Explaining Aesthetic Judgement in Closed-Category Product Design through the Unified Model of Aesthetics and the Categorical-Motivation Model: A Laptop Product Study

2.1 Aesthetic preferences

The study of aesthetic preferences has deep roots in psychology, dating back to ancient Greek philosophy, especially the works of Plato and Aristotle (Phillips et al., 2011; Whitfield & de Destefani, 2011). Over time, the field has evolved from a deductive philosophical approach to Fechner’s pioneering work in experimental aesthetics in 1876, marking a shift from studying aesthetic objects to a scientific methodology (Fechner, 1876). Early aesthetic research primarily centered on “high-end” art fields like painting and sculpture before gradually extending to “low-end” aesthetics like everyday product design (Suhaimi et al., 2023). Aesthetic preferences refer to how individuals assess the visual appeal of a product, often shaped by their past experiences and familiarity with design elements (Ding et al., 2025). Recent studies highlight the significant influence of aesthetics on consumer perception and decision-making. Aesthetic preferences reflect how people judge a product’s appearance based on personal experience and cultural background (Ma et al., 2025). This evaluation not only increases product appeal but also enhances user satisfaction and emotional engagement (Desmet & Hekkert, 2007). According to modern interactionism, aesthetic pleasure emerges from the dynamic interaction between individuals and objects, creating a sense of enjoyment and positive emotions (Blijlevens et al., 2014). Therefore, product design should go beyond functionality to align with consumers’ lifestyles and aesthetic expectations.

Contemporary research on aesthetic preferences focuses on two influential perspectives, Whitfield’s prototype preference theory and Paul Hekkert’s UMA model. Whitfield’s “Preference for Prototype” theory (1979) posits that people show a cognitive bias toward objects that are closely related to the prototypical form of a category. Essentially, products that resemble “classic” examples of their type (e.g., a chair with all the expected features of a “chair”) are perceived as more aesthetically pleasing, primarily due to familiarity and processing fluency (Reber et al., 2004). However, while the theory emphasizes the appeal of familiarity, it initially seemed to contradict Berlyne’s arousal theory (Berlyne & Boudewijns, 1971), which emphasizes novelty and complexity as central to aesthetic pleasure. Whitfield’s subsequent research (CM Model), which recognized that both familiarity and novelty can induce pleasure, sparked further exploration of how these opposing forces can be reconciled (Whitfield, 2000).

The UMA balance the seemingly contradictory emphasis on typicality (from Whitfield) and novelty (from Berlyne). Under UMA, too much familiarity can become boring, while too much novelty can be disorienting, so the most pleasing designs balance these two tendencies (Hekkert, 2014). This perspective not only incorporates Whitfield’s emphasis on prototypes into a larger theoretical construct, but also highlights the interplay between safety and accomplishment, two evolutionary impulses that shape how we view and appreciate designed objects (Ding et al., 2025).

In conclusion, the UMA model provides a more comprehensive and basic theoretical framework for the study of aesthetic preferences. When we process this information fluently, our positive response to aesthetics will also be stronger (Reber et al., 2004). This processing process of aesthetic pleasure experience can be regarded as a kind of happiness evolution drive.

2.2 Categorical-Motivation (CM) model

The CM model, proposed by Whitfield (2000), offers an early framework for explaining how typicality and novelty jointly shape aesthetic judgments, particularly at the cognitive level. It emerged in response to tensions between Berlyne’s collative-motivation theory which emphasized novelty, complexity, and ambiguity as sources of arousal (Berlyne, 1966), and Rosch’s (1978) prototype theory, which highlighted user preference for typical and cognitively fluent stimuli. Prior research in art and design domains supported the preference for prototypical, easily categorized forms, forming the empirical basis of the CM model (Herzog et al., 1976; O’Hare, 1976; Whitfield & Slatter, 1979; Wohlwill, 1976).

To reconcile the preference for both novelty and typicality, Whitfield (2000) proposed the CM model, which introduces three key constructs: categorical salience, motivational arousal, and social significance. In this model, Whitfield distinguishes between two types of feature salience: diagnostic and intensive. Diagnostic features help define an object’s membership within a known category and are closely tied to typicality. They represent category-valid cues and therefore foster recognition and safety. In contrast, intensive features are perceptually striking, arousing curiosity and attention. They are associated with novelty and accomplishment-related motives (Whitfield, 2009). In the CM model, the variables of typicality and novelty are presented as opposing cognitive forces: one representing fluency and safety, the other representing exploration and reward. The model assumes that aesthetic preferences depend on the category structure of the product. In closed categories, those with narrow definitions and low tolerance for deviation (e.g., teacups, pianos)—users generally favor typical, recognizable forms. In contrast, open categories (e.g., chairs, clothing) allow more variation, and users may seek novelty or expressive individuality (Whitfield, 2000, 2009; Tyagi et al., 2013).

Crucially, Whitfield also introduced the notion of social significance as a determinant of aesthetic judgement. Some stimuli carry symbolic or cultural value (e.g., luxury branding), influencing preferences beyond cognitive efficiency or sensory arousal. Although social significance is acknowledged in the CM model, it remains conceptually underdeveloped. While the CM model and associated studies (e.g., Tyagi et al., 2013; Suhaimi et al., 2023) have shown that typicality often dominates aesthetic judgments in closed categories, little is known about how users respond to perceptual coherence or social symbolism within the same categorical structure. For instance, do users tolerate perceptual variety in closed-category products, provided that cognitive typicality is preserved? Does the expectation of group conformity reduce the appeal of autonomy, while enhancing the desire for connectedness? These questions remain empirically underexplored, particularly for technological products such as laptops, which are both functionally constrained and aesthetically competitive.

2.3 Unified model of aesthetics

The UMA integrates research from various fields, including cognitive psychology, social psychology, design, sociology, cognitive neuroscience, philosophy, and art (Hekkert, 2014). This model provides a comprehensive framework for understanding product aesthetics by examining three key levels: perceptual, cognitive, and social. Perception forms the initial impression of a product, which is then processed and refined through cognitive and social evaluations based on personal experiences and environmental influences (Ding et al., 2025). Aesthetic preferences arised from a balance between two fundamental human drives: the need for safety and the desire for accomplishment. Safety-oriented factors include typicality, unity, and connectedness, while accomplishment-driven elements involve novelty, variety, and autonomy (Blijlevens & Hekkert, 2015; Hekkert et al., 2003; Post et al., 2013). UMA emphasized that aesthetic appeal is shaped by the dynamic interplay between these opposing forces at the perceptual, cognitive, and social levels, as well as explaining why people are drawn to certain designs (Suhaimi et al., 2023). This model has been widely used to analyze consumer reactions to product design and has been applied across multiple industries (Ding et al., 2025; Ma et al., 2025; Loos et al., 2022; Post et al., 2023; Suhaimi et al., 2023). The following sections will explore each of these three levels in detail, discussing their specific roles in shaping aesthetic preferences.

At the perceptual level of the UMA model, unity refers to visual consistency, order, and harmony, which help create a sense of familiarity and ease in perception. Wagemans et al. (2012) discovered that Gestalt principles, such as symmetry, repetition, continuity, and closure, allow the human brain to group visual elements into a structured whole, facilitating cognitive processing and pattern recognition. Similarly, Berghman and Hekkert (2017) found that designs with strong structural coherence are perceived as more aesthetically pleasing because they reduce cognitive effort. This smooth processing enhances aesthetic pleasure by reducing cognitive effort. However, when a design is too uniform, it can appear dull or uninteresting, lacking elements that engage the observer (Berlyne & Boudewijns, 1971; Biederman & Vessel, 2006). Variety, on the other hand, introduces complexity, contrast, and change, which stimulate curiosity and encourage exploration. While unity provides stability and comfort, variety adds excitement and engagement, making the balance between the two crucial for an aesthetically appealing design.

The “Unity in Variety” principle in UMA suggests that the most aesthetically pleasing designs achieve a dynamic balance: they are coherent enough to be readily comprehensible yet varied enough to sustain interest (Post et al., 2016; Loos et al., 2022). Scholars have long explored this balance in aesthetic perception. Dresp-Langley (2015) examined how visual processing mechanisms enable the human brain to integrate diverse elements into a unified perceptual experience, emphasizing that variety enriches perceptual complexity without disrupting coherence. Similarly, Loos et al. (2022) analyzed consumer responses to product designs and found that designs with a structured, yet varied composition tend to elicit stronger aesthetic preferences. Further reinforcing this perspective, the Gestalt approach emphasizes that our perceptual system inherently organizes stimuli into structured, meaningful patterns (Au-Yeung et al., 2023).

However, the relative importance of unity and variety is not uniform across all design contexts. Different product types may shift the perceptual weight of these variables in aesthetic evaluations. For example, variety factors significantly impact website aesthetics more than unity (Post et al., 2017). In topological optimization, unity resides at the heart of improving aesthetic appreciation (Loos et al., 2022). In the study of aesthetic preferences for fast-moving consumer goods (soft drink packaging) and emerging technology products (smart watches), unity has a greater influence than variety (Ding et al., 2025; Ma et al., 2025). Thus, the “Unity in Variety” principle should not be regarded as a static aesthetic rule, but rather as one whose balance depends on the categorical structure of the product. The introduction of product category structure not only helps to understand the mechanism of action of each variable in the UMA model, but also provides a theoretical basis for grasping the applicable strategies of “unity and diversity” balance in specific products.

Cognitive processes allow us to understand art, interpret sensory information, and engage in abstract thinking, helping us collect and analyze aesthetic experiences that shape our sensitivity to beauty ( Świątek et al., 2024). Drawing on past experiences, many design studies have explored how cognitive factors influence aesthetic appreciation, particularly focusing on the relationship between typicality and novelty (Ding et al., 2025). One of the most influential principles capturing this tension is Loewy’s Most Advanced Yet Acceptable (MAYA), which suggests that aesthetically successful designs strike a balance between innovation and familiarity (Loewy, 2002). This duality was echoed in Berlyne’s (1973) theory of arousal potential, which posits that aesthetic pleasure peaks at moderate complexity, balancing the competing drives for clarity and surprise. Expanding on this idea, Whitfield’s CM Model (2000) links the emotional response triggered by novelty with the cognitive process of categorization, offering a deeper understanding of how people perceive and evaluate popular designs.

Building on these ideas, the UMA incorporates typicality and novelty as complementary cognitive-level variables. Typicality satisfies our need for safety by providing easily recognized and comprehensible stimuli, while novelty caters to our drive for achievement by offering unexpected elements that spark curiosity (Hekkert, 2014). Because of human evolutionary tendencies, people generally favor products that are easy to recognize and categorize, often preferring designs that feel familiar or prototypical (Hekkert et al., 2003). The UMA model incorporates this idea alongside the MAYA principle, which suggests that the most appealing designs successfully balance familiarity with innovation (Hekkert et al., 2003; Thurgood et al., 2014).

However, the balance between typicality and novelty is not static. Different product types may change the perceived weights of these variables in an aesthetic assessment. For example, novelty factors have a greater impact on industrial boilers than typicality (Suhaimi et al., 2023). In daily necessities toothbrushes, typicality is the core of affecting consumer preferences (Yahaya, 2017). Typicality has a greater impact than novelty when it comes to aesthetic preferences for emerging technology products (smartwatches) (Ma et al., 2025). Therefore, incorporating category typology is essential for explaining how consumers cognitively evaluate products. It provides a theoretical anchor for understanding when and why either typicality or novelty dominates, thereby enhancing the predictive capacity of the UMA in diverse design contexts.

Social cues embedded in product design facilitate interactions between individuals and products, allowing people to feel a sense of belonging while also expressing their autonomy (Blijlevens & Hekkert, 2015). People often use design elements in products as social signals to express group identity and individuality, a balance captured by the concepts of connectedness and autonomy (Ding et al., 2025). Connectedness refers to how well a product aligns with social norms and cultural expectations, making users feel a sense of belonging (Baumeister & Leary, 2017). Research suggests that products can symbolize shared values and group identity, strengthening social bonds (Barrett & Bar, 2009; Markus & Kitayama, 1991). Designs that reflect familiar social cues tend to evoke comfort and security, increasing their aesthetic appeal (Deci & Ryan, 2000; Baumeister & Leary, 2017). For example, Bloch (1995) pointed out that social elements in product design significantly influence aesthetic appreciation by reinforcing a shared visual language. Empirical studies on everyday items such as sunglasses, staplers, backpacks (Blijlevens et al., 2014), smartwatches (Ma et al., 2025), and soft drink packaging (Ding et al., 2025) consistently indicate that higher connectedness leads to stronger aesthetic preferences. In contrast, autonomy reflects the need to stand out and express individuality. While connectedness fulfills the desire for belonging, autonomy addresses the need for personal differentiation (Blijlevens & Hekkert, 2015). Sociologically, autonomy relates to the pursuit of freedom, independence, and self-expression (Deci & Ryan, 2009; Lynn & Harris, 1997). From an aesthetic perspective, products that challenge traditional design norms and showcase uniqueness can attract positive responses, allowing users to express their individuality (Bourdieu, 2018).

The UMA model integrates these social dimensions by asserting that the most attractive designs strike an optimal balance between connectedness and autonomy. This balance is reflected in the “Autonomous yet Connected” principle (Blijlevens & Hekkert, 2015), which suggests that the most aesthetically appealing designs successfully combine the need for social belonging with the desire for individuality. Research has also explored how security and accomplishment influence the relationship between these two factors. For example, Blijlevens and Hekkert (2019) found that in high-social-risk situations, designs emphasizing social conformity tend to enhance aesthetic appeal, whereas in lower-risk contexts, products with greater autonomy may be more attractive. In summary, connectedness and autonomy at the social level play a crucial role in shaping aesthetic experiences. However, the balance between connectedness and autonomy is not static. The relative influence of these variables may vary according to the product environment, so a more detailed category definition is needed to explain their weights and their interactions in different design areas. Inclusion in category types can significantly improve UMA’s explanatory power and help designers better combine product aesthetics with consumer expectations.

Based on these considerations, the present study raises the following questions: How much do the three levels of perception, cognition, and social influence aesthetic preferences when tested simultaneously on laptop? How does the UMA’s aesthetic variables influence the appraisal of the aesthetics of laptop? Does Whitfield’s CM Model have a guiding role in UMA? Finally, we propose three hypotheses:

3.1 Research methodology

This study adopts a quantitative experimental methodology to explore how six aesthetic variables affect users’ aesthetic preferences for laptop designs. As shown in Figure 1, the research process follows a structured multi-stage procedure comprising stimulus and participant selection, data collection, instrumentation, and multilevel statistical analysis.

Figure 1. Overview of research methodology.

This flowchart outlines the stepwise process used in the study, including stimulus selection, participant recruitment, data collection, and multi-level statistical analyses.

3.2 Participants

The current study involved 234 Chinese participants who evaluated aesthetic preferences for laptops, a representative closed-category digital product. To minimize bias, individuals under 18 years of age and those with professional design backgrounds were excluded, as prior research indicates that design experts may respond based on specialized knowledge rather than general fundamental aesthetic evaluation (Whitfield, 2007). Participants were recruited online via Google Forms following standardized guidelines, ensuring consistency across multiple study phases. The sample was stratified into four age groups: 24.2% were18–25 years (n = 57), 32.6% were 26–35 years (n = 76), 19.9% were 36–45 years (n = 47), and 23.3% were 46 years old or above (n = 54). The gender distribution was relatively balanced, with 53.4% male (n = 125) and 46.6% female participants (n = 109).

The use of Chinese participants was appropriate for the present study because China represents an important consumer context for laptop products and provides a meaningful setting for examining aesthetic judgements of consumer electronics. At the same time, this sampling decision limits the cross-cultural generalizability of the findings. Therefore, the results should be interpreted as evidence from a Chinese non-design consumer sample rather than as universal aesthetic principles. Future studies should examine whether similar patterns occur across different cultural groups.

The final sample size of 234 participants was also consistent with previous UMA-based product-aesthetics studies, where sample sizes typically ranged from 85 to 300 participants (Berghman & Hekkert, 2017; Post et al., 2017; Tyagi, 2017; Suhaimi et al., 2023; Ding et al., 2025; Ma et al., 2025). However, no formal a priori power analysis was conducted before data collection. This is acknowledged as a methodological limitation, and future studies should include an a priori power analysis based on the expected effect size and repeated-measures structure. By focusing on a non-expert population, this study aimed to capture general aesthetic judgments that were not shaped by specialized design training. All participants provided informed consent through an online form. Participation was voluntary, and respondents could withdraw at any time without consequences.

3.3 Stimuli

The stimuli for this study consisted of visual images of ten laptop designs, incorporating five commercially available laptops and five conceptually designed models developed by an experienced team of industrial designers, as shown in Figure 2. This hybrid construction approach was deliberately adopted to ensure sufficient variation across the six aesthetic variables defined in the Unified Model of Aesthetics (UMA): unity, variety, typicality, novelty, connectedness, and autonomy. Commercially available laptops alone could not adequately represent all six variables, particularly at the higher levels of novelty, variety, or autonomy. Therefore, conceptually designed stimuli were included to enhance contrast and coverage across the full UMA variables.

Figure 2. Visual stimuli of ten laptop designs used in the study.

Visual stimuli of ten laptop designs used in the study. Each stimulus is numbered from S1 to S10 to facilitate identification in the estimated marginal mean plots and scatter plots.

Although a formal taxonomy was not applied, the conceptually designed stimuli were generated by a professional design team with prior experience in UMA-based design studies. Their design brief was to maximize perceptual contrast along specific UMA dimensions, particularly where commercial products tend to cluster around typicality and unity. The team purposefully introduced distinctive features such as non-traditional silhouettes, asymmetrical layouts, layered configurations, and expressive stylistic elements. The theoretical alignment of the stimuli with the six UMA variables was reviewed by senior design experts to support clear differentiation among the laptop designs.

To reduce potential confounding effects, all ten laptop images were processed using the following standardization procedures. First, all laptops were converted to grayscale to reduce bias from color preference. Second, all brand logos, model identifiers, and operating-system interface cues were digitally removed using Adobe Photoshop. Third, all stimuli were placed against neutral backgrounds to reduce the influence of contextual visual information. However, because the stimulus set included both commercially available laptop images and conceptually rendered designs, complete equivalence in viewing angle, lighting, and rendering style could not be fully achieved. This limitation is acknowledged because differences in orientation or rendering style may have influenced participants’ visual judgements.

This hybrid approach combined ecological validity through real-world products with experimental contrast through conceptually designed stimuli. For example, S1 and S2 represent relatively conventional laptop forms with higher typicality and unity, whereas S9 and S10 feature exaggerated sculptural forms that emphasize novelty and autonomy. S7, with its rounded and user-friendly layout, was intended to enhance perceptions of connectedness, while S6 presented high visual variety through a layered configuration. To facilitate recognition in the estimated marginal mean plots and scatter plots, each stimulus was numbered from S1 to S10 in Figure 3. A descriptive classification of the ten stimuli is provided in Table 1.

Figure 3. Estimated marginal means (EMMs) of aesthetic pleasure ratings across ten laptop designs.

Bars indicate mean ratings of overall aesthetic pleasure on a 7-point Likert scale (1 = strongly disagree, 7 = strongly agree) for each laptop stimulus.

The hybrid stimulus set also introduced a potential confound between real and conceptually designed laptops. Real products may carry residual form familiarity even after brand removal, whereas conceptually designed products may appear less familiar because they are not commercially available. Therefore, participants’ responses may partly reflect differences in recognition or market familiarity in addition to the intended UMA variables. This issue was not separately modeled in the present analysis and is treated as a limitation. Future studies should either use a fully controlled set of newly designed stimuli or include stimulus source type, real versus conceptual, as an additional factor in the analysis.

Before the main study, an independent manipulation check was conducted with 30 participants who did not take part in the formal experiment. Participants evaluated the ten laptop stimuli using the same 7-point rating items for the six UMA variables. Repeated-measures ANOVA showed significant differences among the ten stimuli for all six variables: unity, F(9, 261) = 18.42, p < .001, η_p² = .389; variety, F(9, 261) = 14.76, p < .001, η_p² = .337; typicality, F(9, 261) = 21.35, p < .001, η_p² = .424; novelty, F(9, 261) = 16.89, p < .001, η_p² = .368; connectedness, F(9, 261) = 12.57, p < .001, η_p² = .302; and autonomy, F(9, 261) = 15.94, p < .001, η_p² = .355. These results indicated that the stimulus set produced sufficient perceived variation across the intended UMA dimensions. Conventional laptop forms were generally rated higher in unity and typicality, while unconventional and sculptural forms were rated higher in novelty, variety, and autonomy. Thus, the manipulation check supported the suitability of the stimuli for the main experiment.

3.4 Procedures

The study was conducted through an online questionnaire accessible via web and mobile devices. Participants were instructed to complete the questionnaire in a quiet environment, to view the images carefully, and to use a device with a screen large enough to display the laptop images clearly. However, screen size, display resolution, ambient lighting, and viewing distance were not experimentally controlled. Participants’ visual impairments were also not independently verified. These factors may have introduced variability into the visual aesthetic judgements and are acknowledged as limitations of the online data collection procedure.

Ethical approval for this study was granted by the Ethics Committee for Research Involving Human Subjects of Universiti Putra Malaysia (Jawatankuasa Etika Universiti Penyelidikan Manusia UPM), under Approval No. JKEUPM-2023-1213. The study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. All participants were presented with an online informed consent form prior to commencing the survey, and only those who provided written electronic consent were allowed to participate.

The questionnaire consisted of two main sections. The first section gathered demographic details such as age, gender, and other relevant background information for participant screening and later analysis. The second section focused on a visual evaluation of laptops. Participants viewed ten laptop images, each displayed individually in a randomized order to reduce potential sequence bias. For each image, participants responded to several statements using a 7-point Likert scale, ranging from “strongly disagree” (1) to “strongly agree” (7). Each aesthetic variable was measured using one item per stimulus. Thus, for each of the ten laptop images, participants rated seven statements: unity, variety, typicality, novelty, connectedness, autonomy, and overall aesthetic pleasure.

The items were designed to evaluate the stimuli across the three levels of the Unified Model of Aesthetics (UMA). At the perceptual level, unity and variety were assessed using items such as “this is a unified design” and “this design conveys variety.” At the cognitive level, typicality and novelty were evaluated using items such as “this is a typical design” and “this is a novel design.” At the social level, connectedness and autonomy were examined using items such as “this design makes me feel connected” and “this design emphasizes my individuality.” Overall aesthetic pleasure was measured using the item “this design is pleasing to see.” These measurements were adapted from established aesthetics pleasure and product-aesthetics studies (Blijlevens et al., 2014; Blijlevens et al., 2017). The full questionnaire items are provided in the supplementary material.

3.5 Data analysis

Data analysis was performed using IBM SPSS Statistics (version 26.0; https://www.ibm.com/products/spss-statistics). First, repeated-measures ANOVA was conducted to examine whether participants' ratings differed significantly across the ten laptop stimuli. This analysis was used to identify stimulus-level variation in aesthetic pleasure and in the six UMA variables: unity, variety, typicality, novelty, connectedness, and autonomy.

Second, Generalized Estimating Equations (GEE) were employed to assess the population-averaged effects of the six UMA variables on the dependent variable, aesthetic pleasure. GEE was appropriate because each participant evaluated multiple laptop stimuli, resulting in correlated repeated observations. The six UMA variables were entered as independent variables, and aesthetic pleasure was entered as the dependent variable. The unstandardized GEE β coefficients were used to compare the relative predictive contribution of each aesthetic variable because all predictors were measured using the same 7-point Likert scale.

Third, given the repeated-measures design, in which each participant evaluated multiple laptop stimuli, this study also employed Linear Mixed-Effects Modeling (LMM). LMM is suitable for nested data structures, where ratings are nested within participants and repeated evaluations may vary across stimuli. In the LMM, the six UMA variables were included as fixed effects, and random intercepts for participants were included to account for individual-level baseline differences in aesthetic preference. The LMM results were used to complement the ANOVA and GEE analyses by accounting for participant-level variability in repeated aesthetic judgements.

Given the moderate-to-high correlations among several UMA predictors, multicollinearity diagnostics were also examined before interpreting the regression-based models. Variance inflation factor (VIF) and tolerance values were inspected to assess whether the coefficient estimates were likely to be unstable. Conventional thresholds were used, with VIF values below 5 and tolerance values above .20 indicating that multicollinearity was not severe enough to invalidate the regression-based estimates.