Understanding the decisions of AI-driven systems and the rationale behind such decisions is key to the success of the human-robot team. However, the complexity and the "black-box" nature of many AI algorithms create a barrier for establishing such understanding within their human counterparts. Reinforcement Learning (RL), a machine-learning algorithm based on the simple idea of action-reward mappings, has a rich quantitative representation and a complex iterative reasoning process that present a significant obstacle to human understanding of, for example, how value functions are constructed, how the algorithms update the value functions, and how such updates impact the action/policy chosen by the robot. In this paper, we discuss our work to address this challenge by developing a decision-tree based explainable model for RL to make a robot’s decision-making process more transparent. Set in a human-robot virtual teaming testbed, we conducted a study to assess the impact of the explanations, generated using decision trees, on building transparency, calibrating trust, and improving the overall human-robot team’s performance. We discuss the design of the explainable model and the positive impact of the explanations on outcome measures.

Charisma is a powerful device of communication and persuasion. Researchers have pinpointed specific behaviors that contribute to the perception of charisma. How can we realize such behaviors in a virtual character? In this paper, we discuss our work in the design of charismatic behavior for a virtual human. We developed a series of verbal and nonverbal (with the focus on voice) charismatic strategies based on the analysis of behaviors of charismatic leaders. We developed scripted speech dialogues with the verbal strategies and recorded the speeches with actors using the nonverbal strategies. The dialogue is further implemented in a virtual human, embedded in a virtual classroom, to give a lecture on the human circulatory system. We conducted a study with the virtual human to assess the impact of charismatic verbal and nonverbal behaviors on perceived charisma. The results show the positive impact of the use of verbal strategies and how the use of voice can influence such impact. The results shed light on the next steps needed to automatically generate charismatic speech, voices, and gestures for virtual characters.

Trust calibration is critical to the success of human-agent interaction (HAI). However, individual differences are ubiquitous in people's trust relationships with autonomous systems. To assist its heterogeneous human teammates calibrate their trust in it, an agent must first dynamically model them as individuals, rather than communicating with them all in the same manner. It can then generate expectations of its teammates' behavior and optimize its own communication based on the current state of the trust relationship it has with them. In this work, we examine how an agent can generate accurate expectations given observations of only the teammate's trust-related behaviors (e.g., did the person follow or ignore its advice?). In addition to this limited input, we also seek a specific output: accurately predicting its human teammate's future trust behavior (e.g., will the person follow or ignore my next suggestion?). In this investigation, we construct a model capable of generating such expectations using data gathered in a human-subject study of behavior in a simulated human-robot interaction (HRI) scenario. We first analyze the ability of measures from a pre-survey on trust-related traits to accurately predict subsequent trust behaviors. However, as the interaction progresses, this effect is dwarfed by the direct experience. We therefore analyze the ability of sequences of prior behavior by the teammate to accurately predict subsequent trust behaviors. Such behavioral sequences have shown to be indicative of the subjective beliefs of other teammates, and we show here that they have a predictive power as well.

Digital doppelgangers are virtual humans that highly resemble the real self but behave independently. Digital doppelgangers possess great potential to serve as powerful models for behavioral change. An emerging technology, the Rapid Avatar Capture and Simulation (RACAS) system, enables low-cost and high-speed scanning of a human user and creation of a digital doppelganger that is a fully animatable virtual 3D model of the user. We designed a virtual role-playing game, DELTA, that implements a powerful cognitive dissonance-based paradigm for attitudinal and behavioral change, and integrated it with digital doppelgangers to influence a human user's attitude towards sexism on college campuses. In this paper, we discuss the design and evaluation the RACAS system and the DELTA game-based environment. Results indicate the potential impact of the DELTA game-based environment in creating an immersive virtual experience for attitudinal change.

Trust is a critical factor for achieving the full potential of human-robot teams. Researchers have theorized that people will more accurately trust an autonomous system, such as a robot, if they have a more accurate understanding of its decision-making process. Studies have shown that hand-crafted explanations can help maintain trust when the system is less than 100% reliable. In this work, we leverage existing agent algorithms to provide a domain-independent mechanism for robots to automatically generate such explanations. To measure the explanation mechanism's impact on trust, we collected self-reported survey data and behavioral data in an agent-based online testbed that simulates a human-robot team task. The results demonstrate that the added explanation capability led to improvement in transparency, trust, and team performance. Furthermore, by observing the different outcomes due to variations in the robot's explanation content, we gain valuable insight that can help lead to refinement of explanation algorithms to further improve human-robot trust calibration.

The prevalence of negotiation in social interaction has motivated researchers to develop virtual agents that can understand, facilitate, teach and even carry out negotiations. While much of this research has analyzed how to maximize the objective outcome, there is a growing body of work demonstrating that subjective perceptions of the outcome also play a critical role in human negotiation behavior. People derive subjective value from not only the outcome, but also from the process by which they achieve that outcome, from their relationship with their negotiation partner, etc. The affective responses evoked by these subjective valuations can be very different from what would be evoked by the objective outcome alone. We investigate such subjective valuations within human-agent negotiation in four variations of a wartime negotiation scenario taken from the political science literature. We observe that the objective outcomes of these negotiations are not strongly related to the human negotiators' subjective perceptions, as measured by the Subjective Value Index. We examine the game dynamics and agent behaviors to identify features that induce different subjective values in the participants, even when separated from the effect of the objective outcomes. We thus are able to identify characteristics of the interaction between negotiation and battlefield processes that most impact people's subjective valuations.

The prevalence of negotiation in social interaction has motivated researchers to develop virtual agents that can understand, facilitate, teach and even carry out negotiations. While much of this research has analyzed how to maximize the objective outcome, there is a growing body of work demonstrating that subjective perceptions of the outcome also play a critical role in human negotiation behavior. People derive subjective value from not only the outcome, but also from the process by which they achieve that outcome, from their relationship with their negotiation partner, etc. The affective responses evoked by these subjective valuations can be very different from what would be evoked by the objective outcome alone. We investigate such subjective valuations within human-agent negotiation in four variations of a wartime negotiation game. We observe that the objective outcomes of these negotiations are not strongly correlated with the human negotiators' subjective perceptions, as measured by the Subjective Value Index. We examine the game dynamics and agent behaviors to identify features that induce different subjective values in the participants. We thus are able to identify characteristics of the negotiation process and the agents' behavior that most impact people's subjective valuations in our wartime negotiation games.

Communication is more effective and persuasive when par-ticipants establish rapport. Tickle-Degnen and Rosenthal [57] argue rapport arises when participants exhibit mutual attentiveness, positivity and coordination. In this paper, we investigate how these factors relate to perceptions of rap-port when users interact via avatars in virtual worlds. In this study, participants told a story to what they believed was the avatar of another participant. In fact, the avatar was a computer program that systematically manipulated levels of attentiveness, positivity and coordination. In contrast to Tickel-Degnen and Rosenthal’s findings, high-levels of mutual attentiveness alone can dramatically lower percep-tions of rapport in avatar communication. Indeed, an agent that attempted to maximize mutual attention performed as poorly as an agent that was designed to convey boredom. Adding positivity and coordination to mutual attentiveness, on the other hand, greatly improved rapport. This work un-veils the dependencies between components of rapport and informs the design of agents and avatars in computer me-diated communication. Author Keywords Virtual human, rapport, back-channel, gaze, head nod,

Research show that teacher's nonverbal immediacy can have a positive impact on student's cognitive learning and affect [3]. This paper investigates the effectiveness of nonverbal immediacy using a virtual human. The virtual human attempts to use immediacy feedback to create rapport with the learner. Results show that the virtual human established rapport with learners but did not help them achieve better learning results. The results also suggest that creating rapport is related to higher self-efficacy, and self-efficacy is related to better learning results.

Recent research shows that instructors commonly use politeness strategies to achieve affective scaffolding in educational contexts. The importance of affective factors such as self-confidence and interest that contribute to learner motivation is well recognized. In this paper, we describe the results of a Wizard-of-Oz experiment to study the effect of politeness strategies on both cognitive and motivational factors. We compare the results of two different politeness strategies, direct and polite, in assisting seventeen students in a computer-based learning task. We find that politeness can affect students' motivational state and help students learn difficult concepts. The results of the experiment provide a basis for the design of a polite pedagogical agent and its tutorial intervention strategies.

In this paper, we describe a method for pedagogical agents to choose when to interact with learners in interactive learning environments. This method is based on observations of human tutors coaching students in on-line learning tasks. It takes into account the focus of attention of the learner, the learner's current task, and expected time required to perform the task. A Bayesian network model combines evidence from eye gaze and interface actions to infer learner focus of attention. The attention model is combined with a plan recognizer to detect different types of learner difficulties such as confusion and indecision which warrant intervention. We plan to incorporate this capability into a pedagogical agent able to interact with learners in socially appropriate ways.