This paper proposes a novel framework for empathetic dialogue generation in Japanese large language models (LLMs) by employing LoRA-based adapter tuning. The proposed system incorporates both cognitive and emotional empathy through two key modules: the Next Intent Predictor and the Intent Combiner. These modules dynamically predict, select, and integrate multiple intent representations, enabling empathetic and user-centered responses tailored to the context. Evaluation results highlight significant improvements, including a notable increase in user-friendliness and empathetic quality compared to the base model, achieving a win rate exceeding 70%. Furthermore, the system demonstrates human-level competitiveness, with over 40% of its responses being rated as comparable to human-generated ones. This study demonstrates the effectiveness of combining reinforcement learning with intent-specific fine-tuning to enhance empathetic capabilities in dialogue systems. By focusing on intent-driven responses, this approach paves the way for more user-centric applications, particularly in privacy-sensitive and culturally nuanced contexts.

This study explores how perfume packaging design, particularly surface treatment techniques and ornamentation, influences the perception of luxury. Using cluster analysis, we examined 23 perfume bottle samples to identify key design factors contributing to a luxurious impression. The results indicate that perfume bottles with refined surface treatment, such as cut patterns and glossy finishes, were strongly associated with higher luxury ratings. In contrast, simpler designs were perceived as less luxurious. Additionally, no significant correlation was found between price per milliliter and luxury perception, as participants were not informed about the price during the experiment. This suggests that, for this study, design elements in perfume packaging, rather than price, influenced the perception of luxury. These findings highlight the importance of balancing design elements in creating a luxurious product and provide insights for future research on the relationship between consumer perceptions and product design.

This study examined how avatar contour thickness influences the Proteus effect, particularly in moral self-perception and altruistic behavior. Based on the metaphorical association between thickness and morality-specifically, the inclination to conform to social norms-we hypothesize that avatars with thick contours will enhance users’ perceptions of morality-related personality traits and foster altruistic behavioral intentions. Through two experiments involving avatar manipulation, we found that interacting with an avatar featuring thick contours significantly enhanced participants’ self-perceived openness and conscientiousness, particularly among female college students. Moreover, it increased their willingness to engage in altruistic behavior. Additionally, our findings revealed that, to elicit the Proteus effect, it is essential to employ a design expression that fosters a sense of body ownership without oversimplifying the avatar’s design elements. This research offers practical implications for the design of avatar-based customer service systems, suggesting that visual cues can influence user behavior and potentially enhance service quality.

This study aimed to explore the acoustic parameters of artificially generated sounds associated with thermal sensations and to examine the possibility that the temperature associated with the sounds is judged from timbral impressions. Specifically, an experiment was conducted to investigate the perceived temperature and timbral impressions of complex artificial sounds that varied in harmonic structure, harmonic amplitude, and temporal envelope. The results showed that participants most strongly associated sounds with low harmonic density and non-decaying frequency envelopes with coldness. This finding suggests the presence of statistical correspondences based on acoustic features reminiscent of metallic sounds. In contrast, the results suggest that it is unlikely that the associated temperature is judged based on timbral impressions. Therefore, the temperature associated with artificial complex sounds is more likely influenced by statistical correspondences grounded in multimodal experiential regularities, rather than by semantic or affective correspondences based on timbral impressions.

This study explores the relationship between line-drawing task performance and cognitive characteristics during a simulated right-turn situation, focusing on reaction times and errors. Right-turn accidents in Japan occur twice as often as left-turn accidents, primarily due to cognitive errors like insufficient safety checks. Efficient methods to assess driving-related cognitive traits are crucial for accident prevention. 36 participants performed a tablet-based line drawing task, drawing lines from a point cloud, and a reaction task simulating right turns, each repeated three times. Features of the drawn lines, such as “line stability” and “distance from the point cloud” were analyzed. Line stability negatively correlated with missed responses and reaction time variability, while greater distances positively correlated with false responses. Participants with stable lines exhibited fewer errors and consistent reactions. These results suggest that line drawing tasks can effectively assess driving-related cognitive characteristics, providing a useful tool for driver education and accident prevention.

Ecosystem services provided by community gardens include enhancing environmental sustainability and personal health and well-being, which makes the diffusion of community garden programs important within the broad context of urban planning. Since community garden is becoming popular as a form of outdoor leisure in Japan, this study tested that the leisure constraint model can be applied to community garden programs in Japan by using the results of the online questionnaire (n = 732) about constraints for program participation, showing 1 of 4 activity scenarios emphasizing different themes such as health, community building, learning, and ecology and environment. Confirmatory factor analysis indicated that the leisure constraint model can be applied to the community garden practices in Japan, while many factors extracted by the exploratory factor analyses had combinations of questions having vague and / or inconsistent meanings. Future research is needed for constructing the leisure constraint model specific to community garden programs.