This study referred to the search results of the aforementioned review studies ^{13 – 20} to search journal articles and academic books (written in English) on empirical intervention studies that used visual arts for cultivating observation skills in health professionals’ education and training. Manual searches for additional articles were also conducted because the search results were inconsistent; moreover, it was difficult to collect information from databases by using typical search words such as “visual arts,” “observation skills,” and “medical education.”

Detailed inclusion and exclusion criteria to assess the results of the interventions in which only appreciating visual arts affected observation skills were set, comprising those that used only visual arts to foster observation skills. Therefore, studies that included other activities were excluded, except for pre- and post-tests, such as drawing or creating visual arts and observing medical images. Additionally, studies on “graphic medicine” ²⁴ were also excluded, because these studies used comics that included written dialogues.

Moreover, studies that objectively measured observation skills were included, such as using psychometric scales, quantitative metrics, or grading participants’ statements by predetermined criteria. Thus, studies that subjectively measured observation skills, irrespective of the method used (e.g., participants’ subjective evaluation of their skills were excluded, their free comments on the interventions, and scales of their impressions of the interventions). For example, studies that extracted key concepts from the participants’ comments, such as thematic analysis ²⁵ were excluded because they are not suitable for determining the learning effects for the entire concerned population since the concepts are identified from the comments of only a few participants.

The full texts of the related articles identified by the search results of the aforementioned review studies ^{13 – 20} were collected. Through manual search, the titles of articles that seemed to use visual arts appreciation in health professionals’ education were selected by checking reference lists and the articles listed in PubMed’s sidebar or Google’s search result of related articles. Then, their abstracts were reviewed via websites, such as PubMed and Google. If the abstract met the inclusion criteria, the full-text article was collected. Additionally, if the abstract could not be obtained or was insufficient to determine whether the article met the inclusion criteria, the full-text article was collected. Therefore, if it could be clearly determined from the abstracts that the inclusion criteria were not met, full-text articles were not collected. This process was repeated until no more related articles could be found. I conducted this search from January to June 2022.

After a review of the full texts that were collected, studies that met the inclusion criteria were selected. An exhaustive summary of each study was collated. The summary described the features, research design, and key results of the study (see Table 1).

On the basis of the summary, the features and tendencies were synthesized into the interventions’ goals and methods, research design, and findings of the studies. Then, the features and tendencies were examined in relation to the findings of the studies on expertization of medical diagnostic skills and visual perception, according to review articles on related themes as well as the articles discussed in those reviews. Through the examination, the aforementioned review studies ^{13 – 20} and relevant articles that were collected but excluded in this study were also referenced.

Most interventions instructed participants to carefully observe the entire target. As stated by Agarwal et al., ²⁶ taking time to observe is considered a sign of careful observation. However, observing for a longer period is not necessary for an accurate diagnosis. Experts take less time than novices to diagnose accurately. In contrast, they are more likely to misdiagnose if they take longer. ^{44 – 46}

Moreover, systematic viewing intervention, such as detecting radiograph abnormalities by viewing the images in a given order, or full-coverage viewing intervention, such as mentally dividing each image into nine imaginary segments (3×3) and inspecting each segment separately, did not outperform non-systematic viewing intervention, wherein participants were urged to inspect whatever attracted their attention. ⁴⁷ Systematic and full-coverage viewing took a longer (but not significant) time than non-systematic viewing. Van Geel et al. ⁴⁸ also reported similar results.

Many studies ^{44 , 49 – 51} and reviews ^{52 – 58} on medical image perception have shown that experts’ diagnosis is generally characterized by holistic and quick recognition. This is due to pattern recognition ^{59 – 61} or intuition composed of scripts, schemas, or mental models/images/representations (hereinafter collectively referred to as mental images). ⁶² Experts compare medical images or patients with mental images—which are associated with domain-specific knowledge—of normal and abnormal cases at a single glance. ^{23 , 63 – 65} A study using fMRI ⁶⁶ revealed that when radiologists looked at radiographs, the brain regions related to visual attention and the encoding, storing, and retrieval of visual memory were activated.

Not just medical images, humans generally recognize things by associating them with mental images formed through our daily visual experiences. ⁶⁷ For example, humans can recognize a scene’s gist and gaze into areas of interest through their knowledge of the world. ^{68 – 70} This top-down control of gaze by using mental images is common for humans and is not limited to those who diagnose medical images.

The experts’ intuition is cultivated through the accumulation of domain-specific knowledge and experiences. The number of mammograms observed is proportional to the accuracy of the diagnosis. ⁷¹ Thus, it is not conducive if the intervention studies do not consider the participants’ pre-existing knowledge in measuring observation skills.

For encouraging persistent and detailed observation, some interventions ^{33 , 35 , 72} (included in this article and others) are intended to cultivate tolerance for ambiguity. An increase in speculative thinking—which was observed by Jasani and Saks ³⁴ —implies such cultivation.

However, the intervention studies ^{35 , 72} showed that using visual arts contributes very little or does not contribute toward building tolerance for ambiguity.

The intervention studies enabled the participants to describe the visual features in the paintings and medical images in greater detail. However, the observation and description of more visual features or signs in medical images or patient photographs do not lead to accurate diagnoses for novices and experts. Novices, such as medical school students, generate hypotheses based on constant information input and are unable to select appropriate hypotheses. ⁷³ An experiment that presented electrocardiograms to non-medical laypersons revealed that participants usually misdiagnosed the problem because they were unable to overlook irrelevant features. ⁷⁴

Not just in the context of medical images, experts generally diagnose and make decisions based on mental images. ⁷⁵ They can extract more useful information about the given situation, immediately generate appropriate hypotheses, eliminate irrelevant hypotheses, and test the hypotheses. ^{62 , 63 , 76} Thus, expert general practitioners can make a correct diagnosis without discovering all the features of a disease. ⁶⁰

Goodman and Kelleher ²⁹ asserted that reducing misdiagnoses is one of the benefits of visual training, such as appreciating visual arts. Though they did provide an explicit reason for why visual training would reduce misdiagnoses, it can be asserted that misdiagnoses arise from careless observation, and appreciating visual arts can teach one to look carefully.

However, the causes are complex ⁷⁷ and may not always be due to mere carelessness. In general, medical errors can be classified into search errors caused by not looking at the abnormality, recognition errors caused by not identifying an abnormality, and decision errors that occur in the process of clinical reasoning. ⁷⁸

In the context of novices, Brunyé et al.’s ⁷⁸ review showed that search errors generally occur when novices overlook definitive features with subtle visual features and look repeatedly at areas that are not relevant for a diagnosis. Misdiagnoses is also called when signs of extremely low prevalence are overlooked by novices (and experts). Novices are also likely to make recognition errors due to insufficient mental images and knowledge. ^{49 , 65} This implies that novices’ errors are caused by the lack of top-down control of gaze due to insufficient mental images.

On the other hand, experts can diagnose quickly and accurately by using mental images. Mental images are compromises between fitting the data and minimizing the complexity of the model, ⁶⁷ which leads experts’ perceptions toward certain classes of stimuli in the domain. ²³ Therefore, many expert radiologists overlook the gorilla inserted into chest CTs. ⁷⁹ This demonstrates that experts’ errors, such as premature closure, may be a byproduct of expertization. ^{23 , 80}

Furthermore, human and environmental factors that induce decision errors cannot be ignored. ⁸¹ Therefore, cultivation of novices’ mental images, organizational/systematic support or use of cognitive tools/strategies ⁸² may be more effective than appreciating visual arts to reduce medical errors.

The intervention studies intended to transfer observation skills for the visual arts to apply to medical examination, based on the assumption that there are similarities between these fields. However, the transfer occurs only to a very limited extent. Experimental studies showed that in terms of radiology, expertise was positively transferred to the task of finding small low-contrast dots in phantom X-ray images ⁸³ but not to the task of finding hidden targets in pictorial scenes. ⁸⁴ The same can be said for visual recollection memory. Expert cytologists and radiologists were not better at recognizing scenes or isolated objects than non-medical participants. ⁸⁵

The limitations of the transfer depend on the domain-specificity of mental images. Whether looking at medical images, patients, landscapes, or paintings, each requires unique domain-specific knowledge and experiences to form mental images.

I suspect that the effectiveness of Goodman and Kelleher’s ²⁹ intervention is unsatisfactory. Their task was too easy, as making a diagnosis is more difficult than identifying abnormalities. ⁸⁶ There may be more efficient or effective ways of fostering diagnosing skills than appreciating visual arts. For example, learning modules, based on theories of perceptual learning, such as Perceptual and Adaptive Learning Modules (PALM), have been developed to foster pattern recognition of medical images. ^{61 , 87} All the relevant studies have reported excellent results of diagnostic accuracy for medical school students and/or residents ^{22 , 88 – 94} and non-medical participants, ^{95 – 98} however, it is possible such excellent results across the board could be due to publication bias (see also the review by Guégan et al. ⁹⁹ ). Although it is difficult to directly compare the studies due to different testing methods, these modules seem superior to Goodman and Kelleher’s intervention. ²⁹ It is also notable that the outcome measures used by these modules included fluency, or the speed of recognition. This is consistent with research findings on expertization.

Likewise, teaching a specific eye movement search pattern that corresponds with the characteristics of the area shown in the radiographs ¹⁰⁰ or demonstrating an expert model that searches visually and interprets symptoms (eye-movement modeling examples) ¹⁰¹ could also be effective.

It would be more accurate to say that the interventions that use the appreciation of visual arts foster verbalizing skills. It also may be useful to teach concepts of visual elements, such as color, light, and texture in order to foster such skills. Verbalizing accurately and lucidly is important for various reasons in medical practice. ¹⁰²

Group discussions can also act as an opportunity to practice and reflect on one’s verbalization skills. This can contribute toward the development of communication skills that are required for teamwork. Klugman et al. ³⁵ intended to develop communication skills and claimed that their intervention achieved the same objective. However, they did not present any empirical data to support their claim.

The use of visual arts may also be useful for group discussions across specialties. For example, Katz and Khoshbin ¹⁰³ reported that their curriculum included visual arts to facilitate multidisciplinary team building. One possible reason for this was that the visual arts free the participants from the hierarchy of medical personnel and ensure their equal participation in discussions. It has been reported that even children who do not perform well academically or usually speak, actively participated in the VTS sessions. ⁴³

Thus far, I have shown that appreciating visual arts may not have a direct effect on making an accurate diagnosis. Meanwhile, Jasani and Saks ³⁴ observed an increase in descriptions of the scope of interpretation. Understanding the patient’s background and context—which are invisible—may be necessary for patient interviews. This corresponds with Shapiro et al.’s supposition ¹⁰⁴ that teaching through cases would be effective in fostering pattern recognition, while teaching through paintings would promote interpretation skills based on a broader context, including a patient’s subjective perception and understanding of diseases.

Perspective-taking as cognitive empathy, such as knowing and postulating how another is thinking and feeling, ¹⁰⁵ is also essential for patient interviews. Appreciating visual arts and group discussions about observations and interpretations of them may foster such perspective-taking. Although there are several reports on such practice, ^{106 – 111} none presented empirical evidence. On the other hand, Gurwin et al. ³² used the Reading the Mind in the Eyes Test, ¹¹² which assesses the ability to recognize emotions based upon photographs of actors’ eyes, and reported no significant difference, possibly due to the ceiling effect. It would be beneficial to consider using other measures, such as the Theory of Mind Test ¹¹³ and Faces Test, ¹¹⁴ as exemplified by Pino and Mazza. ¹¹⁵

Understanding patients’ facial expressions is also likely to enhance perspective-taking for medical professions. ¹⁸ It may be worthwhile to train such understanding because of the functional specialization for face perception in the brain. ¹¹⁶ While several studies have reported on such practices, ^{117 , 118} they did not provide any supporting empirical evidence.

This article has several limitations. First, I excluded articles published in languages other than English and those that subjectively measured observation skills. These studies may have included information that could have contributed to this article.

Second, I found only one study that directly examined the impact of appreciating visual arts on diagnostic skills. The conclusions of this article are tentative, while the findings of the expertization studies are sufficiently reliable.

Third, I was not able to fully interpret the results of Goodman and Kelleher, ²⁹ specifically, the small transfer of visual arts appreciation to the identification of abnormalities in the radiograph. Although it may be due to adapting to the test format, it is not clear by what process the transfer occurred, if at all.

I offer the following suggestions for future research. First, while studies on expertization reveal qualitative differences between novices and experts, the process of expertization is not understood very well. ^{74 , 127} It is likely, albeit to a lesser extent, that listing more visual features and hypotheses is unnecessary for the process of expertization. Hence, it may be promising to investigate whether listing such features and hypotheses by appreciating visual arts is effective in scaffolding diagnosis skills. For example, such appreciation might accelerate learning through the learning module based on perceptual learning. Thus, longitudinal studies may be required ¹³ because the effects of such appreciation may be delayed.

Second, to measure the effectiveness of appreciating visual arts, it is recommended to use the same tests that have been used to measure the effectiveness of learning modules on perceptual learning. Diagnostic accuracy and fluency should also be examined, as asserted by Kellman et al. ²² Additionally, the influence of pre-existing knowledge should be considered.

Third, as mentioned by Perry et al., ¹⁵ the intervention methods must also be reported adequately. It is unlikely that merely looking at visual art develops the skills required for medical diagnoses. Rather, the design of the experience, such as selection of the paintings, instruction, and facilitation and activities within peers/groups is required because the outcome of the intervention should vary based on the type and number of paintings that have been viewed and on the activities or reflections mentioned as instructions. ¹²⁸ It is also necessary to mention the responses that were evoked by the subjects (or paintings), prompts, and questions to identify whether the outcomes were intended or derived. Moreover, I found the following deficiencies in the studies used in this article: the number of missing samples from the participants; what did the control group do; details of the content and format of the lectures that the control participants were subjected to; maximum possible values of the scales or items for considering the ceiling effect; details of the scoring criteria; how many schools/organizations did the participants belong to; how many images were used for the test; and timing of the pre- and post-test. Thus, items listed in Table 1 should be reported in future studies.

Fourth, it is necessary to investigate whether the skills and knowledge that would be fostered by appreciating visual arts enhance diagnosis skills. This article found that appreciating visual arts may enhance verbalizing skills, teamwork skills, interpretive skills, and perspective-taking. Mukunda et al.’s ¹⁶ review also found a paucity of rigorous studies on training modules that use arts to promote empathy, team building, communication skills, wellness/resilience, or cultural sensitivity.

Fifth, research is needed that directly compares the skills fostered by visual arts appreciation with other methods that attempt to foster the same skills, such as learning modules applied by perpetual learning and using other forms of arts. For example, we can use only visual arts and also performing arts for teamwork and communication skill development among health professionals. ¹²⁹ Due to time constraints imposed by the many demands of medical education and training, it is necessary to identify a more efficient method among those available.

Finally, as Osman et al. ¹³⁰ suggested, further research exploring the process of learning through visual arts appreciation is required, as extant literature only focuses on outcomes. Dominant research methods such as controlled studies and psychometric scales do not allow for such investigation ^{128 , 131} ; thus, further research should include alternative methods and outcome measurements, such as qualitative methods. ^{15 , 132}

I also offer the following suggestions to practice fostering diagnostic skills. Given the time constraints, adopting visual art appreciation is not encouraged. Taylor ⁶⁴ claimed that education in medical school should provide students with many instances of medical cases and appropriate knowledge to form adequate mental images. Such provision must be made in an appropriate manner, as in PALM, to evoke implicit learning in order to extract subtle differences in the cases. This should be prioritized over visual art appreciation.

Furthermore, since expert perception is a combination of observation and interpretation through intuition, separating observation and interpretation, as instructed in several intervention studies, seems to be unnatural and could be avoided. Jaarsma et al. ¹³³ showed that in clinical reasoning, experts used less descriptive terms, like expressing colors and shapes, and used more comparative terms to interpret findings in terms of normal/abnormal or typical/atypical.