Randolph M. Jones

Soar Technology, Inc.

Traditional classification approaches are straightforward: collect data, apply classification algorithms, then generate classification results. However, such approaches depend on data being amply available, which is not always the case. This paper describes an approach to maximize the utility of collected data through intelligent guidance of the da...

To support training for offensive and defensive cyber operations, we focus on giving the trainee a realistic ecosystem to train in. This ecosystem includes models of attackers, defenders, and users. The high-level goals for adaptation in this ecosystem are of two types: realism in behavior and tailoring of training. In terms of realism, real-world...

In designing and developing intelligent automation systems, there is often tension between the computational capabilities of the automated system and its usability and understandability. This paper presents a case study in which this tension was manifest and how we attempted to resolve it in a particular application. The application requires intell...

Humans tend to represent and to understand the world in terms of stories, while computer reasoning tends to require formal, mathematical representations. This paper describes a research prototype that enables computers to parse human stories and use collections of those stories to inform causal modeling of political, military, economic, social, inf...

We describe the development of a serious game designed to illustrate the impact of a supervisory control technology. The technology is designed to help human operators deliver better training experiences in simulation. It automates low-level control tasks, thus reducing operator workload. The serious game embeds the technology and simulation in a d...

Constructive simulation-based training does not always go according to plan. Instructors observe scenarios and direct adaptive responses to unexpected events. Simulation operators typically use a graphical interface to monitor the scenario and generate specific interventions that implement the instructor's intent. These activities require the opera...

Scenario-based training provides valuable opportunities for practice and assessment of cross-cultural skills in representative environments. Cross-cultural training that is presented within scenarios can help to motivate trainees and to increase perceptions of relevance and validity. Further, with immersive computer simulations, a sufficiently rich...

Semi-automated Forces (SAFs) are commonly used in training simulation. SAFs often require human intervention to ensure that appropriate, individual training opportunities are presented to trainees. We cast this situation as a supervisory control challenge and are developing automation designed to support human operators, reduce workload, and improv...

Most agent research seeks insights about a single technology, and problems are chosen to allow this focus. In contrast, many real-world applications do not lend themselves to a single technology, but require multiple tools. In an applied AI company, each tool often has its own advocate, whose specialized knowledge may lead her to overestimate her t...

The patterns of life (POL) present a unique challenge for computational modeling because systems must produce both meaningful regularities throughout a large population and also believable behaviors at the level of each individual. We describe how computational POL modeling integrates diverse artificial intelligence research areas and provides inte...

The modeling and simulation of human performance forces the analyst to confront a range of well-known but difficult challenges. One challenge the analyst does not seem to face is a shortage of human performance modeling tools. But because there is no uniform framework for expressing the content and structure of a human performance model, it is diff...

Patterns of life (POL) are emergent properties of com- plex social systems such as neighborhoods or even cities. Accurately modeling POL is not only an academic pursuit; complex training and analysis efforts rely upon these models. Computational POL models will benefit academic researchers by giving them new tools to generate and validate theories...

This paper argues that integration is an essential approach to advancing the state of the art in cognitive systems. Integration is the path to aggregating requirements for cognition, as well as the solutions to those requirements. However, evaluation of integrated cognitive systems has proven difficult and expensive, which has hindered scientific a...

This paper argues the position that an essential approach to the advancement of the state of the art in cognitive systems is to focus on systems that deeply integrate knowledge representations, cognitive capabilities, and knowledge content. Integration is the path to aggregating constraints in ways that improve the science of cognitive systems. How...

This paper discusses representations and processes for agents and behavior models that encode large knowledge bases, are long-lived, and exhibit high degrees of competence and flexibility while interacting with complex environments. There are many different approaches to building such agents, and understanding the important commonalities and differ...

With the increasing need for flexibility and adaptivity in computerized systems, the application of fuzzy expert systems is becoming increasingly commonplace in today's industry. Fuzzy logic expert systems often improve performance by allowing knowledge to generalize without requiring the knowledge engineer to anticipate all possible situations. Th...

Cognitive architectures provide a definition of an abstract machine to support programming of cognitive models and intelligent systems. The point of the abstract machine is to provide the most useful set of processes and representations for developing such models, and the machine usually comes hand in hand with a programming language. However, most...

INTRODUCTION The Soar architecture was created to explore the requirements for general intelligence and to demonstrate general intelligent behavior (Laird, Newell, & Rosenbloom, 1987; Laird & Rosenbloom, 1995; Newell, 1990). As a platform for developing intelligent systems, Soar has been used across a wide spectrum of domains and applications, incl...

This paper describesEureka, a problem-solving architecture that operates under strong constraints on its memory and processes. Most significantly,Eurekadoes not assume free access to its entire long-term memory. That is, failures in problem solving may arise not only from missing knowledge, but from the (possibly temporary) inability to retrieve ap...

AAAI presented the AAAI-04 workshop program on July 25-26, 2004 in San Jose, California. This program included twelve workshops covering a wide range of topics in artificial intelligence. The titles of the workshops were as follows: (1) Adaptive Text Extraction and Mining; (2) Agent Organizations: Theory and Practice; (3) Anchoring Symbols to Senso...

The Soar architecture was created to explore the requirements for general intelligence and to demonstrate general intelligent behavior (Laird, Newell, & Rosenbloom, 1987; Laird & Rosenbloom, 1995; Newell, 1990). As a platform for developing intelligent systems, Soar has been used across a wide spectrum of domains and applications, including expert...

This report describes the High-bevel Symbolic Representation (HLSR) project for the U.S. Air Force PRDA 03-01-HE: Human Performance in Modeling and Simulation, Technical Area 2: Opposing Force Behaviors. This report summarizes the work done on Defense Modeling Simulation contract F33615-03-C-6343 to develop a high level symbolic representation (HLS...

Increasingly, across the broad spectrum of modeling and simulation, as well as in battlefield information and control systems, autonomous reasoning ability is required, and the "intelligence" developed and tested in simulation is migrating to real-world entities. Cognitive and agent architectures represent maturing computational approaches to intel...

Coalition formation methods allow agents to join together and are thus necessary in cases where tasks can only be performed cooperatively by groups. This is the case in the Request For Proposal (RFP) domain, where some requester business agent issues ...

This paper describes the implementation and evaluation of a framework for modeling emotions in complex, decision-making agents. Sponsored by U.S. Army Research Institute (ARI), the objective of this research is to make the decision-making process of complex agents less predictable and more realistic, by incorporating emotional factors that affect h...

This paper describes the general requirements for such intelligent opponents, our approach to developing intelligent opponents using the Soar cognitive architecture, our current software system, which employs a commercial computer game as a prototype development environment, and a discussion of the research issues in cognitive science and artificia...

This paper describes the implementation and evaluation of a framework for modeling emotions in complex, decision-making agents. Sponsored by U.S. Army Research Institute (ARI), the objective of this research is to make the decision-making process of complex agents less predictable and more realistic, by incorporating emotional factors that affect h...

Computational modeling plays a central role in cognitive science. This book provides a comprehensive introduction to computational models of human cognition. It covers major approaches and architectures, both neural network and symbolic; major theoretical issues; and specific computational models of a variety of cognitive processes, ranging from lo...

The variety of communications employed by CGFs requires a flexible communications infrastructure that can be readily modified for new applications, new communication devices, and even user preferences. Understanding the variety of CGF communications has led us to re-design the communications infrastructure of TacAir-Soar, a realtime aircraft CGF. T...

Our research focuses on complex agents that are capable of interacting with their environments in ways that are increasingly similar to individual humans. In this article we describe a cognitive architecture for an interactive decisionmaking agent with emotions. The primary goal of this work is to make the decision-making process of complex agents...

In order for agent-based technologies to play an effective role in simulation and trainingsynthetic force applications, they must carry outgenerate realistic and believable behaviors that are virtually indistinguishable from what a human actor might do under the same circumstances. However, systems that are capable of such complex behavior necessar...

This paper presents an initial evaluation of an emotions model developed for a sophisticated synthetic forces model. Sponsored by ARI, the objective of this research is to make the decision-making process of complex agents less predictable and more realistic, by incorporating emotional factors that affect humans. To this end, we have adopted an app...

This work describes a problem, knowledge contention, that arises when agent knowledge learned via knowledge compilation contends with the agent's original task decomposition knowledge in the execution of agent actions. We show the circumstances under which the problem occurs, present a solution that avoids knowledge contention, and provide empirica...

Recent research demonstrates that people learn to solve problems more effectively when presented with a series of faded examples and problems, than when presented with completely worked out examples and completely unsolved problems alone. We propose an explanation for this effect, based on the Cascade model. Cascade was originally built to model th...

In concert with the I/ITSEC '01 theme, "Warfighting Readiness Through Innovative Training Technology", this paper explores an innovative approach to enhancing the realism and hence the efficacy of training - developing the capacity for synthetic forces to act and respond emotionally. Emotions, along with moods and dispositions, have been shown to b...

This paper presents a course in the design and implementation of computer games, offered as an upper-division computer science course at Colby College during the winter semester, 1999. The paper describes the material, topics, and projects included in the course. More generally, I argue that this course provides an ideal environment for students to...

Hierarchical execution of domain knowledge is a useful approach for intelligent, real-time systems in complex domains. In addition, well-known techniques for knowledge compilation allow the reorganization of knowledge hierarchies into more ecient forms. However, these techniques have been developed in the context of systems that work in static doma...

This paper presents a course in the design and implementation of computer games, offered as an upper-division computer science course at Colby College during the winter semester, 1999. The paper describes the material, topics, and projects included in the course. More generally, I argue that this course provides an ideal environment for students to...

The TacAir-Soar system is a computer program that generates human-like behavior flying simulated aircraft in tactical air combat training scenarios. The design of the system has been driven by functional concerns, allowing the system to generate a wide range of appropriate behaviors in severely time-limited situations. The combination of constraint...

trace of behavior in Soar 4. Use goals Using knowledge to deliberately select and apply operators is an important capability for an intelligent entity, but action must be in the service of goals, and the determination of goals itself can be quite complex. Soar unifies the reasoning about action and goals by treating goals as abstract operators to b...

TacAir-Soar is an intelligent, rule-based system that generates believable "human-like" behavior for large scale, distributed military simulations. The innovation of the application is primarily a matter of scale and integration. The system is capable of executing most of the airborne missions that the United States military flies in fixed-wing air...

: We have created a graphical interface tool to aid in the development and deployment of intelligent synthetic forces. The Situational Awareness Panel provides a number of views "inside the mind" of a synthetic agent. The views update continuously during the lifetime of the agent, as long as the panel is displayed, so users can inspect the reasonin...

Many results and techniques applicable to human-computer interaction (HCI) have been discovered by using cognitive modelling. However, few of these lessons have been applied to improve the explanation and illustration of cognitive models themselves. We have started to redress this imbalance by developing for a well-known cognitive architecture (Soa...

In many domains, intelligent agents must coordinate their activities in order for them to be successful both individually and collectively. Over the last ten years, research in distributed artificial intelligence has emphasized building knowledge-lean systems, where coordination emerges either from simple rules of behavior or from a deep understand...

Several studies have found that learning is more e#ective when students explain examples to themselves. Although these studies show that learning and self-explanation co-occur, they do not reveal why. Three explanations have been proposed and computational models have been built for each. The gap-#lling explanation is that self-explanation causes s...

When solving homework exercises, human students often notice that the problem they are about to solve is similar to an example. They then deliberate over whether to refer to the example or to solve the problem without looking at the example. We present protocol analyses showing that effective human learners prefer not to use analogical problem solv...

In recent years, there has been a fair amount of research directed toward the goal of developing virtual, human-like characters for simulation environments. Much of this work has focused on creating high-fidelity graphical animations that represent realistic human forms and movement. We are approaching the same goal from a different angle, focusing...

TacAir-Soar is an intelligent, rule-based system that generates believable "human-like" behavior for large scale, distributed military simulations. The innovation of the application is primarily a matter of scale and integration. The system is capable of executing most of the airborne missions that the United States military flies in fixed-wing air...

EUREKA is a problem-solving system that operates through a form of analogical reasoning. The system was designed to study how relatively low-level memory, reasoning, and learning mechanisms can account for high-level learning in human problem solvers. Thus, EUREKA's design has focused on issues of memory representation and retrieval of analogies, a...

Hierarchical execution of domain knowledge is a useful approach for intelligent, real-time systems in complex domains. In addition, well-known techniques for knowledge compilation allow the reorganization of knowledge hierarchies into more efficient forms. However, these techniques have been developed in the context of systems that work in static d...

Training in flight simulators will be more effective if the agents involved in the simulation behave realistically. Accomplishing this requires that the automated agents be under autonomous, intelligent control. We are using the Soar cognitive architecture to implement intelligent agents that behave as much like humans as possible. In order to appr...

The domains that computer-generated forces address (such as tactical flight) are more complex than have generally been used in artificial-intelligence research. A particular characteristic of this complexity is that a reasonable agent must attend to a large number of goals at the same time. Moreover, some of these goals are independent, while other...

TacAir-Soar is a reactive system that uses recognition-driven problem solving to plan and generate behavior in the domain of tactical air combat simulation. Our current research efforts focus on integrating more deliberative planning and learning mechanisms into the system. This paper discusses characteristics of the domain that influence potential...

This paper discusses some of the issues involved in creating realistic intelligent, automated agents for simulation and training. In addition, it presents our efforts at constructing such an agent for the domain of tactical flight.

Hierarchical execution of domain knowledge is a useful approach for intelligent, real-time systems in complex domains. In addition, well-known techniques for knowledge compilation allow the reorganization of knowledge hierarchies into more efficient forms. However, these techniques have been developed in the context of systems that work in static d...

Since the summer of 1992, the Soar/IFR research group has been building intelligent automated agents for tactical air simulation. The Soar/IFOR research project exists at three sites, the University of Michigan, the University of Southern California, and Carnegie Mellon University. The ultimate goal of this project is to develop automated pilots wh...

Interactive simulation environments constitute one of today's promising emerging technologies, with applications in areas such as education, manufacturing, entertainment, and training. These environments are also rich domains for building and investigating intelligent automated agents, with requirements for the integration of a variety of agent cap...

The fielding of large numbers of autonomous computer-generated forces requires that these forces be able to coordinate their behaviors. Within the military, there are many levels of coordination, from the high-level management of a theater of war, down to the low-level interactions of individual soldiers. TacAir-Soar represents a data point at this...

When children learn to add, they count on their fingers, beginning with the simple Sum Strategy and gradually developing the more sophisticated and efficient Min strategy. The shift from Sum to Min provides an ideal domain for the study of naturally occurring discovery processes in cognitive skill acquisition. The Sum -to- Min transition poses a nu...

Since the summer of 1992, the Soar/IFOR research group has been building intelligent automated agents for tactical air simulation. The ultimate goal of this project is to develop automated pilots whose behavior in simulated engagements is indistinguishable from that of human pilots. This technical report is a collection of the research paper's that...

This chapter discusses the integration of analogical search control and explanation-based learning of correctness. A theory revision system, CASCADE was developed as a simulation of human students learning college physics. CASCADE is based on the heuristic that an impasse is worth repairing only if the system is on a solution path when the impasse...

Autonomous systems must operate in dynamic, unpredictable environments in real time. The task of flying a plane is an example of an environment in which the agent must respond quickly to unexpected events while pursuing goals at different levels of complexity and granularity. We present a system, Air-Soar, that achieves intelligent control through...

Autonomous systems must be able to deal with dy- namic, unpredictable environments in real time. Our video describes a system for intelligent control of an airplane, within a realistic flight simulator (the Sili- con Graphics flight simulator). The simulator allows asynchronous control of the plane's throttle, ailerons, elevator and other control s...

Several studies have found that learning is more eeective when students explain examples to themselves. Although these studies show that learning and self-explanation co-occur, they do not reveal why. Three explanations have been proposed and computational models have been built for each. The gap--lling explanation is that self-explanation causes s...

Several investigations have found that students learn more when they explain examples to themselves while studying them. Moreover, they refer less often to the examples while solving problems, and they read less of the example each time they refer to it. These findings, collectively called the self- explanation effect, have been reproduced by our c...

Several investigators have taken protocols of students learning sophisticated skills, such as physics problem solving and LISP coding, by studying examples and solving problems. These investigations uncovered the self-explanation effect: Students who explain examples to themselves learn better, make more accurate self-assessments of their understan...

Cascade models humans learning college physics by studying examples and solving problems. It simulates the main qualitative phenomena visible in human protocols of learning, including several strategies for analogical and non-analogical problem solving, and two strategies for studying examples. It learns at the knowledge level by acquiring new phys...

GIPS is a problem solving system that models the strategy shifts of children learning to add. The system uses a generalized form of means-ends analysis as its reasoning algorithm, and it learns probabilistic selection and execution concepts for its operators. With this combination, GIPS models the SUM-to-MIN transition that children exhibit when le...

The SUM-to-MIN transition that children exhibit when learning to add provides an ideal domain for studying naturally occurring discovery processes. We discuss a computational model that accounts for this transition, including the appropriate intermediate strategies. In order to account for all of these shifts, the model must sometimes learn without...

In this chapter we describe Eureka, a problem solver that uses analogy as its basic reasoning and learning process. Eureka introduces a learning mechanism called analogical search control, and uses a model of memory based on spreading activation to retrieve analogies and solve problems. These relatively simple mechanisms allow the system to account...

This paper is based on our experiences in integrating a real-time high-fidelity model of human behavior in advanced distributed battlefield simulations. Our experiences are drawn from our participation in the Synthetic Theater of War-1997 (STOW-97) in which we developed behaviors for all U.S. Navy, Air Force, and Marine fixed wing aircraft missions...

This paper examines some of the constraints on cognition assumed and imposed by the ACT-R and Soar cognitive architectures. In particular, we study how these constraints either encourage or require particular types of "modeling idioms" in the form of programming patterns that commonly appear in implemented models. Because of the nature of the mappi...

There have been a few efforts to evaluate the robustness of performance of the Soar cognitive architecture, with positive results. However, previous efforts have focused primarily on running the architecture with agent models that are research systems, designed specifically for Soar performance evaluation, or otherwise limited in capability. This p...