Form Follows Content: An Empirical Study on Symbol-Content (In)Congruences in Thematic Maps

Through signs and symbols, maps represent geographic space in a generalized and abstracted way. Cartographic research is, therefore, concerned with establishing a mutually shared set of signs and semiotic rules to communicate geospatial information successfully. While cartographers generally strive for cognitively congruent maps, empirical research has only started to explore the different facets and levels of correspondences between external cartographic representations and processes of human cognition. This research, therefore, draws attention to the principle of contextual congruence to study the correspondences between shape symbols and different geospatial content. An empirical study was carried out to explore the (in)congruence of cartographic point symbols with respect to positive, neutral, and negative geospatial topics in monothematic maps. In an online survey, 72 thematic maps (i.e., 12 map topics × 6 symbols) were evaluated by 116 participants in a between-groups design. The point symbols comprised five symmetric shapes (i.e., Circle, Triangle, Square, Rhomb, Star) and one Asymmetric Star shape. The study revealed detailed symbol-content congruences for each map topic as well as on an aggregated level, i.e., by positive, neutral, and negative topic clusters. Asymmetric Star symbols generally showed to be highly incongruent with positive and neutral topics, while highly congruent with negative map topics. Symmetric shapes, on the other hand, emerged to be of high congruence with positive and neutral map topics, whilst incongruent with negative topics. As the meaning of point symbols showed to be susceptible to context, the findings lead to the conclusion that cognitively congruent maps require profound context-specific considerations when designing and employing map symbols.


Introduction
When we look at maps, "we see symbols spread out on the space of a document, on paper or a computer screen" and we expect these symbols to refer to geographic space [1] (p. 21). Through the mediation of signs, maps enable us to relate to spatial phenomena and events from a viewpoint beyond direct experience [2]. By this, maps enable us to imagine near as well as distant events, some of which we would never encounter otherwise. As such, cartographic maps are a form of communication, a means of conceiving, articulating, and structuring the human world [3].
However, albeit a map may be designed to convey a single focus of interest, it does not convey a single universal message [4]. The elements of a map are independent associative signs and symbols with a reference fixed by convention but not fixed by a single, unequivocal reference [5,6]. The map-making process must, therefore, follow particularly informed choices and deliberate decisions since the variety of methods and visual variables available to represent information allow for strikingly different results based on a single dataset [4].
One of the first decisions made by the cartographer concerns the type of symbolization to employ for a given topic [7]. Yet, to this day, the cartographer faces the challenge-as well as the between sign and meaning. Theorists, however, further emphasize the dimension of sense as a crucial factor in cartographic communication processes, such that the meaning of signs may vary by context. Such semiotic differentiations have, however, often been neglected in cartographic research and applications of semiotics. "Despite a large number of papers dealing with communication in cartography, relatively few have pursued in detail the analysis of map symbols, and the relationships between map symbols and semiotic theory" [2] (p. 179). Consequently, "the difference of what a map sign means and what it represents has become blurred" [32] (p. 245). Research highlights the persisting need for a differentiated understanding of how visual variables can be used to congruently represent geospatial information by taking into consideration the different dimensions of meaning. The present research, therefore, draws attention to cartographic point symbols employed in thematic maps and explores their congruence with different geospatial contexts.

Materials and Methods
An empirical study was carried out to explore how congruent-as well as incongruent-cartographic point symbols are experienced in different geospatial contexts. Geospatial context, in the present study, referred to three types of content, i.e., geospatial topics of positive, neutral, or negative character. For each of the three content groups, monothematic maps were created, in which point symbols for representing a given map topic were systematically varied. The final set of monothematic maps was used as stimulus material and the perceived symbol-content congruence of each map was rated by study participants. The following sections describe the materials and methods of this empirical study in detail.

Materials and Study Design
An online survey was set up to empirically assess the (in)congruence of abstract map symbols with respect to different thematic map content. Twelve map topics were selected a priori by the experimenter, i.e., four positive, four neutral, and four negative topics, to test for symbol-content congruences (see Table 1). For each map topic, a set of six thematic maps were created, where each map depicted one of six abstract shape symbols, i.e., Circle, Triangle, Square, Rhomb, Star, and Asymmetric Star (for examples see Figure 1). The map symbols used in the present study refer to commonly used symmetric shapes in thematic cartography and related visual disciplines [5,45]. In addition, one asymmetric star shape was incorporated into the stimulus set, as related literature indicates particular perceptual qualities of asymmetric shapes, such as quicker detection times in visual search tasks [46], and distinct negative hedonic qualities of asymmetric star symbols [44]. Map symbols were displayed in black in 50 percent transparency, on light-hued basemaps. Basemaps were created to thematically correspond with each map topic. As a result, the basemaps used for the twelve map topics varied by content and scale. For example, the basemap for illustrating a "Traffic accident" displayed a street network, while the basemap for locating a "Lookout tower" indicated urban structures and settlements. Topics that referred to events of a single location were depicted by maps displaying one map symbol, while topics referring to events of multiple locations were represented by three map symbols (see Figure 1 for examples). The final stimulus set comprised 72 maps, as a result of six map stimuli employed for each of the twelve map topics. ISPRS Int. J. Geo-Inf. 2020, 9, x FOR PEER REVIEW 5 of 23 perceptual qualities of asymmetric shapes, such as quicker detection times in visual search tasks [46], and distinct negative hedonic qualities of asymmetric star symbols [44]. Map symbols were displayed in black in 50 percent transparency, on light-hued basemaps. Basemaps were created to thematically correspond with each map topic. As a result, the basemaps used for the twelve map topics varied by content and scale. For example, the basemap for illustrating a "Traffic accident" displayed a street network, while the basemap for locating a "Lookout tower" indicated urban structures and settlements. Topics that referred to events of a single location were depicted by maps displaying one map symbol, while topics referring to events of multiple locations were represented by three map symbols (see Figure 1 for examples). The final stimulus set comprised 72 maps, as a result of six map stimuli employed for each of the twelve map topics. A between-groups design was applied, which randomly assigned participants to one of two stimulus groups. This approach aimed to minimize learning and response transfer across the stimulus conditions. As a result, each participant assessed half of the stimulus set, i.e., 36 maps. The online survey was carried out in the German language by using the software LimeSurvey [47].
After the survey briefly introduced the aim of the study, which was phrased as to explore how shape symbols and maps are subjectively perceived, participants were instructed to proceed with map evaluation tasks. In these tasks, participants were asked to rate the (in)congruence between each map topic and its cartographic representation on a unipolar 6-point rating scale. The rating scale ranged from 0 (i.e., the cartographic representation does not suit the topic at all) to 5 (i.e., the representation suits the topic very well). Each map stimulus was presented individually, consecutively, and in a randomized order, at a size of 5.
After the map evaluation task, participants rated each map topic on an 11-point Likert scale, from −5 (very negative) to +5 (very positive). These topic ratings were later subjected to empirically identify homogeneous topic clusters of positive, neutral, and negative valence (see Section 4.1.).
Sociodemographic data were gathered at the end of the survey, regarding the participants' affinity for graphic design (self-evaluation on a unipolar 4-point rating scale, from "not at all" to "very affine", with the additional option for "no answer"), their affinity for maps or cartography (unipolar 4-point rating scale, from "not at all" to "very affine", or "no answer"), age, and gender. At the end of the survey, participants were also asked to indicate the devices they used for completing the survey (i.e., desktop PC, laptop, tablet, or smartphone). A between-groups design was applied, which randomly assigned participants to one of two stimulus groups. This approach aimed to minimize learning and response transfer across the stimulus conditions. As a result, each participant assessed half of the stimulus set, i.e., 36 maps. The online survey was carried out in the German language by using the software LimeSurvey [47].
After the survey briefly introduced the aim of the study, which was phrased as to explore how shape symbols and maps are subjectively perceived, participants were instructed to proceed with map evaluation tasks. In these tasks, participants were asked to rate the (in)congruence between each map topic and its cartographic representation on a unipolar 6-point rating scale. The rating scale ranged from 0 (i.e., the cartographic representation does not suit the topic at all) to 5 (i.e., the representation suits the topic very well). Each map stimulus was presented individually, consecutively, and in a randomized order, at a size of 5.
After the map evaluation task, participants rated each map topic on an 11-point Likert scale, from −5 (very negative) to +5 (very positive). These topic ratings were later subjected to empirically identify homogeneous topic clusters of positive, neutral, and negative valence (see Section 4.1).
Sociodemographic data were gathered at the end of the survey, regarding the participants' affinity for graphic design (self-evaluation on a unipolar 4-point rating scale, from "not at all" to "very affine", with the additional option for "no answer"), their affinity for maps or cartography (unipolar 4-point rating scale, from "not at all" to "very affine", or "no answer"), age, and gender. At the end of the survey, participants were also asked to indicate the devices they used for completing the survey (i.e., desktop PC, laptop, tablet, or smartphone).

Participants
Bachelor's students of Regional Planning were recruited from a course on "Thematic Cartography in Regional Planning" held in winter term 2019 at TU Wien-Vienna University of Technology, Austria. Students participated voluntarily and received course credits in the form of bonus points, which counted towards their final grades. The online survey was sent to students in December 2019.
In total, 116 students completed the survey (49 males, 64 females, one person of diverse gender, two missing responses). Participants were randomly assigned to one of two stimulus groups, resulting in 63 participants who completed stimulus material A and 53 individuals who completed stimulus material B, which both comprised a set of 36 map stimuli. Of all participants, 109 persons indicated their age (M = 22.10, SD = 3.29, Min = 18, Max = 35). The majority of participants used laptops to complete the questionnaire (85.3%), followed by desktop PCs (10.3%), tablets (3.4%), and smartphones (0.9%, i.e., one person). Most participants indicated their affinity for graphic design to be moderate to high (somewhat affine = 29.3%; quite affine = 37.9%; very affine = 25.9%), while one person reported to have no interest in graphic design (0.9%). Seven individuals (6.0%) did not answer this question. The participants' affinity for cartographic design showed to be moderate in most cases (somewhat affine = 30.2%; quite affine = 50.9%) and high in 12.9% of the cases. Seven individuals (6.0%) did not report their affinity for cartographic design.

Results
The study collected symbol-content congruency ratings for 72 map stimuli (12 maps × 6 stimulus conditions) from 116 participants in a between-groups study design. It also collected valence ratings for each of the twelve topics. These two sets of data were statistically analyzed by using the statistical software package SPSS [48] and XLStat addon for Microsoft Excel [49].
First, valence ratings were subjected to statistical analyses to identify homogeneous topic clusters (see Section 4.1. for details). These topic clusters were subsequently used to analyze the participants' symbol-content congruency ratings on a cluster-level as well as individually by each map topic (see Section 4.2).

Topic Analysis
Twelve map topics were used in the present study and were selected a priori. To empirically assess each topic's connotative quality, the participants' valence ratings for each topic were subjected to statistical analyses. The topics' valence ratings generally ranged from very negative (−5) to very positive (+5). The descriptive statistics of these ratings are summarized in the Supplementary Material Table S1.
A hierarchical cluster analysis was performed based on the topics' valence ratings to statistically identify topic clusters of similar qualities. An agglomerative clustering approach was applied, which starts with each item as a cluster and progressively links the items based on their estimate of distances to one another. A cluster distance measure of average linkage was employed, which balances the limitations of single and complete linkage methods, i.e., using information about all pairs of distances to assign cluster membership not just the nearest or the farthest item pairs.
Results indicate a three-cluster solution (see Figure 2) and confirm the three clusters of positive, neutral, and negative topics that had been initially compiled a priori by the experimenter (see Table 1). The cluster analysis dendrogram further indicates a metaproximity between neutral topics and positive topics, suggesting a more proximal relation between the selected neutral and positive topics and a more distant relation of both towards the negative topic cluster. The findings from hierarchical cluster analysis were subsequently used for the next stages of data analysis, where the twelve thematic maps were assigned to their corresponding positive, neutral, or negative topic cluster. In the next stages of analysis, differences between symbol-content congruency ratings were, first, statistically analyzed by each topic cluster, before they were assessed individually by each map topic (see Section 4.2.).

Map Analysis
Based on participants' symbol-content congruency ratings of the 72 thematic map stimuli (i.e., 12 map topics × 6 symbols), mean ratings were computed for each map and each map symbol. Figure  3 visualizes the results, indicating fairly similar ratings for symmetric map symbols when representing positive or negative content, while the Asymmetric Star symbol appeared to be rated least suitable for referring to positive or neutral map content. In contrast, Asymmetric Star symbols showed to be rated as most congruent for cartographically relating to negative geospatial topics. To explore for significant differences between the map stimuli, participants' symbol-content (in)congruency ratings were subjected to statistical analyses. First, participants' responses were The findings from hierarchical cluster analysis were subsequently used for the next stages of data analysis, where the twelve thematic maps were assigned to their corresponding positive, neutral, or negative topic cluster. In the next stages of analysis, differences between symbol-content congruency ratings were, first, statistically analyzed by each topic cluster, before they were assessed individually by each map topic (see Section 4.2).

Map Analysis
Based on participants' symbol-content congruency ratings of the 72 thematic map stimuli (i.e., 12 map topics × 6 symbols), mean ratings were computed for each map and each map symbol. Figure 3 visualizes the results, indicating fairly similar ratings for symmetric map symbols when representing positive or negative content, while the Asymmetric Star symbol appeared to be rated least suitable for referring to positive or neutral map content. In contrast, Asymmetric Star symbols showed to be rated as most congruent for cartographically relating to negative geospatial topics. To explore for significant differences between the map stimuli, participants' symbol-content (in)congruency ratings were subjected to statistical analyses. First, participants' responses were analyzed on an aggregated level, i.e., according to the three topic clusters identified by agglomerative hierarchical cluster analysis (see Section 4.2.1). Second, symbol-content congruences were analyzed individually, i.e., according to each of the twelve map topics (see Section 4.2.2, Section 4.2.3, and Section 4.2.4).
Both stages of analysis followed a two-step procedure. Symbol-content (in)congruency ratings were first subjected to a Durbin-Skillings-Mack test for nonparametric data to identify overall significant differences. The Durbin-Skillings-Mack procedure reveals if the null hypothesis must be rejected, i.e., if at least one result of symbol-content congruence significantly differs from another. The Durbin-Skillings-Mack test can be considered as an extension of the Friedman test. While the Friedman test can only be used in cases of complete paired samples, the Durbin-Skillings-Mack test applies a block design and, thus, allows to compare paired samples of unequal size [50]. To follow up on significant findings, post hoc Conover-Iman analyses were applied subsequently to identify subsets of homogeneous symbol-content congruences in greater detail. The post hoc Conover-Iman procedure performs multiple pairwise comparisons for nonparametric data based on rank differences [51]. Post hoc procedures generally explore the data for between-group differences by comparing all possible combinations of pairs [52]. As such statistical procedures conduct multiple pairwise comparisons, they also correct the significance levels such that the overall Type I error rate (i.e., significance level α, which is the probability of rejecting the null hypothesis given that it is true) across all comparisons remains 0.05 [52]. Hence, a Bonferroni correction was applied as part of the post hoc Conover-Iman analyses to ensure that despite multiple pairwise comparisons the cumulative Type I error remains below 0.05.

Topic Clusters and Symbol Congruence
In the first stage of analysis, participants' congruency ratings were aggregated and analyzed according to the three positive, neutral, and negative topic clusters, which had been identified by agglomerative hierarchical cluster analysis. For each of the three valence clusters, a Durbin-Skillings-Mack procedure was performed, revealing highly significant differences (p < 0.001) between the symbol congruency ratings for either of the three topic clusters (see Table 2). To follow up on these findings, Conover-Iman post hoc tests were performed for each of the three map clusters. As part of the analysis, a Bonferroni correction was applied which controls for Type I error by dividing α by the number of pairwise comparisons. The ratings of six map symbols were pairwise compared within each map cluster and resulted in 15 pairwise combinations. The Bonferroni correction, therefore, set the significance levels at α = 0.0033 (i.e., Bonferroni corrected α = 0.5/15), and differences between the multiple pairwise comparisons were interpreted accordingly (for detailed Conover-Iman test results see Supplementary Material Table S2).
Based on the results of the Conover-Iman post hoc procedure, Demšar graphs were computed to visualize the results of critical differences within and between the map symbols for each topic cluster (see Figure 4) [53]. Demšar graphs illustrate critical differences between mean ranks in a way that groups that are not significantly different are connected by horizontal lines. In those cases, the null hypothesis cannot be rejected, i.e., that the pairs' mean ranks must be considered equal. In contrast, any pair which is not connected with a horizontal line can be considered to differ with statistical significance. Positive topic cluster. Based on the set of six shape symbols, a Conover-Iman post hoc analysis suggests three homogeneous symbol subsets when depicting positive map topics (see Group A, B, and C in Figure 4 top). The results showed that symmetric shapes were rated most suitable for depicting positive content. Among the symmetric shapes, two particular symbol subsets emerged (Group B and Group C). Group C refers to map symbols of highest congruency ratings, i.e., Circle, Rhomb, Star, and Triangle, followed by Group B of second highest congruence, i.e., Star, Triangle, and Square. Conover-Iman also discloses some similarities between Group B and C, illustrated by overlaps of Star and Triangle in Figure 4. The third subset comprises the Asymmetric Star symbol, which was rated as least suitable for depicting positive map topics (see Group A).
Neutral topic cluster. When depicting neutral map topics, Conover-Iman post hoc analysis suggests four homogenous subsets among the six-point symbols studied (see Figure 4 center). The Positive topic cluster. Based on the set of six shape symbols, a Conover-Iman post hoc analysis suggests three homogeneous symbol subsets when depicting positive map topics (see Group A, B, and C in Figure 4 top). The results showed that symmetric shapes were rated most suitable for depicting positive content. Among the symmetric shapes, two particular symbol subsets emerged (Group B and Group C). Group C refers to map symbols of highest congruency ratings, i.e., Circle, Rhomb, Star, and Triangle, followed by Group B of second highest congruence, i.e., Star, Triangle, and Square. Conover-Iman also discloses some similarities between Group B and C, illustrated by overlaps of Star and Triangle in Figure 4. The third subset comprises the Asymmetric Star symbol, which was rated as least suitable for depicting positive map topics (see Group A).
Neutral topic cluster. When depicting neutral map topics, Conover-Iman post hoc analysis suggests four homogenous subsets among the six-point symbols studied (see Figure 4 center). The first subset comprises Circles, which were rated as most congruent with neutral map topics (see Group D). The four symmetric polygonal shapes were assigned to two homogeneous, yet partially overlapping subsets (see Group C and D). The results further found that Asymmetric Stars were rated as significantly least congruent for representing neutral geospatial topics (see Group A).
Negative topic cluster. For the negative map topic cluster, the Conover-Iman analysis suggests four homogeneous symbol subsets (see Figure 4 bottom). The results revealed that the Asymmetric Stars were generally rated most congruent for depicting negative map topics (see Group D). Two subsets of moderate congruence further emerged, i.e., Circle and Star (see Group C) and Triangle and Circle (see Group B), with partial overlaps between them. The fourth group referred to the map symbols Square and Rhomb, which formed a distinct subset of lowest congruency ratings when to cartographically depicting negative topics (see Group A).

Positive Map Topics and Symbol Congruence
At the next stage, statistical analyses were performed for each map topic to explore topic-symbol correspondences in greater detail. Findings from Durbin-Skillings-Mack tests suggest highly significant differences (p < 0.001) between the six map symbols for each of the four positive map topics (see Table 3 for details). To follow up on these findings, congruency ratings were subsequently subjected to Conover-Iman post hoc analyses to specifically identify which of the shape symbols significantly differed. Detailed results are discussed for each map topic in the following.
Map #1: City awarded for clean energy. Conover-Iman test suggests three subsets of congruences for the given map topic (for detailed test results see Supplementary Material Table S2). Demšar significance graphs further visualize the critical differences between these three congruence subsets (see Group A, B, C in Figure 5). The results show that the Asymmetric Star significantly differed from all other shapes, characterized by the lowest congruency ratings for the given map topic (see Group A). The other two homogeneous subsets comprise the five symmetric shape symbols (see Groups B and C). Group C revealed to comprise the symbol subset of highest content-congruency ratings, i.e., Star, Circle, and Rhomb, followed by the subset of Circle, Rhomb, Triangle, and Square (see Group B). These results, however, further indicate some similarities between Groups B and C, as illustrated in Figure 5 (see Rhomb and Circle). Map #2: National parks. Conover-Iman's multiple pairwise comparisons suggest three homogeneous symbol-subsets for the given map topic (for detailed test results see Supplementary Material Table S4). Demšar graphs further illustrate these results (see Group A, B, C in Figure 6). The Asymmetric Star was found to form a distinct subset (see Group A), characterized by significantly lowest congruency ratings. In contrast, the five symmetric shape symbols were rated to correspond significantly better for depicting the map topic "National parks" (see Groups B and C). Overall, the Triangular shape emerged to be most congruent with the given map topic, followed by Rhomb, Circle, and Square (see Group C). Conover-Iman results further suggest some similarities between Group B and Group C, as indicated by overlaps between some of the shapes of these two subsets (see Figure 6). Map #3: Urban park with high biodiversity. Three homogeneous symbol subsets were identified by Conover-Iman post hoc analysis for this map topic, yet, also revealing some overlaps between the three groups (for detailed test results see Supplementary Material Table S5). Demšar graphs illustrate the critical differences between the three subsets (see Figure 7). The first subset of lowest congruency ratings comprises the Asymmetric Star, followed by Square and Triangle (see Group A). The second subset of moderate to high congruence refers to the Square, Triangle, Circle, and Rhomb (see Group B). The third homogeneous symbol subset comprises the asymmetric Star, which was rated as most congruent with the given map topic, followed by Rhomb, Circle, and Triangle (see Group C). As mentioned, despite uncovering three homogeneous subsets, the results Map #2: National parks. Conover-Iman's multiple pairwise comparisons suggest three homogeneous symbol-subsets for the given map topic (for detailed test results see Supplementary Material Table S4). Demšar graphs further illustrate these results (see Group A, B, C in Figure 6). The Asymmetric Star was found to form a distinct subset (see Group A), characterized by significantly lowest congruency ratings. In contrast, the five symmetric shape symbols were rated to correspond significantly better for depicting the map topic "National parks" (see Groups B and C). Overall, the Triangular shape emerged to be most congruent with the given map topic, followed by Rhomb, Circle, and Square (see Group C). Conover-Iman results further suggest some similarities between Group B and Group C, as indicated by overlaps between some of the shapes of these two subsets (see Figure 6). Map #2: National parks. Conover-Iman's multiple pairwise comparisons suggest three homogeneous symbol-subsets for the given map topic (for detailed test results see Supplementary Material Table S4). Demšar graphs further illustrate these results (see Group A, B, C in Figure 6). The Asymmetric Star was found to form a distinct subset (see Group A), characterized by significantly lowest congruency ratings. In contrast, the five symmetric shape symbols were rated to correspond significantly better for depicting the map topic "National parks" (see Groups B and C). Overall, the Triangular shape emerged to be most congruent with the given map topic, followed by Rhomb, Circle, and Square (see Group C). Conover-Iman results further suggest some similarities between Group B and Group C, as indicated by overlaps between some of the shapes of these two subsets (see Figure 6). Map #3: Urban park with high biodiversity. Three homogeneous symbol subsets were identified by Conover-Iman post hoc analysis for this map topic, yet, also revealing some overlaps between the three groups (for detailed test results see Supplementary Material Table S5). Demšar graphs illustrate the critical differences between the three subsets (see Figure 7). The first subset of lowest congruency ratings comprises the Asymmetric Star, followed by Square and Triangle (see Group A). The second subset of moderate to high congruence refers to the Square, Triangle, Circle, and Rhomb (see Group B). The third homogeneous symbol subset comprises the asymmetric Star, which was rated as most congruent with the given map topic, followed by Rhomb, Circle, and Triangle (see Group C). As mentioned, despite uncovering three homogeneous subsets, the results Map #3: Urban park with high biodiversity. Three homogeneous symbol subsets were identified by Conover-Iman post hoc analysis for this map topic, yet, also revealing some overlaps between the three groups (for detailed test results see Supplementary Material Table S5). Demšar graphs illustrate the critical differences between the three subsets (see Figure 7). The first subset of lowest congruency ratings comprises the Asymmetric Star, followed by Square and Triangle (see Group A). The second subset of moderate to high congruence refers to the Square, Triangle, Circle, and Rhomb (see Group B). The third homogeneous symbol subset comprises the asymmetric Star, which was rated as most congruent with the given map topic, followed by Rhomb, Circle, and Triangle (see Group C).
As mentioned, despite uncovering three homogeneous subsets, the results also indicate partial overlaps, suggesting some similarities between the three symbol groups (see Figure 7). ISPRS Int. J. Geo-Inf. 2020, 9, x FOR PEER REVIEW 12 of 23 also indicate partial overlaps, suggesting some similarities between the three symbol groups (see Figure 7). Map #4: Cities with excellent water quality. Two distinct symbol subsets were found for the given map topic based on a Conover-Iman analysis (for test results see Supplementary Material Table  S6). One subset refers to the Asymmetric Star which is characterized by significantly lowest congruency ratings (see Group A in Figure 8). The other distinct subset comprises the five symmetric shapes which were rated to be significantly more congruent with the given topic (see Group B). Within that group, the Circle was rated as most congruent, followed by Rhomb, Square, Triangle, and Star (see Group B in Figure 8). Hence, the results revealed a significant preference towards symmetric map symbols when representing "Cities with excellent water quality".

Neutral Map Topics and Symbol Congruence
For each of the four neutral topics identified by agglomerative hierarchical cluster analysis (see Section 4.1), Durbin-Skillings-Mack tests were performed to explore for significant differences between their perceived symbol-content congruences. The results revealed highly significant differences within each map set, suggesting that some map symbols were experienced to be more congruent-and others less-within either of the four neutral topics (see Table 4 for details). Map #4: Cities with excellent water quality. Two distinct symbol subsets were found for the given map topic based on a Conover-Iman analysis (for test results see Supplementary Material Table S6). One subset refers to the Asymmetric Star which is characterized by significantly lowest congruency ratings (see Group A in Figure 8). The other distinct subset comprises the five symmetric shapes which were rated to be significantly more congruent with the given topic (see Group B). Within that group, the Circle was rated as most congruent, followed by Rhomb, Square, Triangle, and Star (see Group B in Figure 8). Hence, the results revealed a significant preference towards symmetric map symbols when representing "Cities with excellent water quality". ISPRS Int. J. Geo-Inf. 2020, 9, x FOR PEER REVIEW 12 of 23 also indicate partial overlaps, suggesting some similarities between the three symbol groups (see Figure 7). Map #4: Cities with excellent water quality. Two distinct symbol subsets were found for the given map topic based on a Conover-Iman analysis (for test results see Supplementary Material Table  S6). One subset refers to the Asymmetric Star which is characterized by significantly lowest congruency ratings (see Group A in Figure 8). The other distinct subset comprises the five symmetric shapes which were rated to be significantly more congruent with the given topic (see Group B). Within that group, the Circle was rated as most congruent, followed by Rhomb, Square, Triangle, and Star (see Group B in Figure 8). Hence, the results revealed a significant preference towards symmetric map symbols when representing "Cities with excellent water quality".

Neutral Map Topics and Symbol Congruence
For each of the four neutral topics identified by agglomerative hierarchical cluster analysis (see Section 4.1), Durbin-Skillings-Mack tests were performed to explore for significant differences between their perceived symbol-content congruences. The results revealed highly significant differences within each map set, suggesting that some map symbols were experienced to be more congruent-and others less-within either of the four neutral topics (see Table 4 for details).

Neutral Map Topics and Symbol Congruence
For each of the four neutral topics identified by agglomerative hierarchical cluster analysis (see Section 4.1), Durbin-Skillings-Mack tests were performed to explore for significant differences between their perceived symbol-content congruences. The results revealed highly significant differences within each map set, suggesting that some map symbols were experienced to be more congruent-and others less-within either of the four neutral topics (see Table 4 for details). To follow up on the findings of the Durbin-Skillings-Mack analysis, participants' congruency ratings were subjected to Conover-Iman post hoc tests in the next step of the analysis. With this, the differences between the six map symbols were explored for each map topic in greater detail.
Map #5: Art college. Conover-Iman's multiple pairwise comparisons identified four subsets of homogeneous symbol-content congruency for this map topic (for detailed test results see Supplementary Material Table S7). Demšar diagrams were computed to illustrate the results (see Groups A, B, C, D in Figure 9). The Asymmetric Star was revealed to refer to a distinct subset (see Group A), characterized by significantly lowest congruency ratings. In contrast, the Circular shape was rated as most congruent with the map topic, followed by Rhomb (see Group D). Rhomb, Star, and Square were found to comprise a subset of second highest congruence (Group C), followed by Star, Square, and Triangle (see Group B). The results further imply some similarities between the three homogeneous subsets of symmetric shapes, as partially overlapping graphs in Figure 9 illustrate (see Group B, C, D).  To follow up on the findings of the Durbin-Skillings-Mack analysis, participants' congruency ratings were subjected to Conover-Iman post hoc tests in the next step of the analysis. With this, the differences between the six map symbols were explored for each map topic in greater detail.
Map #5: Art college. Conover-Iman's multiple pairwise comparisons identified four subsets of homogeneous symbol-content congruency for this map topic (for detailed test results see Supplementary Material Table S7). Demšar diagrams were computed to illustrate the results (see Groups A, B, C, D in Figure 9). The Asymmetric Star was revealed to refer to a distinct subset (see Group A), characterized by significantly lowest congruency ratings. In contrast, the Circular shape was rated as most congruent with the map topic, followed by Rhomb (see Group D). Rhomb, Star, and Square were found to comprise a subset of second highest congruence (Group C), followed by Star, Square, and Triangle (see Group B). The results further imply some similarities between the three homogeneous subsets of symmetric shapes, as partially overlapping graphs in Figure 9 illustrate (see Group B, C, D). Map #6: Art galleries. For cartographically representing "Art galleries", Conover-Iman post hoc analysis suggests three distinct homogeneous symbol subsets (for Conover-Iman test results see Supplementary Material Table S8). Demšar graphs, again, illustrate these findings (see Groups A, B, C in Figure 10). The first subset refers to the Asymmetric Star shape (Group A), which revealed significantly lowest congruency ratings. In contrast, the Circular map symbol was found to form a distinct subset of highest congruence (see Group C), whilst the subset of Star, Rhomb, Square, and Triangle was found to be of distinct, moderate congruence (see Group B). Map #6: Art galleries. For cartographically representing "Art galleries", Conover-Iman post hoc analysis suggests three distinct homogeneous symbol subsets (for Conover-Iman test results see Supplementary Material Table S8). Demšar graphs, again, illustrate these findings (see Groups A, B, C in Figure 10). The first subset refers to the Asymmetric Star shape (Group A), which revealed significantly lowest congruency ratings. In contrast, the Circular map symbol was found to form a distinct subset of highest congruence (see Group C), whilst the subset of Star, Rhomb, Square, and Triangle was found to be of distinct, moderate congruence (see Group B). ISPRS Int. J. Geo-Inf. 2020, 9, x FOR PEER REVIEW 14 of 23 Map #7: Lookout tower. Within the set of six map symbols used to cartographically indicate a "Lookout tower", the Conover-Iman test suggests three distinct symbol subsets (for detailed results see Supplementary Material Table S9). Demšar graphs illustrate these results (see Groups A, B, C in Figure 11). The subset comprising the Asymmetric Star was found to be least congruent with the given map topic (see Group A), while the Circular shape revealed to refer to a distinct subset of high congruence (Group C). The four polygonal shapes of Square, Triangle, Rhomb, and Star (Group B) were revealed to form a distinct homogeneous subset of moderate symbol-content congruence. Map #8: Outdoor stages. Within the set of six map symbols used to cartographically refer to "Outdoor stages", a Conover-Iman post hoc analysis revealed three homogeneous symbol-subsets (for detailed results see Supplementary Material Table S10). The critical differences between the three subsets are illustrated in Figure 12 (see Groups A, B, C). The first subset refers to the Asymmetric Star, which revealed significantly lowest congruency ratings (see Group A). In contrast, the Circle was rated as most congruent for representing the map topic, followed by Square, Star, and Rhomb (see Group C). The third homogeneous subset comprises the four polygonal shapes Square, Star, Rhomb, and Triangle (see Group B). The results further suggest overlaps between some of the polygonal shapes of Group C and B (see Square, Star, and Rhomb), which indicate some similarity between these subsets (for details see Demšar graphs of Figure 12). Map #7: Lookout tower. Within the set of six map symbols used to cartographically indicate a "Lookout tower", the Conover-Iman test suggests three distinct symbol subsets (for detailed results see Supplementary Material Table S9). Demšar graphs illustrate these results (see Groups A, B, C in Figure 11). The subset comprising the Asymmetric Star was found to be least congruent with the given map topic (see Group A), while the Circular shape revealed to refer to a distinct subset of high congruence (Group C). The four polygonal shapes of Square, Triangle, Rhomb, and Star (Group B) were revealed to form a distinct homogeneous subset of moderate symbol-content congruence. Map #7: Lookout tower. Within the set of six map symbols used to cartographically indicate a "Lookout tower", the Conover-Iman test suggests three distinct symbol subsets (for detailed results see Supplementary Material Table S9). Demšar graphs illustrate these results (see Groups A, B, C in Figure 11). The subset comprising the Asymmetric Star was found to be least congruent with the given map topic (see Group A), while the Circular shape revealed to refer to a distinct subset of high congruence (Group C). The four polygonal shapes of Square, Triangle, Rhomb, and Star (Group B) were revealed to form a distinct homogeneous subset of moderate symbol-content congruence. Map #8: Outdoor stages. Within the set of six map symbols used to cartographically refer to "Outdoor stages", a Conover-Iman post hoc analysis revealed three homogeneous symbol-subsets (for detailed results see Supplementary Material Table S10). The critical differences between the three subsets are illustrated in Figure 12 (see Groups A, B, C). The first subset refers to the Asymmetric Star, which revealed significantly lowest congruency ratings (see Group A). In contrast, the Circle was rated as most congruent for representing the map topic, followed by Square, Star, and Rhomb (see Group C). The third homogeneous subset comprises the four polygonal shapes Square, Star, Rhomb, and Triangle (see Group B). The results further suggest overlaps between some of the polygonal shapes of Group C and B (see Square, Star, and Rhomb), which indicate some similarity between these subsets (for details see Demšar graphs of Figure 12). Map #8: Outdoor stages. Within the set of six map symbols used to cartographically refer to "Outdoor stages", a Conover-Iman post hoc analysis revealed three homogeneous symbol-subsets (for detailed results see Supplementary Material Table S10). The critical differences between the three subsets are illustrated in Figure 12 (see Groups A, B, C). The first subset refers to the Asymmetric Star, which revealed significantly lowest congruency ratings (see Group A). In contrast, the Circle was rated as most congruent for representing the map topic, followed by Square, Star, and Rhomb (see Group C). The third homogeneous subset comprises the four polygonal shapes Square, Star, Rhomb, and Triangle (see Group B). The results further suggest overlaps between some of the polygonal shapes of Group C and B (see Square, Star, and Rhomb), which indicate some similarity between these subsets (for details see Demšar graphs of Figure 12).

Negative Map Topics and Symbol Congruence
Durbin-Skillings-Mack tests were performed for each of the four negative map topics, to explore for negative topic-symbol relations. The results suggest significant differences within three of the four negative map topics (see Table 5 for details). Highly significant differences between the set of six map symbols were found for the three maps indicating "Floods", Melting glaciers", and "Traffic accident". No statistically significant differences were found between the map symbols depicting "Smog polluted cities", indicating that the six map symbols were experienced to be equally congruent when depicting this particular topic. To follow up on these findings, Conover-Iman post hoc tests were performed for each of the four map topics.
Map #9: Flood affected area. Results of the Conover-Iman analysis suggest two homogeneous symbol subsets when indicating "Flood affected areas" (see Group A and B in Figure 13; for Conover-Iman test results see Supplementary Material Table S11). Symmetric shapes refer to one of the two distinct, homogeneous symbol subset (see Group A in Figure 13), with the Triangle rated as least congruent, followed by Square, Rhomb, Circle, and Star. In contrast, the Asymmetric Star was rated as most congruent with the given map topic, thus, forming a distinct symbol subset of high congruence (see Group B).

Negative Map Topics and Symbol Congruence
Durbin-Skillings-Mack tests were performed for each of the four negative map topics, to explore for negative topic-symbol relations. The results suggest significant differences within three of the four negative map topics (see Table 5 for details). Highly significant differences between the set of six map symbols were found for the three maps indicating "Floods", Melting glaciers", and "Traffic accident". No statistically significant differences were found between the map symbols depicting "Smog polluted cities", indicating that the six map symbols were experienced to be equally congruent when depicting this particular topic. To follow up on these findings, Conover-Iman post hoc tests were performed for each of the four map topics.
Map #9: Flood affected area. Results of the Conover-Iman analysis suggest two homogeneous symbol subsets when indicating "Flood affected areas" (see Group A and B in Figure 13; for Conover-Iman test results see Supplementary Material Table S11). Symmetric shapes refer to one of the two distinct, homogeneous symbol subset (see Group A in Figure 13), with the Triangle rated as least congruent, followed by Square, Rhomb, Circle, and Star. In contrast, the Asymmetric Star was rated as most congruent with the given map topic, thus, forming a distinct symbol subset of high congruence (see Group B).
Map #10: Melting glaciers. Between the six map symbols used to cartographically indicate "Melting glaciers", Conover-Iman multiple pairwise comparisons identified three symbol subsets of homogenous congruence (see Group A, B, C in Figure 14; for detailed Conover-Iman test results see Supplementary Material Table S12). The first subset comprises Square and Rhomb, characterized by low congruence (see Group A). The second symbol subset was found to be of moderate congruence, comprising Rhomb, Circle, Star, and Asymmetric Star (see Group B). The subset of highest congruency ratings emerged to refer to the Triangular symbol, followed by Asymmetric Star and Star (Group C). Despite the homogeneity within each of the three subsets, Conover-Iman further discloses partial overlaps between some symbols of the three subsets, suggesting some similarity between them (for details see Figure 14). Map #10: Melting glaciers. Between the six map symbols used to cartographically indicate "Melting glaciers", Conover-Iman multiple pairwise comparisons identified three symbol subsets of homogenous congruence (see Group A, B, C in Figure 14; for detailed Conover-Iman test results see Supplementary Material Table S12). The first subset comprises Square and Rhomb, characterized by low congruence (see Group A). The second symbol subset was found to be of moderate congruence, comprising Rhomb, Circle, Star, and Asymmetric Star (see Group B). The subset of highest congruency ratings emerged to refer to the Triangular symbol, followed by Asymmetric Star and Star (Group C). Despite the homogeneity within each of the three subsets, Conover-Iman further discloses partial overlaps between some symbols of the three subsets, suggesting some similarity between them (for details see Figure 14). Map #11: Traffic accident. A Conover-Iman post hoc analysis revealed three subsets of homogeneous symbol-topic congruence for the map topic "Traffic accident" (for detailed test results see Supplementary Material Table S13). Demšar graphs further visualize the subsets' critical differences (see Figure 15), suggesting two subsets of particularly low congruency ratings. One of these subsets uncover the Square as a symbol of least congruence, followed by Rhomb, and Triangle (see Group A). The second subset comprises Rhomb, Triangle, and Circle (see Group B), which can be regarded as fairly similar to Group A (see Figure 15 for details). The subset of highest congruency ratings was found to refer to the Asymmetric Star and symmetric Star (see Group C). Hence, in the  Map #10: Melting glaciers. Between the six map symbols used to cartographically indicate "Melting glaciers", Conover-Iman multiple pairwise comparisons identified three symbol subsets of homogenous congruence (see Group A, B, C in Figure 14; for detailed Conover-Iman test results see Supplementary Material Table S12). The first subset comprises Square and Rhomb, characterized by low congruence (see Group A). The second symbol subset was found to be of moderate congruence, comprising Rhomb, Circle, Star, and Asymmetric Star (see Group B). The subset of highest congruency ratings emerged to refer to the Triangular symbol, followed by Asymmetric Star and Star (Group C). Despite the homogeneity within each of the three subsets, Conover-Iman further discloses partial overlaps between some symbols of the three subsets, suggesting some similarity between them (for details see Figure 14). Map #11: Traffic accident. A Conover-Iman post hoc analysis revealed three subsets of homogeneous symbol-topic congruence for the map topic "Traffic accident" (for detailed test results see Supplementary Material Table S13). Demšar graphs further visualize the subsets' critical differences (see Figure 15), suggesting two subsets of particularly low congruency ratings. One of these subsets uncover the Square as a symbol of least congruence, followed by Rhomb, and Triangle (see Group A). The second subset comprises Rhomb, Triangle, and Circle (see Group B), which can be regarded as fairly similar to Group A (see Figure 15 for details). The subset of highest congruency ratings was found to refer to the Asymmetric Star and symmetric Star (see Group C). Hence, in the Map #11: Traffic accident. A Conover-Iman post hoc analysis revealed three subsets of homogeneous symbol-topic congruence for the map topic "Traffic accident" (for detailed test results see Supplementary Material Table S13). Demšar graphs further visualize the subsets' critical differences (see Figure 15), suggesting two subsets of particularly low congruency ratings. One of these subsets uncover the Square as a symbol of least congruence, followed by Rhomb, and Triangle (see Group A). The second subset comprises Rhomb, Triangle, and Circle (see Group B), which can be regarded as fairly similar to Group A (see Figure 15 for details). The subset of highest congruency ratings was found to refer to the Asymmetric Star and symmetric Star (see Group C). Hence, in the present study, the two-star symbols appeared to be most suitable to cartographically depict traffic accidents.
present study, the two-star symbols appeared to be most suitable to cartographically depict traffic accidents. Durbin-Skillings-Mack revealed no significant differences between the participants' symbol-topic congruency ratings (see Table 5). Pairwise comparisons between the six map symbols based on the Conover-Iman analysis support this finding (for detailed results see Supplementary Material Table  S14). Besides the highest congruency ratings for the Asymmetric Star symbol and lowest congruency scores for Rhomb and Square, these ratings cannot be considered to be significantly different, as statistical analysis showed and which Demšar graphs further illustrate (see Figure 16).

Discussion
As visual means of communication, maps employ signs and symbols to relate to geospatial phenomena and events. As such, maps support the communication and imagination of geographic space and related events from a viewpoint beyond direct experience. At the same time, maps are selective and based on a myriad of decisions. The visual variables employed in maps, such as map signs and symbols, are based on abstractions and generalizations. When using visual variables, cartographers generally strive for congruence, where the schematization of the external representation aims to correspond to structures of the internal representation. Cartography is, Durbin-Skillings-Mack revealed no significant differences between the participants' symbol-topic congruency ratings (see Table 5). Pairwise comparisons between the six map symbols based on the Conover-Iman analysis support this finding (for detailed results see Supplementary Material Table S14). Besides the highest congruency ratings for the Asymmetric Star symbol and lowest congruency scores for Rhomb and Square, these ratings cannot be considered to be significantly different, as statistical analysis showed and which Demšar graphs further illustrate (see Figure 16). present study, the two-star symbols appeared to be most suitable to cartographically depict traffic accidents. Durbin-Skillings-Mack revealed no significant differences between the participants' symbol-topic congruency ratings (see Table 5). Pairwise comparisons between the six map symbols based on the Conover-Iman analysis support this finding (for detailed results see Supplementary Material Table  S14). Besides the highest congruency ratings for the Asymmetric Star symbol and lowest congruency scores for Rhomb and Square, these ratings cannot be considered to be significantly different, as statistical analysis showed and which Demšar graphs further illustrate (see Figure 16).

Discussion
As visual means of communication, maps employ signs and symbols to relate to geospatial phenomena and events. As such, maps support the communication and imagination of geographic space and related events from a viewpoint beyond direct experience. At the same time, maps are selective and based on a myriad of decisions. The visual variables employed in maps, such as map signs and symbols, are based on abstractions and generalizations. When using visual variables, cartographers generally strive for congruence, where the schematization of the external representation aims to correspond to structures of the internal representation. Cartography is,

Discussion
As visual means of communication, maps employ signs and symbols to relate to geospatial phenomena and events. As such, maps support the communication and imagination of geographic space and related events from a viewpoint beyond direct experience. At the same time, maps are selective and based on a myriad of decisions. The visual variables employed in maps, such as map signs and symbols, are based on abstractions and generalizations. When using visual variables, cartographers generally strive for congruence, where the schematization of the external representation aims to correspond to structures of the internal representation. Cartography is, therefore, concerned with establishing a mutually shared set of signs and semiotic rules to communicate geospatial information successfully. Yet, in as far as cartographic semiotics provide some guidance for the selection of visual variables from an analytic perspective, such as when to employ visual variables for a given type of data, it does not address contextual congruence, such as when to use a particular shape symbol for a given type of geospatial content. This research, therefore, aimed to contribute to closing this research gap by investigating the contextual congruence of cartographic point symbols with respect to different types of geospatial information.
In detail, the present study empirically assessed the perceived (in)congruence of a set of six-point symbols in monothematic maps of positive, neutral, and negative valence. The six-point symbols comprised five symmetric shapes (i.e., Circle, Triangle, Square, Rhomb, Star) and one Asymmetric Star shape. In an online survey, 72 maps (i.e., twelve map topics in six symbol variations) were evaluated by 116 participants in a between-groups design. Participants evaluated the congruence between each map topic and its depiction by the different point symbols on a 6-point rating scale. Four positive, four neutral, and four negative map topics were assessed by the participants. Participants' congruency ratings were subjected to statistical analyses. First, symbol-content congruency ratings were analyzed on an aggregated level, i.e., by the three positive, neutral, and negative topic clusters. Second, symbol-content congruency ratings were analyzed individually, i.e., by each of the twelve map topics.
The clustered congruency ratings showed that Asymmetric Stars were rated to be least suitable to cartographically represent positive content. In contrast, the five symmetric shapes were generally found to be of high congruence with positive map topics. Congruency ratings for neutral topics revealed similar results, such as Asymmetric Stars were perceived least congruent, while polygonal, symmetric shapes were found to correspond significantly better with neutral map content. Results further disclosed a significant preference for Circular map symbols for depicting neutral map topics. Hence, results suggest that neutral topics may best be represented by Circular point symbols, while symmetric polygonal map symbols may be considered as second-best alternatives. Whilst incongruent with both positive and neutral topics, Asymmetric Star symbols were significantly preferred for depicting negative events. In contrast, Star, Circle, and Triangle were generally rated as moderately congruent with negative topics, while Square and Rhomb were experienced as least congruent for cartographically referring to negative events.
While the present findings on a cluster-level propose general tendencies of symbol-topic relations, some topic-specific visualization preferences were revealed at a more detailed level of analysis. For example, Triangles were rated to best correspond to representing "National parks" and "Melting glaciers", whilst symmetric Stars showed to be the preferred choice for representing "Cities awarded for clean energy" and "Urban parks with high biodiversity". For depicting "Traffic accidents", the Asymmetric Star and symmetric Star were both found to be significantly favored. These findings may imply particular preferences towards iconic map symbols. Iconicity refers to the conceived similarity or analogy between the form of a sign-vehicle and its meaning. Abstract, geometric shapes are generally considered to be of low iconicity, whilst associative and pictorial point symbols are regarded as moderately to highly iconic [32]. The results of the present study indicate that the iconicity of abstract map symbols may have been amplified by the cartographic context. The topic of "Melting glaciers", for example, may have been associated with mountains, which may be visually associated and abstracted by triangular shapes. Traffic accidents, on the other hand, may be associated with collision and conflict, often graphically abstracted by star-like shapes. In the present study, contextually more iconic symbols may have been experienced as more congruent with the given map topics, as participants' draw towards those map symbols indicate. Yet, as the present findings only allow to conjecture the influence of iconicity on congruence, future research is needed to systematically and empirically explore such an influence in depth.
Related research further suggests that when people make evaluative judgments, they draw on a range of different processes [20]. Such processes may relate to cognitive as well as to affective responses [18]. Winkielman et al. even stress that some affective response is always involved in evaluative judgments. These affective responses may be based on stimulus attributes as well as be caused by the dynamics of information processing, i.e., by processing fluency [20].
Processing fluency is considered to be hedonically marked, such as that high fluency elicits positive affective reactions [20]. A tendency towards affectively congruent map symbols may also be supported by the findings of this study. Negative map topics in the present study referred to rather dynamic spatiotemporal events (e.g., smog, floods, traffic accident), where in contrast, positive and neutral topics predominantly related to rather static geospatial occurrences (e.g., national parks, biodiversity, art galleries, lookout tower). These different qualities may have influenced participants' congruence judgments, such as towards favoring symbols of dynamic character for dynamic topics. Osgood et al., for example, referred to such dynamic qualities of stimuli as the affective dimension of activity, for which he found "some relation to physical sharpness or abruptness" [31] (p. 38). Such particular stimulus-affect correspondences are moreover supported by related empirical research on map symbols, which found that asymmetric star symbols were experienced as highly dynamic and of negative valence, whereas symmetric shapes were regarded as neutral or slightly positive and of rather static quality [44]. Related research further showed that point symbols may be experienced visually, associatively, and affectively [43]. Map topic-specific visualization preferences may therefore result due to either of these dimensions. Hence, while the present findings and related literature strongly suggest that the principle of congruence may be a crucial factor for cartographic communication, such as to enhance map design, future research is needed to dissect its dimensions and explore its effects in greater detail.
So far, the present research provides first empirical insights into symbol-content relations of such kind. Yet, the study also encompasses some shortcomings: The present study used a limited set of six map symbols. Besides the attempt to incorporate the most common abstract cartographic point symbols, the possible number of map symbols is sheer unlimited. At the same time, cartographic nominal point symbols used in practice may not only be manipulated by shape but also by color, size, and other qualities. Such symbol properties have been excluded deliberately from the present study in order to allow for a systematic study on the influence of shape by controlling for potential confounders, such as different symbol colors and symbol size. Future research may, however, expand on the set of point symbols by considering additional manipulable symbol qualities. Future studies may also consider increasing the complexity and the interaction of different visual map features, as the findings of the present study, are limited to monothematic maps of low visual complexity. In addition, the number of geospatial topics used in the present study was limited to a selection of twelve topics, which certainly limits the findings' transferability. Hence, despite a greater variety of point symbols and thematic maps of differing visual complexity, a larger set of map topics must also be considered in future studies to expand upon the present findings and to ensure the findings' external validity.
Another shortcoming refers to the selection of map topics, which was undertaken a priori by the experimenter. Topics were selected to be of positive, neutral, and negative character. Whilst statistical analysis confirmed such three homogenous clusters, results further showed a more proximal relation between the neutral and positive topic clusters with a greater distance towards the negative cluster. Mean ratings further showed that neutral topics in the present study were rated slightly positive and may, therefore, not be considered to be purely neutral. The selection of maps towards a set of distinctively positive, neutral, and negative geospatial topics may be improved in future studies, such as by conducting prestudies for prior topic assessments.
Future research also needs to take into account more heterogeneous groups of participants, as this study relied on students only. Participants of the present study can be characterized as young adults, with high education, and a moderate to high affinity to graphic design and cartography. The findings of the present study must, therefore, be considered to be restricted to this fairly homogeneous user group as the study did not embrace the vast diversity of map users. Social semiotics further stress that "people respond culturally and engage emotionally with data and their visualizations" [54] (p. 50). This is supported by research in cartography, which found that cultural backgrounds influence map perception and performance in cartographic visual search tasks [42]. Such findings strongly imply the impact of cultural sign-conventions on map reading performances. It is, therefore crucial to continue exploring the effects of cartographic design decisions from the perspectives and backgrounds of different user groups.

Conclusions
Maps are a form of communication: a means of conceiving, articulating, and structuring the human world [3]. Through cartographic signs and symbols, maps refer to phenomena and entities in geographic space in a generalized and abstracted way. To communicate geospatial information successfully, cartography generally strives for cognitively congruent map design. Cartographic research has, however, only started to understand the myriad of relations between the map as an external representation and internal human processes.
This research, therefore, drew attention to the principle of contextual congruence and empirically explored the cognitive correspondences between cartographic point symbols and different positive, neutral, and negative geospatial contexts. Findings strongly imply that context matters. Overall results revealed high congruences of Asymmetric Star symbols when representing negative map topics, whilst highly incongruent when depicting positive or neutral geospatial content. Symmetric shapes, on the other hand, generally emerged to highly correspond with positive and neutral map topics, whilst incongruent with negative content. As the meaning of point symbols showed to be susceptible to context, the findings lead to the conclusion that cognitive congruence can be enhanced by integrating context-specific information into the cartographic design, such as when form follows content.
Yet, while this research provided first empirical insights, the present findings aim to be expanded upon. Future research is needed to continue exploring the different semiotic dimensions of sign meanings and their role in cartographic communication processes. A profound understanding of the various context-related communication effects of cartographic signs may allow for more deliberate and informed design decisions, and with this, may contribute to enhancing cartographic communication.