Formal Specification of Cognitive Models in CARINA

Computational psychology has the aim to explain human cognition by computational models of cognitive processes. The cognitive architecture Adaptive Control of Thought--Rational (ACT-R) is popular to develop such models. Although ACT-R has a well-defined psychological theory and has been used to explain many cognitive processes, there are two problems that make it hard to reason formally about its cognitive models: First, ACT-R lacks a computational formalization of its underlying production rule system, and, second, there are many different implementations and extensions of ACT-R with many technical artifacts complicating formal reasoning even more. This article describes a formal operational semantics—the very abstract semantics —that abstracts from as many technical details as possible, keeping it open to extensions and different implementations of the ACT-R theory. In a second step, this semantics is refined to define some of its abstract features that are found in many implementations of ACT-R—called the abstract semantics . It concentrates on the procedural core of ACT-R and is suitable for analysis of the general transition system, since it still abstracts from details like timing, the sub-symbolic layer of ACT-R or conflict resolution. Furthermore, a translation of ACT-R models to the declarative programming language Constraint Handling Rules (CHR) is defined. This makes the abstract semantics an executable specification of ACT-R. CHR has been used successfully to embed other rule-based formalisms like graph transformation systems or functional programming. There are many theoretical results and practical tools that support formal reasoning about and analysis of CHR programs. The translation of ACT-R models to CHR is proven sound and complete w.r.t. the abstract operational semantics of ACT-R. This paves the way to analysis of ACT-R models through CHR analysis results and tools. Therefore, to the best of our knowledge, our abstract semantics is the first abstract formulation of ACT-R suitable for both analysis and execution.

ate computer simulations that perform the tasks that humans perform by simulating the way humans process information. These simulations of behavior, called cognitive models, are theories of the knowledge and the mechanisms that give rise to behavior. The sets of mechanisms are assumed to be fixed across tasks, which allow them to be realized as a reusable computer program that corresponds to the architecture of cognition, or cognitive architecture. (More complete explanations are provided by Newell’s [1990] Unified Theories of Cognition, by Anderson’s ACT-R work [Anderson et al., 2004], and by ongoing work with connectionist and neural architectures.) Because cognitive models increasingly allow us to predict behavior and explain the mechanisms behind behavior, they have many applications. They can support design activities, and they serve in many roles where intelligence is needed. As a result, interest in cognitive models and architectures can be found in several areas: Researchers in psychology and cognitive science are interested in them as theories. Researchers in human factors, in synthetic environments, and in intelligent systems are interested in them for applications and design. Researchers in applied domains such as video games and technical applications such as trainers are interested in them as simulated colleagues and opponents. Although some earlier precursors can be found, the main work on cognitive models began in about 1960 (Newell, Shaw, & Simon, 1960). These models have now reached a new level of maturity. For example, a review commissioned by the National Research Council (Pew & Mavor, 1998) found that cognitive models had been developed to a level that made them useful in synthetic environments. A later review (Ritter et al., 2003) examined cognitive architectures created outside the United States and found similar results. Both reviews recommended a list of future projects, which are being undertaken by individual researchers. These and similar projects have been increasingly seen in requests for proposals put out by funding agencies around the world. Results and interest in cognitive models and architectures are rising.

Cognitive modeling is a fundamental tool used to understand the processes that underlie behavior, and has become a standard technique in the cognitive sciences. Cognitive modeling depends on the use of cognitive architectures. CARINA is a metacognitive architecture for artificial intelligent agents, derived from the MISM Metacognitive Metamodel that integrates self-regulation and metamemory with support for the metacognitive mechanisms of introspective monitoring and meta-level control. In this paper we show the cognitive modeling process that allows to develop computerized model for simulating human problem solving and mental task processes in metacognitive architecture CARINA. Finally, a runnable cognitive model of syntactic analysis of a sentence is shown.

What can cognitive architectures do for robotics?

In this paper, Cognitive Artificial Intelligence computing modeling process in Meta cognitive architecture CARINA is implemented. Basically in cognitive sciences, cognitive modeling has become fundamental tool to process. Based on the usage of cognitive architectures, cognitive modeling is designed. For the artificial intelligent agents, CARNIA is most widely used and this is a Meta cognitive architecture which is derived from the Meta cognitive Meta model. This Meta cognitive Meta model is based on the Meta cognitive mechanism which will monitor and control the Meta level. Initially, the cognitive task is selected. Next, the information is described for the cognitive task. By using natural language, the cognitive task is described. GOMS also describes the cognitive task. For the obtained data, decision functions and cognitive functions are described. By using artificial intelligence, the data is computed. Now, the data is transferred from Cognitive Model form GOMS to M + + Language. Now, the data will be performed by using cognitive model in CARNIA. At last testing and maintenance will be done very effectively. From results it can observe that accuracy, cost, errors and observation time gives effective result. KeywordsMeta cognitive architectureCARNIAArtificial IntelligenceGOMS (GoalsOperatorsMethods and Selection)

This chapter presents a brief overview of theories and concepts that ar some well-established and widely used cognitive architectures, such as ACT-R (Anderson et al (2004) Psychol Rev 111(4):1036–1060; Anderson (2007) How can the human mind occur in the physical universe? Oxford series on cognitive models and architectures. Oxford University Press, Oxford) and SOAR (Laird (2012) Soar cognitive architecture. MIT Press, Cambridge). These are computational attempts to model cognition for general and complete tasks rather than for single, small tasks. This chapter also reviews the most known and used cognitive models, KLM and GOMS, which are computational models used for simulations of human performance and behavior (Ritter et al (2000) ACM Trans Comput-Hum Interact 7(2):141–173. https://doi.org/10.1145/353485.353486). We will show how some cognitive architectures that originated within artificial intelligence (AI) have been developed to cover aspects of cognitive science, and vice versa. The relevance of the cognitive approach to HCI can be seen in the successful use of cognitive models in the HCI community to evaluate designs, assist users’ interactions with computers, and substitute users in simulations (Ritter et al (2000) ACM Trans Comput-Hum Interact 7(2):141–173. https://doi.org/10.1145/353485.353486).

Computational models of cognitive processes may be employed in cyber-security tools, experiments, and simulations to address human agency and effective decision-making in keeping computational networks secure. Cognitive modeling can addresses multi-disciplinary cyber-security challenges requiring cross-cutting approaches over the human and computational sciences such as the following: (a) adversarial reasoning and behavioral game theory to predict attacker subjective utilities and decision likelihood distributions, (b) human factors of cyber tools to address human system integration challenges, estimation of defender cognitive states, and opportunities for automation, (c) dynamic simulations involving attacker, defender, and user models to enhance studies of cyber epidemiology and cyber hygiene, and (d) training effectiveness research and training scenarios to address human cyber-security performance, maturation of cyber-security skill sets, and effective decision-making. Models may be initially constructed at the group-level based on mean tendencies of each subject's subgroup, based on known statistics such as specific skill proficiencies, demographic characteristics, and cultural factors. For more precise and accurate predictions, cognitive models may be fine-tuned to each individual attacker, defender, or user profile, and updated over time (based on recorded behavior) via techniques such as model tracing and dynamic parameter fitting.

The Towers of Hanoi is a mathematical problem, which consists of three pegs, and a number “n” of disks of distinct sizes which can slide onto any peg. A cognitive model is a theoretical, empirical and computational representation of mental processes which belong to a cognitive function. A cognitive model generates a human-like performance for developing tasks, correcting errors, using strategies, and acquiring knowledge. A cognitive model constructed in a cognitive architecture is characterized to be runnable and producing specific behaviors. A cognitive architecture is a general-purpose control framework based on scientific theories to specify computational models of human cognitive performance. CARINA is a metacognitive architecture for artificial intelligent agents, obtained from the MISM Metacognitive Meta model. The objective of this paper is to present a cognitive model based on NGOMS-L to solve the algorithm of the Towers of Hanoi that can be runnable in the metacognitive architecture CARINA. The methodology used for the analysis of the cognitive task was: pre-processing stage, processing stage, classification of subjects, description of cognitive task in natural language and finally description of the cognitive task in NGOMS-L. The results obtained showed that of the four subjects originally selected, three of them were able to solve the problem and only one abandoned the problem. In the classification made, a successful and an unsuccessful subject was selected, to represent the cognitive task in natural language and finally express in the NGOMS-L notation the different Goals, Operators, Methods, and Selection Rules.

Language acquisition is supported by phonological awareness, which intentionally makes children aware of phonological units. By understanding the internal processes of children during language acquisition, this study aims to elucidate factors that can correct erroneous phonological generation. Therefore, we developed a cognitive model using innate and experiential factors of the memory retrieval in the cognitive architecture–ACT-R. Furthermore, we performed simulations using Shiritori, a Japanese word game, as an interaction task. The simulation included the observation of effects of the experiential factor of repeating a task and innate factors of different settings. It showed that repeating a single task causes incorrect convergence, and this convergence can be prevented by comprehensive activation of overall phonological knowledge during the interval of Shiritori tasks. Moreover, the simulation in specific innate settings exhibited commonalities with cases of developmental disorder by showing errors like consonant deletion. In the future, we will examine the correlation of the aforementioned findings with actual language development to realize the use of cognitive architecture in real world.

Reviews lssues in Applied Linguistics Vol. 1 No. 1 As an introductory linguistics text, one expects the goal of Contemporary Linguistics to be to get people started in the field of linguistics and related disciplines. In spite of a few minor problems the authors have not only achieved this goal in a more-than-adequate fashion, they have also demonstrated that an introductory text can be comprehensive enough for the non-initiated to become acquainted with the complexities which the study of language entails. For the practicing linguist, Contemporary Linguistics could be a good reference source. For the linguistics teacher, it should serve as an excellent course text. For the student, it is particularly helpful because of the straightforward, explanatory style the authors have adopted. In all, Contemporary Linguistics fills a major gap, since state-of-the-art introductory linguistics texts are not plentifufon the market. REFERENCES Yule, G. (1985). The study of language: An introduction. Cambridge: Cambridge University Press. Stephen Adewole is a Ph.D. student in the Department of Linguistics at UCLA. His interests include the teaching and analysis of African languages. (Received April 25, 1990) Microcognition 1989. 226 pp. by Andy Clark. Cambridge, M A : M I T Press Reviewed by Cheryl Fantuzzi University of California, Los Angeles In the true spirit of integrative, interdisciplinary cognitive science research, philosopher Andy Clark seeks to call a truce to what he aptly refers to as the holy war of cognitive modeling. In complex matters of the mind, it is Clark's contention that since the brain may simply need and use more than one mode of information- processing to perform such tasks as comprehending and producing human language, a more integrative theory of cognition is therefore needed. In Microcognition, Clark introduces the reader to some of the major issues and conundrums of 20th-century philosophy and cognitive science: What is the mind, and what is its relationship to the brain? How does the mind work, and how does it affect behavior? How do seemingly abstract entities, such as personal beliefs and desires, influence bodily movement, and how might sentential propositions actually be instantiated in neural stuff? Clark gives us a quick overview of the modern answer to the mind/body dilemma: the computational model of mind. Most interestingly, he brings us into the heart of the currently lively debate between two competing computer models of the mind: the classical symbol systems of AI (Artificial Intelligence) and the new connectionist movement, also known as PDP (Parallel Distributed Processing). This new debate in cognitive science is reminiscent of the heated discussions between Chomsky and the behaviorists some 30 years ago. One central issue, now as then, concerns the systematicity of thought and language and the explanatory adequacy of a generative grammar for language, as opposed to a 'subsymbolic,' or associationist, view of both cognition and language. The symbolic computational approach of classical AI, in fact, represents the mathematical-syntactic approach of Chomskian generative grammar par excellence, since it conceptualizes the mind as a type of formal logic machine with mental rules operating on abstract, symbolic representations. The subsymbolic connectionist approach, on the other hand, operates without rules and without symbols. It emphasizes the messy, biological substrate of cognition, rather than the rational, logical nature of thinking, and uses a computational architecture closer to the distributed, parallel nature of processing in the brain, rather than the conventional sequential architecture of a digital computer. Clark's thesis is that cognitive modeling may actually require both classical and connectionist 'cognitive architecture' to model two different kinds of information processing: for some tasks, our thinking does appear to be slow, deliberate and serial, while for others, such as visual perception, it is fast, automatic and parallel. Microcognition is divided into two parts: the mind's-eye view, which describes classical AI and some of the philosophical criticisms against it, and the brain's-eye view, which focuses on

Contributors to this volume have explored the ways in which cognitive models or architectures may be helpful or even essential for building simulations. In this epilogue, I shall be considering whether cognitive models are always necessary – is a social simulation necessarily inadequate if it has no or only a very simple model of cognition? If not, is it possible to specify classes of simulations for which cognitive models are necessary or unnecessary?

In this article, I present the ideas and trends that have given rise to the use of cognitive architectures in human factors and provide a cognitive engineering-oriented taxonomy of these architectures and a snapshot of their use for cognitive engineering. Architectures of cognition have had a long history in human factors but a brief past. The long history entails a 50-year preamble, whereas the explosion of work in the current decade reflects the brief past. Understanding this history is key to understanding the current and future prospects for applying cognitive science theory to human factors practice. The review defines three formative eras in cognitive engineering research: the 1950s, 1980s, and now. In the first era, the fledging fields of cognitive science and human factors emphasized characteristics of the dancer the limited capacity or bounded rationality view of the mind, and the ballroom, the task environment. The second era emphasized the dance (i.e., the dynamic interaction between mental operations and task environment). The third era has seen the rise of cognitive architectures as tools for choreographing the dance of mental operations within the complex environments posed by human factors practice. Hybrid architectures present the best vector for introducing cognitive science theories into a renewed engineering-based human factors. The taxonomy provided in this article may provide guidance on when and whether to apply a cognitive science or a hybrid architecture to a human factors issue.

Multimodal behavior has been studied for a long time and in many fields, e.g., in psychology, linguistics, communication studies, education, and ergonomics. One of the main motivations has been to allow humans to use technical systems intuitively, in a way that resembles and fosters human users' natural way of interacting and thinking [Oviatt 2013]. This has sparked early work on multimodal human-computer interfaces, including recent approaches to recognize communicative behavior and even subtle multimodal cues by computer systems. Those approaches, for the most part, rest on machine learning techniques applied to large sets of behavioral data. As datasets grow larger in size and coverage, and computational power increases, suitable data-driven techniques are able to detect correlational behavior patterns that support answering questions like which feature( s) to take into account or how to recognize them in specific contexts. However, natural multimodal interaction in humans entails a plethora of behavioral variations and intricacies (e.g., when to act unimodally vs. multimodally, with which specific behaviors or multi-level coordination between them). Possible underlying patterns are hard to detect, even in large datasets, and often such variations are attributed to context-dependencies or individual differences. How they come about is still hard to explain at the behavioral level.

Whereas game theorists and logicians use formal methods to investigate ideal strategic behavior, many cognitive scientists use computational cognitive models of the human mind to predict and simulate human behavior. In this paper, we aim to bring these fields closer together by creating a generic translation system which, starting from a strategy for a turn-based game represented in formal logic, automatically generates a computational model in the Primitive Information Processing Elements (PRIMs) cognitive architecture, which has been validated on various experiments in cognitive psychology. The PRIMs models can be run and fitted to participants’ data in terms of decisions, response times, and answers to questions. As a proof of concept, we run computational modeling experiments on the basis of a game-theoretic experiment about the turn-based game “Marble Drop with Surprising Opponent”, in which the opponent often starts with a seemingly irrational move. We run such models starting from logical representations of several strategies, such as backward induction and extensive-form rationalizability, as well as different player types according to stance towards risk and level of theory of mind. Hereby, response times and decisions for such centipede-like games are generated, which in turn leads to concrete predictions for future experiments with human participants. Such precise predictions about different aspects, including reaction times, eye movements and active brain areas, cannot be derived on the basis of a strategy logic by itself: the computational cognitive models play a vital role and our generic translation system makes their construction more efficient and systematic than before.

Driving is a multitasking activity that requires drivers to manage their attention among various driving- and non-driving-related tasks. When one models drivers as continuous controllers, the discrete nature of drivers’ control actions is lost and with it an important component for characterizing behavioral variability. A proposal is made for the use of cognitive architectures for developing models of driver behavior that integrate cognitive and perceptual-motor processes in a serial model of task and attention management. A cognitive architecture is a computational framework that incorporates built-in, well-tested parameters and constraints on cognitive and perceptual-motor processes. All driver models implemented in a cognitive architecture necessarily inherit these parameters and constraints, resulting in more predictive and psychologically plausible models than those that do not characterize driving as a multitasking activity. These benefits are demonstrated with a driver model developed in the ACT-R cognitive architecture. The model is validated by comparing its behavior to that of human drivers navigating a four-lane highway with traffic in a fixed-based driving simulator. Results show that the model successfully predicts aspects of both lower-level control, such as steering and eye movements during lane changes, and higher-level cognitive tasks, such as task management and decision making. Many of these predictions are not explicitly built into the model but come from the cognitive architecture as a result of the model’s implementation in the ACT-R architecture.