Gaze is a crucial social cue in any interacting scenario and drives many mechanisms of social cognition (joint and shared attention, predicting human intention and coordinating tasks). Gaze is an indication of social and emotional functions affecting the way the emotions are perceived. Evidence shows that embodied humanoid robots endowed with social abilities can be seen as sophisticated stimuli to study several mechanisms of human social cognition while increasing engagement and ecological validity. In this context, building a robotic perception system to automatically estimate the human gaze only relying on robot’s sensors is still demanding. Main goal of the article is to propose a learning robotic architecture estimating the human gaze direction in table-top scenarios without any external hardware. Table-top tasks are largely used in experimental psychology because they are suitable to implement numerous face-to-face collaborative scenarios. Such an architecture can provide a valuable support in studies where external hardware might represent an obstacle to spontaneous human behaviour, especially in environments less controlled than the laboratory (e.g., in clinical settings). A novel dataset was also collected with the humanoid robot iCub, including images annotated from 24 participants in different gaze conditions.

A key challenge in human-robot interaction research lies in developing robotic systems that can effectively perceive and interpret social cues, facilitating natural and adaptive interactions. In this work, we present a novel framework for enhancing the attention of the iCub humanoid robot by integrating advanced perceptual abilities to recognise social cues, understand surroundings through generative models, such as ChatGPT, and respond with contextually appropriate social behaviour. Specifically, we propose an interaction task implementing a narrative protocol (storytelling task) in which the human and the robot create a short imaginary story together, exchanging in turn cubes with creative images placed on them. To validate the protocol and the framework, experiments were performed to quantify the degree of usability and the quality of experience perceived by participants interacting with the system. Such a system can be beneficial in promoting effective humanrobot collaborations, especially in assistance, education and rehabilitation scenarios where the social awareness and the robot responsiveness play a pivotal role.

Gaze plays a crucial role in human social behavior. Notably, the same applies also to interactions between humans and robots, as gaze can communicate intentions and express interest or aversion similarly to what happens among humans. Besides the direction of gaze (direct vs. averted gaze), its temporal characteristics, such as duration, significantly affect our perception and interpretation of the other’s behavior. In the context of Human-Robot Interaction (HRI), this is still poorly investigated. Thus, the present study aimed to investigate whether, and how, the duration of the robot direct gaze impacts participants’ attitudes towards robots. To do so, participants observed the humanoid robot iCub, whose direct gaze varied in duration between 1 and 8 seconds. Then, they used three Likert scales to rate to what extent the robot gaze made them feel i) comfortable, ii) trustful, and iii) threatened, with participants’ rating operationalizing their attitudes towards the robot. Results showed that, overall, a positive relationship emerged between the duration of the robot gaze and participants’ attitudes, i.e., longer gaze duration led to higher ratings for all three Likert scales.

In this study, we developed a setup allowing two participants ("operators") to jointly control a single humanoid robot body. Specifically, each operator controlled one robot arm, using their anatomically congruent hand (left hand controlling a left robot arm and vice versa). In our task participants had to move each robot arm into one of two possible positions (arm raised or lowered). We used this setup to investigate (1) whether presenting prior information about the relationship between movements performed by each participant ("We-information") can facilitate performance in this task, (2) how joint control over a robot body affects sense of control over the robot and sense of joint agency with the other operator, and (3) how it influences the perceived boundaries between oneself and the others (the robot and the other operator), the so-called "self-other" overlap. We found that (1) "We-information" increased the speed of task performance, but only for simpler configurations, (2) participants experienced high level of sense of control over the robot which increased throughout the task, and (3) a short session of joint control over a humanoid led to pronounced increase in self-other overlap (blurring of boundaries) with both the robot and a co-operator. We discuss implications of our results for understanding of human body representation and how they can inform future applications, such as exoskeletons for individuals affected with hemiplegia.

This study investigates the effects of non-verbal socio-affective feedback provided by two different virtual agents (i.e. a human and a humanoid robot) on the performance of children with Autism Spectrum Disorder (ASD), in a decision-making task. The task, inspired by the “Shell Game”, required participants (N = 29, Mean age = 6.5 years) to guess which cup out of two was hiding a ball. After participants made their choice, the virtual agents provided either positive, negative, or no feedback. Results indicated no significant effects on response times (RTs) but revealed a main effect of virtual agent type on accuracy, with participants performing less accurately when playing with the robot virtual agent compared to the human virtual agent. Furthermore, participants' preference for the robot virtual agent was associated with lower accuracy. No significant effects were found on feedback presence. These findings suggest that the preference for the robot virtual agent may have distracted some children, leading to decreased accuracy during the task. Attentional focus on the virtual agent during the cue presentation may have influenced performance but not overall cognitive processing speed. Understanding individual preferences and potential distractions using virtual agents can help optimize the development of interventions for children with ASD.

This study investigated whether mentalistic text-vignettes of a robot enhances participants’ tendency to adopt the intentional stance. We also investigated whether the valence of a robot's action consequence influences judgements of intentionality and moral responsibility. We presented participants with scenarios from the InStance Test before and after they read the mentalistic text-vignettes. We found that InStance scores were significantly higher after they read the text-vignettes, implying an increased likelihood of adopting the intentional stance. This effect may have arisen from the mentalistic description of the robot, which could have primed participants to mentally simulate its experiences. Additionally, we found that the valence of a robot's action consequence did not significantly affect ratings of intentionality, which replicates previous results. However, unlike previous studies the valence of the action consequence did not affect moral responsibility ratings. Future studies could investigate (1) how the appearance of a robot influences moral responsibility judgements and (2) if text-vignettes using mechanistic descriptions modulate the tendency to adopt the intentional stance.

This paper explores the concept of awareness in the context of embodied artificial intelligence (AI), aiming to provide a practical definition and understanding of this multifaceted term.Acknowledging the diverse interpretations of awareness in various disciplines, the paper focuses specifically on the application of awareness in embodied AI systems.We introduce six foundational elements as essential building blocks for an aware embodied AI.These elements include access to information, information integration, attention, coherence, explainability, and action.The interconnected and interdependent nature of these building blocks is emphasised, forming a minimal base for constructing AI systems with heightened awareness.The paper aims to spark a dialogue within the research community, inviting diverse perspectives to contribute to the evolving discipline of awareness in embodied AI.The proposed insights provide a starting point for further empirical studies and validations in real-world AI applications.

The integration of diffusion models (DMs) into generative AI systems presents an approach with implications for ethical and moral AI development and our understanding of human-AI interaction.This study explores the intersection of generative AI, human values, and neuroscience, emphasizing the significance of valueawareness in AI systems.The methodology involves a behavioral experiment to evaluate the accuracy of DM-generated visual stimuli in capturing human values and related keywords.Results indicate promising match rates, marking stride in aligning AI systems with ethical and moral considerations.Additionally, the study introduces a criterion for selecting stimuli based on an "Aha" moment, setting the stage for an EEG experiment to explore the neural correlates associated with becoming aware of a value.This multidisciplinary study is a step toward the development of procedures to evaluate the effectiveness of Value-Aware AI systems in enhancing the perceived ethical and moral agency.

This article describes a study investigating the effects of decision-making autonomy of a robot, which is teleoperated in a simulated unstable or dangerous environment. It specifically focuses on robot’s autonomy to disregard the user’s command if the robot finds an alternative method of achieving the same goal as pursued by the user, but with significantly reduced risk of failure. Such autonomous control module might prove especially useful under circumstances where human operators cannot access or process all available information quickly enough to make the most optimal decision. We conducted an experiment in which subjects participated in a task of teleoperating either a robot that possesses such autonomous cognitive module, or not. We found that such module significantly reduced their sense of agency over the robot as well as the sense of doing the task together with the robot (sense of joint agency). Most interestingly, it reduced their subjective ratings of performance in the task, when in fact their actual performance was better with such an autonomous robot. A further analysis revealed that this counterintuitive finding was due to an effect of bias: in our study loss of control associated with operating an autonomous robot on average lowered the reported subjective performance by 11% points, with other factors staying equal. These results suggest that this type of assistive autonomy can be beneficial for performance, but might lead to unwanted effects that need to be overcome in order for such system to prove useful in practical applications.

When acting together with another human agent, humans form shared representations with their partner in order to predict and adjust to their partner’s behaviors and ensure a smooth and efficient joint action. Previous work has shown that humans do not form shared representations when acting with partners they do not perceive as intentional, such as computers or robots. In the current study, we investigated the effect in the opposite direction: we asked whether engaging in a task with the humanoid robot iCub, in order to achieve a shared goal, could influence the perceived intentionality of the robot. In our study, participants completed a target tracking and detection task with iCub, in which each agent fulfilled different but complementary roles, thus sharing the task. Participants’ likelihood of perceiving the robot as intentional was assessed before and after the interaction. Results showed that participants were more likely to perceive iCub as an intentional agent after they shared the task with it, suggesting that interacting with a robot to achieve a common goal promotes attribution of intentionality toward the robot.

With the increasing use of social robots and automated machines in our daily lives, roboticists need to design robots that are suitable for human-robot collaboration. Prior work suggests that robots that are perceived to be intentional (i.e., are able to experience mental life capacities), can, in most cases, positively affect human-robot collaboration. With studies highlighting the importance of individual differences and how they drive our perception. We aimed to investigate how individual differences in gender moderate the relationship between subjective perceptions of robots and behavioral performance in a human-robot collaborative task. Participants rated a humanoid robot (i.e., iCub) on whether it can experience mental life capacities and completed a collaborative task with it. We correlated their subjective ratings with the completion time of the collaborative task and found a positive correlation between perceiving iCub to experience basic and social emotion with their performance (i.e., movement times). This relationship, however, was evident for males but not females. The results of this study suggest that perceiving humanoid robots as capable of experiencing mental states influences collaborative performance differently depending on gender. These findings can be relevant for the field of social robotics and to successfully design robot interaction partners for workplaces.

Understanding how and when humans attribute intentionality to artificial agents is a key issue in contemporary human and technological sciences. This paper addresses the question of whether adopting intentional stance can be modulated by exposure to a 3D animated robot character, and whether this depends on the human-likeness of the character's behavior. We report three experiments investigating how appearance and behavioral features of a virtual character affect humans’ attribution of intentionality toward artificial social agents. The results show that adoption of intentional stance can be modulated depending on participants' expectations about the agent. This study brings attention to specific features of virtual agents and insights for further work in the field of virtual interaction.

Humans interpret and predict behavior of others with reference to mental states or, in other words, by adopting the intentional stance. The present study investigated to what extent individuals adopt the intentional stance towards two agents (a humanoid robot and a human). We asked participants to judge whether two different descriptions fit the behaviors of the robot/human displayed in photographic scenarios. We measured acceptance/rejection rate of the descriptions (as an explicit measure) and response times in making the judgment (as an implicit measure). Our results show that at the explicit level, participants are more likely to use mentalistic descriptions for the human agent and mechanistic descriptions for the robot. Interestingly, at the implicit level, we found no difference in response times associated with the robotic agent. We argue that, at the implicit level, both stances are processed as "equally likely" to explain the behavior of a humanoid robot, while at the explicit level there is an asymmetry in the adopted stance. Furthermore, cluster analysis on participants' individual differences in anthropomorphism likelihood revealed that people with a high tendency to anthropomorphize tend to accept faster the mentalistic description. This suggests that the decisional process leading to adoption of one or the other stance to adopt is influenced by individual tendency to anthropomorphize non-human agents.

When humans interact with artificial agents, they adopt various stances towards them. On one side of the spectrum, people might adopt a mechanistic stance towards an agent and explain its behavior using its functional properties. On the other hand, people can adopt the intentional stance towards artificial agents and explain their behavior using mentalistic terms and explain the agents’ behavior using internal states (e.g., thoughts and feelings). While studies continue to investigate under which conditions people adopt the intentional stance towards artificial robots, here, we report a study in which we investigated the effect of social framing during a color-classification task with a humanoid robot, iCub. One group of participants were asked to complete the task with iCub, in collaboration, while the other group completed an identical task with iCub and were told that they were completing the task for themselves. Participants completed a task assessing their level of adoption of the Intentional Stance (the InStance test) prior to - and after completing the task. Results illustrate that participants who "collaborated" with iCub were more likely to adopt the intentional stance towards it after the interaction. These results suggest that social framing can be a powerful method to influence the stance that people adopt towards a robot.

In this study, we examined whether the likelihood of attributing intentionality to robots is influenced by the human-likeness of errors during HRI. To this end, we designed an experimental protocol in which users performed a melody in a joint task with the iCub robot. We programmed the iCub robot to make an error in 60% of the repetitions. For half of the users, in the erroneous trials, the robot displayed a human-like error, i.e. switched one element of the melody by pressing the wrong key. For the other half of users, the robot erred mechanically, i.e., it interrupted to play the melody and moved back and forth between two keys in an "endless" loop. Before and after the joint musical task, we administered the InStance Test to evaluate the likelihood of treating the robot as an intentional agent. Results showed that mechanical errors during HRI reduced intentionality attribution toward the robot.

Understanding the human cognitive processes involved in the interaction with artificial agents is crucial for designing socially capable robots. During social interactions, humans tend to explain and predict others' behavior adopting the intentional stance, that is, assuming that mental states drive behavior. However, the question of whether humans would adopt the same strategy with artificial agents remains unanswered. The present study aimed at identifying whether the type of behavior exhibited by the robot has an impact on the attribution of mentalistic explanations of behavior. We employed the Instance Questionnaire (ISQ) pre and post-observation of two types of behavior (decisive or hesitant). The ISQ probes participants' stance towards a humanoid robot by requiring them to choose the likelihood of an explanation (mentalistic vs. mechanistic) of iCub depicted in sequences of photographs. We found that decisive behavior, with rare and unexpected "hesitant" behaviors, lead to more mentalistic attributions relative to primarily hesitant behavior. Findings suggest that higher expectations regarding the robots' capabilities and unexpected actions might lead to more mentalistic descriptions.

In this paper we present the TobiiGlassesPySuite, an open-source suite we implemented for using the Tobii Pro Glasses 2 wearable eye-tracker in custom eye-tracking studies. We provide a platform-independent solution for controlling the device and for managing the recordings. The software consists of Python modules, integrated into a single package, accompanied by sample scripts and recordings. The proposed solution aims at providing additional methods with respect to the manufacturer's software, for allowing the users to exploit more the device's capabilities and the existing software. Our suite is available for download from the repository indicated in the paper and usable according to the terms of the GNU GPL v3.0 license.

In this study, we examined if humans adapt their performance to delays in robot's actions in a leader-follower interaction scenario. Participants were asked to “teach” a sequence of musical tones to the iCub robot. The robot repeated the sequence with decreasing delay between its own taps and taps performed by the participants. We observed that mean period of participants' tapping behavior was affected by the iCub's performance. This suggests that humans are sensitive to subtle parameters in robot's behavior and they adapt to them in leader-follower contexts.

The present study aims at investigating how eye contact established by a humanoid robot affects engagement in human-robot interaction (HRI). To this end, we combined explicit subjective evaluations with implicit measures, i.e. reaction times and eye tracking. More specifically, we employed a gaze cueing paradigm in HRI protocol involving the iCub robot. Critically, before moving its gaze, iCub either established eye contact or not with the user. We investigated the patterns of fixations of participants' gaze on the robot's face, joint attention and the subjective ratings of engagement as a function of eye contact or no eye contact. We found that eye contact affected implicit measures of engagement, i.e. longer fixation times on the robot's face during eye contact. Moreover, we showed that joint attention was elicited only when the robot established eye contact, whereas no joint attention occurred when it did not. On the contrary, explicit measures of engagement with the robot did not vary across conditions. Our results highlight the value of combining explicit with implicit measures in an HRI protocol in order to unveil underlying human cognitive mechanisms, which might be at stake during the interactions. These mechanisms could be crucial for establishing an effective and engaging HRI, and provide guidelines to the robotics community with respect to better robot design.

The present study highlights the benefits of using well-controlled experimental designs, grounded in experimental psychology research and objective neuroscientific methods, for generating progress in human-robot interaction (HRI) research. More specifically, we aimed at implementing a well-studied paradigm of attentional cueing through gaze (the so-called “joint attention” or “gaze cueing”) in an HRI protocol involving the iCub robot. Similarly to documented results in gaze-cueing research, we found faster response times and enhanced event-related potentials of the EEG signal for discrimination of cued, relative to uncued, targets. These results are informative for the robotics community by showing that a humanoid robot with mechanistic eyes and human-like characteristics of the face is in fact capable of engaging a human in joint attention to a similar extent as another human would do. More generally, we propose that the methodology of combining neuroscience methods with an HRI protocol, contributes to understanding mechanisms of human social cognition in interactions with robots and to improving robot design, thanks to systematic and well-controlled experimentation tapping onto specific cognitive mechanisms of the human, such as joint attention.

Robots will soon enter social environments shared with humans. We need robots that are able to efficiently convey social signals during interactions. At the same time, we need to understand the impact of robots' behavior on the human brain. For this purpose, human behavioral and neural responses to the robot behavior should be quantified offering feedback on how to improve and adjust robot behavior. Under this premise, our approach is to use methods of experimental psychology and cognitive neuroscience to assess the human's reception of a robot in human-robot interaction protocols. As an example of this approach, we report an adaptation of a classical paradigm of experimental cognitive psychology to a naturalistic human-robot interaction scenario. We show the feasibility of such an approach with a validation pilot study, which demonstrated that our design yielded a similar pattern of data to what has been previously observed in experiments within the area of cognitive psychology. Our approach allows for addressing specific mechanisms of human cognition that are elicited during human-robot interaction, and thereby, in a longer-term perspective, it will allow for designing robots that are well-attuned to the workings of the human brain.