In virtual environments, 2D top-down view maps and 3D bird’s eye view maps are two prevalent types of map visual presentations (Li and Giudice, 2013). Lu and Peponis (2014) found that 2D maps allowed users to identify maximally visible locations. Liao et al. (2017) insisted that 2D maps are more effective for spatial knowledge acquisition. In addition, they also found that users could have performed better when using 3D maps than when using 2D maps due to excessive map information. Kray et al. (2003) found that 2D maps allowed for faster identification of initial location and route finding, and enabled users to extract more spatial information in less time. However, users’ processing of spatial information in 2D maps requires a sizeable cognitive load to translate into self-centered physical spatial information (Goodman et al., 2005). Zanola et al. (2009) insisted that the cues provided by 3D maps enabled users’ ability to recognize space and landmarks compared to 2D maps. Many past studies have supported the advantages of 3D maps in wayfinding. For example, for unfamiliar environments, 3D maps can provide near-natural world vision (Kray et al., 2003), simplify user recognition of objects contained in the environment, and significantly improve user comprehension (Wickens, 2000). They also have a highly positive impact on user satisfaction (Kray et al., 2003). Nonetheless, Popelka and Doležalová (2015) found no significant differences in eye-movement metrics between 2D and 3D maps in a visual search task. Too many 3D landmark models on the screen confused map users and slowed their information processing (Delikostidis and Van Elzakker, 2009). Lei et al. (2016) found that 3D maps produced a clustered gaze with longer duration and smaller sweep magnitude, which meant that 3D maps users were less efficient at gathering detailed information. In summary, there is no unified conclusion on the performance of 2D and 3D visual presentation regarding user wayfinding. Moreover, different types of map visual presentations on user wayfinding visual search performance and subjective evaluations in intelligent shopping guide terminals used in shopping malls have not been sufficiently studied. Therefore, this study mainly focuses on the impact of different virtual map visual presentation types on user experience regarding intelligent shopping guide terminals.

Landmarks play a crucial role in the spatial task of wayfinding (Caduff and Timpf, 2006). In the wayfinding process, map labels refer to providing valuable geographic information by annotating geographic phenomena with textual descriptions (Casakin et al., 2000). Map labels not only help the user convey a large amount of semantic information during map reading (Roth et al., 2017), but also reduce the number of wayfinding errors (Deakin, 1996). However, while landmarks can help users process their learning environment, they also require the user’s cognitive resources (Cheng et al., 2022). Subject to users’ limited cognitive abilities, too much landmark information on mobile maps may lead to cognitive overload for users (Baddeley, 1992).

Landmark descriptions speed up the comprehension time when the user’s viewpoint is switched (Dong et al., 2020), giving the user more time to process the overall spatial information (Lee and Tversky, 2005). Vilaplana (2015) suggested improving the readability of map annotations with a conspicuousness-based approach to avoid annotation overlap. Yoeli (1972) emphasized that the placement of annotations is one of the primary elements of map annotation design. Slocum et al. (2022) argued that point feature labels should be horizontal, preferably above and to the right of the point. In addition, Harrie et al. (2022) also suggested that area labels should preferably be placed horizontally within the polygonal elements they represent. Later, Rylov and Reimer (2017) proposed a method for external labeling region features. It is well known that the user’s attention is strongly influenced by the overall layout of the user interface and the visual presentation of the location of target elements (Bojko, 2006). During wayfinding, the user gradually built up a mental representation of space by interacting with the geospatial aspect of the map (Millonig and Schechtner, 2007; Montello et al., 2004). Although there is evidence that different forms of landmark information affect users’ visual search performance in different ways (Ye et al., 2022), the location of landmark descriptions in intelligent shopping guide terminals in shopping malls has not been sufficiently studied regarding users’ wayfinding visual search performance and their subjective perceptions. Therefore, this study focused on the impact of different virtual map landmark descriptions on user experience of intelligent shopping guide terminals.

Previous research has shown that “virtual map visual presentations” and “locations of landmark descriptions” are critical factors influencing user experience (UX). However, limited studies have explored the interaction between these two factors. To bridge this gap, a conceptual model was developed, summarizing the relationship between prior research and users’ wayfinding visual search performance. A 2x3 between subject design (60 participants) was employed to examine how different virtual map types (2D and 3D) and landmark description locations (on, within, or below the map) affect task performance and subjective evaluations. Regarding the sample size of participants commonly adopted in previous related studies, refer to Table 2. This study, conducted through the user interface of an intelligent shopping guide terminal, aimed to answer the following key questions:

Q1: Does the location of landmark descriptions (within the map, on the map, or below the map) impact users’ wayfinding visual search performance?

A purposive sampling method was used to recruit 60 participants (32 males and 28 females) aged 18 to 39, all of whom had prior experience using intelligent shopping guide terminal interfaces. According to the experimental design, we divided them into 6 groups, each comprising 10 participants. The sample comprised 45 undergraduate students (75%), 10 master’s students (17%), and 5 Ph.D. students (8%). Of the participants, 48 (80%) were between the ages of 18 and 28, and 12 (20%) were between 29 and 39. This age group was selected to represent frequent users of intelligent shopping guide terminals in large shopping malls.

In the context of wayfinding, spatial ability is task-related, and information processing capacity is a critical metric for assessing performance (Allen, 1999). wayfinding visual search performance can be evaluated by measuring the time it takes users to complete a task (Chen et al., 2013). Before the experiment commenced, participants were informed that they would be required to perform five wayfinding tasks using the intelligent shopping guide terminal (see Figure 2). The task design involved participants standing in front of the touchscreen, closely replicating the typical operation scenario of consumers in a real shopping mall. During the experiment, participants could easily interact with the touchscreen display within their reach (see Figure 3). Figure 4 provides an illustration of the experimental operations. Participants first completed a consent form and a questionnaire to collect background information. They then proceeded to the experiment, performing the designated wayfinding tasks on the prototype’s user interface. Task completion times were recorded via screen recording software for further analysis. Upon completing the experiment, participants were asked to fill out a questionnaire that included five questions rated on a 7-point Likert scale to provide subjective evaluations of the task performance. Finally, a semi-structured interview was conducted to gather participants’ personal reflections on task difficulty and any additional opinions or suggestions.

The information displayed in the prototypes is mainly standard functions in intelligent shopping guide terminals of shopping malls, including wayfinding instructions, promotion activities, brand searching, and parking payment (see Table 3).

In this study, the collected data regarding the main effects of the virtual map visual presentation and the location of landmark description, and their interaction effects on participants’ wayfinding task completion time (i.e., in seconds), as well as subjective evaluations were analyzed using the SPSS software. The significant effect was further analyzed with the LSD post hoc comparison to help address the differences among the factor levels (see Table 4).

The results of the SUS analysis (Table 5) indicated no significant main effect of virtual map visual presentation (F_{(1, 54)} = 0.70, p = 0.410, η² = 0.03). However, a significant main effect of the location of landmark description was observed (F_{(2, 54)} = 5.01, p = 0.011, η² = 0.19). Post hoc analysis revealed that SUS scores for landmarks located on (M = 71.88, SD = 12.36) and below (M = 72.81, SD = 10.04) the map were significantly higher than for those within the map (M = 63.23, SD = 15.30). No significant interaction effect was found (F_{(2, 54)} = 1.49, p = 0.237, η² = 0.06).

Subjective evaluations were collected using a 7-point Likert scale (1 = least agree, 7 = most agree) after participants completed their assigned tasks. The results from the between subject design for subjective evaluations are presented in Table 6.

The experimental results revealed that the visual presentation of the virtual map had a significant main effect in Task 1, with participants completing the task significantly faster using a 3D map compared to a 2D map. This could be attributed to participants being able to visually perceive the 3D map of the shopping mall more quickly than the 2D version, as landmarks in the 2D map were more visually dispersed. There is an inherent difference in visual perception between central and peripheral fields of vision, influenced by the physiological structure of the human retina (Lei et al., 2016). The visual acuity of the human system is highest in the central region and decreases towards the periphery (Albarrak et al., 2021). As a result, landmarks presented centrally in the 3D map were easier for participants to locate. Additionally, the perspective view in 3D maps facilitates matching the visual information from the map to the physical world (Oulasvirta et al., 2007). These findings support Q2, confirming that virtual map visual presentations can enhance user wayfinding visual search performance in intelligent shopping guide terminals.

In Tasks 2, 4, and 5, which involved visual searches and information comparisons, the location of landmark descriptions significantly impacted performance. Participants navigated faster when landmarks were described on or below the map rather than within the map. However, no significant difference was found between landmarks positioned on and below the map. One explanation could be that numerical labels were used to correlate map regions with corresponding landmark descriptions, simplifying the map’s visual complexity compared to descriptions within the map. Previous studies have demonstrated that increased visual complexity prolongs visual search time (Neider and Zelinsky, 2011). Though prior research indicated quicker response times at the top of the screen (Gong et al., 2012), no significant difference was observed in this study, likely due to the need for frequent eye focus shifts between the map and the external landmark descriptions. These shifts may have offset any advantage in visual attention (Lee and Tversky, 2005). Overall, the results confirm Q1, suggesting that the location of landmark descriptions can influence wayfinding visual search performance.

The study also identified a significant interaction between virtual map presentation and landmark description location in Task 3. Specifically, participants performed faster with the 3D map when landmarks were positioned on or below the map, while the 2D map outperformed the 3D map when landmarks were within the map. This could be due to the higher density of landmark descriptions in 3D maps, which could hinder visual search performance compared to the 2D map. These results align with previous research indicating that 3D maps cluster visual attention and reduce search efficiency (Lei et al., 2016). Reduced visual complexity and a narrower peripheral view in 3D maps, especially when landmarks are placed outside the map, further explain the faster performance with 3D maps. These findings support Q4, showing a significant interaction between virtual map visual presentation and landmark description location.

Additionally, it is worth noting that in our study, we found that the main effect of Virtual Map Visual Presentation was only significant in Task 1 and Task 4. That is, participants completed the tasks significantly faster when using the 3D map compared to the 2D map. However, no such significant main effect was observed in Task 2, Task 3, and Task 5. The possible reasons are as follows: Firstly, although all our tasks align with functions frequently used in daily life, there are differences in task difficulty. Regarding Task 1 and Task 4, they may involve more intuitive spatial navigation or object localization. These tasks can provide richer spatial information and depth perception in a 3D environment, thereby helping participants complete the tasks more quickly. Hence, the 3D map demonstrated a significant advantage in these tasks. In contrast, the characteristics of Task 2, Task 3, and Task 5 are such that the 2D map may have already provided sufficient information, and the additional dimension of the 3D map did not bring about a noticeable improvement in efficiency. It might even slightly slow down the task completion speed due to the increased complexity of information processing.

Analysis of the System Usability Scale (SUS), subjective preference, attractiveness, information clarity, and physical demand consistently showed that landmark description location significantly influenced user experience. Participants rated landmark descriptions on or below the map more favorably than those within the map, with no significant difference between descriptions placed on or below the map. The higher density of labels within the map increases the cognitive load required to process the information (Goldberg and Kotval, 1999). In contrast, placing landmarks outside the map reduced visual complexity and facilitated faster spatial mental model construction. These findings support Q3, confirming that landmark descriptions on or below the map are more attractive to users. From the semi-structured interviews, many participants mentioned that when the landmark descriptions were within the map, they felt overwhelmed by the amount of information. On the other hand, when the descriptions were on or below the map, participants reported a clearer and more organized view.

The subjective evaluation also revealed a significant interaction between virtual map presentation and landmark location. When landmarks were placed on or below the map, 3D maps were rated higher in attractiveness and information clarity compared to 2D maps. However, when landmarks were within the map, 2D maps outperformed 3D maps. This is likely due to the higher density and complexity of 3D maps when landmarks are within the map, which demands greater cognitive resources to process (Popelka and Doležalová, 2015). Furthermore, the high visual complexity of 3D maps can overload the visual system, resulting in serial searches and reducing search fluency (Kahneman, 1973). These findings align with Q4, indicating a clear interaction effect between map presentation and landmark location. The interviews also supported this interaction. When using 3D maps with landmarks on or below, participants found the perspective and depth appealing. However, when landmarks were placed inside the 3D map, the complexity became a deterrent.

While this study provides valuable insights, several limitations should be noted. First, the participant sample was limited to college students aged 18–39, representing a narrow demographic. Although this age group is typical of intelligent shopping guide terminal users, future studies should include broader demographic factors such as age, gender, occupation, and user experience, as these may significantly influence wayfinding visual search performancee. Second, the color scheme of the experimental interface could have affected participants’ visual search and perception during wayfinding, as color is known to influence visual cognition. Future research could employ eye-tracking technology to explore the effects of different color schemes on wayfinding visual search performance. Lastly, the complexity of the virtual environment, including the density of landmark information, plays a crucial role in spatial perception and search efficiency. This study did not simulate highly complex and realistic virtual environments, which may limit the generalizability of the findings. Future studies should explore the impact of environmental complexity in more realistic settings to provide a deeper understanding of its influence on wayfinding behavior.