Evaluating guided viewing strategies to enhance well-being effects of art viewing: a three-arm randomised controlled trial Authors: MacKenzie D. Trupp, Shiqi Yin, Robyn Landau, Katherine Templar Lewis, Aleksandra Igdalova, Corinna Kühnapfel, Giacomo Bignardi, Lawrence Chiles, Joerg Fingerhut, Rebecca Chamberlain, Matthew Pelowski Published: 2026 ABSTRACT Art viewing has been associated with well-being and positive mental health, yet its benefits are not universal across contexts or viewers. A key question is which aspects of art-viewing experiences drive beneficiary effects and whether they can be directly augmented to enhance well-being. To answer this question, this study explored the well-being effects of two art-based theory-informed augmentation videos at the National Gallery, London: one providing guided viewing instructions focused on style and slow, detailed observation, and the other offering a pre-viewing mindful breathing session. In the first large (N = 281) pre-registered study, we compared augmentation videos to a standard gallery visit. Further, in a laboratory-based follow-up (N = 33), we compared one pre-viewing intervention with a filler free-writing task to isolate the potential effects of the video strategy. Both studies used mixed-methods randomised controlled trial designs. Across conditions and studies, significant improvements in well-being were detected from pre- to post-gallery viewing. However, the overall effect of either video intervention did not enhance the art-viewing experience or lead to greater well-being changes. Rather, we found that guided viewing led to greater increases in positive mood than pre-viewing mindfulness and that, among first-time visitors, such strategies may be especially beneficial. We interpret the findings in the context of prior mixed evidence and suggest that future research should adopt a more holistic approach that incorporates multiple interacting mechanisms and more systematically investigates the role of our moderating variables. Receptive art engagement, including leisure activities such as attending performances, visiting museums, cultural institutions, or viewing art, has been consistently linked to higher well-being and positive mental health outcomes in population-based epidemiological studies1–3 and, increasingly, in experimental research evaluating the impact of the arts across various settings such as art galleries4–6, urban spaces7,8, clinical contexts9,10, and online11–13. These studies suggest that art engagements, and even encounters with individual artworks, might provide an accessible, adaptable, and often community-based set of activities to support a range of well-being outcomes for review see14. However, alongside these results, there are several critiques and outstanding questions. Recent reviews have emphasised that effects of arts encounters are not uniform across viewers or contexts14,15, underscoring the need to better understand the factors shaping art experiences and, more so, to develop strategies for optimising their well-being impact. Understanding and enhancing well-being outcomes through art viewing is a growing area of research. While studies have explored various strategies, addressing the role of the individual or viewing context16, strategies remain needed. For instance, there is some evidence that suggests that representational art reduces physiological stress biomarkers more effectively than abstract art17, and mindfulness-focused gallery designs enhance mood relative to traditional gallery spaces18. However, despite their success, such interventions require high logistical investments, such as modifying gallery layouts or changing art frequently, making them more impractical for widespread implementation. A potentially more scalable approach may be to focus on the transitional space at the entrance to an art setting, such as the foyer of a museum. Such moments of transition may provide strategic opportunities for brief yet impactful priming interventions before visitors engage with artworks. Supporting this, prior research suggests that top-down structured viewing instructions, such as teaching viewers to pay attention to details, to focus on style, and to slow down19–21, can help improve art experience aspects, such as pleasantness, liking, or meaningfulness—identified mechanisms that explain whose well-being benefits more from art viewing13,14. However, research using such strategies to enhance well-being during art viewing has had mixed results11,22,23. For example, instructions to reflect on the meaningfulness of the artwork, or to look mindfully or curiously22 have not been found to improve well-being outcomes from art viewing. One reason could be that past implementations have relied on short, text-based prompts, which may lack the depth and engagement necessary for meaningful impact. Alternatively, the present study introduces professionally produced, theory-informed gallery-augmentation videos as novel pre-viewing interventions designed to enhance art engagement, such as liking or meaningfulness, and, through these, well-being. Specifically, we examine two strategies: guided aesthetic viewing instructions and a pre-viewing mindfulness exercise. Both are grounded in theoretical and empirical research presented below, yet have not been systematically compared in video format during a controlled gallery-based assessment. To do so, we conduct two mixed-method studies to evaluate how such gallery-augmentation videos placed before gallery entry may affect well-being outcomes via supporting subjective art-viewing experiences. Guided viewing: the case for slow and aesthetic looking Offering instructions to slow down and pay attention to an artwork’s style as a means of enhancing the experience can be linked to prior theory and evidence through several avenues, suggesting that this strategy is a promising candidate for a gallery augmentation to support well-being outcomes. First, ‘slow looking’ instructions24 encourage deliberate and detailed observation of artworks over extended periods, promoting a deeper engagement at a sensory and cognitive level. For example, a study addressing whether slow looking could impact the art experience has found that providing such instructions to slow down via a historical framing around the art resulted in modest positive effects on self-reported cognitive engagement aspects20. Similarly, concurrent guided slow looking compared to free exploration has been associated with higher ratings of beauty and feelings of being ‘enraptured’ and ‘edified’21. Not only how slowly one views an artwork but also to which aspect of an artwork attention is given seems to have an impact on subjective art experiences. ‘Aesthetic looking’, contrary to slow looking, is a viewing style that emphasises the idea that every picture perception is a twofold experience, involving both the medium and what is depicted in artworks—stemming from the philosophical concept of a twofold experience in such cases25. The simultaneous engagement with an artwork’s stylistic features and its content is especially salient when experiencing representational artworks. It has been argued that art appreciation might be enhanced when attention is directed towards the configurational fold of the experience and, e.g., the surface rather than the content of a painting26,27. This has been directly tested in one study that asked participants to consider either style or content to come up with beauty ratings, showing significantly higher reported beauty ratings for the style condition27. Similarly, another study contrasted viewing orientations, to either pay attention to pragmatic (‘try to understand what is happening’) or aesthetic (‘focus on artistic form and composition’) aspects of artworks while participants were undergoing a functional magnetic resonance imaging (fMRI) brain scan. The results indicated that prioritising aesthetic aspects (style) over pragmatic ones (content) led to enhanced art evaluations and activation in brain areas associated with heightened emotional processing19. Overall, the strategy of slowing down and shifting one’s attention towards style may support well-being outcomes of art viewing through enhancing the meaningfulness, beauty, or pleasantness of the art experience, and ultimately support well-being14,28. Pre-viewing mindfulness: enhancing art engagement through awareness Alternatively, to slow aesthetic-focused looking, pre-viewing mindfulness exercises also have a theoretical foundation for enhancing art experiences and could support a gallery visit. Mindfulness is the practice of being fully present in the moment without judgment29 and can be cultivated through breathing exercises. Several studies indicate a positive relationship between mindfulness and enhanced art experiences. For instance, individuals with higher mindfulness traits report more frequent aesthetic experiences30. Further, a stronger mind-body connection, closely linked to mindfulness and reflected in heightened interoceptive body awareness, has been associated with greater emotional responses to representational paintings31. Beyond associations, a large study found that mindful breathing significantly increased the pleasantness of art-viewing experiences compared to an information-based condition at the Manchester Art Gallery18, although the breathing was not more effective than an as-usual visit, possibly because the purpose-designed gallery space already enhanced the viewing experience. However, mindfulness inductions, such as body scans, have been shown to enhance aesthetic responses in music listening32 and improve observational skills33, suggesting that a pre-viewing mindfulness session may support art viewing and well-being outcomes. Study 1 In Study 1, we conducted a pre-registered three-arm, mixed-methods, randomised control trial (RCT) to evaluate the effectiveness of the two intervention strategies (Aesthetic Looking and Mindful Breathing) introduced above, which were created with a multidisciplinary team in collaboration with or by, respectively, our industry partner Kinda Studios–a company designing evidence-based experiences. These conditions were further compared to a typical gallery visit control condition, all at the National Gallery, London. The first condition, shown in Fig. 1, was a four-minute Aesthetic Looking video, designed and produced based on the past literature for this study, which animated a painting from the National Gallery, London’s collection, to demonstrate how to slow down and pay attention to an artwork’s style, colours, and details, exemplifying slow aesthetic-looking techniques. The second condition, also shown in Fig. 1, was a five-and-a-half-minute Mindful Breathing video that used abstract art images and music to guide viewers through an intentional breathing activity. Finally, the as-usual group had no video and allowed participants to enter the museum directly, representing a more typical gallery experience. As depicted in Fig. 2, and discussed in detail in the Methods section, N = 281 participants were included in the analysis, drawn from an overall sample of N = 337 (see Methods for exclusions, demographics, and art background and interest). Participants were recruited and given a pre-survey assessing demographics and baseline well-being. They were then randomised blindly to either one of the two video conditions or to go directly to the gallery without a video. Once in the gallery, all groups were asked to choose an artwork to view before returning to complete a post-survey that included secondary evaluations of mood and state anxiety, as well as appraisals and evaluations of the art-viewing experience and artwork (for details, see the Methods section). Fig. 1Depiction of videos shown before art viewing. Aesthetic Looking video animation demonstrates how to slow down and pay attention to the details and style of an artwork. Each image is a video frame, showing different focal points, panning and zooming into the painting (Claude, 1604/5?–1682, A View in Rome, 1632, The National Gallery, London). The Mindful Breathing video, “Inspirit”, guides viewers with music and visuals to breathe to the beat on 6–6 blocks, each image is a video frame. For more details, see the Methods section. Images reproduced with permission from the National Gallery, London (panel A) and Kinda Studios (panel B). Depiction of videos shown before art viewing. Aesthetic Looking video animation demonstrates how to slow down and pay attention to the details and style of an artwork. Each image is a video frame, showing different focal points, panning and zooming into the painting (Claude, 1604/5?–1682, A View in Rome, 1632, The National Gallery, London). The Mindful Breathing video, “Inspirit”, guides viewers with music and visuals to breathe to the beat on 6–6 blocks, each image is a video frame. For more details, see the Methods section. Images reproduced with permission from the National Gallery, London (panel A) and Kinda Studios (panel B). We predicted that the Aesthetic Looking video, which explicitly taught viewing techniques shown to trigger candidate mechanisms, followed by the Mindful Breathing video, would result in more positive art-viewing experiences (as measured via liking, beauty, meaning, understanding, self-relevance, goodness) and larger improvements in well-being outcomes compared to the standard gallery visit. Well-being outcomes encompass momentary positive and negative mood (also referred to as experienced or hedonic well-being) and state anxiety. Furthermore, we assessed art viewing behaviour to see if the videos impacted attention towards the art or one’s body, as well as evaluated the video and art viewing experiences via qualitative open text boxes. Specifically, we evaluated whether the Aesthetic Looking video led to larger self-reported shifts in attention towards the style of the artwork and a feeling of looking more slowly. For the Mindful Breathing video, we assessed if participants had greater increases in awareness of their heartbeat and reported being more aware of their bodies in the gallery, in line with past literature suggesting a stronger connection to one’s body after mindfulness activities34,35. In addition to our pre-registered analysis, we report a series of exploratory analyses, including equivalence testing, and examining the relationship between art experience and changes in well-being, in lieu of our registered mediation analysis, to assess whether changes in well-being across conditions were mediated by liking and meaning. Finally, we also explore the moderating effects of art interest, education, and first visit. Fig. 2The design of Study 1 shows the random blinded allocation between conditions and the timeline of participation (pre-survey, allocation, art viewing in the gallery, post-survey). The floor plan shows where participants entered: the research room on the left of the stairs, mid-way up, and Gallery Room Nine, where the art viewing occurred. Four paintings given as options to participants are shown in situ; see the Materials section for artwork details. Images of the video screening room are depicted for both videos. Images of the Gallery taken by authors, published with permission for the National Gallery, London. Image of Aesthetic Looking video: Claude, 1604/5? – 1682, A View in Rome, 1632, The National Gallery, London. Paintings included in gallery from left to right: Paolo Veronese, 1528–1588, The Consecration of Saint Nicholas, 1562, The National Gallery, London; Paolo Veronese, 1528–1588, The Family of Darius before Alexander, 1565–1567, The National Gallery, London; Paolo Veronese, 1528–1588, The Dream of Saint Helena, 1570, The National Gallery, London; Titian, c. 1488/90–1576, Diana and Actaeon, 1556–1559, The National Gallery, London. The design of Study 1 shows the random blinded allocation between conditions and the timeline of participation (pre-survey, allocation, art viewing in the gallery, post-survey). The floor plan shows where participants entered: the research room on the left of the stairs, mid-way up, and Gallery Room Nine, where the art viewing occurred. Four paintings given as options to participants are shown in situ; see the Materials section for artwork details. Images of the video screening room are depicted for both videos. Images of the Gallery taken by authors, published with permission for the National Gallery, London. Image of Aesthetic Looking video: Claude, 1604/5? – 1682, A View in Rome, 1632, The National Gallery, London. Paintings included in gallery from left to right: Paolo Veronese, 1528–1588, The Consecration of Saint Nicholas, 1562, The National Gallery, London; Paolo Veronese, 1528–1588, The Family of Darius before Alexander, 1565–1567, The National Gallery, London; Paolo Veronese, 1528–1588, The Dream of Saint Helena, 1570, The National Gallery, London; Titian, c. 1488/90–1576, Diana and Actaeon, 1556–1559, The National Gallery, London. Results Following our pre-registration, as the assumptions were not met for parametric testing for Study 1 or 2, we used a non-parametric alternative from the MANOVA.RM R package. This method relies on bootstrapping permutation approaches that do not assume homogeneity of covariances or multivariate normality using a Wald-type statistic (WTS)36. An assessment of differences between randomised groups was conducted to compare pre-test levels of well-being variables and demographics across conditions and days (comparing weekday sessions to weekend sessions and examining differences in the average time of day each condition started). See the Supplement for a full overview. The only difference found was that the Aesthetic Looking group started on average ~ 40 min later in the day. Priming video experience and effects Participants generally enjoyed both priming videos but had higher ratings for the Aesthetic Looking video (Aesthetic Looking, enjoyment min = 1, max = 7, M = 6.07, SD = 1.12) as compared to Breathing (M = 4.60, SD = 1.33). We then evaluated whether the videos impacted aspects of art viewing behaviour and participants’ perception of the gallery or their awareness of their own bodies in the gallery. The descriptive statistics found in Table 1 indicated that all conditions reported paying more attention to the style than the content and that the Aesthetic Looking Video participants reported paying the most attention to style, as expected. Similarly, all participants felt they looked slowly at the artwork, and the Aesthetic Looking condition reported the slowest subjectively felt-looking. The overall sample tended to be aware of the location of their body in the gallery, with the Aesthetic Looking condition having the highest levels of awareness. All groups showed a decrease in their awareness of their heartbeat (the difference between pre- and post-heartbeat awareness), with the largest decrease in heartbeat awareness in the Mindful Breathing condition. Table 1Study 1: descriptive statistics for art viewing experience.MeasureAesthetic looking M (SD)Breathing M (SD)Control M (SD)Liking5.51 (1.23)5.09 (1.38)5.29 (1.27)Meaningful4.98 (1.63)4.67 (1.49)5.00 (1.51)Beautiful5.61 (1.40)5.27 (1.63)5.54 (1.54)Self-relevant4.02 (1.70)3.99 (1.92)4.11 (1.95)Understanding4.46 (1.58)4.50 (1.64)4.75 (1.58)Art goodness5.89 (1.17)5.72 (1.32)5.92 (1.19)Twofoldness (check)4.79 (1.59)4.88 (1.79)5.17 (1.66)Slowness (check)3.30 (1.23)3.44 (1.41)3.66 (1.46)Heartbeat awareness change− 0.19 (1.69)− 0.46 (1.25)− 0.23 (1.69)Body location awareness4.47 (1.74)4.05 (1.80)4.07 (1.97)The subsample sizes are NLooking = 94, NBreathing = 90, NControl = 93, if not indicated otherwise. Twofoldness/Slowness NLooking = 94, NBreathing = 93, NControl = 94, Heart beat/Location Awareness NLooking = 94, NBreathing = 92, NControl = 94. Table 1Study 1: descriptive statistics for art viewing experience.MeasureAesthetic looking M (SD)Breathing M (SD)Control M (SD)Liking5.51 (1.23)5.09 (1.38)5.29 (1.27)Meaningful4.98 (1.63)4.67 (1.49)5.00 (1.51)Beautiful5.61 (1.40)5.27 (1.63)5.54 (1.54)Self-relevant4.02 (1.70)3.99 (1.92)4.11 (1.95)Understanding4.46 (1.58)4.50 (1.64)4.75 (1.58)Art goodness5.89 (1.17)5.72 (1.32)5.92 (1.19)Twofoldness (check)4.79 (1.59)4.88 (1.79)5.17 (1.66)Slowness (check)3.30 (1.23)3.44 (1.41)3.66 (1.46)Heartbeat awareness change− 0.19 (1.69)− 0.46 (1.25)− 0.23 (1.69)Body location awareness4.47 (1.74)4.05 (1.80)4.07 (1.97)The subsample sizes are NLooking = 94, NBreathing = 90, NControl = 93, if not indicated otherwise. Twofoldness/Slowness NLooking = 94, NBreathing = 93, NControl = 94, Heart beat/Location Awareness NLooking = 94, NBreathing = 92, NControl = 94. Study 1: descriptive statistics for art viewing experience. The subsample sizes are NLooking = 94, NBreathing = 90, NControl = 93, if not indicated otherwise. Twofoldness/Slowness NLooking = 94, NBreathing = 93, NControl = 94, Heart beat/Location Awareness NLooking = 94, NBreathing = 92, NControl = 94. To statistically evaluate these differences, we conducted two MANOVAs, with condition as the IV and looking speed and attention to style as the DVs. The first MANOVA predicted looking speed and attention to style while viewing the artwork (inter-DV correlation, r = .21, p < .001). The second MANOVA examined a change in awareness of one’s heartbeat and the location of one’s body in the gallery, for which one participant was removed due to missing data (A: 94, B: 92, C: 94, inter-DV correlation r = .01, p = .845). We found no differences between conditions on the linear combination of DVs for either test. First MANOVA (looking speed, style/content): WTS = 5.15, df = 4, p = .272. Second MANOVA (heart and body awareness): WTS = 5.09, df = 4, p = .278. Cohen’s D effect sizes between conditions, reported in the Supplement, were all negligible to small in magnitude. Finally, participants’ written reports about their experience watching the videos via an open text box were assessed using qualitative inductive content analysis, with responses coded and grouped into categories. For detailed descriptions of each category and quotations, see the Supplement. We included only the top three themes for each video. Participants’ reactions to the Aesthetic Looking video were themes: Inspiring Art Appreciation (40% of N = 22 total codes), Educational and Insightful (32%), and Relaxing (9%). The Mindful Breathing video responses were grouped into themes of Relaxing (18% of N = 39 codes), Confusing or Difficult (18%) and Having Criticisms or Suggestions (15%), suggesting different subjective experiences of the videos. Art viewing experience Descriptive statistics for the evaluations of the artworks (including liking, meaning, pleasure, self-relevance, understanding and art goodness) across all three groups are shown in Table 1. All conditions tended to report high levels on all appraisals. Liking, beauty and art goodness were especially high, with average scores above 5 (1–7. max = 7). Liking, meaning, and beauty were highest in the aesthetic-looking condition, which aligned with our prediction. The DVs were significantly correlated, with r = .30 to 0.70 (see Supplement). A MANOVA using condition as IV and the six art appraisals as DVs was used to test potential condition differences. Two multivariate outliers were removed based on Mahalanobis distance scores (analysis sample size N = 275, A: 94, B:90, C:91). Results revealed no significant differences between conditions on the linear combination of DVs (WTS = 12.55, df = 12, p = .403). Cohen’s D effect sizes between conditions, all reported in the Supplement, were all negligible to small in magnitude. Finally, to further compare participants’ art-viewing experiences inside the gallery, we offered another open text box asking participants for keywords that described their experience. For detailed descriptions of each theme, quotations, and a breakdown across conditions, see the Supplement. We again used inductive content analysis and found the same top five themes between conditions. These included: Positive Emotional Experience, Interesting and Curious, Calming, Engaging, and Insightful and Inspirational. We found one notable qualitative difference between conditions, that the control condition included only 5% of responses associated with the theme of Calming, compared to 16% for Aesthetic-looking and Breathing conditions (X2 (2) = 12.56, p = > 0.001). Well-being outcomes Descriptive statistics in Table 1 and depicted in Fig. 3 indicated that all conditions experienced improvements in all well-being variables. We used repeated measures MANOVA (RM.MANOVA) to test potential condition differences between the condition and time (pre/post), which were IVs, and the three well-being variables, which were DVs. N = 8 multivariate outliers were detected and filtered out, reducing the sample to N = 91 in each condition (total N = 273). The main effect of Time was significant (WTS = 24.04, df = 3, p = < 0.001), indicating significant changes from pre- to post-assessment on the linear combination of DVs. A post-hoc Wilcoxon signed-rank test with Bonferroni correction revealed that positive mood increased significantly (Z = 5.19, p < .001) while negative mood (Z = − 5.19, p < .001) and state anxiety decreased significantly at the group level (Z = − 6.99, p < .001). Block effect sizes were all moderate in magnitude (Pos. mood ηp² = 0.32, Neg. mood ηp² = 0.44, anxiety ηp² = 0.33). The Time*Condition interaction, however, was not significant (WTS = 9.71, df = 6, p = .137), indicating no difference in changes in well-being between conditions. Cohen’s Dz effect sizes pre to post in each condition are shown in Table 2. Table 2Study 1: descriptive statistics for well-being variables.MeasurePre M (SD)Post M (SD)ChangeCohen’s DzABCABCABCABCPositive mood3.20 (0.65)3.09 (0.70)3.23 (0.78)3.47 (0.77)3.19 (0.74)3.39 (0.82)+ 0.27+ 0.10+ 0.150.460.190.27Negative mood1.33 (0.39)1.30 (0.31)1.34 (0.35)1.19 (0.36)1.20 (0.27)1.20 (0.27)− 0.15− 0.10− 0.13− 0.43− 0.38− 0.46Anxiety1.64 (0.48)1.64 (0.42)1.67 (0.47)1.47 (0.46)1.50 (0.39)1.60 (0.43)− 0.17− 0.15− 0.07− 0.43− 0.36− 0.19A = Aesthetic Looking Condition (N = 91), B = Breathing Condition (N = 91), C = Control Condition (N = 91). Dz effect size interpretation, 0.2 = small, 0.5 = medium, 0.8 = large. Table 2Study 1: descriptive statistics for well-being variables.MeasurePre M (SD)Post M (SD)ChangeCohen’s DzABCABCABCABCPositive mood3.20 (0.65)3.09 (0.70)3.23 (0.78)3.47 (0.77)3.19 (0.74)3.39 (0.82)+ 0.27+ 0.10+ 0.150.460.190.27Negative mood1.33 (0.39)1.30 (0.31)1.34 (0.35)1.19 (0.36)1.20 (0.27)1.20 (0.27)− 0.15− 0.10− 0.13− 0.43− 0.38− 0.46Anxiety1.64 (0.48)1.64 (0.42)1.67 (0.47)1.47 (0.46)1.50 (0.39)1.60 (0.43)− 0.17− 0.15− 0.07− 0.43− 0.36− 0.19A = Aesthetic Looking Condition (N = 91), B = Breathing Condition (N = 91), C = Control Condition (N = 91). Dz effect size interpretation, 0.2 = small, 0.5 = medium, 0.8 = large. Study 1: descriptive statistics for well-being variables. A = Aesthetic Looking Condition (N = 91), B = Breathing Condition (N = 91), C = Control Condition (N = 91). Dz effect size interpretation, 0.2 = small, 0.5 = medium, 0.8 = large. Fig. 3Overall well-being variables change from pre- to post and breakdown by condition, showing a significant group change in each variable but no significant differences between conditions. Lower panel shows the 95% confidence intervals centered around the mean. Figure created based on R Raincloud plots package37. Overall well-being variables change from pre- to post and breakdown by condition, showing a significant group change in each variable but no significant differences between conditions. Lower panel shows the 95% confidence intervals centered around the mean. Figure created based on R Raincloud plots package37. Exploratory analysis Equivalence testing. Equivalence testing using the Two One-Sided Tests (TOST) procedure, with equivalence bounds set a priori at a Standardised Mean Difference (SMD) of ± 0.15, was used to assess equivalence between conditions for well-being outcomes. As reported in the Supplement, across all pairwise Estimated Marginal Mean (EMM) condition comparisons and blocks, none of the SMD confidence intervals fell entirely within the ± 0.15 equivalence bounds (see Tables S2-4). Rather, this analysis revealed a significant difference in the change in positive mood between the Aesthetic-looking and Breathing conditions, whereas the prior experienced larger increases in positive mood (EMM difference estimate = 0.17, SMD = 0.44, p = .035; non-significant after a Bonferroni correction). These results were consistent with the exploratory Linear mixed models (LMM) reported in the supplement. The relationship between art experience and well-being changes. Because the groups did not differ in changes in well-being across all conditions, we did not evaluate the pre-registered mediating effects of liking and meaningfulness. However, Pearson correlations between subjective art experiences and change scores in well-being revealed that after controlling for multiple comparisons (Bonferroni), we found that higher perceived meaning (r = .30), liking (r = .28), and beauty (r = .28) were moderately positively associated with increases in positive mood (see Supplement). Moderation analysis of art interest, education and first visit. To determine whether either video could have affected a subset of participants, we assessed whether individual differences (art interest, art education (none vs. at least one class), and visitor type (first time vs. return visitor) moderated the impact of the experimental conditions on changes in well-being or art experiences. An LMM examining visitor type as a moderator revealed a significant three-way interaction between condition, block, and visitor type for negative mood (see Supplement), suggesting that first-time visitors benefited more from the video conditions (Aesthetic Looking and Breathing) relative to Control, whereas return visitors had similar reductions across groups. Further marginal moderation effects are reported in the Supplement. Study 1: discussion Although results trended in the predicted direction, neither priming video produced significant differences in art experience or well-being relative to the as-usual gallery visit, even though all conditions led to significantly improved well-being across all outcomes. Equivalence testing further indicated that the study lacked sufficient precision to draw firm conclusions, as the effects were neither large enough to confirm effectiveness nor small enough to establish equivalence, calling for replication with greater statistical power. Despite these, some interesting exploratory results emerged: for example, the Aesthetic-looking condition led to greater increases in positive mood than the Breathing video, and both videos were better able to support decreases in negative mood among first-time visitors compared to the control condition. Similarly, qualitative data offer some insight into possible differences. For instance, those in the control condition described their art-viewing experiences significantly less often within the theme of Calming. Yet we did not observe significant differences in variables we anticipated would be higher across video conditions, such as looking more slowly at the style in the Aesthetic looking group, and greater body awareness in the Breathing group. Despite being in the direction of the effect being consistent with our predictions, this could indicate that our videos did not have the indended effects. Similarly, one weakness of the study is that the as-usual control condition proceeded directly to the gallery, introducing some bias, which would further cloud the interpretation of the qualitative or moderation findings that the video groups supported well-being outcomes better than the control. Overall, this study showed that although the results aligned with predictions, the effect sizes were smaller than expected compared to the control. One interpretation is that the gallery setting alone encourages more deliberate, aesthetic viewing and greater awareness of one’s body and feelings, as it removes social demands of daily life. Study 2 Study 2 was intended to validate and extend the findings of Study 1 by using a more rigorous control condition in a laboratory environment. Specifically, as shown in Fig. 4, we replaced the as-usual gallery visit with an active control condition—a free-writing filler task—designed to match the Aesthetic Looking video in duration but was not expected, based on theory or prior research, to influence art viewing. In this condition, participants were instructed to write about anything they wished for four minutes. This approach allowed us to better isolate the effects of the Aesthetic Looking video on art appraisal and well-being changes, beyond the extra time elapsed and additional focused attention that video participants had between the pre- and post-test, compared to the as-usual viewing group that proceeded directly to the art viewing. We conducted a two-arm mixed-methods RCT, opting for a two-condition design to maximise our limited statistical power, omitting the Mindful Breathing video. High-quality replica images of the same artworks from Study 1 were displayed on an iPad in a pseudo-gallery room, maintaining consistency in materials and procedures across studies (see Methods section). Although this study had a smaller sample size due to recruitment challenges at the university laboratory compared to the National Gallery, London, our primary interest was in comparing effect sizes to see if they were consistent with those from Study 1. Of the initial sample (N = 41), n = 33 were included in the analysis (see Methods section for exclusions, demographics, and past art education interest). Fig. 4Design of Study 2, showing randomisation of blinded participants to either the Aesthetic Looking condition or the Free-Writing Filler tasks active control condition followed by an art viewing period, where participants can choose one of four artworks from Study 1 on an iPad, followed by a post-test. A = Aesthetic Looking video, F = Free-Writing task. Image of Aesthetic Looking video: Claude, 1604/5? – 1682, A View in Rome, 1632, The National Gallery, London. Paintings included from left to right: Paolo Veronese, 1528–1588, The Consecration of Saint Nicholas, 1562, The National Gallery, London; Paolo Veronese, 1528–1588, The Family of Darius before Alexander, 1565–1567, The National Gallery, London; Paolo Veronese, 1528–1588, The Dream of Saint Helena, 1570, The National Gallery, London; Titian, c. 1488/90–1576, Diana and Actaeon, 1556–1559, The National Gallery, London. Reproduction of paintings with permission for the National Gallery, London. Design of Study 2, showing randomisation of blinded participants to either the Aesthetic Looking condition or the Free-Writing Filler tasks active control condition followed by an art viewing period, where participants can choose one of four artworks from Study 1 on an iPad, followed by a post-test. A = Aesthetic Looking video, F = Free-Writing task. Image of Aesthetic Looking video: Claude, 1604/5? – 1682, A View in Rome, 1632, The National Gallery, London. Paintings included from left to right: Paolo Veronese, 1528–1588, The Consecration of Saint Nicholas, 1562, The National Gallery, London; Paolo Veronese, 1528–1588, The Family of Darius before Alexander, 1565–1567, The National Gallery, London; Paolo Veronese, 1528–1588, The Dream of Saint Helena, 1570, The National Gallery, London; Titian, c. 1488/90–1576, Diana and Actaeon, 1556–1559, The National Gallery, London. Reproduction of paintings with permission for the National Gallery, London. Results Video effects First, we addressed the effects of the video on viewing behaviour. The descriptive statistics for art-viewing behaviour, also shown in Table 3, including the extent to which individuals reported focusing on style over content and how slowly they felt they were looking, suggested that the Aesthetic Looking group reported paying more attention to the style and reported slightly slower subjective-looking. To compare these, a MANOVA with the condition as IV and speed of looking and attention to style as DVs, was conducted. The results indicate a non-significant effect of the condition (WTS = 0.324, df = 2, p = .850). Effect size differences between conditions were negligible (See Supplement for all Cohen’s values). Table 3Study 2: descriptive statistics for art experience variables.MeasureAesthetic looking M(SD)Free writing M(SD)Liking4.50 (1.20)4.60 (1.40)Meaningful4.44 (1.46)4.07 (1.53)Beautiful4.89 (1.78)5.07 (0.88)Self-relevant3.11 (1.53)3.07 (1.67)Understanding3.67 (1.03)3.53 (1.55)Art goodness5.17 (1.25)5.13 (0.83)Twofoldness (check)4.72 (1.41)4.93 (1.39)Slowness (check)3.28 (1.32)3.20 (1.08)The subsample sizes are Nlooking = 18, Nwriting = 15. Table 3Study 2: descriptive statistics for art experience variables.MeasureAesthetic looking M(SD)Free writing M(SD)Liking4.50 (1.20)4.60 (1.40)Meaningful4.44 (1.46)4.07 (1.53)Beautiful4.89 (1.78)5.07 (0.88)Self-relevant3.11 (1.53)3.07 (1.67)Understanding3.67 (1.03)3.53 (1.55)Art goodness5.17 (1.25)5.13 (0.83)Twofoldness (check)4.72 (1.41)4.93 (1.39)Slowness (check)3.28 (1.32)3.20 (1.08)The subsample sizes are Nlooking = 18, Nwriting = 15. Study 2: descriptive statistics for art experience variables. The subsample sizes are Nlooking = 18, Nwriting = 15. Art viewing experience Descriptive statistics, shown in Table 3, indicated, consistent with our hypothesis, that the Aesthetic Looking video group had higher scores on almost all art appraisals except for liking. To compare the art-viewing experiences of the two groups, we followed the protocol for Study 1. We ran a MANOVA with the condition as IV and variables liking, meaning, beauty, goodness, understanding and relevance as DVs (Inter-DV correlations ranging from 0.04 to 0.85). The results indicate a non-significant effect of condition on the linear combination of DVs (WTS = 2.43, df = 6, p = .877). Effect size differences between conditions were negligible, except for a small effect size for meaning, where the aesthetic-looking group had higher scores (see Supplement for all Cohen’s D values). Finally, a comparison of the qualitative descriptions of the art-viewing experiences, following the same procedure as Study 1, suggested that the three highest codes for the Aesthetic Looking group were Curiosity and Interest (18%), Peaceful (15%), and Positive Feelings (15%). For the free-writing group, the top themes were related to Curiosity and Interest (23% of codes), Focused (14%), Peaceful (14%), and Calming (14%) (see Supplement for details). Differences in well-being changes Finally, we compared the changes in well-being variables between conditions. We similarly did this with a repeated measures MANOVA with condition and time (pre/post) as IVs, and positive mood, negative mood, and anxiety as DVs (Inter DV correlations ranging from − 0.19 to 0.66). N = 1 multivariate outlier was removed. The main effect time was significant (WTS = 19.27, df = 3, p < .001). The interaction time*condition was not significant (WTS = 1.84, df = 3, p = .606). A post-hoc Wilcoxon signed-rank test with Bonferroni correction revealed that positive mood did not increase significantly (Z = 2.19, p = .089); however, negative mood (Z = − 3.01, p = .008) and state anxiety did decrease significantly at the group level (Z = − 2.89, p = .011). Block effect effect sizes were moderate to large (Pos. mood ηp² = 0.37, Neg. mood ηp² = 0.54, ηp² anxiety = 0.55). Effect sizes and mean changes from pre to post within conditions ranged from small to medium. Cohen’s Dz are listed in Table 4. We did not conduct exploratory TOST or moderation analyses due to the small sample size. Table 4Study 2: descriptive statistics for well-being variables.MeasurePre M(SD)Post M(SD)ChangeCohen’s DzAFAFAFAFPositive mood2.45 (0.79)2.43 (0.81)2.59 (0.70)2.69 (0.93)+ 0.15+ 0.260.310.55Negative mood1.41 (0.37)1.70 (0.51)1.25 (0.27)1.45 (0.37)− 0.16− 0.25− 0.46− 0.80Anxiety1.95 (0.59)2.27 (0.56)1.65 (0.34)1.98 (0.50)− 0.30− 0.29− 0.45− 0.68A = Aesthetic Looking Condition (N = 17), F = Free Writing Condition (N = 15). Dz effect size interpretation, 0.2 = small, 0.5 = medium, 0.8 = large. Table 4Study 2: descriptive statistics for well-being variables.MeasurePre M(SD)Post M(SD)ChangeCohen’s DzAFAFAFAFPositive mood2.45 (0.79)2.43 (0.81)2.59 (0.70)2.69 (0.93)+ 0.15+ 0.260.310.55Negative mood1.41 (0.37)1.70 (0.51)1.25 (0.27)1.45 (0.37)− 0.16− 0.25− 0.46− 0.80Anxiety1.95 (0.59)2.27 (0.56)1.65 (0.34)1.98 (0.50)− 0.30− 0.29− 0.45− 0.68A = Aesthetic Looking Condition (N = 17), F = Free Writing Condition (N = 15). Dz effect size interpretation, 0.2 = small, 0.5 = medium, 0.8 = large. Study 2: descriptive statistics for well-being variables. A = Aesthetic Looking Condition (N = 17), F = Free Writing Condition (N = 15). Dz effect size interpretation, 0.2 = small, 0.5 = medium, 0.8 = large. Study 2: discussion Study 2 reported similar effect sizes when comparing the Aesthetic Looking and Free Writing conditions to Study 1. Furthermore, considering the overall qualitative similarities between art-viewing experiences, we conclude that there is no evidence that the Aesthetic Viewing video causes measurable differences in art-viewing experiences or well-being effects, despite how enjoyable it was. Overall discussion Across two studies, contrary to our hypotheses, we did not find robust evidence of group-level enhancements in well-being or art experience resulting from our theory-informed, professionally produced pre-viewing gallery-augmentation videos. Specifically, neither guided aesthetic viewing nor mindful breathing led to greater improvements in well-being from pre- to post-viewing compared with either an as-usual gallery visit or a free-writing filler task. Rather, across both studies, significant increases in well-being and high levels of positive art appraisal were observed in all conditions, regardless of whether a pre-viewing augmentation was present. These medium-magnitude effects on negative mood and anxiety align well with prior research, while effects on positive mood have been rarely documented in single-painting paradigms12,13,38. Although promising, pre-to-post changes in well-being cannot be attributed solely to art viewing, as confounds such as researcher interaction, placebo effects, or regression to the mean may contribute. Our main findings should therefore be contextualised by focusing on between-condition contrasts, which provide a more controlled comparison by including an as-usual gallery visit and a filler-task control condition. Overall, our findings were inconclusive regarding whether such augmentation strategies are useful for enhancing well-being. Despite finding no evidence of a difference relative to controls, we also found no evidence of equivalence, suggesting that meaningful, albeit small, effects cannot be ruled out. Several explanations are possible. Most notably, the videos may not have functioned as intended, as suggested by the absence of between-condition differences on manipulation checks assessing slow looking, attention to style, and body and heartbeat awareness. Alternatively, the effects of the videos on art experience, such as increases in liking or meaningfulness, which were significantly associated with larger improvements in positive mood, may have been too small to translate into detectable changes in well-being outcomes. Regardless of the reasons, our main result aligns with findings from some recent studies, while it is inconsistent with others. For instance, several studies have reported positive effects from pre-viewing instructions on art appraisal19–21, which are known to predict well-being changes from art viewing4,7,12,13. These studies, reviewed in the introduction, evaluated how top-down modulation, in the form of instructions about how to view the artwork, or what aspects to pay attention to. However, other studies reported that providing instructions–such as to reflect on the meaningfulness of the artwork, or to look mindfully or curiously–did not significantly enhance mood-related outcomes22,23. In contrast to the former, and in line with the latter, our study, which included a large sample and experientially-based didactic approach to teaching aesthetic-looking techniques, suggests that gallery augmentation instructions given before viewing may not be effective for supporting well-being outcomes across all gallery visitors, but rather may be supportive for only some participants, such as those visiting for the first time. Among first-time visitors, both videos were associated with larger reductions in negative mood relative to the control condition. This may reflect the value of instructional scaffolding for viewers unfamiliar with the gallery context: first-time visitors may benefit from structured guidance that supports processing fluency and aesthetic engagement, whereas returning visitors, already familiar with the setting, may require less priming. Given the exploratory nature of these findings, caution in interpretation is warranted, and further validation is needed. Notwithstanding the absence of evidence for augmentation effects on well-being or art experience overall, it remains an open question whether pre-viewing interventions may facilitate other outcomes, such as learning. This possibility is particularly relevant given our qualitative findings, in which participants described the Aesthetic Looking video as educational and insightful. To facilitate such outcomes, future intervention development could more holistically address the candidate mechanisms proposed to explain the impact of art viewing on well-being. For instance, a recent review14 summarised several such pathways, including learning and memory evocation, stress reduction (aligning with qualitative findings), and social processes that may facilitate social bonding, reduce isolation, and foster belonging. An alternative approach might involve concurrent or ‘online’ modifications to the art-viewing experience, such as audio guides or tours, rather than pre-viewing activities, despite their greater logistical demands. For example, Igdalova et al. 5 successfully enhanced the well-being effects of art viewing by modifying social interactions during a gallery visit, supporting the view that social processes during art-viewing experiences contribute to well-being outcomes. Limitations Despite the strengths of our design, which include high-quality stimuli, RCT methodology, mixed methods, and both ecological and laboratory settings, several limitations and future directions warrant note. First, it remains unclear whether the Breathing and Aesthetic Looking videos produced their intended effects, or which mechanisms they may have triggered, due to inadequate sensitivity in the manipulation checks. Although single-item checks were included, objective data on viewing behaviour, session length, or physiological responses were not collected to further quantify if the videos had an effect on targeted pathways (i.e., slowing down looking). Future research could address this by incorporating eye-tracking, time-on-artwork measures, and body- or movement-tracking, which may capture effects not detectable through self-report alone. Second, although important, the ecological validity of the setting may have introduced ceiling effects. The artworks were masterpieces displayed in a world-renowned gallery, and perceived beauty and liking may already have been near the ceiling prior to the intervention. Under such conditions, shifting attention to style or slowing down may yield insufficient incremental improvements to translate into measurable effects on well-being. Future researchers might consider anchoring Likert-scale endpoints more broadly. For example, replacing “a lot” with “more than I have ever experienced” allows for more normally distributed responses and greater sensitivity to change. Third, despite our large sample size of both natural and recruited visitors from diverse countries, statistical power was limited, particularly in the second study. Given the consistent direction of effects, the qualitative finding that more participants in the video conditions reported calming art-viewing experiences, and the exploratory result suggesting benefits for first-time visitors, a well-powered follow-up study is needed. Effect sizes are reported throughout to facilitate future power analyses. Conclusion This study evaluated the impact of theoretically informed pre-viewing gallery-augmentation strategies compared with an as-usual gallery visit and a free-writing filler task. Although participants liked the videos, evidence that they improved art-viewing experiences or well-being outcomes was inconclusive. Well-being improved significantly across all conditions, with limited indication that first-time visitors may derive particular benefit from pre-viewing augmentation. These findings offer directions for future research into how individual differences may interact with intervention design to optimise the well-being benefits of art viewing. Methods Study 1: methods This study fell under the blanket ethics approval for the EU Horizon 2020 consortium program ARTIS (Art and Research on Transformations of Individuals and Society), as submitted and positively reviewed by the University of Vienna Ethics Review Board. The study followed the guidelines and regulations of the Ethics Board. We pre-registered our hypothesis and the quantitative analysis for Study 1 (https://osf.io/2enkd). The analysis was done in R (R Studio, 2021.9.1). Participants A total of N = 281 participants were included in the analysis, drawn from an overall sample of N = 337. To be eligible for participation, individuals had to have normal or corrected to normal vision, be over 18 years of age, and feel confident in their English ability to complete the survey and follow instructions. We excluded participants based on several pre-registered criteria, such as having incomplete data (N = 12) or failing to pass the attention check questions concealed in the surveys (N = 44). Participants’ ages ranged from 18 to 96 (M = 37.46 years, Mdn = 31 years), and most participants were female (N = 167; 60%) or male (N = 104, 37%; N = 6 preferred not to say, N = 3 non-binary). Participants represented a diverse array of nationalities, including being from countries in the UK (N = 116, ~ 42%), Europe (N = 77), North America (N = 34), Asia (N = 29), Australasia (N = 12), South/Central America (N = 6), and Africa (N = 3); one participant reported being a dual national (UK/Asia). Of the participants, 164 had no formal past art education, 79 had taken an art class in school, 14 were either pursuing or had pursued a minor or special focus on art in college or university, 19 individuals were seeking or had completed an art degree, and five were professionals in art or art history. Most of the sample (N = 121) visited the National Gallery, London, for the first time on the day of testing, while 71 individuals had been there more than three years ago, 24 within the last year, 34 within the previous six months, and 31 within the previous month. Overall, the sample generally reported above midpoint interest in the arts (M = 5.62, SD = 1.30, rated from 1-not at all interested to 7-very interested). Power analysis A power analysis was conducted to determine the needed sample size to detect differences in changes in well-being outcomes between conditions. Using G*Power and based on an RM-MANOVA with small to moderate effect sizes, a within*between participant interaction (F effect size = 0.15–0.25, 3 groups, two measurements, and three DVs) at 80% power indicated a total sample size of N = 158–432. Given this information and considering the study’s budget to compensate participants, we set an a priori stopping rule at N = 350 participants, which we were slightly short of during our 10-day data collection window (N = 337). Randomisation and blinding To reduce the risk of bias, we randomised participants into conditions and blinded them to the condition they were assigned to. Initially, the study was designed to double-blind the researchers who were recruiting participants from the gallery stairs and the researchers who gave the viewing instructions in the gallery to prevent any researcher effects. However, after two days (of ten), we discovered that this was not possible, as we often had to switch roles of the research team (recruiter, gallery instructor, or research room manager) to maintain smooth management. Thus, this study is only single-blind. Randomisation was done by time block using a random number generator to select the order of conditions, meaning that different periods (day 1 morning, day 1 afternoon, day 2 morning, etc.) were randomised rather than the actual participants. This was implemented to prevent participants from realising they were not offered the chance to watch a video (i.e. contamination). The extent to which randomisation was assessed and no significant differences were found at baseline or in individual difference factors, except that the Aesthetic Looking condition started significantly later by ~ 45 min. Starting time was not a significant covariate (see Supplement for more details). Procedure Participants for this study were recruited through pre-booking an appointment on Eventbrite (https://www.eventbrite.com) and on-the-day recruitment. The Eventbrite invitation was sent to several London universities, including Goldsmiths, University of London, University College London, Royal Holloway, University of London, The London College of Fashion, the University of Cambridge, Queen Mary, University of London, and the National Gallery’s members list, and our team’s networks. Potential participants were informed that the study aimed to explore how art makes people feel and that the results would help museums and policymakers better understand the impact of art on well-being. Moreover, participants were told that they would receive a free exhibition ticket to the Winslow Homer: Force of Nature exhibition. To recruit participants on the data collection days (10 days in December 2022 between 10 am and 4:30 pm), researchers were positioned on Level 2 of the National Gallery Sainsbury Wing, midway up the stairs between the entrance and the gallery rooms. A sign was placed to draw attention, and researchers approached visitors in groups of 1–3 people to ask if they would be interested in a free exhibition ticket. After gaining interest, the researchers explained the study duration and aims and answered any questions the visitors may have had. Participants were unaware of the video aspect of the study until after they had completed the entire procedure. All participants were equally told they would participate in a study testing a “special viewing experience.” After being recruited either after entering the museum or after arriving for their pre-booked appointment, participants were informed about the study and gave their informed consent. The study management then gave participants a participant number on a small piece of paper that they carried with them throughout the study, and a letter that was written on the pre-study table and communicated verbally as the “letter of the day” which allowed for a record of their condition. Participants were asked to complete a pre-survey on Qualtrics (https://www.qualtrics.com/) before continuing to their viewing experience, which they accessed via a QR code on their personal smartphones or paper and pen. The pre-survey included questions regarding demographics, the participant’s last visit to the National Gallery, their perception of the National Gallery, baseline assessments of well-being and how much they felt their heart rate at that moment. If the participants were allocated to a video condition, they were taken from the survey section of the room to a separate area divided by screens to watch the condition-specific video. When allocated to the as-usual gallery visit control participants, they were directly led to gallery room nine. The video viewing area consisted of five chairs arranged in front of a large screen (see Fig. 1), with individual headphones and volume settings that were sterilised between participants to maintain hygiene. After being comfortably seated, participants were instructed to view the video until the end, and if they were in a group, they were asked not to talk to each other during the viewing. Once the video ended, the researcher returned and took them to the gallery room nine. Participants were told to call out in case of technical difficulties, and the researcher would come to assist them (this only happened once). Although the participants were left alone to view the video, they were still under the supervision of the researchers, who were able to observe them from a distance. Once in gallery room nine, a 30-second walk from the research room, the researcher verbally instructed the participants to spend as long as they wanted, such as a few minutes, viewing one painting of their choice from a particular wall. Participants were instructed not to leave the room or view any other paintings. After viewing the painting, they were instructed to return to complete a post-survey. At all times, the participants were supervised by a research team member. The post-survey, also accessed either via QR code on a personal smartphone or via paper and pen, included questions regarding which painting they chose, their experience while viewing the artwork in the gallery, a second assessment of their momentary well-being as well as open text boxes about the video (if they saw one) and the gallery viewing experience. Additional piloting, exploratory and museum-requested questions were also asked. Once finished, participants were debriefed and given their compensation. Materials Aesthetic looking video. To design and produce a gallery augmentation to enhance the viewing experience, we reviewed established theories and research to pinpoint engagement styles that significantly increase aesthetic experiences such as pleasure and meaningfulness. Past literature (reviewed in the intro) revealed that spending more time observing the same painting20 and focusing on the style rather than the content19 are strategies to enrich the art viewing experience, potentially improving mood and reducing anxiety more than a typical viewing experience via enhanced aesthetic experience. We developed a 4-minute introductory video in an interdisciplinary team to implement these two engagement styles (see Fig. 1). The video animated the painting “A View of Rome” (Claude, 1604/5? – 1682, A View in Rome, 1632, The National Gallery, London); professional animation on commission by Jack Field Worldwide) from the gallery’s collection, portraying a cityscape of Rome on the left and an imaginary ruin in the right foreground, featuring five figures, including a man giving change to women. By animating the image, the video demonstrated how one might view it through zooming in and changing perspective, showcasing methods of observation that encouraged viewers to slow down and focus on the style and details of the artwork while connecting with their own experiences. The narrator guides viewers in examining the painting, highlighting the colours, brushstrokes, and textures. The video starts with a free exploration and then points out the many layers of the painting. As it zooms in and out, the brushstrokes and use of paint become the focus, demonstrating how standing closer to or further from a painting can drastically alter one’s perception and experience. For instance, as the video zooms in on a section of the painting depicting buildings, the narrator urges visitors to observe how each building comprises lighter and darker, nearly abstract rectangles, forming a recognisable city street from a distance. The video concludes by asking the viewer to think about what the artist is trying to express through the colours, brushstrokes, and textures, ending with the message, “Next time you are in front of a painting, stop for a moment, slow down, and explore for yourself,” encouraging visitors to apply these techniques during their visit. Breathing video. The five-and-a-half-minute mindful breathing video was included to provide an embodied condition that did not focus on teaching viewing techniques, instead offering a space for breathing before entering the gallery room. The video, “Inspirit,” produced by Kinda Studios (a preview can be viewed at https://www.kindastudios.com/inspirit), uses art and sound to guide viewers in 6–6 resonance breathing (in for six beats, out for six). This video was designed to reduce stress and support the interoceptive connection between the mind and the body. The video begins by welcoming viewers, inviting them to get comfortable, guiding them into body sensations, followed by controlled breathwork, in rhythm with the beat of the music composed by Nailah Hunter, as the narrator counts, “in, 2, 3, 4, 5, 6, out, 2, 3, 4, 5, 6.” The visuals by artist Lucy Hardcastle consisted of vibrant abstract images that shift with the music (see Fig. 1). A comparison of the two videos can be found in Table 5. Table 5Comparison between video conditions.Intervention mappingDistinct to aesthetic lookingSimilar in bothDistinct to mindful breathing1. Contain instructions on where and how to look slowly and deeply at the surface and style of painting1. Contain instructions to follow during the video (look here, notice this, or breathe in, breathe out)1. Deep Breathing interoception intervention2. Representational painting (from National Gallery, London’s Collection).2. Contain animated artwork2. Contain bi-neural music and beat to breathe in time3. Contain calming narrator3. Abstract non-representational digital artwork4. Moment to sit and connect to themselves before entering the gallery (4–6 m)4. Improves interoceptive awareness (heartbeat awareness)5. Contain aspects of mindfulness (focus on breathing or focus on details in painting)6. Both will likely evoke aesthetic experiences, and initial impacts on mood/anxiety Table 5Comparison between video conditions.Intervention mappingDistinct to aesthetic lookingSimilar in bothDistinct to mindful breathing1. Contain instructions on where and how to look slowly and deeply at the surface and style of painting1. Contain instructions to follow during the video (look here, notice this, or breathe in, breathe out)1. Deep Breathing interoception intervention2. Representational painting (from National Gallery, London’s Collection).2. Contain animated artwork2. Contain bi-neural music and beat to breathe in time3. Contain calming narrator3. Abstract non-representational digital artwork4. Moment to sit and connect to themselves before entering the gallery (4–6 m)4. Improves interoceptive awareness (heartbeat awareness)5. Contain aspects of mindfulness (focus on breathing or focus on details in painting)6. Both will likely evoke aesthetic experiences, and initial impacts on mood/anxiety Comparison between video conditions. Art viewing setting and paintings. Participants had the chance to choose from four paintings located beside one another on one corner walls of room nine at the National Gallery, London. The choice of painting was offered to increase the participant’s sense of autonomy in the gallery, which was more ecologically valid. Furthermore, we chose this group of artworks due to their location close to the research room and their relatively lesser-known profiles than other more famous pieces in the Gallery (i.e. paintings by J. M. W. Turner or C. Monet, among many others). The first painting was “The Consecration of Saint Nicholas” by Paolo Veronese (1562, 286.5 × 175.3 cm). The painting depicts Saint Nicholas being consecrated as bishop in front of spectators. The second painting is also by Paolo Veronese, titled “The Family of Darius before Alexander” (1565–1567, 236.2 × 474.9 cm). It portrays Alexander the Great meeting the family of King Darius III of Persia after his defeat in battle, with Alexander showing mercy and insisting the family continue to be treated as royalty. The third painting was “The Dream of Saint Helena” by Paolo Veronese (1570, 197.5 × 115.6 cm). It depicts Saint Helena, the mother of Emperor Constantine, having a dream in which she is told to go to the location of the True Cross, as revealed by the angle. Finally, the fourth painting was “Diana and Actaeon” by Titian (1556–1559, 184.5 × 202.2 cm). The painting depicts the moment when the goddess Diana discovers the hunter Actaeon spying on her while bathing and punishes him by transforming him into a stag. See Fig. 1 for a photo showing the hanging of the paintings in situ. Most of the participants chose to view “The Dream of Saint Helena” (N = 101), followed by “The Family of Darius before Alexander” (N = 84), then “Diana and Actaeon” (N = 67), and “The Consecration of Saint Nicholas” (N = 29). Measures Demographic information about the participants and their previous experiences with art and the National Gallery were collected, including the year of birth, gender (male, female, non-binary, prefer not to say), nationality, and level of interest in art on a 7-point scale (1 = not at all, 7 = very much). Participants were also asked about their formal education in fine art or art history, with options ranging from none to being a professional. To assess their previous experience with the National Gallery, the participants were asked when they last visited, with options ranging from within the last month to within the last six months, one year, more than three years ago, or it was their first time. They were also asked to describe their perception of the National Gallery on a 5-point Likert scale, with options ranging from very negative to very positive. Subjective art experience appraisals and states. The study included questions about the participants’ art experience and appraisal. The questions were designed to gather information about the participants’ art experience and perception. The post-survey began with a question asking participants to select the painting they viewed in the gallery (by choosing the photo of the painting out of the four options). Then, the participants were asked to rate their emotional response to the painting on a scale of 1 to 7 (1 = not at all, 7 = very much) using keywords such as confident, absorbed, and grateful (not included in this paper). The next set of questions asked participants to rate their experience while viewing the artwork in the gallery on a scale of 1 to 7 (1 = not at all, 7 = very much). They were asked to rate their liking, meaning, beauty, and whether the artwork related to their identity, their experiences or the events that they felt defined them as a person (self-relevance). Additionally, participants were asked to rate whether the artwork was good and whether they understood it on a scale of 1 to 7 (1 = not at all, 7 = very much). Finally, at the end of the survey, we asked them if they saw a video as part of the study (as a last question, do not confound the other scales) and if they selected yes, we asked how much they enjoyed the video (1 = not at all, 7 = very much). Well-being measures. Additionally, two standardised measures of well-being were included: the State-Trait Anxiety Inventory STAI39 and the Positive and Negative Affect Schedule PANAS40. The STAI assessed participants’ current and general anxiety levels on a 4-point scale, while the PANAS assessed their positive and negative affect on a 5-point scale. Research has shown that the PANAS and STAI can detect changes in effect following interventions and exposure to art14 with high reliability and validity39,40. Video manipulation checks. Several bespoke single-item questions were designed to gather information about how participants might have been impacted by the videos. Looking speed was assessed by asking how fast participants felt they were looking at the art (bipolar scale, 1 = slow, 7 = fast). Similarly, to estimate which aspects participants paid more attention to (twofoldness check), they were asked which aspects they felt they looked at more (bipolar, 1 = style, 7 = content). Further participants were asked both pre and post to estimate how aware they were of their own heart rate in that moment without checking their pulse with their fingers, but sensing their inner bodies. This item was used to compute a change score that was then used in the analysis. Finally, to estimate how aware of their bodies participants were in the gallery, they reported how aware they were of the location or position of their bodies while viewing the artwork (1 = not at all, 7 = very much). Qualitative data and analysis. Three open text boxes were included to collect qualitative data to enrich the study. To evaluate the gallery experience, participants were asked “to add a few keywords (or more) that describe your experience in the gallery space while looking at the artwork (adjectives/descriptive words)” and about the video, if they viewed one. In addition, for other purposes, we asked participants to add keywords about how their entrance experience made them feel before the pre-survey. We used an inductive content analysis method known as bottom-up analysis to analyse the qualitative data. This method involves examining response data by developing categories during the coding process, grounded in the actual content of the data set to describe and understand the phenomenon under investigation. Unlike deductive theory-driven approaches, such as framework synthesis, this approach does not aim to generate theory. The study adopted an emergent coding approach consisting of three steps: (1) familiarisation with the responses; (2) coding by assigning at least one code to all statements, often resulting in multiple codes per statement; and (3) synthesis, where related codes were identified and merged into broader semantic themes. A flat framework was used, meaning all codes maintained the same specificity and significance level. This iterative process allowed for new aspects to be discerned from the data as more responses were analysed, with answers being recoded as necessary. These steps were completed by an independent coder (author C.K.) who was unaware of the condition in which the participants were involved. Once each response was coded, a second reviewer (author M.T.), aware of the conditions, checked and synthesised the codes per condition, counted them, and selected example quotes. The results section presents categories as percentages, indicating the number of mentions (codes) divided by the total codes found across all categories. This does not represent the percentage of people who reported the category, as many responses were double-coded. Study 2: Methods This study was approved by the Ethics Review Board at Goldsmiths, University of London and followed the guidelines and regulations of the Ethics Board. Participants Of the initial sample (N = 41), n = 33 were included in the analysis with eight individuals excluded for failing the attention checks. Participants’ ages ranged from 19 to 47 years (M = 25.91, mdn = 24, SD = 5.04), where 75% women (Nwomen = 25, Nmale + 4, Nnon−binary = 4). Most participants were from countries in Asia (N = 24), followed by Europe (N = 3), North America (N = 4) and the UK (N = 1), South America (N = 1). N = 14 had no formal past art education, 10 had taken an art class in school, five were either pursuing or had pursued a minor or special focus on art in college or university, two individuals were seeking or had completed an art degree, and 2 were professionals in art or art history. Overall, the sample again generally exhibited interest in the arts (M = 5.91, SD = 0.95, 1-not at all interested to 7-very interested). The sample size was based on logistic and resource evaluations rather than power. Randomisation and blinding Participants were randomly assigned to conditions after giving their informed consent and blinded to the conditions they were allocated to. Procedure The study was conducted in person at the standard testing cubicles at Goldsmiths, University of London. Upon arrival, participants were randomly assigned to one of two groups (control/video viewing) that was unbeknownst to them. After reviewing the participant information sheet, understanding data protection regulations, and signing a consent form, participants scanned a QR code on their personal smartphones linking to the pre-survey on Qualtrics. Initially, they provided demographic information such as nationality, age, gender, and level of art education. Next, they completed a pre-viewing assessment of well-being variables. To mitigate potential bias from the order effect, the pre- and post-survey question items were randomised. Participants then engaged in either a four-minute free-writing session or watched the Aesthetic Looking video, both lasting the same duration to ensure a comparable baseline. They were then instructed to choose one of four paintings displayed on an iPad and not view any of the other artworks. They used their fingers to zoom in and out to simulate moving closer to or further away from the painting in a gallery setting. When they felt satisfied with their viewing experience, they exited the exhibition room, returned to the testing room, and scanned the QR code for the post-survey. The post-survey asked participants to identify the painting they had selected earlier, questions about their art experience, and the same questions about their well-being for a second time. They also indicated their willingness to visit the original paintings and described their art-viewing experience using one or two words. Participants in the intervention group were additionally asked to rate the video they had watched. To conclude, the survey included a question about their interest in participating in a raffle for Amazon shopping vouchers as a thank you. Materials The paintings were high-resolution digital reproductions of those in Study 1, but they were displayed on a smart tablet. Similarly, the Aesthetic Looking video was also utilised; see Study 1 for a detailed description. This study introduced a novel writing prompt for the free-writing condition: Participants were asked to “Please write freely for four minutes about how you feel at this moment. This written content will not be used for data collection in the experiment, and you can keep it to yourself after writing.” Most of the participants choose to view “Diana and Actaeon” (N = 11) or “The Dream of Saint Helena” (N = 10), followed by “The Family of Darius before Alexander” (N = 8) and “The Consecration of Saint Nicholas” (N = 4). Measures The same measures from Study 1 were used to assess changes in well-being variables (PANAS, STAI). The art-viewing experience was also evaluated similarly, with questions regarding liking, meaning, beauty, understanding, relevance, and goodness. Furthermore, the same questions were employed to evaluate the impact of the video on the art viewing session, inquiring about the speed of looking and how much participants focused on the style of the artwork over the content. Additional questions were posed for the purpose of conducting the second author’s master’s Thesis project.