Large language models increasingly serve as conversational agents that adopt personas and role-play characters at user request.This capability, while valuable, raises concerns about sycophancy: the tendency to provide responses that validate users rather than prioritize factual accuracy.While prior work has established that sycophancy poses risks to AI safety and alignment, the relationship between specific personality traits of adopted personas and the degree of sycophantic behavior remains unexplored.We present a systematic investigation of how persona agreeableness influences sycophancy across 13 small, open-weight language models ranging from 0.6B to 20B parameters.We develop a benchmark comprising 275 personas evaluated on NEO-IPIP agreeableness subscales and expose each persona to 4,950 sycophancy-eliciting prompts spanning 33 topic categories.Our analysis reveals that 9 of 13 models exhibit statistically significant positive correlations between persona agreeableness and sycophancy rates, with Pearson correlations reaching r = 0.87 and effect sizes as large as Cohen's d = 2.33.These findings demonstrate that agreeableness functions as a reliable predictor of persona-induced sycophancy, with direct implications for the deployment of role-playing AI systems and the development of alignment strategies that account for personality-mediated deceptive behaviors.

As AI systems increasingly mediate negotiations, understanding how the number of negotiated issues impacts human performance is crucial for maintaining human agency. We designed a human–AI negotiation case study in a realistic property rental scenario, varying the number of negotiated issues; empirical findings show that without support, performance stays stable up to three issues but declines as additional issues increase cognitive load. To address this, we introduce a novel uncertainty-based visualization driven by Bayesian estimation of agreement probability. It shows how the space of mutually acceptable agreements narrows as negotiation progresses, helping users identify promising options. In a within-subjects experiment (N=32), it improved human outcomes and efficiency, preserved human control, and avoided redistributing value. Our findings surface practical limits on the complexity people can manage in human–AI negotiation, advance theory on human performance in complex negotiations, and offer validated design guidance for interactive systems.

The average attention span of humans is decreasing. To adapt to the change, researchers investigate visual and auditory techniques that alter the perception of time passing. However, the auditory technique is underexplored in comparison to visual techniques, which are becoming increasingly important as voice assistants, speech interfaces or time-aware computational/robotic systems proliferate. This paper investigated the effect of pitch and interval variables on time perception. Three pitch levels (static, falling, rising) and two interval patterns (static, narrowing) for a total of six auditory stimuli were compared. Principal findings indicated that rising/falling pitch and interval narrowing, as well as their combinations, can be used to shorten the perception of time. This research contributes to a better understanding of end-users' subjective temporal estimation of auditory stimuli, which can be used to further develop audio-based interactive devices/agents.

Abstract Identifying those at risk for depression is a crucial issue and social media provides an excellent platform for examining the linguistic patterns of depressed individuals. A significant challenge in depression classification problems is ensuring that prediction models are not overly dependent on topic keywords (i.e., depression keywords) such that it fails to predict when such keywords are unavailable. One promising approach is masking—that is, by selectively masking various words and asking the model to predict the masked words, the model is forced to learn the inherent language patterns of depression. This study evaluates seven masking techniques. Moreover, predicting the masked words during the pre-training or fine-tuning phase was also examined. Last, six class imbalanced ratios were compared to determine the robustness of masked words selection methods. Key findings demonstrate that selective masking outperforms random masking in terms of F1-score. The most accurate and robust models are identified. Our research also indicates that reconstructing the masked words during the pre-training phase is more advantageous than during the fine-tuning phase. Further discussion and implications are discussed. This is the first study to comprehensively compare masked words selection methods, which has broad implications for the field of depression classification and general NLP. Our code can be found at: https://github.com/chanapapan/Depression-Detection.

Current gesture interaction paradigm mainly involves a one-to-one gesture-command mapping. This leads to memorability issues regarding (1) the mapping - as each new command requires a new gesture, and (2) the gestures specifics (e.g., motion paths) - that can be complex to leverage the recognition of several gestures. We explore the concept of combining 3D gestures when interacting in smart environments. We first propose a design space to characterize the temporal and spatial combination aspects, and the gesture types used by the combination. We then report results from three user studies in the context of smart TV interaction. The first study reveals that end-users can create gesture sets with combinations fully optimized to reuse gestures. The second study shows that combining gestures can lead to improved memorability compared to single gestures. The third study reveals that preferences for gestures combination appear when single gestures have an abstract gesture-command mapping.

Mindfulness practices are well-known for their benefits to mental and physical well-being. Given the prevalence of smartphones, mindfulness applications have attracted growing global interest. However, the majority of existing applications use guided meditation that is not adaptable to each user's unique needs or pace. This article proposes a novel framework called Attention Regulation Framework (ARF) , which studies how more flexible and adaptable mindfulness applications could be designed, beyond guided meditation and toward self-regulated meditation. ARF proposes mindfulness interaction design guidelines and interfaces whereby practitioners naturally and constantly bring their attention back to the present moment and develop non-judgmental awareness. This is achieved by the performance of subtle movements, which are supported by non-intrusive detection-feedback mechanisms. We used two design cases to demonstrate ARF in static and kinetic meditation conditions. We conducted four user evaluation studies in unique situations where ARF is particularly effective, vis-à-vis mindfulness practice in busy environments and mindfulness interfaces that adapt to the pace of the user. The studies show that the design cases, compared with guided meditation applications, are more effective in improving attention, mindfulness, mood, well-being, and physical balance. Our work contributes to the development of self-regulated mindfulness technologies.

In our fast-paced society, stress and anxiety have become increasingly common. Meditation for relaxation has received much attention. Meditation apps exploit various senses, e.g., touch, audio and vision, but the relationship between human senses and interactive meditation is not well understood. This paper empirically evaluates the effects of single and combined human senses on interactive meditation. We found that the effectiveness of human senses can be defined by their respective roles in maintaining the balance between relaxation and focus. This work is the first to attempt to understand these relationships. The findings have broad implications for the field of multi-modal interaction and interactive meditation applications.

Predicting how users learn new or changed interfaces is a long-standing objective in HCI research. This paper contributes to understanding of visual search and learning in text entry. With a goal of explaining variance in novices' typing performance that is attributable to visual search, a model was designed to predict how users learn to locate keys on a keyboard: initially relying on visual short-term memory but then transitioning to recall-based search. This allows predicting search times and visual search patterns for completely and partially new layouts. The model complements models of motor performance and learning in text entry by predicting change in visual search patterns over time. Practitioners can use it for estimating how long it takes to reach the desired level of performance with a given layout.

We are often overwhelmed by everyday stressors. Mindfulness meditation can help slow things down and bring one's attention into the present moment. Given the prevalence of smartphones, mindfulness-based mobile applications (MBMAs) have received much attention. Current MBMAs mainly use the guided meditation method which may not be always effective, e.g., users may not be able to follow the pace of instructions and they need a private environment. This paper presents a framework for interactive MBMAs which allows users to self-regulate their attention according to their abilities and conditions. The framework is described by an Attention-Regulation Process and has two components: (1) Relaxation Response and (2) Attention Restoration Theory. The framework is validated by our experiment. It also informs future development for interactive meditation and has broad implications for designing mindfulness and well-being.

Mid-air gestures enable intuitive and natural interactions. However, few studies have investigated the use of mid-air gestures for blind people. TV interactions are one promising use of mid-air gestures for blind people, as "listening"' to TV is one of their most common activities. Thus, we investigated mid-air TV gestures for blind people through two studies. Study 1 used a user-elicitation approach where blind people were asked to define gestures given a set of commands. Then, we present a classification of gesture types and the frequency of body parts usage. Nevertheless, our participants had difficulty imagining gestures for some commands. Thus, we conducted Study 2 that used a choice-based elicitation approach where the participants selected their favorite gesture from a predefined list of choices. We found that providing choices help guide users to discover suitable gestures for unfamiliar commands. We discuss concrete design guidelines for mid-air TV gestures for blind people.

Engagement is a key reason for introducing gamification to learning and thus serves as an important measurement of its effectiveness. Based on a literature review and meta-synthesis, this paper proposes a comprehensive framework of engagement in gamification for learning. The framework sketches out the connections among gamification strategies, dimensions of engagement, and the ultimate learning outcome. It also elicits other task - and user - related factors that may potentially impact the effect of gamification on learner engagement. To verify and further strengthen the framework, we conducted a user study to demonstrate that: 1) different gamification strategies can trigger different facets of engagement; 2) the three dimensions of engagement have varying effects on skill acquisition and transfer; and 3) task nature and learner characteristics that were overlooked in previous studies can influence the engagement process. Our framework provides an in-depth understanding of the mechanism of gamification for learning, and can serve as a theoretical foundation for future research and design.

Motion gestures enable natural and intuitive input in video games. However, game gestures designed by developers may not always be the optimal gestures for players. A key challenge in designing appropriate game gestures lies in the interaction-intensive nature of video games, i.e., several actions/commands may need to be executed concurrently using different body parts. This study analyzes user preferences in game gestures, with the aim of accommodating high interactivity during gameplay. Two user-elicitation studies were conducted: first, to determine user preferences, participants were asked to define gestures for common game actions/commands; second, to develop effective combined-gestures, participants were asked to define possible game gestures using each body part (one and two hands, one and two legs, head, eyes, and torso). Our study presents a set of suitable and alternative body parts for common game actions/commands. We also present some simultaneously applied game gestures that assist interaction in highly interactive game situations (e.g., selecting a weapon with the feet while shooting with the hand). Interesting design implications are further discussed, e.g., transferability between hand and leg gestures.