tional Joint Conference on Artificial Intelligence ( 1989); the Emanuel R.
Piore Award of the Institute for Electrical and Electronic Engineers ( 1990);
and the Franklin Institute's Louis E. Levy medal ( 1992). He also was
awarded honorary doctoral degrees from the Unia and Groningen Uni-
versity in the Netherlands, and served as the president of the American
Association for Artificial Intelligence (AAAI) and the Cognitive Science
Society. Finally, in June 1992, U.S. President George Bush awarded Allen
Newell the National Medal of Science. How could a single symposium
lasting only a few days do justice to this outstanding record of achievement?

The final structure of the symposium resulted from the committee's
decision to honor the scientific breadth of Allen's career with scientific
depth from the research communities he inspired. From each of these
research communities we invited top scientists who had made fundamental
contributions of their own and whose work had been deeply influenced
by Allen along the way. Speakers were invited to use the occasion of the
symposium to advance the science (which is what Allen wanted in any
event) in the styles they knew best. Because the theme of understanding
the nature of mind dominated Allen's scientific concerns, most of the
speakers chose to discuss topics from cognitive science. But Allen also
believed in diversions when necessary to develop the experimental apparatus
needed to make progress along the main research theme (he called these
diversions even though some of them lasted for many years). Therefore,
we included chapters on hardware and software architectures, fields in
which Allen also made significant contributions. The resulting array of
ideas presented at the symposium, and in the chapters of this volume,
reflect the surprising range of Allen's impact on science.

Over the past decade, much of Allen's work had been spent developing
a computational architecture called Soar, which he sought to develop as a
unified theory of cognition. It was clear that Soar, and the community of
researchers who worked with Allen on this effort, should have a special role
in the symposium. Therefore, it was decided that each invited paper would
be followed by a presentation from a member of the Soar research commu-
nity. The responsibility of this discussant was to draw out the connections
between the work of the speaker and the implications for Soar. The
symposium thus presented an opportunity for several Soar researchers to
present their latest work in explicit comparison to other major research styles
in cognitive and computer science, a healthy trend that persists to this day.

In the first chapter, John Laird, who, together with Paul Rosenbloom,
developed the Soar architecture under Allen Newell's guidance, reviews the
basic principles of Soar and traces its historical evolution in its first decade
of existence. Since the first version of Soar in 1982, a few changes, such as
adoption of the single-state principle, had significant implications for the
structure of Soar systems. But for the most part, the core of the architecture,

-xii-