On Minds

Plt;bPgt;Toward a deeper understanding of minds, whether natural or artificial, and of how they are studied and built across AI, AGI, and cognitive science.Plt;/bPgt;Plt;brPgt;Plt;brPgt;Minds are functional entities that enable intelligent behavior. In Plt;iPgt;On MindsPlt;/iPgt;, Paul Rosenbloom offers reflections and perspectives on the study of minds arising from a multi-decade career focused largely on cognitive architecturesP#8212;hypotheses concerning the fixed structures and processes that yield a mind, whether natural or artificial. Cognitive architectures must be integrative in accommodating the full range of capabilities required of a mind. They must also be general, in applying across many, if not all, problems and domains. This breadth of investigation, across a sequence of three such architecturesP#8212;XAPS, Soar, and SigmaP#8212;and over many years, has inspired a set of reflections on both cognitive architectures and three core disciplines within which such work occursP#8212;artificial intelligence (AI), artificial general intelligence (AGI), and cognitive science.Plt;brPgt;Plt;brPgt;The bookP#8217;s reflections are of three types. The first explores four concepts critical to understanding mindsP#8212;architectures, symbols, intelligence, and theoriesP#8212;to provide new perspectives and insights on them. The second rethinks the structure of the space of technologies that underlies work on minds, yielding multidimensional maps that clarify their similarities and differences plus detailed insights into the tradeoffs that exist across both individual technologies and integrated systemsP#8212;whether symbolic, neural, or hybridP#8212;built from them. The third traces the authorP#8217;s decades of work on architecturesP#8212;including cognitive architectures, the Common Model of Cognition, and the nature and structure of computing as a great scientific domainP#8212;interwoven with personal reflections and anecdotes.