Elizabeth A. Plant, Ph.D. - US grants

Florida State University will study the effectiveness of pedagogical agents as social models to influence girls' beliefs and stereotypes about engineering, the potential for using pedagogical agents to change beliefs.

Many females have negative and unconstructive beliefs regarding engineering. As stated by Muller (2002), these misperceptions are prompted by a social fabric that pervades our society and results in stereotypic perceptions of engineering and scientific fields as "geeky" and particularly inappropriate for girls and women. Consequently, young women need more exposure to productive beliefs about engineering, ideally through some form of a social model (e.g., Bandura, 1986). While human models can help to bridge the gap, the logistics of arranging such opportunities are difficult to coordinate, and available mentors may not necessarily best match a particular young women's needs. By implementing computer-based social models, factors can be controlled that cannot be in human mentors, with a greater potential for outreach through avenues such as the Internet. Pedagogical agents, which are three-dimensional, animated computer-based characters, are particularly advantageous for serving as such models given that there is strong evidence that students, particularly females, easily develop social relationships with them. However, there is limited evidence to guide the design of such agents for the purpose of influencing beliefs about engineering.

The goals of this interdisciplinary project are as follows: 1) to systematically investigate
the effectiveness of pedagogical agents as social models to influence girls' beliefs and stereotypes about engineering; and, in parallel, 2) to use pedagogical agents as a vehicle to systematically examine the nature of the belief-changing process. Two populations, representing two age groups (middle school girls, and freshman female undergraduates from both a traditional and a historically-Black institution), will be sampled for the research. Three types of research designs (Choice, Construction, and System-Controlled Studies (experimental and adaptive)) will be implemented to provide a more integrative understanding. Specifically, the first year will focus on investigating agent appearance; the second year on agent message/delivery; and, the third year on agent persona (combined factors). Dependent variables include student beliefs about and motivation toward engineering, intentions to pursue engineering, and perceptions of the agent. Dependent variables will be assessed by a combination of self-report (e.g., Likert-scale items), performance, and behaviorally-indicated measures.

Intellectual Merit. Given that little is known as to the affordances and learner preferences for pedagogical agents as social models, this project is critical for advancing several interdisciplinary fields (e.g., instructional systems, social psychology, human-computer interaction, education/training, ecommerce and marketing, and game designers). By integrating multiple research approaches, the project is creative and original and draws on the two PIs' experience conducting pedagogical agent and social psychology research. Importantly, it will provide insight for influencing females' beliefs and stereotypes regarding engineering. Results will also complement ongoing development of pedagogical agents and other technology-mediated learning, such as intelligent tutoring systems. Along this line, the research will also help better illuminate the social impact of agents as an information technology, which is important given that society is becoming increasingly more information-driven.

Broader Impact. It is expected that the project will build a strong empirical research base in this area with guidelines for designing agents to support engineering (or other types of) belief adoption. These findings will have the potential to broadly influence computer interface design, agent implementations in learning environments, and have marketing-related implications for other projects that involve "selling" engineering to females. Based on the findings, the next step would be to develop and implement a highly competent (e.g., artificially intelligent) and persuasive agent that would positively impact young women's beliefs about engineering.

In today's world, police officers face difficult split-second decisions that require them to determine whether criminal suspects are armed and constitute an imminent threat. When suspects are members of ethnic minorities there may be a greater tendency for police officers to believe that they are armed. Examples of this include the shooting of an unarmed young Black man by police officers who mistook the cellular phone that he was carrying for a weapon. When tragedies like this occur, the question arises as to whether police officers' split-second decisions to shoot may be influenced by the suspect's race. There is mounting evidence that police officers are more likely to mistakenly shoot unarmed Black suspects compared to unarmed White suspects. This research examines the efficacy of a training program to eliminate this type of racial bias in responses to criminal suspects. The goals of this research are to: 1) Develop a deeper understanding of how to eliminate racial bias in police responding; 2) Identify the behaviors and responses that can be influenced by training on a bias reduction simulation; and 3) Explore whether there are factors that affect either the degree of racial bias initially present or the efficacy of a training simulation. A series of studies employing computer simulations is proposed to programmatically examine each of these issues. In this work, law enforcement personnel participate, and complete a computer simulation where they must decide whether to shoot at suspects who appear on screen. Here, the race of suspect is unrelated to weapon possession such that attending to race and being influenced by race impairs performance on the simulation. Initial evidence indicates that this approach is effective in eliminating racial biases in responses to the simulation even 24 hours after training.

Humans, like many other vertebrate species, developed a physiological immune system designed to detect and destroy harmful organisms that have entered the body. Recent research suggests that, in addition to this physiological immune system, there evolved a behavioral immune system comprised of psychological and behavioral mechanisms designed to detect and defend against pathogen threats even before they enter the body. For example, people are quick to notice and avoid others who display cues of infection (e.g., coughing, lesions). The studies in this proposal will examine the behavioral immune system in depth, and link its action to multiple aspects of physiological and psychological functioning.

Past research has treated the physiological and behavioral immune systems independently. Yet, an evolutionary perspective suggests that there should be an adaptive relationship between physiological and psychological functioning. Thus, the studies in this proposal test the prediction that concerns about disease increase behaviors designed to prevent initial contagion (e.g., avoiding others who appear sick), as well as increase physiological processes designed to destroy pathogens that have entered the body (e.g., secretion of cytokines). Activation of both behavioral and physiological immune systems would provide the best possibility of overcoming a disease threat.

This research also examines the specific processes underlying behavioral immune system activation. To be effective, the behavioral immune system must be sensitive to the possibility of contagion. Some physical characteristics are more associated with disease than other physical characteristics (e.g., the presence of a rash more than hair color). In addition, sometimes people feel more susceptible to disease than at other times (e.g., after touching a bloody cut versus after washing their hands). We propose that both of these factors - implicit disease associations and perceptions of disease susceptibility - influence the degree to which the behavioral immune system is activated.

Last, this research examines ways of down-regulating behavioral immune system activation. Prior research suggests that disease concerns can lead people to avoid and have negative attitudes toward individuals who, although are not actually contagious, display physical characteristics that are stereotypically associated with disease (e.g., obesity). Thus, understanding ways of down-regulating such disease-based anti-social behavior is key for reducing prejudice and discrimination. Our studies examine how minimizing disease-associations (e.g., decreasing the link between disease and obesity) as well as reducing perceptions of disease susceptibility (e.g., making people feel immunized against disease) can reduce behavioral immune system activation and therefore decrease prejudice toward groups stereotypically associated with disease.

In sum, the studies in this proposal integrate research from social cognition, immunology, and evolutionary psychology to test three broad hypotheses: (1) Behavioral immune responses to disease threat are complemented by anticipatory physiological immune responses (2) Behavioral immune system activation is mediated by those physiological responses and by specific implicit cognitions, and (3) The behavioral immune system can be down-regulated by manipulating those implicit cognitions. The results of these studies will have an important impact on theoretical and empirical work in social psychology, cognitive science, and other related domains of study. The proposed research is therefore interdisciplinary in nature, and will foster synergistic connections between research in domains that historically have had relatively little contact but are increasingly being integrated (e.g., psychophysiology and prejudice). Additionally, the execution of the proposed research will involve training for both graduate and undergraduate students
from groups underrepresented in science careers.