In human-robot collaboration, effective communication of robot intentions and actions is crucial for safety and efficiency. While various explanation methods have been proposed, a unified framework integrating different explanation approaches from a theoretical perspective remains a challenge. This paper presents a three-style framework that formally integrates mechanical, teleological, and emotional explanations for robot actions. Building upon traditional mechanical explanations describing causal chains and teleological explanations focusing on goals, we formalized emotional explanations based on utility changes, grounded in Appraisal theory. These three complementary styles were unified through a common mathematical foundation of state transitions and utility functions, enabling systematic selection and integration of explanations according to the situation. We provide guidelines for selecting appropriate explanation styles based on task phases and explanatory purposes, while discussing implementation challenges and future directions including personalization and real-time adaptation mechanisms.

Attitude formation involves both rational beliefs ("should do") and personal desires ("want to do"). We developed and evaluated an AI dialogue system for personalized persuasion addressing both rational and desire dimensions by integrating Multi-Attribute Utility Model and Elaboration Likelihood Model. Our system identifies each individual’s high-priority perspectives and adapts persuasive messages accordingly, leveraging both central-route and peripheral-route cues. Using nuclear power plant restarts in Japan as a test case, our experiment (N=148) compared three strategies: Positive, Negative, and Neutral. The Positive condition significantly increased both rational "Should" and subjective "Want" dimensions, while the Negative condition decreased them; the Neutral condition produced no notable changes. Results indicate that tailoring messages to core values is critical for effective persuasion, though overly forceful approaches risk psychological reactance. Future work should address dynamic persuasive route selection, balance persuasive intensity with user autonomy, and consider broader demographic diversity.

People often anthropomorphize agents and show social concern for the agents' goals. Whereas this can be useful to build human-agent cooperation in some settings, in others it can be counterproductive - e.g., when people risk themselves to help a robot. A mechanism, thus, is needed to regulate how much cooperation people show towards agents, according to the context. Here, we show that emotion expressions can be a powerful mechanism to help people identify the appropriate level of cooperation given the situation. In the present study, participants (n=379) engaged in a 20-round iterated prisoner's dilemma game with agents that showed emotional expressions that reflected a preference for maximizing its own interests versus maximizing the participants' interests. Accordingly, the results showed that participants focused significantly more on their own interests when facing the agent that expressed emotions favoring the participants' outcome; moreover, this treatment was more successful in steering the participants' focus to their own interests than showing no emotion. These findings reveal that, in addition to helping build cooperation, as shown in prior work, emotion expression can play a central role in mitigating some negative consequences of anthropomorphizing agents.

Emotional expressions serve as important communicative tools in human negotiations, and prior work has shown that artificial agents can use synthetic expressions to enhance negotiation outcomes and to train negotiation skills. These prior findings have focused on virtual agents and little is known about the effect of expressions when negotiating with physical robots. Therefore, in this study, we compared how participants negotiated with emotionally expressive virtual agents and android robots. Participants$(\mathrm{n}={82})$, as a proposer, played a nonverbal version of a four-issue ultimatum bargaining game with a counterpart who was either a virtual agent or an android robot. Before negotiating, participants observed their counterpart's emotional reactions to potential deals. The results showed that participants were better able to estimate the preferences of virtual counterparts compared with robotic counterpart, and thereby achieve better win-win solutions. We find this effect was mediated by uncanniness: participants found the emotional robot to be uncanny, and this undermined their ability to extract information from the robot's expressions. We discuss theoretical mplications for our understanding of human-robot negotiation and practical implications for the design of effective robot negotiators.

In this study, we conducted multi-label annotation based on textual entailment for text data related to incident cases that occurred at an electric power company, utilizing GPT, a general-purpose LLM, without additional training. The experiment examined GPT’s zero-shot textual entailment performance and, for comparison, its one-shot textual entailment performance using prompt engineering. Furthermore, in this study, the abstract category labels for the causes of incidents used in the annotation task were also extracted zero-shot from GPT-4, and these were approved by human participants to determine the labels. The results of the experiment showed that, particularly in the one-shot approach using prompt engineering, GPT exhibited strong generalization capabilities and demonstrated promising performance, approaching the level of human annotators in certain evaluation metrics. However, it was also suggested that when dealing with highly specialized and multifaceted cases like those in this study, careful adjustments in model choice and prompt settings are required.

Behavioral trust is to entrust resources to the trustee, expecting a high return while accepting the risk of betrayal. Previous studies have demonstrated bias on behavioral trust, that trust is not necessarily rationally to the expected values, and is inconsistent to human vs. non-human counterparts. People on the one hand seem to be aversive to intentional betrayal (Betrayal Aversion), but on the other hand, people avoid being dependent on algorithms rather than human (Algorithm Aversion). Yet, these aversions are not comprehensively investigated. The present study conducted a well-controlled behavioral game experiment that systematically explored the entrusting (risk-taking) behavior when facing a counterpart (human vs. AI) or a natural risk, and further explored the effect of the counterpart’s computational ability (intentional for human and algorithmic for AI). Participants (n=284) played a trust game with (a) a human with intentional decisions, (b) an AI with algorithmic decisions, (c) a human with random decisions, or (d) an AI with random decisions whose probability of return was known, as well as a lottery task structurally equivalent to the trust game. Entrusting decisions to different levels of probability of return were measured to compute the Minimum Acceptable Probability (MAP) as a quantitative measure of trust. The results showed that participants were more trusting of a counterpart than a lottery machine, yet this tendency did not differ by counterparts (i.e., human vs. AI counterparts), or between high computational ability or lower (i.e., intentional/algorithmic decision vs. random decisions). The results suggest an overtrust bias, rather than an aversive bias to a counterpart with agency, whether they are AIs or humans, regardless of their intentions.

Individual differences in preferences allow for integrative (win–win) solutions in negotiations. However, reaching an integrative solution is difficult as each party's preferences and limits are private and must be inferred. We hypothesized that teaching people to infer a generative model of how individuals appraise outcomes and express them as emotional expressions, i.e., an appraisal model, contributes to improving the capability of mental state inference and thus facilitates integrative solutions. In the present study, we compared participants' performance in a 4-issue negotiation after training participants to infer appraisal model during three 2-issue negotiations with a visualized appraisal process and text feedback with the performance of those without inference learning. The results showed that training participants to infer appraisal model helped them better estimate their counterpart's preferences but did not lead them to negotiate more integrative solutions.

Research has demonstrated the effectiveness of personalization in persuasive agents, recommendation agents, and nudge agents. Ultimate personalization targets the presentation of information tailored to individual’s nuanced understanding and preferences, instead of relying on broad attributes such as personality traits, age, or gender. In this paper, we propose a method for persuading people to accept controversial technologies, such as a fully automated vehicle, by presenting personalized positive counter-propositions to address their concerns about the technology.

In recent years, robots have been equipped with the ability to express emotions and have begun building social relationships with people. However, the significance and effectiveness of incorporating emotion in industrial robots, which have a strong instrumental nature, is not fully understood. We investigated how emotional expressions of an industrial robot influence human collaborative decision-making. The participants (n=52), in a laboratory experiment, engaged in a dessert survival task with an arm robot in a 2 (emotion expression: present vs. absent) × 2 (competence: high vs. low) between-participants study. Emotion was expressed using color through a LED strip of lights - e.g., anger was conveyed by flashing red. The results showed that emotion expression and competence did not influence the final agreement and, in fact, emotion expressions made the interaction longer, emphasizing the difficulty in communicating emotion and the reason for those expressions. We discuss lessons learnt and provide insight on improving the value of emotion expression in industrial robots.

Negotiation is a social interaction aimed at reaching a mutually beneficial agreement among all participants in a conflict situation. Unfortunately, parties often find negotiations threatening or aversive, undermining the chances of reaching good agreements. Politeness strategies are means of communicating one's demands to a counterpart without threatening the counterpart's "face" by using tactical phrasing. Politeness strategies are classified into positive, negative, and off-record strategies depending on how they avoid face-threatening acts. In the present study, we investigated whether differences in the politeness strategies used by a virtual agent impact negotiated outcomes in a non-zero-sum situation. The participants (n=106) engaged in an online multi-issue negotiation with one of three agents (using the positive, off-record, or no politeness strategies, while the negative strategy was excluded because of validation failure). The results showed that the agents who used the off-record strategy were able to extract greater concessions from their human partners, whereas positive politeness, which does not threaten the other's face, led to fairer negotiated agreements. The human participants were comfortable exploiting agents who failed to adopt any politeness in their language. Politeness is a part of the toolbox that people use to manage the social rewards and punishments associated with all interactions, and our work highlights that agents can use this important social tool.

ASD children are characterised by a lack of intentionality. We analysed nonverbal and verbal information, associated with heart rate and emotional feeling, respectively, in ASD and neurotypical children. Analogies of heart rate between ASD and neurotypical children were expressed when the human was ‘passive’ actor and the robot was ‘active’ actor; disanalogies were released when the human was ‘active’ actor. Only ASD children reported better emotional feeling ‘after’ than ‘before’ the interaction with the robot. These results suggest that ASD children might be more reliable to low-level intentionality represented by robots, than to high-level intentionality associated with humans.

Conventional studies have suggested that emotion representation of a robot affects human action decision in human-robot communication. The influence of emotional expression of the robot to the human decision-making can be represented by a simple fuzzy inference model. However, they often use emotion expression by mimicking human facial expression on a robot, however, emotion expression by no facial expression but the motion of limbs, LEDs, and audios of a humanoid robot has not been studied for the evaluation of human action decision effect. This study proposes an experiment design based on a finite iterated prisoner's dilemma game with a humanoid robot that shows multimodal emotion expression during the game in order to investigate the effect of the emotion expression of the robot on cooperative and/or selfish action decision making of the human individuals. The experimental results are analyzed to show the influence of the decision making and impression on the robot of the people according to the emotional expression of the humanoid robot.

Assuming that a mind is a source of variety in behavior is important in a non-zero-sum situation in which cooperators and competitors (free riders) are mixed. In such a context, one should rapidly differentiate between competitors and cooperators and avoid useless battles against competitors. Once an actor's mind (intentions) and current situation have been identified, the strategy of assuming a mind enables one to infer the actor's future behavior even if the situation is different. In the present study, we conducted an experiment to assess whether humans predict the future behavior of an agent (a 1 degree of freedom (DOF) stick) in a different situation by attributing malice or benevolence. We developed a stick-mediated interactive hole that provides minimum modal interaction in a non-zero-sum game situation. Participants were asked to insert as many sticks as they could into the hole within two minutes. The motor behind the hole produced cooperative or obstructive actions. The results show that participants who performed the task with a cooperative hole attributed benevolence and predicted future cooperative behavior in a different task and that participants who performed the task with an obstructive hole attributed malice but did not predict future obstructive behavior.

The purpose of this work is to investigate which of an agent's properties determines leader-follower relationships in cooperative tasks performed by a human and an agent (a computer). The possible factors of an agent are intelligence, obstinance, and appearance. In this paper, we focused on intelligence and obstinance and conducted a psychological experiment using a mark matching game with a declaration phase, which enables us to observe who becomes the leader in a cooperative task. Experimental results showed that humans tend to follow an agent who has low intelligence and more obstinance rather than an agent who has high intelligence and less obstinance, and we found that obstinance is more important than intelligence in being a leader in human-computer interaction.

Although humans perceive animacy in blinking fireflies, humans do not perceive animacy in flashing traffic lights. In the present study, we conducted a psychological experiment to investigate which characteristics of temporal changes of luminance affect animacy perception in humans. Unpredictability in temporal change might be one of the key factors that cause animacy perception. We focused on the periodicities and chaos intensities of time series because both characteristics are related to the subjective impression of predictability. Furthermore, we investigated the effects of top-down categorical knowledge on the animacy perception of an entity. The participants were asked to rate videos in which the luminance of a target entity (i.e., a squid, charcoal fire, or power lamp) changed according to pre-defined periodicities and chaos intensities. The experimental results indicated that the participants perceived animacy strongly based on luminance changes with periodic and high chaos intensities and that animacy was not affected by the chaos intensity of aperiodic change. These results imply that the periodicity of a time series contributes more strongly to the cause of human animacy perception than the chaos characteristics.

Researchers in artificial intelligence and robotics have long debated whether robots are capable of possessing minds. We hypothesize that the mind is an abstract internal representation of an agent's input-output relationships, acquired through evolution to interact with others in a non-zero-sum game environment. Attributing mental states to others, based on their complex behaviors, enables an agent to understand another agent's current behavior and predict its future behavior. Therefore, behavioral complexity, i.e., complex sensory input and motor output, might be an essential cue in attributing abstract mental states to others. To test this theory, we conducted experiments in which participants were asked to control a robot that exhibits either simple or complex input-output relationships in its behavior to achieve goals by pushing a button switch on a remote control device. We then measured participants' subjective impressions of the robot after a sudden change in the mapping between the button switch and motor output during the goal-oriented task. The results indicate that the complex relationship between inputs and a robot's behavioral output requires greater abstraction and induces humans to attribute mental states to the robot in contrast to a simple relationship scenario.

Human-interactive robots are assessed according to various factors, such as behavior, appearance, and quality of interaction. In the present study, we investigated the hypothesis that impressions of an unattractive robot will be improved by emotional interaction with physical touch with the robot. An experiment with human subjects confirmed that the evaluations of the intimacy factor of unattractive robots were improved after two minutes of physical and emotional interaction with such robots.

We investigated how people communicate in a situation where there is no shared languages among them. We designed "image sharing game", in which players must cooperate to solve it. The only communication media available for the players is the exchange of gestures represented by skeleton diagrams. In this game, some of the player-pairs developed not only signals for objects, but also meta-signals that controlled communication. Those with meta-signals were observed to perform better than those without meta-signals.

Attributing intention to others is a difficult task for individuals with autism spectrum disorders (ASD). We hypothesize that individuals with ASD ascribe a lower level of abstraction to observed behavior than typically developing individuals. In this study, we discuss the development of a robot that gradually increases the level deviation from a fixed pattern behavior, inducing individuals to ascribe a greater level of abstraction to observed behavior.

We are investigating on how to build a human-agent interface based on mind-reading mechanism. The research on the emergence of communication sheds light on mind-reading mechanism. However, theses researches so far are conducted only in cooperative environment. We investigated how communication emerges using a series of non-zero-sum games as a task. The task is such that the conflict of interests exists but cooperation is necessary for overall good performance. The only communication media is sending of monotonic sound. For comparison, the experiments are also conducted on soundless condition, i.e., no communication is possible between the players. It is confirmed that the existence of communication media improved the performance even under the conflict of interests. On sound condition, some of the subjects could establish communication and earned nearly optimal points. But others could not communicate well with the partner, and eventually could earn only small points. The task is very simple but various strategies for communication could be observed. The emergence of varieties of strategies is a key point for understanding of mind-reading mechanism.

In the present study, human altruistic behavior was used to measure a robot's ability to express emotional states through facial expressions. An experimental investigation was conducted, in which human participants interacted with a humanoid robot and an LCD monitor was used to display facial expressions. A simple line drawing of a face was used to remove realistic, biological human features from the agent's face. Participants were asked to play the ultimatum bargaining game, which is usually used to measure human altruistic behavior. All participants were assigned as the proposer and were instructed to decide their offer within one minute by controlling a slider bar. The corners of the mouth of the line drawing simply moved upward or downward depending on the position of the slider bar. The results suggest that the change of facial expressions of a simple line drawing of a face significantly affected altruistic behavior. Offers were increased by 13% by showing contingent changes of facial expression.

We conducted an experimental investigation on human adaptation to change in an agent's strategy through a competitive two-player game. Modeling the process of human adaptation to agents is important for designing intelligent interface agents and adaptive user interfaces that learn a user's preferences and behavior strategy. However, few studies on human adaptation to such an agent have been done. We propose a human adaptation model for a two-player game. We prepared an on-line experimental system in which a participant and an agent play a repeated penny-matching game with a bonus round. We then conducted experiments in which different opponent agents (human or robot) change their strategy during the game. The experimental results indicated that, as expected, there is an adaptation phase when a human is confronted with a change in the opponent agent's strategy, and adaptation is faster when a human is competing with robot than with another human.

We investigated how humans establish communication in a novel environment using an artificial cooperative task. Subjects are asked to solve a kind of 8-puzzle game cooperatively. One subject (director) can see the design of the puzzle and instructs how to move a tile. The other (operator) cannot see the design but can manipulate the puzzle. The director can communicate his intention only through his body movement. We conducted two experiments. In Experiment I, the operator role is played by the experimenter, who follows the pre-determined algorithms for establishing communication. The director role is played by a recruited subject. In Experiment II, both the director and the operator roles are played by recruited subjects. In both the experiments, simple languages for communication were developed between the players, but the strategies adopted were quite different. The results correspond to two approaches to future Human-Agent communication - human-controlled communication and mind-reading communication. We should re-consider the role of mind-reading communication in designing Human-Agent interface.

We propose a novel method of expressing emotions for a robot by dynamically changing the color luminosity of its body. To identify color and dynamic parameters for representing emotions, we asked subjects to select hue values and frequencies and waveforms of blinking to represent the 24 emotions of Plutchik's Wheel of Emotions. The results indicate that the hue value represents basic types of emotion, i.e., anger-red, anticipation-amber, joy-yellow, trust-green, fear-violet, surprise-red, sadness-blue, and disgust-purple, and the duration and waveform represent emotion intensity, i.e., a rect-angular waveform with high frequency represents intense emotion. Another experiment validated that emotions of a robot expressed using our model impressed other subjects appropriately.

How can humans come to share the meaning of signals when only very limited media are available and there are no pre-defined meanings to signals? To answer the above question, we designed two-player games, which require the players' cooperation to play. The only communication means are to send color (hue) signals in one game, or monotonic sound signals in another. The player must assign a necessary meaning to an available signal, and send it to the partner. The partner must infer its meaning (sender's intention) and act cooperatively. Using these games, the process of sharing the meaning of signals is investigated, and some interesting common features are found. The process is based on mind-reading of the partner's intention, which is a key ability for any types of human communication. The mechanism is analyzed in the relevance theory framework. Our findings can be used for improving human-agent communication where no pre-defined languages are available.

In the present study, we investigated whether a robot is able to deceive a human by producing a behavior against him/her prediction. A feeling of being deceived by a robot would be a strong indicator that the human treat the robot as an intentional entity. We conducted a psychological experiment in which a subject played Darumasan ga Koronda, a Japanese children's game, with a robot. A main strategy to deceive a subject was to make his/her mind believe that the robot is stupid so as not to be able to move quickly. The experimental result indicated that unexpected change of a robot behavior gave rise to an impression of being deceived by the robot.

Whether or not humans can construe the behaviours of entities depends on their psychological stance. The philosopher Dennett proposed human cognitive strategies (three stances) according to which humans construe the behaviour of other animated objects, including other humans, artifacts and physical phenomena. The three stances are `intentional', `design' and `physical'. In the present study, we propose a novel method for detecting the underlying stance adopted when humans construe the behaviour of entities. In our method, the subjects were asked to select the most suitable action sequence shown in three animations, each of which represents one of Dennett's three stances. We then conducted an experiment to investigate which stance is adopted for different kinds of behaviour of various entities. Subjects were asked to select, from 60 short movies, the most suitable animation that represents the behaviour of an entity. The results indicate that the subjects did not focuse on their knowledge about the entity, but instead focused on motion characteristics per se, owing to the simple and typical motion of an abstractly shaped object.

The paper investigates the effect of environmental context and attention on how humans attribute goals to the behavior of man-made objects. We presented participants with moving chairpsilas motion along a B-spline trajectory. We manipulated the context by whether an table was placed in the environment or not. We also investigated the effect of chairpsilas attention toward an object on goal attribution. The results indicates that existence of objects per se did not affect inferred goals. However, directing attention toward objects contributed to control the attributed goals which related to the object. These results also suggested that not only taking the intentional stance but also taking the design stance is effective to construe artifactspsila intention.

An artifact’s behavior must be easily construed and interpreted as meaningful signals in a social or working context. In order to design such an artifact’s behavior, we could exploit human psychological functions - theory of mind (ToM) - the ability to interpret other people’s behavior in terms of intentional causal mental states such as beliefs, desires and intentions. In order to apply theory of mind to human-robot interaction, the mechanism that trigger intention attribution must be revealed. The present study examined the effect of reactive movements performed by a non-humanoid robot, including different shaped artifact: chair and cube, on the intention attribution. The result indicated that whether or not humans could construe behaviors of an artifact in terms of its goal depends on how human could attribute intention to the artifact and that reactive movements would be a cue for such mental state attribution.

The present study examined the effect of artifact's direction of attention detector (DAD) stimulating actions on the human psychological stance to the artifact. The DAD is a specialized brain function used to determine the attention target by combining information from separate detectors, e.g., direction of eye, head, body and locomotion. We designed sequences of DAD stimulating movement of a chair which can represent its attention to a subject. The chair was controlled by a remote experimenter. The subjects were randomly assigned to one of the two conditions: the DAD stimulating condition and the random action condition (control condition). The result indicated that the DAD stimulating actions changed the subject's stance and enabled them to discern its intention

Vision based analysis of human action is a fundamental issue for a robot to interact intelligently with a human. We present a technique for 3D human head tracking using stereo camera. We apply the particle filter algorithm to the problem of head tracking in a sequence of depth map received from a stereo camera. Posterior distribution of human head pose is represented by a set of hypothesized polygons. A human head is modeled by polygon mesh which contains statistical color information such as skin or hair color. Head tracking and orientation estimation are solved independently for reduction of the number of particles. Tracking experiments demonstrate the validity of our approach.

A method for behavior acquisition that considers embodiment by associating tactile information to visual input is described. An agent that acts in the physical world always suffers constraints derived from embodiment. On the other hand, embodiment plays a very important role in the formation of visual function. Philosophical and clinical medicine findings assert that vision does not function without learning through experiences of haptic motion. In this paper, we discuss the relation between vision, embodiment, and behavior. We develop a method for behavior acquisition through associating vision and tactile sensors. We perform experiments of obstacle avoidance using a computer simulation and a real agent to test the validity of our method.

We propose a new architecture for recognizing objects based on a concept of "embodiment" as one of primitive functions for a cognitive robot. We define the term "embodiment" as the size and shape of the agent's body, locomotive ability and its sensor. According to embodiment, an object is represented by reaching action paths, which correspond to a set of sequences of movements taken by the agent for reaching the object. Visual information is used to obtain sensorimotor mapping which represents the relationship between the change of object's appearance and the movement of the agent. On the other hands, tactile information is utilized to evaluate the change of physical condition of the object caused by such movement. By means of this method, the agent can recognize an object without depending on its position and orientation in the environment. The experimental result of computer simulation is shown to validate the method.

To promote robotic soccer research, we need a low cost and portable robot with some sensors and a communication device. To date, there is no platform for robotic soccer. Therefore, each research must build his own robots or utilize robots which are commercially available. This paper describes how to construct a robot system which includes a lightweight and low-cost mobile robot with visual, tactile sensors, TCP/IP communication device, and portable PC where Linux is running. An example of the developed soccer robot system and preliminary experimental results are also shown.

An agent must acquire internal representation appropriate for its task, environment, and sensors. As a learning algorithm, reinforcement learning is often utilized to acquire the relation between sensory input and action. Learning agents in the real world using visual sensors are often confronted with the critical problem of how to build a necessary and sufficient state space for the agent to execute the task. We propose the acquisition of a relation between vision and action using the visual state-action map (VSAM). VSAM is the application of a self-organizing map (SOM). Input image data is mapped on the node of the learned VSAM. Then VSAM outputs the appropriate action for the state. We applied VSAM to a real robot. The experimental result shows that a real robot avoids the wall while moving around the environment.