Dynamics of Affect and Cognition in Simulated Agents: Bridging the Gap between Experimental and Simulation Research

Dynamics of Affect and Cognition in Simulated Agents: Bridging the Gap between Experimental and Simulation Research

Ruben Mancha (Department of Business Administration, Trinity University, San Antonio, TX, USA), Carol Y. Yoder (Department of Psychology, Trinity University, San Antonio, TX, USA) and Jan G. Clark (Department of Information Systems and Cyber Security, University of Texas at San Antonio, San Antonio, TX, USA)
Copyright: © 2013 |Pages: 19
DOI: 10.4018/jats.2013040104
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

The purpose of this article is to propose a simulation framework combining Soft-System Methodology, System Dynamics, and the Cognitive Affective Personality System model, to facilitate the design and development of agents (e.g., agent based models, software agents) with interacting affective and cognitive units. A review of the literature supports the building of a third-order positive causal-loop model between the constructs Affectivity, Self-Efficacy, and Perceived Task Complexity. The model is evaluated to exemplify the use of this framework to study affective states in simulated agents. The behavior of the model is consistent with previous research, corroborating its utility as a tool for endowing complex agents in simulations with mechanisms of human affectivity, and as a computational artifact to develop affective computing systems. The framework, incorporating soft-constructs into simulation models and supporting the study of their process-based (dynamic) interactions, serves to bridge the gap between experimental and simulation research. The limitations of the approach and directions for future research are also discussed.
Article Preview

Introduction

The purpose of this article is to propose a simulation framework combining Soft-System Methodology, System Dynamics, and the Cognitive Affective Personality System model, to facilitate the design and development of simulated agents (e.g., agent based models) with interacting affective and cognitive units.

Simulated agents benefit from having affective components but, historically, computer systems did not possess, recognize, or respond to affective states (Calvo & DMello, 2010). Advances in cognitive psychology, simulation, affective computing, and the study of behavioral economics, where emotional and cognitive processes are integrated (Kahneman, 2003), have led to an increased focus on the role of emotion in human-system interaction by means of emotional coupling (Shibata, Yoshida, & Yamato, 1997). Researchers have attempted, and progressively achieved, integrating emotions in the human-machine interaction (Ammar et al., 2010; Asano & Wachsmuth, 2009; Burleson & Picard, 2004; Picard, 1997, 2003), and in the behavior of the system itself (Gratch & Marsella, 2004). This interaction between the human and the computer system has been coined the affective loop (Conati, Marsella, & Paiva, 2005) or, more recently, the affective loop experience (Höök, 2009). In many different ways, simulation and artificial intelligence researchers have expressed that the affective capabilities of humans are desired characteristics of computer systems (Simon, 1967), and even questioned if “machines” could really be intelligent without emotions (Minsky, 1985).

Theoretical research offers an approach to account for the multiplicity of behavioral outcomes resulting from a limited number of interacting cognitive and affective factors—those inferred constructs such as motivation or creativity that guide the modeling of artificial agents. Michel and Shoda (1995) developed a theoretical framework to explain the paradox of having invariant personality profiles and variable behaviors across situations. Their Cognitive-Affective System Theory of Personality defines a system with several mediating units that dynamically interact to give seemingly complex patterns of individual situation-behavior relations. Affectivity “is an emotion-based trait dimension (Watson, Clark, and Tellegen, 1988) that creates a cognitive bias through which individuals approach and understand experiences” (Naquin & Holton, 2002; p. 359). The Cognitive-Affective Personality System (CAPS) model (Figure 1) relies on Freud’s psychodynamic theory and is a contextualization of personality to give meaning to its multiple outcomes. According to their explanations, the “situational features are encoded by a given mediating unit, which activates specific subsets of other mediating units, generating distinctive cognition, affect, and behavior in response to different situations” (Mischel and Shoda, 1995; p. 254).

Figure 1.

The CAPS model: Comparison of two CAPS models with different mediating units and activation patterns. Adapted from Mischel and Shoda (1995).

Mischel and Shoda’s dynamic patterns of activations follow the basic premises of General Systems Theory (Bertalanffy, 1951; Ashby, 1956; Wiener, 1961). The mediating units in their CAPS Theory include: encodings, expectations and beliefs, affects, goals and values, and competencies and self-regulatory plans.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 9: 2 Issues (2017): Forthcoming, Available for Pre-Order
Volume 8: 1 Issue (2016)
Volume 7: 3 Issues (2015)
Volume 6: 4 Issues (2014)
Volume 5: 4 Issues (2013)
Volume 4: 4 Issues (2012)
Volume 3: 4 Issues (2011)
Volume 2: 4 Issues (2010)
Volume 1: 4 Issues (2009)
View Complete Journal Contents Listing