• Go to issue no. 2 (Summer, 2002)
• Go to issue no. 3/4 (Special Issue, Fall/Winter, 2002)

The changing meanings of force

Christos Ioannides (U. of Cyprus) & Stella Vosniadou (U. of Athens, Greece)

pp. 5-62

The research reported in this paper investigated developmental changes in the meaning of force in 105 children ranging in age from 4 to 15 years. The subjects attended the same school in Thessaloniki, in the North of Greece, and came from predominately middle class backgrounds. In individual interviews the children were shown 27 drawings of physical objects in combinations of different sizes and kinetic states, and were asked to determine which forces were being exerted on these objects, if any. Children's responses to these questions were analyzed following a methodology developed by Vosniadou & Brewer (1992, 1994). The results showed that most of the children (88.6%) made use of a small number of relatively well-defined and internally consistent interpretations of force. The discovered meanings of force varied significantly with age. The younger children thought that force is an internal property of objects related to their weight (internal force meaning) while the older children thought that force is an acquired property of objects that move, as the result of an agent pushing or pulling them (acquired force meaning). The acquired force meaning was well established by the age of twelve years and not substantially changed despite the systematic instruction in Newtonian mechanics that takes place in Greek high schools. Under the influence of instruction children added the force of push/ pull and the force of gravity interpretations to the existing acquired force meaning creating synthetic meanings of force. The implications of these results for a theory of conceptual change are discussed.

Examples, rules and strategies in the control of dynamic systems

Wolfgang Schoppek (U. of Bayreuth, Germany)

pp. 63-92

Two main types of knowledge are considered relevant to successful control of dynamic systems: input-output knowledge (I-O-knowledge), which represents specific input values together with the corresponding output values, and structural knowledge, defined as general knowledge about the variables of a system and their causal relations. While I'O-knowledge has proven important for the control of small systems, structural knowledge is expected to enhance performance when dealing with more complex systems. In an experiment, structural knowledge about a complex system was manipulated. Although the experimental group had better structural knowledge, the control group was equally successful in reaching new goals. That seems to contradict other studies where effects of structural knowledge on performance have been found. To resolve these contradictions, the consideration of a third type of knowledge - strategic knowledge - is suggested. The postulated effects of different levels of structural and strategic knowledge are explored with a computational model. The three knowledge types are used to interpret the variety of findings within a unitary conceptual framework.

Modularity, relativism, and neural constructivism

Josefa Toribio (U. of Sussex, UK)

pp. 93-106

Fodor (1983) claims that the modularity of mind (the relatively encapsulated, insulated, special-purpose nature of the psychological mechanisms of perception) helps undermine relativism in various forms. I shall show first, that the modular vision of mind provides insufficient support for the rejection of (most forms of) relativism, and second, that an alternative ('neural constructivist') model may, in fact, provide a better empirical response to the relativist challenge.

Go to Top of Page

Local Associations and Global Reason: Fodor's Frame Problem
and Second-Order Search

Andy Clark (U. of Sussex, UK)

pp. 115-140

Kleinberg (1997) describes a novel procedure for efficient search in a dense hyper-linked environment, such as the world wide web. The procedure exploits information implicit in the links between pages so as to identify patterns of connectivity indicative of 'authoritative
sources'. At a more general level, the trick is to use this connectivity signature to rapidly and cheaply identify the knowledge-structures most likely to be relevant given a specific input. I shall argue that Kleinberg's procedure is suggestive of a new, and demonstrably computationally viable, kind of solution to (at least one incarnation of at least one major aspect of) the so-called 'Frame Problem' in cognitive science. For Kleinberg's procedure suggests a new variety of 'fast and frugal heuristic' (Gigerenzer & Todd, 1999) capable of pressing maximum utility from the vast bodies of information and associations commanded by the biological brain. The paper thus takes up the challenge laid down by Fodor (1983, 2000), who depicts the problem of global knowledge-based reason as the point source of many paradigmatic failings of contemporary computational theories of mind. Whether the specific solution suggested is neurally plausible remains open to question. But at the very least I display an existence proof of the viability of a specific computational solution to the problem of search in a massive data-base. Moreover, it is a solution that may be suggestive, in some more general way, depending as it does upon the use of a rather novel kind of 'second order search'.

'Know the Method your Subject is using'' and 'Never Average over Methods': An Application of Newell's Admonition to Letter-Matching

Josette Marquer and Maria Pereira (U. of Paris 5 and C.N.R.S., France)

pp. 141-162

A new procedure based upon the analysis of verbal reports checked against apti-tude profiles and RTs was used. It suggests that subjects implement a much wider range of cognitive strategies in letter-matching than expected according to previ-ous work. The direct consequence is that in tasks of this type, and probably in others, whenever RT and error means are computed on whole samples they are averaged over a variety of methods and hence do not adequately reflect any of them.

Whether Skill Acquisition is Rule or Instance Based is determined by the Structure of the Task

Niels A. Taatgen and Dieter Wallach (U. of Groningen, Netherlands and U. of Basel, Switzerland)

pp. 163 -204

The traditional view of skill acquisition is that it can be explained by a gradual transition from behavior based on declarative rules in the form of examples or instructions towards general knowledge represented by procedural rules. This view is challenged by Logan's instance theory, which specifies that skill acquisition can be explained by the accumulation of examples or instances of the skill. The position defended in this paper is that both types of
learning can occur ' but their success will depend on the respective task. In the Sugar Factory task, it is very hard to determine the rule guiding the system, rule learning will thus fail while instance learning dominates. In the Anderson-Fincham task, mainly rule learning occurs, but variations in the task show evidence for some instance learning as well. Experiments involving both tasks are modeled using ACT-R, a hybrid cogni-tive architecture whose adaptive learning mechanisms seem to be well suited for modeling two very different tasks using the same methods.

Simulated Task Environments: The Role of High-Fidelity
Simulations, Scaled Worlds, Synthetic Environments, and
Laboratory Tasks in Basic and Applied Cog-nitive Research

Wayne D. Gray (GMU, Fairfax, VA, United States of America)

pp. 205-227

Simulated task environments provide a setting that adds controlled complexity to experimental tasks performed by human subjects in laboratory research. Researchers whose
problems are mostly applied may find that their problems are easier to study in a simulated task environment than in the actual task environ-ment. Researchers whose theories have been nurtured in the simple environments of the typical laboratory study may find that adding controlled complexity will allow them to study how the theoretical constructs they have studied in isolation interact with other constructs in a more complex task environment. In
this article I define a taxonomy and three dimensions of simulated task environments. The di-mensions are based on viewing simulated task environments from the perspectives of the researcher, the task, and the participants. Research on complex systems is inherently complex. It is my hope that the terms and distinctions introduced in this article will further the scientific enterprise by enabling us to spend less time explaining our paradigms and more time communicating our results.

Go to Top of Page

Editorial: Modeling Motivational Representations

Maria Miceli & Cristiano Castelfranchi (Institute of Cognitive Science and Technologies, National Research Council of Italy)

Using Human Behavior Models to Improve the Realism of Synthetic Agents

Barry G. Silverman, Michael Johns, Ransom Weaver, Kevin O?Brien, Rachel Silverman and Jason Cornwell (ACASA, U. of Pennsylvania, United States of America)

This paper describes an effort to integrate human behavior models from a range of ability, stress, emotion, decision theoretic, and motivation literatures into a game-theoretic framework. Our goal is to create a common mathematical framework (CMF) and a simulation environment that allows one to research and explore alternative behavior models to add realism to embodied software agents, e.g., human reaction to noise and stress, bounded rationality, emotive states, and cultural influences. Our CMF is based on Markov chains representing states of the world that the agents can think about and act upon. In these worlds the agents' utilities (payoffs) are derived by a deep model of cognitive appraisal of intention achievement including assessment of emotional activation/decay relative to concern ontologies, and subject to (integrated) stress and related constraints. We present the progress to date on the mathematical framework, and on an environment for editing the various elements of the cognitive appraiser, utility generators, concern ontologies, and Markov chains. We summarize a prototype of an example training game for counter-terrorism and crowd management in which bounded rationality of the micro-decision making leads to emergence of macro-behavior. Future re-search needs are elaborated including validity issues and the gaps in the behavioral literatures that agent developers must struggle with.

Commitments: A Game-Theoretic and Logical Perspective

Vincent Buskens and Lambèr Royakkers (U. of Utrecht and Eindhoven University of Technology, Netherlands)

Many social interactions between agents demand the use of commitments to reach socially efficient or avoid socially inefficient outcomes. Commitments express the desires, goals, or intentions of the agents in an interaction. In this article, we distinguish between unilateral and
bilateral commitments, and between whether or not an agent has to agree with a commitment made by the other agent before the commitment becomes effective. Using a game-theoretic model, we will show that, depending on the incentive structure, different interactions require
dif-ferent types of commitments to reach socially efficient outcomes. Based on these results, we discuss whether existing (or slightly adapted) logical formalizations are adequate for the description of certain types of commitments and which formal-ization is suitable for
reaching a socially efficient outcome in a specific interaction. We claim that a logical formalization of commitment aiming at a socially efficient outcome should be based on assumptions about the type of interaction and the suitable type of commitment. A more general conclusion of this article is that game-theoretic arguments can help to provide specifications for logical for-malizations of systems of more agents if one has an idea about the incentive structure of the interaction.

Belief-Desire-Intention Agency in a General Cognitive Architecture

Renée Elio (U. of Alberta, Canada)

The term beliefs-desires-intentions (BDI) has been used to denote a position on theoretically useful mental state distinctions, particular models of how these mental states affect reasoning, and a genre of architectures or frameworks for developing software agents. Beyond the software agent community, elements of this BDI perspective receive a distinctly different realization in computational theories of human cognition. This paper examines whether and how a general cognitive architecture achieves the functionality attributed to a BDI model of agency, particularly in matters related to deliberation, goal formation, goal values, and the meta-level choice between deliberation and action. Finding the same themes of rational agency emerging at architectures specified at different levels of abstraction reinforces the intuitions driving these theories and provides new insights into how these intuitions translate into process models of cognition.

From Desires, Obligations and Norms to Goals

Frank Dignum and David Kinny
(U. of Utrecht, the Netherlands, and U. of Melbourne, Australia)

Traditional models of agents based on Beliefs, Desires and Intentions usually only include either desires or goals. Therefore the process whereby goals arise from desires is given scant attention. In this paper we argue that the inclusion of both desires and goals in the same model can be important, particularly in a Multi-Agent System context, where other sources of individual motivation such as obligations and norms may be present. This leads us to propose an extended BDI ar-chitecture in which obligations, norms and desires are distinguished from goals and explicitly represented. In this paper we consider suitable logical representations for and properties of these elements, and describe the basic method of operation of the architecture, focusing on how goal generation and goal maintenance may occur.

An Action Selection Mechanism for 'Conscious' Software Agents

Aregahegn S. Negatu and Stan Franklin (Institute of Intelligent Systems and Department of Computer Science, U. of Memphis, United States of America)

In this paper we describe an action selection mechanism for a "conscious" soft-ware agent. We discuss briefly the main cognitive modules of our agent architec-ture. Our focus is
on the operational/functional details of the "consciousness" module and the action selection mechanism and how these two work together. We describe how events come to "consciousness", how "conscious" events prime and bind to relevant actions/action-plans, and how the most relevant behavior/action is selected and executed. Our mechanisms provide the flexibility to interleave actions that operate under different tasks. We consider the
descriptions in this paper a listing of the various hypotheses about human cognition resulting from our design of the agent described.

Attention as a Minimal Criterion of Intentionality in Robots

Lars Kopp and Peter Gärdenfors (Lund University, Sweden)

In this paper, we present a robot which exhibits behavior that at a first glance may seem intentional. However, when the robot is confronted with a situation where more then
object is present on the scene, the fact that its behavior is determined merely by S-R rules becomes apparent. As a consequence, the robot has problems attending to a specific object. A truly attentive system must be able to search for and identify relevant objects in the scene; select one of the identified objects; direct its sensors towards the selected object; and maintain its focus on the selected object. We suggest that the capacity of attention is a
minimal criterion of intentio-nality in robots, since the attentional capacity involves a first level of goal representations. This criterion also seems to be useful when discussing intentionality in animal behavior.

EMIB -- Computational Architecture Based on Emotion and Motivation for Intentional Selection and Configuration of Behaviour-Producing Modules

François Michaud (U. de Sherbrooke, Canada)

Over the years, intelligence has been the subject of studies by many different fields, contributing to reveal some of its mysteries. Computational architectures try to exploit in various ways these aspects for designing artificial systems. The biggest challenge is to integrate more and more properties and principles associated with intelligence, combining their advantages to minimise their limitations. With this objective in mind, we propose a
computational architecture that tries to synthesise concepts about intelligence, while making sure that the underlying principles of these concepts, such as emergence, are preserved. The architecture is based on intentional selection and configuration of behaviour-producing modules. Behaviour-producing modules are used as basic control components that are selected and modified dynamically according to the intentions of the system. These intentions are influenced by the situation perceived, the need to accomplish specific goals over time, and knowledge innate or acquired about the world. Motivational and emotional variables are used to monitor the goals and the overall states of the system. The EMIB architecture is applied in designing intelligent autonomous mobile robots, as illustrated in the three experimental cases presented in this paper.

No Inferiority Complex in the Study of Emotion Complexity:
A Cognitive Neuro-science Computational Architecture of Emotion

David Sander and Olivier Koenig (U. of Lyon 25, France)

The aim of the present paper is to propose, on the basis of a survey of the relevant literature, an explicitly-described computational architecture of the emotion system. We first argue that cognitive scientists have the legitimacy to study emotions and that cognitive neuroscience concepts and methods are critical for the elabora-tion of such an architecture. Then, we propose some functional, computational and nervous system related principles that can constrain the elaboration of this architecture. Finally, this framework leads us to the description of the subsystems that constitute the postulated computational architecture.

A Game-Theoretic Approach to Norms and Agents

Guido Boella and Leonardo Lesmo (U. of Torino)

The behavior of an agent in a MAS should be suitably constrained to make it socially compatible with the community of agents. In this paper, we propose a model for obligations and rules which is inspired by E. Goffman's work in sociology, and which can be integrated in existing BDI (Belief, Desire, Intention) agent architectures. Decision Theory and Anticipatory Coordination can be the basic building blocks for an agent model where obligations are associated with sanctions.

Goal Generation in the BOID Architecture

Jan Broersen, Mehdi Dastani, Joris Hulstijn and Leendert van der Torre
(Vrije University and U. of Utrecht, the Netherlands)

In this paper we consider goal generation in cognitive agent architectures. We show how goal generation can be described in terms of interaction between mental attitudes biased by agent types such as realistic, social, selfish and stable. We introduce a generic BOID architecture and agent type specific BOID architec-tures, in which goals are generated from conditional beliefs, obligations, intentions and desires. We implement the BOID architectures by relating conditional mental attitudes to components, goal generation to extension construction, and
agent types to constraints on priority functions.

Distributed Cognition through Active Mental Entities: an Argumentation-Based Approach

P.Baroni, D. Fogli, M. Giacomin, G. Guida (U. di Brescia, Italy)

This paper introduces the active mental entities approach to the design of intelligent autonomous agents. In this approach, mental attitudes are conceived as active computational entities and the overall mental activity of an agent results from their independent operation and interaction. In order to provide a formal support to this approach, we propose argumentation theory as a basis for the definition of interaction mechanisms among active
mental entities. To this purpose, a novel distributed argumentation algorithm is devised. Finally, we give some operation examples in the context of a robotic application.

Temporal Analysis of the Dynamics of Beliefs, Desires, and Intentions

Catholijn M. Jonker, Jan Treur and Wieke de Vries
(Vrije University and U. of Utrecht, the Netherlands)

In this paper temporal relationships are expressed that provide an external temporal grounding of intentional notions. Justifying conditions are presented that for-malise criteria that a (candidate) statement must satisfy in order to qualify as an external representation
of a belief, desire or intention. Using these external representations, anticipatory reasoning about intentional dynamics can be performed.

Acknowledgements to our reviewers for vol. 2 of CSQ

Go to Top of Page