategies, the number is at least an
order of magnitude greater.
24. But see Clancey (1985) and Lenat (1983) for two exceptions, as well as the collec-
tion of papers in Groner, Groner and Bischof (1983).
25. McDermott and Forgy (1978).
26. The term was introduced by Anderson (1983) and it has come into widespread
use as a convenient verbal label for the innate cognitive equipment, that is, the
totality of the cognitive structures and processes that are not acquired through
learning. See Cooper (2006) for a philosophical discussion of this concept.
27. VanLehn (1991).
28. James (1890, vols. 1 and 2).
29. Thorndike (1898).
30. Bryan and Harter (1897, 1899). This pioneering study has had more impact on
micro-economics than on psychology; see, e.g., Jovanovic and Nyarko (1995) and
Reis (1991).
31. Ebbinghaus (1885/1964).
32. Watson (1913). Hilgard and Bower (1966) is the classic and most comprehensive
review of behaviorist learning theories.
33. See Woodworth (1938) for the summary of complex skill acquisition (pp. 156–175)
and for a discussion of the learning curve (pp. 170–173).
34. Woodworth (1938, p. 164).
35. Gardner (1985, Chap. 2). See also the introduction to Wiener (1948, pp. 7–39), the
historical observations in Miller (2003) and the historical addendum in Newell
and Simon (1972a, pp. 873–889).
36. Wiener (1948); see Conway and Siegelman (2005) and Heims (1991) for the his-
tory of the cybernetic movement.
37. Boden (1972, Chap. 4), Miller, Galanter and Pribram (1960), Moray (1987), Powers
(1973), Smith (1966) and Smith and Smith (1966).
38. Conway and Siegelman (2005).
39. See Crevier (1993) and Gardner (1985) for the history of cognitive science and
Newell, Shaw and Simon (1958) for the original article.
40. Ausubel (1968) and Gagné (1965). See R. C. Anderson (1967) for a review of the
period.
41. Fitts (1964) and Welford (1968).
42. Anzai and Simon (1979).
43. Anderson (1982, 1983, 1987) and Anderson, Kline and Beasley (1978, 1979).
44. See Rosenbloom, Laird and Newell (1993) and Rosenbloom and Newell (1986,
1987). The original Soar model learned only by capturing paths to subgoals dis-
covered during problem solving, but Nason and Laird (2005) extended the sys-
tem with feedback-based (reinforcement) learning.
45. Sun, Merrill and Peterson (2001); and Sun, Slusarz and Terry (2005).
46. VanLehn (1999), VanLehn and Jones (1993), and VanLehn, Jones and Chi (1992).
Cascade grew out of a prior model of learning from examples in arithmetic
(VanLehn, 1990).
47. For a technical review, see Ohlsson (2008a).
48. Bradshaw (1992).