Cognitive Architectures and Multi-agent Social Simulation

As we know, a cognitive architecture is a domain-generic computational cognitive model that may be used for a broad analysis of cognition and behavior. Cognitive architectures embody theories of cognition in computer algorithms and programs. Social simulation with multi-agent systems can benefit from incorporating cognitive architectures, as they provide a realistic basis for modeling individual agents (as argued in Sun 2001). In this survey, an example cognitive architecture will be given, and its application to social simulation will be sketched. 1 Defining Cognitive Architectures As we know, a cognitive architecture is a broadly-scoped, domain-generic computational cognitive model, capturing essential structures and processes of the mind, to be used for a broad, multiple-level, multiple-domain analysis of cognition and behavior (Newell 1990, Sun 2002). The architecture for a building consists of its overall framework and its overall design, as well as roofs, foundations, walls, windows, floors, and so on. Furniture and appliances can be easily rearranged and/or replaced and therefore they are not part of the architecture. By the same token, a cognitive architecture includes overall structures, essential divisions of modules, relations between modules, basic representations, essential algorithms, and a variety of other aspects (Sun 2004). In general, an architecture includes those aspects of a system that are relatively invariant across time, domains, and individuals. It deals with componential processes of cognition in a structurally and mechanistically well defined way. For cognitive science (i.e., in relation to understanding the human mind), a cognitive architecture provides a concrete framework for more detailed modeling of cognitive phenomena, through specifying essential structures, divisions of modules, relations between modules, and so on. Its function is to provide an essential framework to facilitate more detailed modeling and understanding of various components and processes of the mind. Research in computational cognitive modeling explores the essence of cognition and various cognitive functionalities through developing detailed, process-based understanding by specifying computational models of mechanisms and processes. It embodies descriptions of cognition in computer algorithms and programs. That is, it produces runnable computational models. Detailed simulations are then conducted based on the computational models. In this enterprise, a cognitive architecture may be used for a broad, multiple-level, multiple-domain analysis of cognition. In relation to building intelligent systems, a cognitive architecture specifies the underlying infrastructure for intelligent systems, which includes a variety of capabilities, modules, and subsystems. On that basis, application systems can be more easily developed. A cognitive architecture carries also with it theories of cognition and understanding of intelligence gained from studying the human mind. Therefore, the development of intelligent systems can be more cognitively grounded, which may be advantageous in many circumstances. 2 The Importance of Cognitive Architectures This work is specifically concerned with psychologically oriented cognitive architectures (as opposed to software engineering oriented “cognitive” architectures): their importance and their applications. Psychologically oriented cognitive architectures are particularly important because (1) they are “intelligent” systems that are cognitively realistic (relatively speaking) and therefore they are more human-like in many ways, (2) they shed new light on human cognition and therefore they are useful tools for advancing the science of cognition, (3) furthermore, they may (in part) serve as a foundation for understanding collective human behavior and social phenomena (to be detailed later). Let us examine the importance of this type of cognitive architecture. For cognitive science, the importance of such cognitive architectures lie in the fact that they are enormously useful in terms of understanding the human mind. In understanding cognitive phenomena, the use of computational simulation on the basis of cognitive architectures forces one to think in terms of process, and in terms of detail. Instead of using vague, purely conceptual theories, cognitive architectures force theoreticians to think clearly. They are critical tools in the study of the mind. Researchers who use cognitive architectures must specify a cognitive mechanism in sufficient detail to allow the resulting models to be implemented on computers and run as simulations. This approach requires that important elements of the models be spelled out explicitly, thus aiding in developing better, conceptually clearer theories. An architecture serves as an initial set of assumptions to be used for further modeling of cognition. These assumptions, in reality, may be based on either available scientific data (for example, psychological or biological data), philosophical thoughts and arguments, or ad hoc working hypotheses (including computationally inspired such hypotheses). An architecture is useful and important precisely because it provides a comprehensive initial framework for further modeling in a variety of task domains. Cognitive architectures also provide a deeper level of explanation. Instead of a model specifically designed for a specific task (often in an ad hoc way), using a cognitive architecture forces modelers to think in terms of the mechanisms and processes available within a generic cognitive architecture that are not specifically designed for a particular task, and thereby to generate explanations of the task that is not centered on superficial, high-level features of a task, that is, explanations of a deeper kind. To describe a task in terms of available mechanisms and processes of a cognitive architecture is to generate explanations centered on primitives of cognition as envisioned in the cognitive architecture, and therefore such explanations are deeper explanations. Because of the nature of such deeper explanations, this style of theorizing is also more likely to lead to unified explanations for a large variety of data and/or phenomena, because potentially a large variety of task data and phenomena can be explained on the basis of the same set of primitives provided by the same cognitive architecture. Therefore, using cognitive architectures leads to comprehensive theories of the mind (Newell 1990, Anderson and Lebiere 1998, Sun 2002). On the other hand, for the fields of artificial intelligence and computational intelligence (AI/CI), the importance of cognitive architectures lies in the fact that they support the central goal of AI/CI—building artificial systems that are as capable as human beings. Cognitive architectures help us to reverse engineer the only truly intelligent system around—the human being, and in particular, the human mind. They constitute a solid basis for building truly intelligent systems, because they are well motivated by, and properly grounded in, existing cognitive research. The use of cognitive architectures in building intelligent systems may also facilitate the interaction between humans and artificially intelligent systems because of the similarity between humans and cognitively based intelligent systems. 3 Levels of Explanations A broader perspective on the social and behavioral sciences may lead to a view of multiple “levels” of analysis encompassing multiple disciplines in the social and cognitive sciences. That is, a set of related disciplines, may be readily cast as a set of different levels of analysis, from the most macroscopic to the most microscopic. These different levels include: the sociological level, the psychological level, the componential level, and the physiological level. In other words, as has been argued in Sun et al (2005), one may view different disciplines as different levels of abstraction in the process of exploring essentially the same broad set of questions (cf. Newell 1990). See Figure 1. level object of analysis type of analysis model 1 inter-agent processes social/cultural collections of agent models 2 agents psychological individual agent models 3 intra-agent processes componential modular constr. of agent models 4 substrates physiological biological realization of modules Fig. 1. A hierarchy of four levels. First of all, there is the sociological level, which includes collective behaviors of agents, inter-agent processes, sociocultural processes, social structures and organizations, as well as interactions between agents and their (physical and sociocultural) environments. Although studied extensively by sociology, anthropology, political science, and economics, this level has traditionally been very much ignored in cognitive science. Only recently, cognitive science, as a whole, has come to grip with the fact that cognition is, at least in part, a sociocultural process. 1 The next level is the psychological level, which covers individual experiences, individual behaviors, individual performance, as well as beliefs, concepts, and skills employed by individual agents. In relation to the sociological level, the relationship of individual beliefs, concepts, and skills with those of the society and the culture, and the processes of change of these beliefs, concepts, and skills, independent of or in relation to those of the society and the culture, may be investigated (in inter-related and mutually influential ways). At this level, one may examine human behavioral data, compared with models (which may be based on cognitive architectures) and with insights from the sociological level and details from the lower levels. The third level is the componential level. At this level, one studies and models cognitive agents in terms of components (e.g., in the form of a cognitive architecture), with the theoretical language of a particular paradigm (for example, symbolic computation or connectionist networks, or their combinations thereof). At this level, one may specify computationally an overall architecture consisting of multiple components therein. One may also specify some essential computational processes of each component as well as essential connections among components. That is, one imputes a computational process onto a cognitive function. Ideas and data from the psychological level (that is, the psychological constraints from above), which bear significantly on the division of components and their possible implementations, are among the most important considerations. This level may also incorporate biological/physiological facts regarding plausible divisions and their implementations (that is, it can incorporate ideas from the next level down — the physiological level, which offers the biological constraints). This level results in mechanisms (though they are computational and thus somewhat abstract compared with physiological-level details). 2 Although this level is essentially in terms of intra-agent processes, computational models (cognitive architectures) developed therein may be used to capture processes at higher levels, including interaction at a sociological level whereby multiple individuals are involved. This can be accomplished, for example, by examining interactions of multiple copies of individual agent models (based on the same cognitive architecture) or those of different individual agent models (based on different cognitive architectures). One may use computation as a means for 1 See Sun (2001) for a more detailed argument for the relevance of sociocultural processes to cognition and vice versa. 2 The importance of this level has been argued for, for example, in Anderson and Lebiere (1998), and Sun et al (2004). constructing cognitive architectures at a sub-agent level (the componential level), but one may go up from there to the psychological level and to the sociological level (see the discussion regarding mixing levels in Sun et al 2005). The lowest level of analysis is the physiological level, that is, the biological substrate, or the biological implementation, of computation. This level is the focus of a range of disciplines including biology, physiology, computational neuroscience, cognitive neuroscience, and so on. Although biological substrates are not our main concern here, they may nevertheless provide useful input as to what kind of computation is likely employed and what a plausible architecture (at a higher level) should be like. The main utility of this level is to facilitate analysis at higher levels, that is, analysis using low-level information to narrow down choices in selecting computational architectures as well as choices in implementing componential computation. 3 In this enterprise of multiple levels in cognitive and social sciences, a cognitive architecture may serve as a centerpiece, tying together various strands of research. It may serve this purpose due to the comprehensiveness of its functionality and the depth with which it has been developed (at least for some psychologically oriented/grounded cognitive architectures). Thus, detailed mechanisms are developed within a cognitive architecture, which may be tied to low-level cognitive processes, while a cognitive architecture as a whole may function at a very high level of cognitive and social processes. 4 An Example Cognitive Architecture