The definitive presentation of Soar, one AI's most enduring architectures, offering comprehensive descriptions of fundamental aspects and new components. In development for thirty years, Soar is a general cognitive architecture that integrates knowledge-intensive reasoning, reactive execution, hierarchical reasoning, planning, and learning from experience, with the goal of creating a general computational system that has the same cognitive abilities as humans. In contrast, most AI systems are designed to solve only one type of problem, such as playing chess, searching the Internet, or scheduling aircraft departures. Soar is both a software system for agent development and a theory of what computational structures are necessary to support human-level agents. Over the years, both software system and theory have evolved. This book offers the definitive presentation of Soar from theoretical and practical perspectives, providing comprehensive descriptions of fundamental aspects and new components. The current version of Soar features major extensions, adding reinforcement learning, semantic memory, episodic memory, mental imagery, and an appraisal-based model of emotion. This book describes details of Soar's component memories and processes and offers demonstrations of individual components, components working in combination, and real-world applications. Beyond these functional considerations, the book also proposes requirements for general cognitive architectures and explicitly evaluates how well Soar meets those requirements.
Soar: A Cognitive Architecture in Perspective represents a European perspective on Soar with the exception of the special contribution from Allen Newell arguing for Unified Theories of Cognition. The various papers derive from the work of the Soar Research Group that has been active at the University of Groningen, The Netherlands, since 1987. The work reported here has been inspired in particular by two topics that precipitated the group's interest in Soar in the first place -- road user behavior and the temporal organization of behavior, more specifically planning. At the same time, the various contributions go well beyond the simple use of Soar as a convenient medium for modeling human cognitive activity. In every paper one or more fundamental issues are raised that touch upon the very nature and consistency of Soar as an intelligent architecture. As a result the reader will learn about the operator implementation problem, chunking, multitasking, the need to constrain the depth of the goal stack, and induction, etc. Soar is still at a relatively early stage of development. It does, nevertheless, constitute an important breakthrough in the area of computer architectures for general intelligence. Soar shows one important direction that future efforts to build intelligent systems should take if they aim for a comprehensive, and psychologically meaningful, theory of cognition. This is argued in a powerful way by Newell in his contribution to this volume. For this reason, the Soar system will probably play an important integrative role within cognitive science in bringing together important subdomains of psychology, computer science, linguistics, and the neurosciences. Although Soar is not the only `architecture for intelligence', it is one of the most advanced and theoretically best motivated architectures presently available. Soar: A Cognitive Architecture in Perspective is of special interest to researchers in the domains of cognitive science, computer science and artificial intelligence, cognitive psychology, and the philosophy of mind.
How to Build a Brain provides a detailed exploration of a new cognitive architecture - the Semantic Pointer Architecture - that takes biological detail seriously, while addressing cognitive phenomena. Topics ranging from semantics and syntax, to neural coding and spike-timing-dependent plasticity are integrated to develop the world's largest functional brain model.
A concise introduction to a complex field, bringing together recent work in cognitive science and cognitive robotics to offer a solid grounding on key issues. This book offers a concise and accessible introduction to the emerging field of artificial cognitive systems. Cognition, both natural and artificial, is about anticipating the need for action and developing the capacity to predict the outcome of those actions. Drawing on artificial intelligence, developmental psychology, and cognitive neuroscience, the field of artificial cognitive systems has as its ultimate goal the creation of computer-based systems that can interact with humans and serve society in a variety of ways. This primer brings together recent work in cognitive science and cognitive robotics to offer readers a solid grounding on key issues. The book first develops a working definition of cognitive systems—broad enough to encompass multiple views of the subject and deep enough to help in the formulation of theories and models. It surveys the cognitivist, emergent, and hybrid paradigms of cognitive science and discusses cognitive architectures derived from them. It then turns to the key issues, with chapters devoted to autonomy, embodiment, learning and development, memory and prospection, knowledge and representation, and social cognition. Ideas are introduced in an intuitive, natural order, with an emphasis on the relationships among ideas and building to an overview of the field. The main text is straightforward and succinct; sidenotes drill deeper on specific topics and provide contextual links to further reading.
The book focuses on a conceptual flaw in contemporary artificial intelligence and cognitive science. Many people have discovered diverse manifestations and facets of this flaw, but the central conceptual impasse is at best only partially perceived. Its consequences, nevertheless, visit themselves asdistortions and failures of multiple research projects - and make impossible the ultimate aspirations of the fields. The impasse concerns a presupposition concerning the nature of representation - that all representation has the nature of encodings: encodingism. Encodings certainly exist, butencodingism is at root logically incoherent; any programmatic research predicted on it is doomed too distortion and ultimate failure. The impasse and its consequences - and steps away from that impasse - are explored in a large number of projects and approaches. These include SOAR, CYC, PDP, situated cognition, subsumption architecture robotics, and the frame problems - a general survey of the current research in AI and Cognitive Science emerges. Interactivism, an alternative model of representation, is proposed and examined.
From the Foreword: "In this book Joscha Bach introduces Dietrich Dörner's PSI architecture and Joscha's implementation of the MicroPSI architecture. These architectures and their implementation have several lessons for other architectures and models. Most notably, the PSI architecture includes drives and thus directly addresses questions of emotional behavior. An architecture including drives helps clarify how emotions could arise. It also changes the way that the architecture works on a fundamental level, providing an architecture more suited for behaving autonomously in a simulated world. PSI includes three types of drives, physiological (e.g., hunger), social (i.e., affiliation needs), and cognitive (i.e., reduction of uncertainty and expression of competency). These drives routinely influence goal formation and knowledge selection and application. The resulting architecture generates new kinds of behaviors, including context dependent memories, socially motivated behavior, and internally motivated task switching. This architecture illustrates how emotions and physical drives can be included in an embodied cognitive architecture. The PSI architecture, while including perceptual, motor, learning, and cognitive processing components, also includes several novel knowledge representations: temporal structures, spatial memories, and several new information processing mechanisms and behaviors, including progress through types of knowledge sources when problem solving (the Rasmussen ladder), and knowledge-based hierarchical active vision. These mechanisms and representations suggest ways for making other architectures more realistic, more accurate, and easier to use. The architecture is demonstrated in the Island simulated environment. While it may look like a simple game, it was carefully designed to allow multiple tasks to be pursued and provides ways to satisfy the multiple drives. It would be useful in its own right for developing other architectures interested in multi-tasking, long-term learning, social interaction, embodied architectures, and related aspects of behavior that arise in a complex but tractable real-time environment. The resulting models are not presented as validated cognitive models, but as theoretical explorations in the space of architectures for generating behavior. The sweep of the architecture can thus be larger-it presents a new cognitive architecture attempting to provide a unified theory of cognition. It attempts to cover perhaps the largest number of phenomena to date. This is not a typical cognitive modeling work, but one that I believe that we can learn much from." --Frank E. Ritter, Series Editor Although computational models of cognition have become very popular, these models are relatively limited in their coverage of cognition-- they usually only emphasize problem solving and reasoning, or treat perception and motivation as isolated modules. The first architecture to cover cognition more broadly is PSI theory, developed by Dietrich Dorner. By integrating motivation and emotion with perception and reasoning, and including grounded neuro-symbolic representations, PSI contributes significantly to an integrated understanding of the mind. It provides a conceptual framework that highlights the relationships between perception and memory, language and mental representation, reasoning and motivation, emotion and cognition, autonomy and social behavior. It is, however, unfortunate that PSI's origin in psychology, its methodology, and its lack of documentation have limited its impact. The proposed book adapts Psi theory to cognitive science and artificial intelligence, by elucidating both its theoretical and technical frameworks, and clarifying its contribution to how we have come to understand cognition.
"The question for me is how can the human mind occur in the physical universe. We now know that the world is governed by physics. We now understand the way biology nestles comfortably within that. The issue is how will the mind do that as well."--Allen Newell, December 4, 1991, Carnegie Mellon University The argument John Anderson gives in this book was inspired by the passage above, from the last lecture by one of the pioneers of cognitive science. Newell describes what, for him, is the pivotal question of scientific inquiry, and Anderson gives an answer that is emerging from the study of brain and behavior. Humans share the same basic cognitive architecture with all primates, but they have evolved abilities to exercise abstract control over cognition and process more complex relational patterns. The human cognitive architecture consists of a set of largely independent modules associated with different brain regions. In this book, Anderson discusses in detail how these various modules can combine to produce behaviors as varied as driving a car and solving an algebraic equation, but focuses principally on two of the modules: the declarative and procedural. The declarative module involves a memory system that, moment by moment, attempts to give each person the most appropriate possible window into his or her past. The procedural module involves a central system that strives to develop a set of productions that will enable the most adaptive response from any state of the modules. Newell argued that the answer to his question must take the form of a cognitive architecture, and Anderson organizes his answer around the ACT-R architecture, but broadens it by bringing in research from all areas of cognitive science, including how recent work in brain imaging maps onto the cognitive architecture.
Stan Franklin is the perfect tour guide through the contemporary interdisciplinary matrix of artificial intelligence, cognitive science, cognitive neuroscience, artificial neural networks, artificial life, and robotics that is producing a new paradigm of mind. Along the way, Franklin makes the case for a perspective that rejects a rigid distinction between mind and non-mind in favor of a continuum from less to more mind.
