Introduction

With the continuous advancement of urbanization and the continuous rise in housing prices in first-tier cities, young people are experiencing significant housing pressure. To meet this challenge, the co-living community model, which primarily involves young people as residents, has emerged as a new trend in urban living1,2. The “Implementation Plan for New Urbanization during the 14th Five-Year Plan Period”, released in 2021, along with the “Opinions on the Construction of Youth Development-Oriented Cities” issued in 2022, provide substantial policy support and encouragement for the development of youth co-living community models in China. To foster neighborly relationships and enhance the effective use of social resources3, such communities share the use of meeting, recreation, and reading spaces4, highlighting the significance of shared spaces, including the environment of the community’s public space and the associated furniture.

Despite the achievements of this model in resource sharing, addressing emotional health issues has become a pressing challenge, necessitating urgent attention due to the increasing social and psychological pressures on young people. In contemporary society, depression and anxiety disorders have become the most prevalent mental health issues globally5, particularly among the youth6. According to the National Mental Health Development Report of China (2021-2022), 24.1% of individuals aged 18 to 24 exhibit a risk of depression, while the overall detection rate of anxiety in the adult population stands at 15.8%7. These statistics underscore the unmet needs of the young people for effective emotion regulation and psychological relief when facing pressures from work, study, and social interactions. Emotion regulation, defined as an individual’s ability to assess and manage their emotional experiences and expressions, is crucial for mental health and social adjustment8. Positive emotion regulation helps individuals in coping with stress, enhances socialization, and positively impacts various life domains. Conversely, negative emotional distress can lead to challenges in emotional health, social functioning, and overall quality of life9. In addition to medication and cognitive behavioral therapy (CBT), favorable environmental factors can also contribute to emotion regulation10. In this context, the study of emotional healing design for public seating within youth co-living communities is particularly significant. Emotional healing design is an emotion-centered concept that emphasizes the stimulation of positive emotions through environments, products, and interactions, focusing on emotional resonance and long-term mental health11. Compared with private furniture, public seating has greater potential to improve the emotional state of community residents due to its high frequency of use and social attributes. Therefore, it has become an urgent issue to explore how to assist users in making positive emotion regulation, forming long-term emotional healing experiences, and enhancing mental health through the design of community public seating.

However, existing public seating designs often prioritize functionality and durability, while neglecting considerations related to social interaction and the promotion of emotional well-being12. For instance, research on auditorium chairs has highlighted that public furniture development lags behind that of home and office furniture, with designs overly reliant on past experiences and lacking scientific guidance and psycho-functional considerations13. Additionally, a study conducted by Hu et al.14 revealed that 63% of elderly participants perceived the colors of public seating as dull and negative, while 64% felt that the materials used in existing public seating lacked comfort and failed to incorporate elements that facilitate emotional adjustment. Furthermore, these designs do not integrate multi-sensory features that are beneficial for emotion regulation. Regarding strategies for enhancing emotional healing effects, environmental psychology has demonstrated that environmental factors such as color, material, and spatial form can significantly influence human emotions and mental health. The Attention Restoration Theory (ART) proposed by Kaplan, R. and Kaplan, S. (1989)15 represents a milestone in environmental psychology. Based on the premise of limited cognitive resources, the theory posits that natural environments can help restore directed attention fatigue, a state of cognitive depletion caused by prolonged focus. Subsequent research has extended ART across multiple domains. For instance, DeMello (2016)16 examined how nature-based healing environments can improve both the mental and physical health of the elderly. Similarly, Pasini et al. (2021)17 discovered that the integration of natural materials, greenery, and natural light significantly enhances the restorative qualities of workplaces. In parallel, Ulrich’s (1983)18 Stress Reduction Theory (SRT) posits that natural elements in the environment, such as natural elements, natural light, open spatial layouts, soft materials, and soothing color schemes, serve as restorative cues that can reduce anxiety and promote emotional balance. Over time, this theory has expanded beyond natural settings to influence indoor environmental design, supporting the conclusion that well-matched environmental design elements foster psychological comfort and emotional equilibrium, whereas inappropriate configurations may intensify stress and anxiety. Additionally, Rodriguez’s research indicates that engaging multiple senses (e.g., visual, auditory, olfactory, and tactile) can significantly influence emotional experiences19. This paper posits that incorporating multi-sensory elements, such as visual color coordination, tactile material selection, and even auditory soundscapes, into public seating can not only enhance community residents’ willingness to use public spaces but also foster a more relaxing and therapeutic environment. The interaction between public seating and bodily senses plays a crucial role in evoking positive emotional responses and facilitating emotion regulation.

Multi-sensory design typically begins with the human senses, including vision, hearing, smell, touch, and taste stimulating sensory functions at multiple levels. This approach enables participants to express and interact with their external environment in a comprehensive manner20. The essence of multi-sensory design lies in facilitating emotional release and resonance through the environment and objects, allowing individuals to internalize their capacity for emotion regulation21. In his book “Emotional Design,” Norman categorizes emotional design into three levels: visceral, behavioral, and reflective, which mentions the significance of multi-sensory design and highlights the pivotal role of multi-sensory experiences in enhancing users’ emotional experiences, which carries profound implications for the field of design22. Presently, researchers and scholars across various countries are investigating how multi-sensory design can aid in emotion regulation, particularly in healing spaces and healing products. These healing contents focus on converting negative energy emotions into positive energy emotions, so as to obtain autonomous experiences of emotional pleasure and psychological relief, and are progressively being implemented in education23, rehabilitation24, and various fields of society. Zheng (2022)25 noted that the visual system constructed within a multi-modal healing space is more effective than purely visual or auditory healing methods. In his book “Design in Design,” Japanese designer Kenya Hara discusses the “awakening of the five senses” and illustrates how multi-sensory design can emotionally regulate patients, exemplified by the visual guide system at Umeda Hospital26. Many scholars have also begun to examine the influence of healing product design on mental health, exploring various methods for achieving emotion regulation27. Through the manipulation of color28, form29, and material30, healing products can induce multi-sensory stimulation, providing emotional relief through visual, auditory, and tactile experiences. Tu et al. (2019)31 proposed design principles for desktop healing products, demonstrating that stimulating the visual, auditory, and tactile senses can offer psychological comfort and release for individuals with diverse lifestyles. Furthermore, Wu et al. (2022)32 employed the FCM model to cluster healing goods and subsequently proposed a multi-sensory design strategy aimed at developing and evaluating healing products for anxious users during the COVID-19.

In recent years, scholars have increasingly explored the healing properties of furniture. Furniture products that offer a multi-sensory experience are not only aesthetically pleasing and functional, but they also assist individuals in alleviating stress, relaxing, and achieving emotion regulation. Chen et al. (2016)33 demonstrated that a positive olfactory experience fosters the development of positive emotions and feelings. Yao et al. (2021)34 proposed design elements and principles for children’s furniture based on tactile perception, focusing on four essential elements: decoration, form, material, and function. This approach is conducive to enhancing preschool children’s cognitive abilities and promoting their physical and mental well-being. Additionally, Hao (2024)35 found that soft furnishings designed with an emphasis on tactile and visual sensations can have a calming effect on individuals with autism, thereby facilitating psychological healing. Categorizing the multi-sensory emotional healing characteristics of existing healing furniture can deepen the understanding of multi-sensory healing furniture design. However, while some studies have established a connection between design and emotion regulation, most remain confined to specific contexts, such as hospitals36, children’s centers37, and commercial spaces38. There is a notable lack of methodologies addressing the design and evaluation of public seating in community spaces aimed at emotional healing, particularly concerning the youth group39. Therefore, this study focuses on exploring the emotional healing design of public seating in youth communities from a multi-sensory perspective, with particular attention to whether it fosters interaction and communication within the community, and to evaluating the effectiveness of healing community seating in supporting residents’ emotion regulation.

Against this background, the present study focuses on public seating in youth communities, exploring the core question of “Which sensory design elements contribute to enhancing emotion regulation and social facilitation in public seating for youth communities?” This study aims to explore how the integration of multi-sensory design elements can construct a novel healing scenario that enhances emotional well-being. The research will be conducted in three phases: “research preparation”, “analysis and outputs”, and “assessment and validation”. Firstly, the DASS questionnaire is employed to screen key participants, followed by the compilation of a sample database for the healing seating. Subsequently, the FCM method is utilized to cluster the multi-sensory emotional healing characteristics of 50 healing seating, thereby obtaining representative samples. User interviews are conducted to establish essential design elements, which results in the following eight community public seating design intentions. This study also innovatively incorporates generative artificial intelligence technology to facilitate the visualization of design options. Ultimately, a three-dimensional scale for assessing the emotional perception of public seating is constructed, which derives feedback and ranks emotion regulation efficacy. The effectiveness of the multi-sensory design strategy in regulating users’ emotions was verified, aiming to provide contemporary young people with a novel healing scenario and offering research insights into enhancing the healing and interactivity of public seating in youth communities.

Methods

Research procedure

This study is a cross-sectional investigation aimed at examining the emotional impact of healing community seating on users. The proposed research framework consists of three main phases: research preparation, analysis and outputs, and assessment and validation. The specific implementation steps are outlined as follows, and the overall research procedure is illustrated in Fig. 1.

Fig. 1
figure 1

Research procedure. Image generated using Adobe Illustrator 2020 (https://www.adobe.com/products/illustrator.html).

Full size image

Participants

The study targeted urban youth as the primary user group, including individuals such as recent graduates entering the workforce, freelancers, and students, who were experiencing uncertainty in decision-making. These individuals are more prone to experiencing anxiety, stress, and depressive symptoms, making them representative users with a strong need for emotion regulation. In-depth perceptual feedback from such purposefully selected users can provide higher-quality evidence for therapeutic design. Participants completed the Depression Anxiety Stress Scales (DASS-21) questionnaire via an online platform (see Appendix A). A total of 67 responses were collected. Based on the questionnaire scores and participant profiles, 10 individuals (five males and five females) were selected as key participants to assist in identifying and evaluating healing design elements. The basic information of the key participants is presented in Table 1, and their emotional states are shown in Table 2. To qualify for this study, participants had to meet the following criteria: (1) be between the ages of 18 and 32, (2) have experience living in a shared community, and (3) have at least one score categorized as moderate or severe on the DASS questionnaire. All participants provided informed consent before the experiment, and approval was obtained from the Ethics Review Committee of East China University of Science and Technology.

In the “Assessment and validation” phase, an additional 15 experts were invited to join the original group of 10 key participants, increasing the total number of evaluators to 25. The expert group comprised 10 furniture designers and 5 psychologists, who collaboratively assessed the seating design options. A statistical power analysis conducted using G*Power revealed that, based on an estimated effect size from previous studies (Cohen’s d = 0.5), the sample size (N = 25) would achieve a statistical power of 80% at a significance level of \(\alpha\) = 0.05.

Table 1 The basic information of key participants.
Full size table
Table 2 The emotional state of key participants.
Full size table

Research preparation

DASS-21 questionnaire screening

The Depression, Anxiety, and Stress Scale (DASS-21) questionnaire40, originally developed by Lovibond and colleagues in 1995, comprises three self-report scales designed to assess the emotional states of depression, anxiety, and stress experienced over the past week. This instrument aids in determining whether a participant is experiencing emotional distress41,42. Each of the three groups in the scale comprises seven questions scored on a four-point Likert scale, with item scores ranging from 0 (“does not meet”) to 3 (“always meets”). The scores for each subscale were multiplied by two to yield a score for that subscale, with higher scores indicating a greater presence of the corresponding emotion. The intensity of the DASS was categorized into five levels, ranging from normal to very severe43. In this study, the DASS-21 questionnaire was used to identify participants experiencing moderate or severe emotional distress during the preparation phase of the design. The scale demonstrated a Cronbach’ \(\alpha\) of 0.891, a KMO value of 0.925, and the Cronbach’s \(\alpha\) for the subscales of depression, anxiety, and stress were 0.774, 0.743, and 0.752, respectively, indicating a strong level of reliability40.

Development of a healing seating sample database

After extensive internet research and expert screening, 50 samples of healing seating products were selected from a multi-sensory perspective. These samples encompass a variety of colors, forms, materials, and functions, featuring sensory design characteristics, and their applications extend to, but are not limited to, public seating. Basic information for each sample was summarized, and an information sheet was created that includes the name, image, description, and material composition of each healing seating, which will serve as a design reference for subsequent clustering. (Refer to Table 3 for details on the healing seating samples and Appendix B for the complete table.)

Table 3 Sample Information Sheet for Healing Seating (partial).
Full size table

Analyses and outputs

Fuzzy C-means clustering

This study proposes seven basic characteristics based on human senses from five perspectives: form, color, material, interaction, and flavor. These characteristics include: the sense of healing of the form seen, the sense of healing of the color seen, the sense of healing of the material seen, the sense of healing of the material heard, the sense of healing of the material touched, the healing of the interaction touched, and the healing of the material smelled, as illustrated in Fig. 2. Ten key Participants used these characteristics to evaluate the emotional impact of 50 healing seating samples.

Fig. 2
figure 2

Basic characteristics describing a multi-sensory healing seating sample. Image generated using Flourish (https://flourish.studio/).

Full size image

During the Visual Presentation Standardization processing stage, several controls were implemented in the experimental presentation to ensure a consistent experimental environment and minimize the influence of extraneous variables:

  1. 1.

    Standardized grayscale background processing: The experiment primarily focused on 50 seating samples with emotional healing efficacy. All images were processed through standardization, with the background converted to a standardized grayscale to highlight the furniture and avoid any influence of background color on participants’ color perception.

  2. 2.

    Screen Brightness Control: All participants used display devices of the same specification (resolution 1920\(\times\)1080, brightness fixed at 120 cd/m²) to minimize potential color and lighting deviations caused by different screen settings.

  3. 3.

    Fabric Matching: To enhance the evaluation of the tactile dimension, fabric samples corresponding to the furniture materials were provided during the experiment. Participants simultaneously touched the actual fabric while viewing the images, thereby improving sensory consistency.

The following steps and control measures were implemented during the participant scoring process:

  1. 1.

    Pre-test and Screen Adaptation: Participants viewed a color calibration chart and adapted to the screen brightness and color temperature to ensure consistent visual conditions.

  2. 2.

    Presentation of Seating Samples: The fifth seating samples were presented, with each sample displayed for 120 seconds. During this period, participants read the basic information and viewed the images of the seating samples while simultaneously touching the corresponding fabric samples and listening to the sounds emitted by the materials, thereby enhancing the multi-sensory perceptual dimension.

  3. 3.

    Scoring Phase: Participants rated the multi-sensory emotional healing characteristics of each piece of seating based on seven basic characteristics. The evaluation time for each sample was 60 seconds. The scores were made on a continuous scale ranging from 0 to 1, with precision to five decimal places, to quantify the emotional healing degree of each characteristic.

Each presentation and scoring of a seating sample constituted one trial, and this process was repeated 50 times to complete the rating of all samples. To control for order effects, the sequence of seating samples presented to each participant was randomized, ensuring the validity and reliability of the study. After the experiment, the internal consistency of the participants’ ratings was tested, with a Cronbach’s alpha coefficient of 0.961, indicating reliable ratings.

After standardizing and normalizing the obtained scoring matrix, the experiment employed the Fuzzy C-Means (FCM) clustering analysis method to classify the multi-sensory emotional healing characteristics of the samples. FCM is a clustering algorithm that incorporates fuzzy theory and is categorized as an unsupervised classification method. It is commonly utilized to cluster similar information within the data44. The Fuzzy C-Means algorithm was developed by Bezdek in 1981, building upon the Hard C-means (HCM) algorithm45. This algorithm optimizes the objective function to determine the degree of affiliation of each sample point to all clustering centers, utilizing this degree of affiliation to express the probability that a sample belongs to a specific class. This process facilitates the automatic classification of samples, and the clustering results produced by FCM are notably more flexible46,47, making it particularly beneficial in fields such as computer science48, medicine49, and business economics50. FCM achieves clustering by minimizing the objective function J (1) and its associated constraints (2).

$$\begin{aligned} J = \sum _{i=1}^{c} \sum _{j=1}^{n} u_{ij}^m \left\| x_j - c_i \right\| ^2 \end{aligned}$$
(1)
$$\begin{aligned} \sum _{i=1}^{c} u_{ij} = 1, \quad j = 1, 2, \ldots , n \end{aligned}$$
(2)

In the formula, ‘c’ denotes the number of cluster centers, ‘n’ represents the number of samples, and ‘m’ is a fuzziness coefficient of membership, which can take values such as 2, 3, etc., indicating the degree of importance of a sample’s belonging to a certain class. cj denotes the j-th cluster center, uij is the degree of affiliation of the data point xi to the cluster cj, and xi is the eigenvector of the data point. According to equation (1), the objective function is defined by the product of the affiliation degree of the corresponding sample and the distance from the sample to the center of each class. Equation (2) stipulates that the sum of affiliations of a sample across all classes must equal 1. Consequently, the closer a sample is to the cluster centroid, the higher its affiliation, whereas greater distances result in lower affiliations. The FCM algorithm employs an iterative computation process that minimizes the objective function while continuously updating the affiliations (uij) and cluster centers (cj), as detailed in equations (3) and (4).

$$\begin{aligned} u_{ij} = \frac{1}{\sum _{k=1}^{c} \left( \frac{\left\| x_i - c_j \right\| }{\left\| x_i - c_k \right\| } \right) ^{\left( \frac{2}{m-1} \right) }} \end{aligned}$$
(3)
$$\begin{aligned} c_j = \frac{\sum _{i=1}^{N} u_{ij}^m x_i}{\sum _{i=1}^{N} u_{ij}^m} \end{aligned}$$
(4)

Grounded theory for interview

Research conducted semi-structured interviews with key participants, employing Grounded Theory to systematically code the interview data, aiming to identify participants’ multi-sensory emotional healing needs regarding public seating in community spaces51. The interview questions explored participants’ understanding of healing seating, the desired characteristics of such seating, and the methodological pathways they believed could facilitate a positive emotional experience, as detailed in Table 4. Grounded Theory extracts concepts and categories from raw data through three primary coding processes (open coding, axial coding, and selective coding)52, further explores the connections between initial categories, inductively derives main categories, and ultimately analyzes the relationships among these categories to construct a relevant theory grounded in actual data. In this study, interviews were categorized by themes, and keywords associated with emotion regulation were identified as design requirements. Additionally, frequency statistics of these keywords were calculated to provide robust support for subsequent decision-making, product enhancement, or research reporting.

Table 4 Semi-structured interview questions.
Full size table

Development of multi-sensory healing seating

Based on the representative samples derived from clustering and the sensory design elements extracted from the interviews, a multi-sensory healing public seating design solution was developed through the following steps:

  1. 1.

    Integration of design requirements: Keywords extracted from the interviews were synthesized into a hierarchical model of key design elements for healing public seating, with representative samples from cluster analysis serving as reference points to guide the design output.

  2. 2.

    Multi-sensory design generation: Based on design requirements and the weights in the hierarchical model, descriptive sentences were constructed (e.g., “using soft, light-colored materials combined with a semi-enclosed design”). Eight innovative design options were generated with the assistance of generative artificial intelligence tools, such as GPT-4 and Midjourney. Each design incorporates various sensory design elements and references a representative sample.

  3. 3.

    Design iteration and optimization: Following the generation of initial design options, the research team collaborated with design experts to conduct multiple iterations, optimizing multi-sensory design elements such as materials, forms, and colors to achieve a functional and aesthetic balance in each design option.

Assessment and validation

Design of the emotional perception assessment scale

The emotional perception assessment scale for public seating in youth communities was adapted from the Emotional Salience Questionnaire developed in Massimiliano’s previous research53. A total of 25 evaluators rated the eight design options based on three different dimensions: validity (positive/negative), arousal (high/low)54,55, and emotion/mood. The emotion-related dimensions were proposed by American psychologists Russell and Mehrabian in the 1970s. This model has become one of the key theoretical frameworks in the study of emotions and is widely employed to characterize and analyze emotional states and their variations. The scale utilized in this article employs a series of bipolar adjectives to measure individuals’ attitudes or perceptions regarding a specific topic, concept, or object. It operates on a nine-point scale, with scores for each item ranging from −4 (indicating “negative emotions”) to 4 (indicating “positive emotions”). This contrasting scale allows evaluators to express the intensity of their feelings or perceptions about the option being assessed. Six of the twelve adjectives from the Emotional Salience Questionnaire were selected for this evaluation questionnaire. These adjectives include “unpleasant-pleasant”, “unattractive-attractive”, and “boring-stimulating”. To evaluate the communicative and communal attributes of public seating in youth communities56, the questionnaire incorporated the adjective pair “non-communicative-communicative”. Additionally, to assess the affective responses elicited by the seating, adjectives describing psychological feelings in specific environments55,57 were added, specifically “anxious-relaxed” and “oppressive-tranquil”. The final questionnaire comprised the following items: valence dimensions (“unpleasant-pleasant”, “unattractive-attractive”); arousal dimensions (“boring-stimulating”, “non-communicative-communicative”); and emotion dimensions (“anxious-relaxed”, “oppressive-tranquil”). Detailed information regarding the questionnaire can be found in Appendix C.

To ensure a consistent experimental environment and minimize the influence of extraneous variables, this study employed the same Visual Presentation Standardization treatment as utilized in the FCM clustering stage. The reliability of the questionnaire results was validated using the Cronbach’s \(\alpha\) coefficient, which was found to be 0.929, thus confirming the reliability of the questionnaire results.

Statistical analysis

Data were analyzed using IBM SPSS Statistics 29.0 for Windows. To systematically examine participants’ emotional perception ratings of different design options, a range of statistical methods were employed. Initially, descriptive statistical analysis was conducted for each design option, encompassing measures such as mean, standard deviation (SD), variance, kurtosis, skewness, and coefficient of variation (CV) to elucidate the overall distribution characteristics of the data. Prior to performing parametric tests, the normality of the data was evaluated using the Shapiro–Wilk test, and statistical comparisons across groups were conducted using analysis of variance (ANOVA) alongside effect size indicators.

Ethics approval

All methods were performed in accordance with the Declaration of Helsinki. All inform consent and assent forms as well as all study procedures were approved by the Ethics Review Committee of East China University of Science and Technology (Approval number: ECUST-2024-r011). All participants were informed about the study and voluntarily consented to participate. Ethical norms were strictly adhered to throughout the research, ensuring the protection of participants’ privacy and data security. Furthermore, all data collected were utilized solely for research purposes.

Results

Cluster analysis of healing seating samples

Through FCM clustering analysis, this study categorized 50 multi-sensory healing seating samples into three clusters, each represented by a distinct color. The × points represent the centroids of each cluster, indicating the weighted average position of all data points within the cluster, as shown in Fig. 3. The relevant parameters of the clustering process were as follows: number of iterations = 100, threshold = 0.00001, fuzziness parameter m = 2, and average silhouette coefficient = 0.5044, indicating a reasonable clustering effect. Each cluster represents distinct design features and is categorized as follows: Cluster 1, the ‘Multi-Sensory Integrated Design Type’ (blue dots), exhibits balanced and effective emotional healing effects from various sensory perspectives, such as vision and touch, in terms of style, color, and materials. Cluster 2, the ‘Color and Form Design Dominant Type’ (yellow dots), is dominated by color and form design, featuring products that soothe users’ moods through their interesting shapes and color schemes. Cluster 3, the ‘Tactile and Interactive Design Dominant Type’ (red dots), is dominated by tactile and interactive design and includes products that alleviate users’ anxiety through unique interactive tactile sensations. This study selected the two samples closest to the center of each cluster as representative samples, along with the two highest-scoring samples, totaling eight representative samples. The information regarding these samples is shown in Table 5.

Fig. 3
figure 3

Results of cluster analysis of healing seating. Image generated using MATLAB R2023a (https://ww2.mathworks.cn/products/matlab.html).

Full size image
Table 5 Representative Sample Information Sheet.
Full size table

Construction of design elements for healing public seating

During the interviews, participants reported that the combination of multi-sensory design, particularly at the visual and tactile levels, significantly alleviate their stress and anxiety. Several participants noted that interactive elements (e.g., N35) and enclosing features (e.g., N47) in the design fostered a sense of psychological safety and helped them regulate their emotions more effectively. Below are some of the participants’ feedback:

One participant stated, “I believe the sofa that heals me is very soft. I appreciate the beige color, which appears gentle and inviting. It can be wrapped around me and is suitable for sitting and relaxing, as well as for sleeping. This design makes me feel more relaxed and secure.”

Another participant stated, “For me, the most important aspect is the healing sensation derived from the interaction. I hope that this sofa will provide my body with comfortable support and, at the same time, foster a sense of fun and mischief when I use this furniture with my friends. I envision this piece of furniture as a generator of emotional resonance between myself and my friends.”

Keywords such as visually soothing colors (e.g., beige), soft-textured materials (e.g., fabrics), and wraparound form designs frequently appeared in these interviews, indicating strong participant recognition of the emotion regulation effects of these sensory design elements. Through open coding and axial coding, this study identified four primary design elements: the soothing nature of color, the softness of touch, and the enveloping nature of form, and the furniture-user interaction. The frequency of keywords mentioned within these categories was analyzed to construct into a hierarchical model of key design elements for healing public seating in youth communities, and assigned weights according to word frequency, as shown in Table 6.

Table 6 Keyword list of design elements.
Full size table

Multi-sensory healing seating design generation

Combining the clustered representative samples and the hierarchical model of key design elements for healing public seating, eight design options are generated. Each option references one representative sample along with a random selection of design indicators, as illustrated in Fig. 4. These design options will serve as the stimulus materials for the subsequent emotional perception assessment experiment. The primary design element of each option is determined by the emotion regulation design element that has the highest proportion of applied design keywords in that option. The allocation of design options to each primary design element is based on the weights presented in Table 6.

Fig. 4
figure 4

Design generation. Image generated using Chat-GPT4 and Adobe Illustrator (https://chatgpt.com/).

Full size image

Assessment of emotional perception

Validity analysis

Ten key participants and fifteen experts evaluated the validity of the eight design options based on valence, arousal, and emotion. To ensure the reliability of the assessment tools and findings, a validity analysis was performed, indicating that all study items had commonality values exceeding 0.4. This indicates that the information from the study items can be effectively extracted. Furthermore, the KMO value was found to be 0.793 (>0.6), suggesting that the data can be effectively analyzed. Additionally, the variance explained for the factors was 61.427%, with a cumulative variance explained after rotation also exceeding 50%. This demonstrates that the information from the research items can be effectively extracted, as illustrated in Table 7.

Table 7 Validity analysis.
Full size table

Descriptive statistical analysis

In this study, a comprehensive descriptive statistical analysis was performed on the eight design options (P1-P8), focusing on key indicators such as the mean, standard deviation(SD), variance, 95% confidence intervals, interquartile range (IQR), kurtosis, skewness, and coefficient of variation (CV), as presented in Table 8.

Based on the mean ± SD, a key metric for evaluators’ emotional perception of the design options, the eight options were ranked in descending order: P3 > P4 > P8 > P2 > P7 > P5 > P1 > P6. Overall, all eight healing public seating options for youth communities received positive mean scores.

The top-ranked P3 option exhibited a mean score of 2.453, a standard deviation of 1.031, and a CV of 42.039%, demonstrating its superior efficacy in emotion regulation. The second-ranked P4 option recorded a mean score of 2.327, with a standard deviation of 0.917 and a CV of 39.424%. Both P3 and P4 displayed the lowest CV, indicating a high stability in program scores, which suggests that evaluators largely agreed on the positive effects of these programs on emotion regulation. P8 option ranked third, with kurtosis (−0.871) and skewness (−0.271) showing a symmetrical distribution of scores without significant skew. The 95% confidence interval for P8, [1.372, 2.175], suggests that score fluctuations remain within a manageable range. In contrast, the 95% confidence interval for P2, [1.197, 2.230], is relatively wide, indicating greater variability in scores and suggesting that evaluators had differing opinions regarding the program’s effectiveness. The scores for P5 and P7 were more dispersed and left-skewed, implying that evaluators may have been predisposed to assign lower scores, leading to an uneven distribution. Conversely, P1 and P6 exhibited the least stability, with P1 having the highest SD (1.527) and P6 showing the highest CV (204.299%), and both programs presented wide 95% confidence intervals, indicating significant bias in evaluators’ perceptions of emotion regulation for P1 and P6, with more negative emotional perceptions obtained overall.

Table 8 Descriptive statistical analysis.
Full size table

Repeated measures ANOVA

Repeated measures ANOVA was performed on six pairs of bipolar adjectives in three emotional dimensions. For the dimension “unpleasant–pleasant,” the Mauchly’s test of sphericity yielded a p-value greater than 0.05, indicating that the assumption of sphericity was not violated. Therefore, the results under the “Sphericity Assumed” row were interpreted. For all other dimensions, the assumption of sphericity was violated (Mauchly’s test p < 0.05), with sphericity W < 0.75. As a result, the results presented in the row labeled as Greenhouse-Geisser will be interpreted. It can be seen that for tests using Greenhouse-Geisser correction, SPSS adjusted the degrees of freedom by multiplying the corresponding epsilon value with the degrees of freedom for the sphericity assumed condition.

The results indicated that each emotional dimension exhibited a significant main effect across the design schemes (p < 0.05), suggesting substantial variability in emotional perception ratings among the seating designs. Following the application of Greenhouse–Geisser corrections, the partial eta-squared (\(\eta ^2\)) values revealed large effect sizes across all dimensions (see Table 9). Notably, the dimension “unattractive–attractive” displayed the highest effect size (partial \(\eta ^2\) = 0.289) and the largest F-value (F = 9.74), indicating that perceived attractiveness was the most discriminative factor and exerted the greatest influence on participants’ satisfaction with public seating designs. Furthermore, the dimensions “Anxious–Relaxed” (partial \(\eta ^2\) = 0.275), “Non-communicative–Communicative” (partial \(\eta ^2\) = 0.246), and “Oppressive–Tranquil” (partial \(\eta ^2\) = 0.210) also demonstrated substantial effects on emotional regulation. Other variables, such as “Unpleasant–Pleasant” (partial \(\eta ^2\) = 0.142) and “Boring–Stimulating” (partial \(\eta ^2\) = 0.160), were found to exert moderate influence and may be considered secondary design references.

These findings imply that future design strategies for public seating should prioritize elements with demonstrable healing potential, particularly those that enhance visual appeal, promote relaxation, encourage social interaction, and foster a sense of tranquility, to more effectively support users’ emotion regulation in co-living communities.

Table 9 Repeated Measures ANOVA Results for Six Bipolar Adjective Pairs.
Full size table

Figure 5 illustrates the mean scores of each design scheme across the six emotional adjective pairs, accompanied by the results of significance groupings. The post hoc Bonferroni test indicated that the differences in user responses among the design schemes were statistically significant across multiple emotional dimensions. In most dimensions, the P6 design received significantly lower scores compared to designs such as P3 and P4 (p < 0.05), with P3 demonstrating the most favorable emotional response overall. Distinct letters (e.g., a, b, c) are employed to denote statistically significant differences between groups.

Fig. 5
figure 5

Mean Ratings and Significance Groups for Each Dimension. Image generated using PyCharm (https://pycharm.710down.com/).

Full size image

Discussion and limitations

Discussion of key findings and contribution

This study focuses on public seating in youth communities and examines the potential and effectiveness of multi-sensory design in supporting emotion regulation. This study integrated FCM clustering, grounded theory, AI-assisted design generation, and empirical evaluation to effectively align existing healing furniture design experiences with the specific needs of users for community public seating. In this process, a hierarchical model of key design elements for healing public seating in youth communities was constructed, along with an emotional perception assessment framework. These contributions enhanced the operability and repeatability of multi-sensory design.

Based on the evaluation results, all eight design options generated using reference samples and key design indicators, elicited positive emotional responses. Among the multi-sensory features tested, soothing colors (e.g., beige and macaron tones), soft textures (e.g., cushioned fabrics), semi-enclosed, semi-partitioned forms, and modular structures that facilitate interaction were identified as the most effective in promoting emotional relaxation and social engagement. The P3 and P4 designs, which integrated these key elements, demonstrated the strongest overall performance. In particular, the P3 design option achieved the highest mean score (2.453), with a standard deviation of 1.031, a coefficient of variation of 42.039%, and a 95% confidence interval of [2.049, 2.858]. These elements enhance feelings of relaxation and safety among young users, thereby improving emotional stability and social engagement in shared spaces, highlighting the advantages of a multi-sensory design approach. In the context of public seating for youth communities, these design elements contribute to spatial identification and positive emotional feedback, addressing the limitations highlighted by Gifford, who criticized conventional public furniture for prioritizing functionality and durability at the expense of psychological and emotional needs58. This is consistent with the findings of Küller et al. (2009)59 regarding the impact of colored rooms on people’s emotions. Moreover, modular and interactive furniture designs significantly enhance spatial shareability and willingness to engage in social interactions, offering youth users spaces that balance privacy with social potential. This supports Kaplan and Kaplan’s (1989)15 Environmental Preference Theory, particularly the dimensions of legibility and mystery.

Statistical analysis further confirmed this trend. Repeated measures ANOVA indicated that “attractiveness” (\(\eta ^2\) = 0.289), “relaxation” (\(\eta ^2\) = 0.275), and “communication facilitation” (\(\eta ^2\) = 0.246) were the most influential emotional dimensions, suggesting that public seating design should prioritize visual appeal, emotional comfort, and social engagement. User interviews supported these findings, with most participants identifying light color palettes, soft materials, semi-enclosed forms, and playful interaction as key features enhancing both attractiveness and relaxation. The P3 design option referenced 12 design indicator elements, which was the highest number of reference indicators among the eight schemes, followed by P4 with 11 indicators. These results validated the effectiveness and applicability of the proposed sensory design criteria for developing emotionally supportive public furniture in future community settings.

At the sensory level, the integration of multiple modalities, such as visual, tactile, and interactive elements, was found to effectively elicit positive emotional responses, whereas overly formal or single-sense designs were less effective or even counterproductive. For instance, although P6 offered some enclosure, its use of dark, rigid materials and a confined structure resulted in the lowest scores on the “anxious–relaxed” and “oppressive–tranquil” scales (M = 0.673). Designs incorporating natural cues (e.g., music, organic materials, greenery) received some positive feedback, aligning with Ulrich’s (1983) Stress Reduction Theory18, as previously applied in offices, hospitals, and restaurants (e.g., Kathryn, 201560; Urquiola’s Openest, 202361; Eftekhari, 202362). However, participants in this study perceived such elements as less impactful than tactile and visual features. This difference may stem from the fact that these effects are mostly environmental factors rather than furniture, and based the limitations of image-based evaluation. Future research should explore these effects in real-world, immersive seating.

The emotional evaluation framework proposed in this study is grounded in Massimiliano’s emotional salience questionnaire53 and the PAD emotional dimensional model proposed by Mehrabian and Russell54. It includes six pairs of bipolar adjectives covering three dimensions: valence (positive/negative), arousal (high/low), and emotion/mood. This framework addresses the need for capturing authentic user feedback and emotional responses in experience evaluation (Bartholomew et al., 201663; Følstad A., 201764). Preliminary validation in this study indicates acceptable reliability and validity, offering a practical tool for assessing user-oriented multisensory furniture design.

In summary, this study demonstrates the effectiveness of a healing design approach informed by clustering analysis and user needs, while clarifying the relationship between sensory perception and emotional outcomes in multi-sensory design. It extends the theoretical scope of emotional healing design beyond clinical settings by establishing perceptual-emotional correspondence mechanisms. Compared to prior research focused on clinical settings (e.g., hospitals65,66) or private spaces (e.g., homes67 and workspaces), this study addresses a gap in emotional support within community public furniture. It contributes to the humanization and emotionalization of shared spaces and fosters interdisciplinary integration across design, urban planning, and environmental psychology.

Limitations

This study has several limitations. First, the sample consisted of only ten young individuals with high emotion regulation needs, limiting generalizability. While the target group was clearly defined, future research should include more diverse participants across age, gender, and cultural backgrounds to evaluate the broader applicability of the findings. Second, due to cost and spatial constraints, participant evaluations were based on screen-displayed images of the seating rather than physical interaction, which may have affected the accuracy of emotional responses. Additionally, auditory and olfactory elements received relatively low emotional feedback, suggesting that current furniture designs may lack effectiveness in these modalities. Future studies should explore multi-sensory interactions and their long-term effects on emotion regulation in physical community environments. Lastly, the emotion assessment framework developed in this study remains exploratory. A full validation process with larger samples is needed to establish its psychometric reliability and structural validity.

Conclusion

This study examines seating design in youth shared community spaces and evaluates the role of multi-sensory strategies in supporting emotion regulation and social interaction. By developing a three-dimensional emotional perception assessment framework (valence, arousal, emotion/mood), the study identifies key sensory elements, such as soothing colors (e.g., macaron tones), soft textures (e.g., cushioned fabrics), and semi-enclosed, semi-partitioned forms, that promote relaxation, safety, and social engagement. Modular and interactive configurations further enhance social initiative and spatial inclusivity. Compared to existing literature, this study not only validates the link between emotional perception and spatial seating design but also innovatively transforms multisensory design into assessable and implementable design strategies in the specific context of “youth co-living communities,” filling the gap in research on emotion-supportive public furniture in non-clinical environments. In conclusion, this research offers actionable tools for designers, community managers, and mental health professionals seeking to integrate emotional well-being into youth co-living communities.