Alexis Morris

A system dynamics view of stress: Towards human-factor modeling with computer agents

Human-factor models are important for computer systems to i) make such systems more human aware (ie. better estimators of human behavior) and ii) make such systems demonstrate more realistic human behaviors (ie. display more human-like AI). These models expand the research horizons in domains such as multi-agent organizational simulation, as the impact of various human-factors can be investigated. This is particularly important in safety and security as organizational conflicts are largely impacted by human failures. Human-factor calculations, however, are difficult to quantify, validate, and encode because human behavior is both fuzzy and complex. This paper applies system dynamics, a modeling technique for understanding complex systems, to human-factors (stress in particular). It is a way that is computationally useful, and may be validated by experts in human studies. We first model stress as a causal loop diagram to discuss relationships between key components, and then produce a stocks and flows diagram to simulate behavior which, in future work, would be used for “mental models” in computer programs and agent simulations.

Modelling culture in multi-agent organizations

We introduce a novel way to model and visualize culture in multi-agent organizations exploring the multi-dimensionality of culture and cultural modelling from a complex systems and multi-agent systems standpoint. The need for performing such modelling and simulation is evident since in-vivo organizational experiments are costly, not easily generalizable, and may not be feasible in critical situations. The proposed model enables one to point to strategies for organizational transformation/evolution by i) developing a unique approach to culture modelling from a holistic and systems-theoretic perspective according to seven dimensions, and ii) simulating cultural interactions as a multi-agent system that achieves an equilibrium of beliefs. Incipient results with a simple model reveal the dynamics of emergent culture of an agent organization having distinct roles and influences that develop as new individuals are added to the system.

Modelling Culture with Complex, Multi-dimensional, Multi-agent Systems

Culture plays a significant role in human civilizations as a key determinant of relationships and organization formation; however, its role, key properties, and mechanisms are not yet fully understood. This work explores culture and cultural modelling from a complex systems, multi-dimensional, and multi-agency standpoint. The need for performing such modelling and simulation is evident since in-vivo organizational experiments are costly, not easily generalizable, and require lengthy analyses that may not be feasible in critical situations. Exploring the role and influence of culture on organizations is the aim of this chapter, whereby definitions, dimensions, and experiments are introduced in order to show the evolution and emergence of culture as a complex, distributed, social system. This work contributes to culture studies by (a) adding to the literature of culture as a complex system, (b) presenting a new seven-dimensional model to describe and encapsulate culture, and (c) simulating cultural interactions using a multi-agent system of high-functioning agents that achieve an equilibrium of beliefs.

FRIEND: A human-aware BDI agent architecture

Effective relationships between people and technology are essential for organizational response (agility, adaptation, and innovation). Advances in computing power, and the rise of complex ubiquitous systems, raise a challenge for managing this relationship given limits of human physical and cognitive capacities. To re-align people with computing technology involves either improved human training, or streamlining technologies to fit human needs, abilities, and perceptions. This paper looks at this socio-technical gap and makes a case for intelligent agent mediation through passive human-input monitoring (human-context awareness) and basic models of human behavior. The target audience is interdisciplinary, involving the cognitive informatics, agent systems, bodynet, socio-technical systems, and human-computer-interface communities. The overall contribution is in the combination of socio-technical systems engineering and human factors concepts with the agent-based paradigm and cognitive sensing technologies towards new, “Human-tech” friendly agent applications for everyday socio-technical systems. As such an early architectural design for such agents is presented, as well as future research directions toward its development.

RECON: An Adaptive Human-Machine System for Supporting Intelligence Analysis

This position paper presents the initial designs for a novel functional, context-aware, case-based recommendation system to enhance the current intelligence capability of military intelligence analysts. Its central objective is to support these analysts during the collection, processing, and analysis phases of the intelligence cycle through load minimization and improved human-machine synergy. This involves sense-making from both explicit and implicit contextual information with a nexus of technologies and processes, including software modelling and simulation, meta-level modelling and recommendation, and psycho-physiological modelling and monitoring. The proposed system, RECON, will contribute a new architecture based on five core components: i) brain-computer interfaces, ii) human-computer interaction, iii) data, iv) context, and v) case-based recommendation. Together with human analysts these form the basis of an adaptive system that will advance toward a powerful future intelligence analysis capability.

FRIENDs: Brain-monitoring agents for adaptive socio-technical systems

Brain-monitoring is quickly becoming an important field of research, with potentially significant impacts on how people will interact with technology. As understandings of the inner-workings of the brain become more accurate technologies are becoming more advanced, smaller, cheaper, and ubiquitous. It is expected that new forms of computing that take advantage of brain states will be developed. This will enable systems to be highly aware of user mental contexts emotions, intentions, and moods. These systems would display higher autonomic behavior and would streamline user-interaction while managing the use of brain context data for applications and services. There are few studies of how to develop and make use of agent architectures in this new domain. Current approaches target a single user and application situation. To be ubiquitous it is unrealistic for applications to have specialized overhead for individual users. Personalizable, but distributed approaches are needed. To realize a general purpose agent for brain-monitoring and management of brain context is the goal of this work. This involves the selection of a brain-monitoring paradigm, the selection of an agent architecture paradigm, an inferencing mechanism, and the combination of the three towards a unified framework. Core motivations are discussed, and an early agent framework design FRIEND is presented, along with proposed proof-of-concept applications for using brain context.

A Multi-Agent Context-Management System for RECON Intelligence Analysis

Adaptive systems require technologies to enable high synchronicity between its users and their unfolding situation dynamics, in concert with system response actions. To be effective, a multi-dimensional view of context must be considered and incorporated. This work advances the development of such a system for RECON, an initiative to support intelligence analysts with a novel context-management and case-based recommendation capability. The central concepts involved in the management of explicit and implicit contexts are presented and are developed into a novel multi-agent approach. In particular a new context-sensitive cognitive model and a community of expert service-oriented agents are proposed to facilitate and improve system adaptations to user-specific, situational, and system states. These designs pave the way towards future developments and experiments in improving human–machine interaction with adaptive context-management systems.

Discovering best practices in emergency response with combined physical and organizational simulation

Simulations that incorporate both physical (hard) as well as psycho-social human behaviours (soft) are expected to improve situational awareness during critical events, yet are not readily available. Because of the complex nature of emergency response and, in particular, because of the potential for unforeseen interdependencies in the physical environment, it is difficult to objectively assess the effectiveness of organizational response practices. To overcome this challenge, both the reality of the physical environment and the impact of human organizations must be represented explicitly. We combine two existing simulators to offer a holistic picture that integrates the “hard” and “soft” factors. We discuss the interaction of these simulators for both “as-is” and “what-if” scenarios related to an actual case study. It is expected that this integrated simulation approach will help to discover best practices and policies that would improve the effectiveness of operations for real-world organizations in disaster situations.