gms | German Medical Science

Pictorial material is one of the oldest human ways of expressing inner psychic dynamics and social processes. Cave painting and engraving from the Early Aurignacian period, are commonly cited as evidence for this human artistic activity [1]. Over time, artistic activity developed from simplicity to technical and graphical complexity. Today, pictorial material is a medium of expression in a vast array of areas, including culture, politics, sociology and psychology. Indeed, pictorial material forms an important topic of psychological research in the fields of episodic memory [2], [3], perception tasks [4] , [5], [6] and non-verbal understanding of language [7].

However, pictorial material as part of psychological assessments and projective tests, remains ambiguous. These tests are summarized as a group of tests “which derive an indirect and global assessment of personality through the analysis of meaning or structure freely imposed by the subject upon unstructured or ambiguous materials” [8]. This material is supposed to project inner feelings, conflicts or interests in the outer structure, interpretation or design. Following the taxonomy of Lindzey, outer forms of expression can be examined by construction techniques, association techniques, completion methods, arrangement or selection methods and expression techniques. Such test procedures have a far-reaching significance in the interdisciplinary collaboration between psychology and art therapy, and numerous modules for evaluating pictorial material exist which have been summarized in [9].

As a special area of projective tests, figure drawings such as the House Tree Person Test (HTP) or the Tree Drawing Test (TDT) are essential parts of assessment processes in the field of clinical psychology [10]. For instance, the HTP shows moderate to good validity in diagnosis of forensic psychological items, such as psychiatry hospitalization [11], or in the diagnosis of child abuse [12] or in the diagnosis of attention-deficit hyperactivity disorder (ADHD) [13]. Likewise, the TDT has served as a diagnostic tool for detecting chronic diseases, dementia, schizophrenia or post-traumatic stress disorder [14]. Further fundamental aspects of projective methods can be found in [15] or [16].

For this study, an innovative interpretation of drawing is considered relating to fundamental aspects of projective testing: Since nature and humans are in an aesthetic relationship [17], the cultivation and forming of a landscape and thus the creation of a “landscape art” is considered to be a special type of landscape “drawing process” [18]. Materials such as plants, bricks or wood are “drawn” and arranged in a plain landscape canvas. In this context, a new kind of landscape picture is created. This aligns with Encyclopædia Britannica, where gardening is defined as “the laying out and care of a plot of ground devoted partially or wholly to the growing of plants such as flowers, herbs, or vegetables”. Encyclopædia Britannica concludes that “gardening can be considered both as an art, concerned with arranging plants harmoniously in their surroundings, and as a science, encompassing the principles and techniques of plant cultivation.” [19]. This idea has been integrated in art therapy, where Land Art is described as a creative therapeutic process in a public space with natural materials, for example in the therapy garden of a hospital [20]. Further aspects of nature in psychotherapy and arts therapies are described in detail in [21] and [22] where the holistic and environmental friendly therapeutic approach is demonstrated in various examples.

Apart from the therapeutic aspect, the aesthetic approach to landscape and gardens is also emphasized in landscape architecture or landscape design. Here, aesthetic sensibility both in composition, production and in the consideration of landscape design is a principle that makes landscape in its interaction with nature and multimedia components a work of art [23]. It is assumed that art triggers aesthetic sensation equal to a fundamental basic emotion in human beings [24]. Thus, this kind of creative landscaping process and its outcome (the final landscape picture) might be used for diagnostic processes[25] .

In this study, we present a picture ranking test approach using the photographs of the final art works. In psychological literature, picture ranking tests are a recognized method of collecting intuitive knowledge and are particularly suitable for visual items, compared to card ranking test which use written statements or symbols [26], [27]. Picture ranking tests were first introduced in 1935 [28] as a categorize facial expressions. Due to its low-threshold approach and its non-verbal character the picture ranking test has gained increased attention in psychology and the social sciences [29]. Currently, ranking tasks are also used in perceptual psychology, such as food preferences [30] or typicality ranking tasks in children, to assess their judgements or implicit knowledge when verbal expression is not as developed as visual perception or ability to judge based on other sensory experiences such as audio [31], [32].

In this study, the picture ranking test is used as an external assessment, during which experts rate the mental state of the drawer.

This pilot study aims at investigating the diagnostic potential of diverse pictorial material (photographs), e.g. of a landscape modeling, using a picture ranking task approach.

The following hypotheses were tested:

1.

Raters of art therapy/theater education or psychology/psychotherapy or horticulture/landscaping can make gradual assessments of a person’s condition based on a) photographs of the garden artwork, b) photographs of tree drawings, and c) movement trajectories.

2.

Raters of art therapy/theater education make a more precise statement about the condition of a person than raters of horticulture/landscaping or psychology/psychotherapy.

3.

There is a gender effect according the picture ranking task, especially with ranking of movement trajectories.

15 participants (8 women, none diverse) aged 19 and 60 took part in this pilot study. All participants were members or acquaintances of the Witten/Herdecke University (e.g. staff, students). Psychology students received credit points for participating. Detailed information about basic demographic data of the participants is shown in Table 1 [Tab. 1].

Prior to the gardening and drawing tasks, all participants were asked to complete a questionnaire including demographic categories such as age, and gender, as well as creativity and horticultural skills assessed on a 10-point visual analogue scale. Moreover, the test battery was completed by the Brief Symptom Inventory BSI-18 [33], [34] to detect the current mental state (expressed by the global severity index GSI as a sum value), and the State-Trait Anxiety Inventory (here state version STAI-S) [35] to detect the current anxiety level, expressed in STAI-S total score. The BSI-18 show satisfactory to high internal consistencies across representative German samples with a Cronbach’s alpha of 0.88 for students, 0.82 for a non-clinical sample, 0.90 for a clinical sample with somatoform disorders, 0.88 for a clinical sample with anxiety, and 0.89 for a clinical sample with depression [34]. Likewise, the STAI-S shows high consistencies with a Cronbach's alpha between 0.88 and 0.94 for patients with anxious or depressive symptoms [36].

Afterwards, all participants were asked to complete the Tree Drawing Test according to Karl Koch [37] on a 40x40 cm white paper with a pencil of hardness 4B and a rubber. After completing these two tasks, participants were asked to do a half-hour landscaping modeling task in the outdoor area. Instructions were given to “create a garden from the given material in the assigned area”. Participants were instructed to start from an identical point of view and give a sign when they had finished the task. They were asked to finish the task with a three minute reminder before the time limit of 30 minutes. Both the tree drawings and the landscaping artwork were photographed with a commercially available camera from an identical point of view (tree drawings – top view; landscaping artworks – 45° from participant’s starting point).

The picture ranking test consists of three dimensions:

1.

colored photographs of the garden artwork (see Figure 1 [Fig. 1]),

2.

black-and-white photographs of the tree drawing (see Figure 2 [Fig. 2]), and

3.

colored spatial movement trajectories of the participants (see Figure 3 [Fig. 3]).

Movement trajectories were traced manually by transferring the observed movement to a blank DIN A4 sheet of paper printed with a square in the middle (representing the corner of the sandpit) while watching the video. Therefore, a psychology student (female) was asked, to watch the videos on a notebook via VLC Player in real-time speed. The task was, to observe the person’s movement trajectories based on the position of the persons ankle and copy them on the sheet of paper. Breaks or interruptions were marked with a point. Because of overlapping movement trajectories, the person was asked to take a new blanked sheet of paper every 10^th minute. The position at beginning and finishing the task was marked with a cross. The papers were then scanned and placed on top of one another using Photoshop software. This is how the final image of the movement patterns was created, which was used for the picture ranking test. All pictures were printed on DIN A4 cardboard and laminated for protection.

Five raters each from the three areas of a) psychology/psychotherapy, b) art therapy/theater education, and c) horticulture/landscaping performed the picture ranking test. The raters were acquired from the Witten/Herdecke University, the University of Applied Sciences and Arts, Ottersberg and the Technical College for Horticulture, Essen via social media and personal contact.

The three categories of the picture ranking test were sorted according to a reference system from “0=healthy” to “15=mentally stressed”. This reference is based on the GSI, which was corrected by the STAI-S total sum value (GSI_corr.) to avoid ranks ties, so that each participant occupies their own rank. Participants were asked to “sort the pictures on a scale from ‘0=healthy’ to ‘15=mentally stressed’”. First, raters ranked the photographs of the garden artwork, followed by the photographs of the tree drawings and finally the movement trajectories. All ranking material was presented in a randomized order. Raters were told to judge how good the items fit the ranking and not to base their judgement of how much they liked the photographs. As soon as the rater finished, the person gave a sign. There was no time limit for finishing the task. After each completed ranking, the pictures were collected and the next ranking started with the new items of the next category. It was therefore a closed process with simple implementation and without specific direction following the instructions given in the typicality ranking by Djalal and colleagues [32]. However, the time for finishing the task was not registered, nor were the number of corrections to the element’s position.

Data was analysed by using SPSS Statistics 26. The average rating (MEAN) of the raters of the three professions was calculated for each category of the picture ranking test. Based on these averaged ratings, Spearman’s correlation coefficient between raters’ judgments and GSI_corr. was calculated to investigate the agreement of all raters as well as the average rating of all five raters of each professional area. To determine the rater-measure agreement, the intra-class correlation coefficient (ICC) was calculated. The ICCs of the respective rater groups and all rater groups were calculated for each of the three picture ranking categories. To explore the difficulty level of the ranking task, a binary variable of each expert’s ranking was defined by dividing their ranking list into two categories (“1=healthy”, “0=mentally stressed”) using a median-split of the ranks with the cut-off-score of 8. Inter-rater reliability on this data was calculated using Cohen’s Kappa coefficient. Finally, a multiple linear regression was used to predict the original ranking based on the GSI_corr. including the independent variables of the three rater’s profession, average ratings, and gender of the pilot study participants. The level of significance was α=0.05.

The research project was approved by the Ethics Committee of the Nürtingen-Geislingen University. All participants and all raters gave a written informed declaration of consent prior to admission.

A total of 15 raters took part in the picture ranking test. Each rater assigned themself to one of three professions: a) psychology/psychotherapy, b) art therapy/theater education, and c) horticulture/landscaping. Each group consisted out of five members.

The raters of group a) (5 male) on average were 31.4 years old (SD=14.89) and had an average of 8 years of professional experience (SD=15.34). Three of the raters were studying for a master’s degree, one rater was training to be a psychotherapist and one rater was self-employed as a psychotherapist. Both personal gardening skills (=2.4, min=2, max=3) and personal artistic skills (=4.6, min=3, max=6) were rated by the raters in the lower average range.

The raters of group b) (5 women) on average were 53.4 years old (SD=3,507) and had an average of 19.0 years of professional experience (SD=8.19). Two of the raters were self-employed, and three of them were employed. All raters had completed a university degree in art therapy at least at bachelor level. The raters rated their personal gardening skills (=6.0, min=4, max=9) in the average range and their personal artistic abilities (=7.6, min=6, max=9) in the upper average range.

The raters of group c) (1 female, 4 male) on average were 23.4 years old (SD=.548) and had an average of 5 years of professional experience (SD=1). All raters completed one year of further education after completing practical vocational training. Both the personal gardening skills (=6.6, min=5, max=9) and the personal artistic skills (=5.0, min=3, max=9) were assessed by the raters in the average range.

Ranking of garden artwork, Table 2 [Tab. 2], shows the rater’s individual measures of agreement with regard to the three image categories, sorted by profession. Results indicate high inconsistencies in all raters from the three professions. However, while no rater of group c) showed significant correlations, high to very high correlation was observed in three raters of group b) – ranking the garden artwork. One rater (SH14) showed high correlations in the ranking of all three picture categories while another rater (UO25) did not correlate in any of the ranking with the true ranking. In group a), three raters showed high correlations with the tree drawings, and one rater (SH09) showed high correlations in the assignment of the other two picture categories.

Two raters (SH09 and SH14) show high levels of agreement both within their own rating and with regard to the true ranking list. They are therefore described in more details below in order to identify similarities.

Rater SH09 is a 26 years old male psychology student in a master’s course. He has no previous professional experience and rates his personal horticultural skills with 3/10 and his personal artistic skills with 4/10.

Rater SH14 is a 49 years old female art therapist with 20 years of professional experience. She rates her personal horticultural skills a 4/10 and her personal artistic skills a 7/10.

For reference, the complete ranking results of the most successful rater SH09 and the least successful rater SE14 with regard to the true ranking (GSI_corr) divided by the categories is provided in the supplementary material (see Attachment 1 [Attach. 1]).

Table 3 [Tab. 3] shows intra-class coefficients for testing the rater-measure agreement. A cut-off according to [38] was set to r>.5, which corresponds to an average agreement. A total of two raters each from group a) psychology/psychotherapy (LB07, SH09) and group b) art therapy/theater (SH14, UO25) achieve this value (see Table 3 [Tab. 3]). The low level of agreement both on individual measurement as well as on average measurement raises the question of the difficulty of the task.

Table 4 [Tab. 4] shows the rater-measure agreement according to Cohen’s Kappa. Following Landis and Koch [39], results show moderate and substantial correlations with respect to the ranking of garden artwork by raters of group b) art therapy and c) landscape design. Furthermore, results show moderate agreement with respect to the ranking of tree drawings by raters of group a) psychology/psychotherapy. There are no significant agreements with respect to the ranking of movement trajectories.

Figure 4 [Fig. 4], Figure 5 [Fig. 5], and Figure 6 [Fig. 6] show all ratings for the pictorial material according to the true ranking list (GSI_corr.), in addition to gender-specific regression lines. For the garden artworks, the regressions show an explained variance of R²=.164 for women and R²=.505 for men. For the tree drawings, the explained variance is R²=.162 for women and R²=.161 for men. For the movement trajectories the explained variance is R²=.040 for women and R²=.169 for men.

Table 5 [Tab. 5] shows three different multiple linear regressions with the original ranking list (GSI_corr.) as the independent variable, and the average rating by each profession (MEAN_profession) and sex of the pilot study participants as the predictors. Regressions were calculated according to the categories of the pictorial material. With respect to the ranking of the garden artwork, results show a significant impact of group b) art therapy. Regarding the ranking of the tree drawings, results show a significant impact of group a) psychology/psychotherapy. Regarding the ranking of the movement trajectories, results show a significant impact of group a) and group b). Neither predictor group c) horticulture/landscaping, nor gender, showed significant impact of any ranking task.

The aim of this study was to examine the possibility of using the picture ranking test at hand to detect psychological stress levels in terms of feasibility and reliability. The results indicate that a picture ranking test might be of use in the field of clinical-psychological diagnostics. However, data at present is too heterogeneous to draw final conclusions and further research is needed to strengthen the results and the value of the ranking test.

With respect to the first hypothesis, this study shows that raters of psychology/psychotherapy or art therapy/theater education or horticulture/landscaping can make gradual assessments of a person’s condition based on a) photographs of the garden artwork, b) photographs of tree drawings and c) movement trajectories. For all surveyed raters the picture ranking task was a manageable task. However, not all expert groups came to valid results.

In support of the second hypothesis, this study found some evidence that raters of art therapy/theater education can use a) photographs of garden artwork, b) photographs of tree drawings and c) movement trajectories to make more precise statements about the condition of a person than raters of horticulture/landscaping or psychology/psychotherapy. At the same time, raters of psychology/psychotherapy did the tasks similarly well while raters of horticulture/landscaping performed poorly in comparison, even when the analysis was simplified using the median split.

Considering the third hypothesis, a gender effect could not be replicated at the level of the outcome evaluation. This is somehow contradictory to another analysis with data from this pilot study where spatial movement process was analyzed [40], [41]. In these analyses, results indicated a negative correlation between spatial movement and mental state for female participants and a positive correlation for male participants. In conclusion, women express mental stress with increased spatial movement activity whilst men restrict movement activity. However, this behavior was not visible in the final pictures of the ranking test applied here. Nevertheless, the results of this study contribute to the scientific call [42] to examine the gender differences in gardening behavior.

Two raters (one from art therapy/theater education and one from psychology/psychotherapy) reached an acceptable rater-measure agreement. Results revealed they showed both high correlation with all three categories of pictorial material as well as high ICCs. Both showed a heterogeneous background according to demographic data such as gender or professional experiences. Thus and in accordance to [43] it remains unclear, if and under which conditions , reliable ranking results can be expected. At this pilot stage, it would be unreasonable to pose indications especially for practical clinical implementation. Furthermore, regression analysis showed an impact of the correct ranking by the profession psychology/psychotherapy in the tree drawings, while in the garden artwork, regressions show a positive impact of the profession art therapy/theater education. Two professions (art therapy/theater education and psychology/psychotherapy) do have an impact on the assessment of movement trajectories. In comparison, raters from horticulture/landscaping did not show significant performance in any of the three pictorial categories. None of the raters showed a significant match with the ranking even within their own rating. Statistical values did improve by simplifying the analysis (as it is displayed in the Kappa coefficients), but only in the category garden artwork, which is well-known material for these raters.

Results indicate that raters predicted the picture owners’ mental state by assessing pictorial material best that belongs to their profession: Experts of psychology/psychotherapy performed well with projective tests such as the TDT as did experts of art therapy/theater education with drawn landscapes. This indicated an ability to transfer their disorder-specific and therapeutic knowledge. Likewise, experts of art therapy/theater education are familiar with artwork analysis. These raters have effectively developed a diagnostic view, even at a students’ level.

Given these findings, experienced garden therapists might show better performance in the ranking task due to their specific therapeutic knowledge in the field of landscape and garden design. Furthermore, experts in the field of movement therapy, dance therapy or eurhythmy may have a high degree of agreement in the ranking of photographs of movement trajectories. Semiotic, and perceptional aspects, (e.g. the aesthetic response), delight the interpretation of a picture and should therefore be considered in future. However, it would be beneficial for students of art therapy/theater education to come into contact with projective tests during their studies.

In a further evaluation, video recordings of the gardening task process were analyzed by experts from the three professions above. Subsequent qualitative data analysis showed four preliminary categories for assessing a person’s mental stress level: movement patterns, attachment with the task, selection of material and hesitation/interruptions [44]. Once again, the difference between work analysis and process analysis should be explicitly pointed out. These considerations emphasize the multimedia aspect of this pilot study. In the clinical context, it also makes sense in terms of the multi-professions of a clinical-therapeutic team to include multimodal tests and evaluation instruments in the diagnostic process to produce a more holistic assessment of a person.

Interestingly, photographs of the garden artwork were best ranked by raters of art therapy/theater education. Being a rater of this group was the only significant predictor in the regression model. Once more, affiliation towards a certain pictorial material can be assumed. At the individual level, there were three raters of art therapy/theater education and one rater of psychology/psychotherapy who completed the ranking close to the original ranking list. Raters of art therapy/theater education and raters of horticulture/landscaping might benefit from making the task easier in terms of sorting the photographs into two categories as it was shown in the analysis of the kappa coefficients.

Though, being a rater of horticulture/landscaping is no predictor for ranking pictorial material not even for photographs of garden artwork.

In Figure 4 [Fig. 4] a gender effect could be assumed but in the multiple regression analysis the effect was not present. In the evaluation of the artwork as an end product of a gardening task the gender effect found in literature [45] could not be verified. At this point process evaluation and outcome evaluation clearly differ. In literature, women are more likely to plant ornamentals than men, who prefer to grow crop plants [44]. However, as a limitation, it is difficult to compare plant richness in real-world gardens with an experimental garden setting. In particular, material, time and attitude towards designing a garden for a small period of time (i.e. 30 min) instead of creating a garden for at least one season, is a reasonable bias to recognize when interpreting these results.

Raters of psychology/psychotherapy achieved the best results in assessing tree drawings. Average rating level, as well as individual level, showed high correlations to the mental state of the painters. Especially rater SH09 who fits the original ranking best with this pictorial material. Regression for the tree drawing ranking task showed an impact only when being a rater of psychology/psychotherapy. This result was stable when the difficulty level was reduced notionally to a binary outcome. As the TDT is well known in psychology, it is appropriate to assume an affinity towards this kind of drawing or test in the rater of psychology/psychotherapy.

Movement trajectories do have a weak impact on the ranking task. Although the raters were able to complete this task, it seems that the pictorial material of the movement trajectories is quite abstract and hard to access. Even the simplification of the ranking task towards a binary outcome did not improve the raters ranking skills. Yet, the regression shows a significant impact of raters’ profession regarding art therapy/theater education and psychology/psychotherapy.

In contrast to results here, digital assessed movement trajectories over time are more informative for clinical diagnostic as was shown in [40] Presumably, the higher precision of generated movement trajectories and the underlying numerical data could show that on the process-level men had less spatial movement activity in the case of psychological stress than women [40]. Other covariates, (e.g. age), showed no significant effects in multivariate analysis in this study. Compared to the approach in [40], [41] the current design does not involve a direct and objective measurements of the movement patterns, and thus may lead to inconclusive results.

Although the picture ranking task shows clear results there are some limitations. First of all, picture ranking relies on the background and expertise of the raters [46].

To get a clearer impression on the impact of expertise, it would be desirable to include more than five raters for each profession. This could increase the variance of demographic aspects and work experience.

From a technical point of view, the pictorial material (especially the garden artwork) lacks high quality image format and resolution. Thus, important details may have been lost, which might have led to different rating results. Future research should also investigate more advanced approaches to create movement trajectories [40], [41]. Detailed information about how the test material was constructed, could improve rater’s judgement. This also represents an approach closer to a clinical situation with more background knowledge about a person.

From the methodological part, the complexity of ranking 15 pictures simultaneously was quite high with a total of 15!=1,307,674,368,000 possible combinations. Thus and in accordance to the recommendations of [47] stepwise binary ranking or a free ranking task into i.e. healthy vs. non healthy might produce more robust results in future research. Although this study attempted to simplify the ranking task to a binary outcome, the underlying ranking task of the complete pictures might be the reason, why no improvements were observed in this part of the analysis.

Independent of the type of ranking task, the reasons why participants decided for a specific ranking should be obtained to get a clearer picture of the underlying decision processes and this is also recommended by [27]. In addition, the time needed for ranking each of the fifteen pictures should be considered. In a similar task in [48] it took a “striking” longer time for untrained participants to place the first picture compared to the time of trained participants.

Although the potential for bias in this study has to be take into account as our observations could only cover certain aspects, our study adds to the existing literature on approaches to assess human behavior in nature and landscape, and builds bridges between research in art therapy and landscape design research [17], [49]. In particular, the combination of qualitative and quantitative aspects by means of a picture-ranking approach and a quantitative analysis of rater-measure agreement, might serve as a promising basis for further research [50]. Future research should, however, focus on improving the ranking test and determining the characteristics of successful raters. If applied to patients with mental disorders, comprehensive diagnostics should also be carried out alongside the process and outcome evaluation.

The authors declare that they have no competing interests.

Christina Niedermann was funded by the Andreas Tobias Kind Foundation (funding number: 2017/08) for the implementation of the pilot study and received a graduate scholarship from the University of Applied Sciences and Arts, Ottersberg. The Scholarship was part of the research focus Artistic Interventions in Health Promotion and Prevention funded by the Volkswagen Foundation and by the Lower Saxony Ministry for Science and Culture.

A big thank you goes to Linda Lösser and Marvin Beringer for actively supporting this pilot study at the Witten/Herdecke University as part of their studies in psychology, to Bethan Fox for proofreading the article, and to all the raters who contributed to the results of this study.