Designing Emotions as Mediators of Attitude Change from Socially-Focused Art Exhibitions: Two Studies of Matched Artist/Curator Intentions and Viewer Response on the Refugee Crisis and Climate Awareness

Emotion, Emotional Journeys, and (Socially-Focused) Art

Emotion is also deeply connected to art. Since the Greek and Latin treatises on poetics, art has always been regarded as inextricably linked to emotions (Menninghaus et al., 2019; Swanger, 1993; Zangwill, 1999; see also Fingerhut & Prinz, 2018, 2020 or Pelowski et al., 2022 for reviews). Most models of art processing (see Pelowski et al., 2016, 2017 for reviews) foreground emotional reactions––positive/negative as well as more profound or esoteric states––as key outputs and also contributing factors throughout the art perception process. Similarly, for artists, imparting emotional reactions has been argued to be a guiding point for artworks that find resonance with an audience (Kozbelt, 2006; Tinio, 2013; Yegorova, 2018). Multiple studies do suggest that the more emotions are imparted, the deeper, the more sincere, and the more pleasurable the experience (Eerola et al., 2016; Gerger et al., 2018; Miu & Baltes, 2012; Zickfeld et al., 2017; although see also Rodriguez et al., 2024 suggesting that more complex emotional experiences are not necessarily more satisfactory or the result of more time spent viewing).

These different aspects of emotions are often linked to the intentions and to the designs of curators and artists. By both reflecting on their own feelings and by intuiting a future perceiver's response, artists may seek to better create their own work and target specific reactions in the perceiver's experience (Pelowski et al., 2020b). This may translate in viewer response (see Barwell, 1986; Pelowski et al., 2020b) to a combination of both intellectually recognizing how an artist may have wanted one to feel, or decoding emotional content, as well as to one's actual feeling states, both of which have been tied to more positive, self-related outputs (Pelowski et al., 2022). These emotional targets are also connected, potentially, to attitude change. Stamkou (2022), for example (see also Belfiore & Bennett, 2007; Keltner & Oatley, 2022), suggests that “when engaging with a work of art, people [may] experience and develop a shared understanding of the emotions communicated through the artwork” and that “the arts …enable individuals to understand a culture's [or a topic's] emotional patterns,” that may “have downstream consequences” on behavior.

The patterns or processes related to designed emotion may also be quite complex. Artists may often anticipate multiple reactions from an audience, thinking of an engagement as a nuanced mixture or journey that one might go through to a desired end. Such patterns, which could of course occur in many ways, are in fact one of the main design elements suggested for many socially-focused installations with art (see e.g., for empirical/theoretical studies, Curtis et al., 2014; Klöckner & Sommer, 2021; Marks et al., 2016; Roeser, 2012; Sommer & Klöckner, 2021; for examples of actual artist/curator statements, see again Figure 1).

Scholars suggest that one of the arts’ most powerful potentialities is such a degree of nuance or “openness” (Hegenbart, 2021, p. 126), allowing them to achieve what traditional “propaganda” or “political activism does not” (see also Groys, 2014; Gabriela Léon, as quoted in Hegenbart, 2021; Weibel, 2015). ² Emotional progressions are also increasingly considered by curators, who similarly employ emotion as a “tool” (Norris & Tisdale, 2017) for exhibition design, using these to engage viewers and to drive awareness and impact of content (see also Falk, 2021; or Varutti, 2023 for discussion of an “affective turn” in curation; as well as Munro, 2014 who explicitly connect this design approach to curatorial focus on societal challenges). These mechanistic aspects, once again, operate in tandem with the other arguments highlighting the arts: enhancing memorability, meaningfulness (Pinsky & Sommer, 2020; Sommer et al., 2019); providing a necessary “aesthetic dimension” (Curtis, 2010, p. 78), where more didactic arguments “can turn people off”.

Empirical Evidence for Emotion as Mediator for Attitude Change in Art

Studies have begun to investigate emotion-related aspects, often at least suggesting complex patterns and their potential as a mediator of impact. In general, recent reviews of socially-focused behaviors (Brosch, 2021; see also Myers et al., 2023; Rees & Bamberg, 2014; Stoll-Kleemann et al., 2022) have shown connections to reported emotions in regards to these topics. Studies have also shown evidence from more experimental designs that explicitly test the role of emotions in ‘art-like’ media contexts. Rees et al. (2015) report an intervention in which individuals were presented with a short text noting that either directly human-caused (e.g., air pollution) or indirectly human-/nature-caused issues (e.g., earthquakes, volcanos) were threats to the environment (compared between groups), and with the former expected to elicit more “moral emotions” (guilt and shame). Reported feelings of these emotions, but also anger and sadness, significantly, and each independently, mediated intentions to act (assessed post-intervention). “Emotional coldness” showed opposite effects. Swim and Bloodhart (2015) report an online study in which adults were presented with pictures of polar bears with a caption describing the threat of climate change to these animals. This was accompanied by a written prompt to either encourage an “emotional” perspective, where participants were asked to imagine how the animals felt, or a more “detached” prime, with participants asked to focus on the image quality, as well as by both no-prompt or no-intervention conditions (all contrasted between participants). Although they reported no main effect of the interventions, participants in the emotional perspective were more likely to donate a portion of their payment ($1.00 USD) to environmental causes. Participants who reported feeling greater magnitudes of emotions (again a blend of, for example, empathy, distress, worry, hope, guilt, shame), as well as less boredom, were more likely to donate and to support environmental groups. Lai et al. (2016) reported a comparative investigation of 18 interventions meant to reduce implicit biases, several of which were formally similar to art (i.e., stories; positively or negative-associated pictures). They suggested that, while varying widely in design, the most effective interventions were emotionally evocative and allowed individuals to personally share the “vivid” experiences of others (see also Appel & Richter, 2010; Richeson & Nussbaum, 2004 for similar results).

A handful of studies have also recently begun assessing art contexts. Sommer et al. (2019) in their questionnaire-based study of visitors’ reactions to an art installation, “Pollution Pods,” designed to allow individuals to experience different degrees of air pollution in cities (see Footnote 1 above), showed that, beyond a small general effect on increased agreement with the intention to “do something to actively prevent climate change/environmental problems in the future,” those who reported experiencing more emotions (sadness, helplessness, anger, awe, inspiration, surprise, guilt, shame, happiness, pride, etc., each considered independently) were significantly more likely to report attitude change. Klöckner and Sommer (2021) reported survey evidence of audience responses from 37 visual arts events collected at ArtCOP21in conjunction with the 21^st UN Climate Summit. Participants were asked to report on different emotions felt in their engagements. Multilevel-structural equation modeling suggested that both positive and negative responses significantly related to perceived artwork quality, heightened reflection, and policy support. Similarly, Kühnapfel et al. (2025) report on the impact of an exhibition visit on neighborhood connectedness, which suggested that most participants reported a blend of positive but also mixed emotions.

See also Marks et al. (2016) for qualitative evidence from an environmentally-focused land art festival, which suggested, again, moderate general impact (of N = 120 respondents, 41% indicated that they “intend[ed] to change their behaviour”). At the same time––and unlike other factors such as length of time, number of visits, or demographics––emotion showed relation to professed change in behavior (see Curtis, 2010 for a similar case report). ³

Complex Emotion Characterizations; Connecting Viewers to Curators and Artists––Promise but also Limitations in Art Emotion Research

At the same time, research is still not without outstanding questions. First and foremost, much as with the general topic of interventions discussed above, most studies still only involve correlations between reported feelings and intentions as stated by an audience (e.g., Rees & Bamberg, 2014) or are only anecdotal and assessed post-intervention (Curtis, 2010; Klöckner & Sommer, 2021; Marks et al., 2016; Rees et al., 2015; Swim & Bloodhart, 2015). Results have also not revealed clear, systematic evidence for the specific contributions from emotion factors. In the “Pollution Pods” art installation above, Sommer et al. (2019) report that, although a significant contribution to attitude data, emotional effects only increase the adjusted R² over a model with a number of basic demographic and pollution-focused factors by 3%. See also Pitt (2019), who reported a pre/post study on music meant to “emotionally” connect viewers with the impact of climate change, contrasted against a scientific article, a presentation of both article and music, and an unrelated musical piece. Contrary to expectations, they found no between-conditions difference.

Potentially related, studies tend to consider rather limited sets of emotions. Most studies have considered only one or a handful of specific (i.e., social) responses, or may otherwise have a rather limited approach for how emotions are selected, combined, or considered. For example, Rees et al. (2015) reported impacts from their target emotion of guilt but also noted relations to sadness and anger, but not pride, which, surprisingly, showed opposite effects. Sommer et al. (2019) reported sadness, helplessness, anger, but also inspiration, surprise, guilt, happiness––all of which were considered individually and with most showing some relations to attitude change. The study by Klöckner and Sommer (2021), while asking participants to report on a number of specific emotions, ultimately only looked at their combinations into basic positive or negative valence, both of which again showed effects (see also Myers et al., 2023). Thus, it may be that such studies have missed or obscured impacts from more complex emotional experiences.

More theoretically, these results, and even the accompanying interventions, could feed into counter-arguments against emotion as a means for attitude change in the first place. Much as artworks or other communications that are too direct or ‘on the nose’ regarding how individuals should think or act could lead to underwhelming or even opposite results, similar effects might be found from art that too overtly seeks to determine what should be felt or that does evoke certain specific feelings (see e.g., Groys, 2014; Hegenbart, 2021). For example, interventions that confront individuals with specific unwanted felt emotions could push them away or terminate further thoughts on a topic due to a desire to avoid the associated distress (Moser & Dilling, 2007; Slovic et al., 2007; Swim & Bloodhart, 2015). Notably, such effects may have been found in the Sommer et al. (2019) study where, while certain emotions (sadness, helplessness, anger) did show a positive relation to attitude change, so-called “moral emotions” of guilt and shame showed negative trends. Hegenbart (2021) makes a similar argument, taking as an example the Holocaust memorial in Berlin, which, she argues, may be too easily framed as only a “monument of shame” and dismissed. See also literature on the topic of “reactance” or tendency to push back against didactice or manipulative messaging (Brehm & Brehm, 2013; see Ratcliff, 2021 for recent review). See Swim and Bloodhart (2015) for an art-based, as well as Allpress et al. (2010; also Nurmis, 2016; Rees & Bamberg, 2014; Reser & Bradley, 2017; Täuber & Van Zomeren, 2013) for similar ambiguous findings across non-arts interventions.

Also related to the above argument, suggestions have been made regarding potential key differences in emotions that are felt versus simply recognized, especially again in the context of overly narrow, emotion-eliciting art. A longstanding discussion in art theory (often under the umbrella-term of ‘Kitsch,’ see Pelowski et al., 2020a for review), suggests that artworks that are too obvious or clearly about specific feeling may have limited impact. As put by Gombrich (1985, p. 62), this is “not because [one] has failed ‘to pick up the message’ but because [they] understand it all too well” (see also Adorno, 1997). Past studies with climate change-related art (Hastings et al., 2004; see O’Neill & Nicholson-Cole, 2009 for review) suggest, for example, that especially laboratory-based intervention often lead to individuals being able to state or ‘understand’ particular emotion-centric approaches but without personally being moved by such messages (see Moser & Dilling, 2007; O’Keefe, 2002 for similar arguments in general climate change research). Thus, once again, these approaches may be overlooking one of the key features, especially of art, involving a mixed ‘profile’ of affective experience. It may be the nuanced emotional mixtures––as suggested in the work of curators or artists––that might better lead to meaningful results. At the same time, current studies have completely omitted the actual intentions of the artist or other designers or curators. We do not in fact know what were such emotion-related aims or what specific form these may have taken.

On the other hand, arguments can also be made against artist or curator guidance. Especially in their discussions of the openness to engagement that might be important with art, scholars also argue that this may require that an audience actively guide their own affective or meaning-making process. As argued recently by Hegenbart (2021; see also Bourriaud, 2020; Groys, 2014; Marini et al., 2012; Nakajima, 2012), successful attitude-changing art “need not be driven by a goal-oriented artistic intention,” or aimed at “a direct …effect on the audience”. A similar argument has also been made specifically for the working artist. The nineteenth century idea of “Art for art's sake” (Pelowski et al., 2022; Zangwill, 1999 for reviews), suggests that a “true” (or maximally successful) artist should not care what the audience feels, or try to lead their experience along a designed emotional path. Worrying too much about this could lead to sub-standard, minimally impactful, art.

It could also be that emotions, in general, whether intended/designed or not, could get in the way of art's or other interventions’ intellectual impact. A longstanding counterargument to emotion-centered discourse, again related to art's ability to deliver knowledge or “truth,” is that individuals must find themselves in a more analytic, evaluative frame (Greppi, 2021). This so-called “aesthetic” mode (see e.g., Pelowski et al., 2017) is often connected to disinterested, intellectual engagement, free of visceral or individual-centered affect––a position often connected to art with more “conceptual” approaches or aims.

Stimuli/Setting—“Show Me Your Wound,” Dom Museum, Vienna

For Study 1, we employed an art exhibition, “Show Me Your Wound” (“Zeig mir deine Wunde” in German), at the Dom Museum in Vienna (20 September 2018–25 August 2019). This addressed the topic of the vulnerability of humans, both at the level of the psyche and the body, as well as the implications for politics and society. As put by the curators in a book accompanying the show (Schwanberg, 2018, pp. 36–40), “everybody sustains injuries or feels threatened by the prospect of harm in the course of her or his own life.” In “recent years,” they continued, “tides of refugees and terrorist attacks in the public realm have increasingly pushed this dimension to the fore” (p. 32). “Through the exposure of [such] vulnerability,” they thus hoped, the exhibition might “bring about positive change” on this topic (pp. 37–38).

The exhibition (Figures 2, 3) included 40 + artworks in a range of visual and multisensory media, displaying examples of physical/mental wounding and with the aim of articulating the interconnectedness between physical and mental pain, as well as artworks constituting a main message of the show suggesting “that wounds are frequently caused by struggles for power and violent conflicts originating in political, religious, or gender-based oppression” (p. 43). This included examples (Figure 2, B) such as a bomb disposal suit from the wars in Afghanistan, or a video piece in which a young, deaf Syrian child conveyed through body gestures and movements his experience of the occupation of his village and his flight from ISIL (Figure 3). The last section showed other examples of wounding as well as showing more hopeful works with the aim of conveying the message that “wounds will only heal if the individual …gets actively involved” (p. 44).

OPEN IN VIEWER

Figure 2.

Overview of Study Components and Exhibition Design, “Show Me Your Wound” (“Zeig Mir Deine Wunde”), Dom Museum Vienna. (A, Exhibition Poster/Catalogue Cover (Artwork; Anders Krisár, Arm (Right), 2006), image used with permission from Dom Museum Wien, artwork image used with permission of the artist. B, Exhibition View with Artworks from ‘The Politics of Wounding'/‘Instruments of Wounding’ Sections of the Exhibition: Top, Unknown Artist, Crucifixion Scene from the Episcopal Palace, Early 19^th C.; Bottom, Anders Krisár, Bomb Suit, 2006–2007; artwork image used with permission of the artist, photo taken by the paper authors. C, Curator Responses (Inset of Full Survey) Showing Emotions that they Wished to Elicit in a Viewer in the Course of the Engagement with the Exhibition. D, Museum Entrance/Street-Level Participant Recruiting Location. E, Exhibition Floorplan. Photo from the Paper Authors).

OPEN IN VIEWER

Figure 3.

Artworks from “Show Me Your Wound” (“Zeig Mir Deine Wunde”), Dom Museum Vienna. (A, Erkan Özgen, Wonderland, 2016, Still From 4 Min. Video of Boy Recounting the Violent Occupation of his Village During the Syrian Civil War. B, Raphaël Dallaporta, Antipersonnel, 2004, Showing Photos of Landmines with Accompanying Descriptions of their Engineering so as to Maximize Damage to the Body. C, Anders Krisár, Bomb Suit, 2006–2007. D, Unknown Artist, Crucifixion Scene from the Episcopal Palace, Early 19^th C., Detail Showing Damage (Stabs into the Canvas) by Nazi Troupes/Hitler Youth, October 1938, Vienna. E, Hana Usui, Black Rain, 2014. (All artwork images used with permission of artists. Artwork E also courtesy of Marcello Farabegoli Projects)).

Part 1: Curator Intentions Regarding Attitude Change and Emotional Experience

To identify target attitudes or behaviors that the curators wished to address through their exhibition design, we conducted an interview (K. Speidel), with the main question: “Imagine an ideal visit to the exhibition. What attitudes, behaviors, or other types of reflection would you like a viewer to report doing or changing as a result of their visit?” Answers, which echoed those also noted in the printed book discussed above (Schwanberg, 2018), included keywords such as “acceptance of others” (especially of refugees), “compassion,” “xenophobia,” “empathic concern,” and “prosocialness.”

To identify the emotions that the curator had specifically targeted for viewers in the exhibition design, we then provided a list of 36 items, with the instructions that the curator should again imagine an ideal visit to the exhibition and identify (by ticking a box, see Figure 2, C) those responses, if any, they would like a visitor to report feeling. The procedure was based on the previous studies of Pelowski et al. (2020, 2022) and included an emotion list (see Figure 4, below) designed to give a broad range of possible feelings, including classic ‘basic’ emotions, from positive to generally negative valence, as well as epistemic and also social or more ‘moral’ emotions as used in many past intervention studies. In addition, the curator had the option of suggesting additional items––in this case, two (“vulnerable,” “compassion”)––which were added, leading to a final total of 38 emotions.

OPEN IN VIEWER

Figure 4.

Emotions Personally Felt When Viewing Artworks and also Emotions Perceived to Be Intended by the Curator (“Show Me Your Wound,” Dom Museum Vienna). (Results from N = 41 Exhibition Visitors. Dotted Line Represents Group Mean of Reported Subjectively Felt Emotions; Flattened Ovals Represents % of Viewers Who Selected Each Emotion as ‘Intended by the Curator to be Evoked in a Viewer’, Regardless of Whether Viewer had Actually Felt this Way in their Experience; Red Circles Represent Emotions Actually Intended by the Curator. *Difference Between Mean Reported Felt Emotions Significant At p < .001, Based on Paired T Test).

From the final list, the curator selected 17 items (45%) as intended (e.g., see Figure 2). This included a mix of rather intense and/or social responses (“vulnerable,” “compassion,” “moved,” “shocked,” “sad,” “overwhelmed”), items related to meaning or reflection (“self-aware,” “epiphany,” “changed mind about meaning,” “changed something about myself,” “transformed,” “novelty,” “insight”), as well as “absorbed” and three other negative emotions (“angry,” “need to leave/stop looking”, “disgusted”), which were noted as not being aimed at per se, but which might occur temporarily as a consequence of visitors’ engagement with the topics addressed.

Note also, this exhibition—considering general artwork appraisals and population-wide pre-post attitude change—was previously reported in Pelowski et al. (2024), which used this data as a comparison for an additional ecological momentary assessment study of a different cohort of visitors. However, the assessment of the main questions here, regarding curator's intentions and visitor emotional experience, was not previously considered.

Part 2: Exhibition Visitor Assessment

Procedure

The study employed a matched pre-/post-viewing design (see, e.g., Pelowski, 2015; Pelowski et al., 2020). Participants were first stopped outside the museum by a researcher and asked to participate. If they agreed, they were then led into the lobby of the museum and brought to a small lounge area in the back of this space. They were given a basic explanation of the research aim, informing them that we were interested in art appreciation in real-life exhibition settings and that they would be asked to look at the exhibition and to report on their experience. Notably, no mention was given to any specific emotions, appraisals, exhibition aims, or possible behavioral/attitude changes related to the art visit (see footnote for full instruction text). ⁴

Participants filled out a pre-test (see below) via Qualtrics (Qualtrics Int., Seattle WA, USA), which could be accessed on participants’ own smartphones, via QR code, or via a tablet computer provided by our research team. Upon completion of the survey, which took about 15 min, they were asked to then visit the exhibition. Participants were told that they could freely move through the exhibition space and engage the artworks in any order, manner, and time duration that they wished. However, they were instructed to proceed directly to the exhibition and to not visit other parts of the gallery. Upon completion of their exhibition visit, they were asked to immediately return to the testing area to fill out a post-test (taking roughly 20 min). Once finished, participants were free to visit or revisit any parts of the gallery that they wished. Participants were not observed while inside the gallery.

Pre-survey—Target Attitudes

The pre-test employed a selection of standardized batteries targeting the range of attitudes or behaviors that had been identified by the curators as aims for the exhibition visit. This included: (1) Xenophobia, assessed via a 5-item battery (Van der Veer, et al., 2011) argued to be a cross-culturally applicable measurement of attitudes regarding immigration or acceptance of others (e.g., “Immigration in this country is out of control,” 1 = strongly disagree, 7 = strongly agree). Additionally, (2) we used the Empathic Concern subscale of the “Interpersonal Reactivity Index” (Davis, 1983), which presents seven statements relating to concern for others and especially the less fortunate (e.g., “Other people's misfortunes do not usually disturb me,” 1 = does not describe me well, 5 = describes me very well). (3) We assessed more general Prosocialness via the “Prosocialness Scale for Adults” (Caprara, 2005), which asks for agreement to 16 statements relating to having helpful behaviors (“I try to help others”) and understanding others’ situations (“I intensely feel what others feel,” 1 = not true, 5 = completely true). (4) We also included the Social Value Orientation (SVO) task (Murphy et al., 2011), which asks individuals to hypothetically divide up money between themselves and a “randomly paired,” previously unknown, “other” person. This task is argued to measure the propensity to maximize one's own gain at the expense of others.

Additionally, (5) Mood was assessed by three items (“How do you feel right now?”, 1 = not at all, 7 = very much) assessing “positive mood,” “negative mood,” and “emotionally aroused,” following a circumplex model of mood/emotion (Russell, 1980). Order of batteries was fixed as above. However the order of individual items within batteries was randomized.

Post-survey

The post-viewing survey, completed immediately after the visit to the exhibition, included a matched set of mood questions (answered first), followed by several additional assessments about the artworks/exhibition and the emotional experience.

Exhibition Appraisal: Participants were asked to evaluate the exhibition using seven bipolar scales (“Overall the exhibition was____”, e.g., 1 = “good,” 7 = “bad;” see full list in Figure 5). These scales (see Pelowski, 2015 for previous use with art) were selected to cover main evaluative/hedonic ratings of quality and enjoyment (bad-good, ugly-beautiful, meaningless-meaningful, boring-interesting) as well as terms connected to activity and potency of the stimuli/experience (potent-impotent, active-passive, strong-weak). Participants were also asked “Would you pay to see the exhibition again?” (1 = not at all, 7 = strongly agree).

OPEN IN VIEWER

Figure 5.

Participant-Wide Artwork Ratings, Understanding of the Curator, Mood, and Empathic, Prosocial, and Xenophobic Attitudes, Pre- and Post-Visit (“Show Me Your Wound,” Dom Museum Vienna). (Results Based Reports from N = 41 Participants. * Denotes Significant Difference (p < .05), Pre-/Post-Visit, from Repeat Measures T-Tests with Time as Within-Participant Factor (See Familywise Bonferroni Correction Information in Table A1).

Felt Emotional Experience: Participants were then asked to report on their emotional experiences. Participants were presented with the same list of 38 terms as used with the curator in Part 1 and asked to report the incidence and the magnitude of their feelings (“When I was experiencing the exhibition, I felt [x],” 0 = not at all, 7 = very strong; see Pelowski et al., 2020, 2022 for similar usage). Importantly, this section explicitly asked participants to consider and report on their actual feelings rather than any attributions that might tie to the art (e.g., a ‘sad’ painting).

Perceived Curators’ Emotion Intentions: Following the above assessment, participants were then presented with the same emotion list and asked to report (via ‘yes’/'no’ tick box) which of the feelings they thought had specifically been intended for them to feel, regardless of whether the participant had personally felt in that way (following again Pelowski et al., 2020, 2022). Participants were free to decide how many emotions to select. No selections was also possible.

Awareness or Thoughts Regarding the Curator: Viewers were also asked to report on five questions regarding their awareness of the curator's and/or artists’ choices (following Pelowski et al., 2020): (1) “Did you have a sense of the curators’/artists’ presence?” (2) “Did you think about the way that the artists/curators must have felt when making/selecting the art?” (3) “Did you have a sense of what the curator was thinking when making the exhibition?” (4) “Do you think the curator wanted to communicate something to the viewer?” (5) “Was there a sense of unity between the artworks?” (all, 1 = not at all; 7 = very much). This last question was included to assess the perceived level of coherence between the art pieces and the presence of a unified message.

Matched Attitude Measures: Participants then answered the same prosocial/empathy batteries, as well as one additional question (assessed post-intervention only) about the general impact from the exhibition in terms of its ability to incite reflection (“Did the exhibition cause you to reflect on or rethink anything about yourself, your expectations, or the world?” see e.g., Pelowski, 2015; Klöckner & Sommer, 2021).

Other Background/Control Questions: Participants then answered several background questions that could potentially impact their engagement, including: (1) political orientation (1 = left wing; 7 = right wing); (2) how important religion was to them (1 = not at all; 7 = very much) and their specific religion (free answer), included due to the religious themes of some of the art (see Dunham et al., 2014); (3–4) general art interest and objective art knowledge, via the Vienna Art Interest and Art Knowledge questionnaire (Specker et al., 2020); (5) whether participants had previously studied or were currently studying art/art history; (6) frequency of visits to museums/galleries (following Leder et al., 2014); (7) basic demographic information; (8) whether individuals had read the wall texts (‘Yes/No’). In addition, (9) we assessed trait empathy via the Questionnaire of Cognitive and Affective Empathy (QCAE, Reniers et al., 2011), following the Pelowski et al. (2020, 2022) studies of viewers-artist emotion matching, which had, however not found strong relations to emotion sharing, identification, or art experience. ⁵

Exhibition Appraisal and Prior Mood

Participants’ appraisals of the exhibition (Figure 5) suggested that the art was generally evaluated positively, however with widish distributions across the entire sample. Answers to the question of whether the exhibition had caused individuals to re-think about themselves also showed a wide distribution. Participants strongly agreed that the exhibition had a communicative intent (M = 5.76) and a perceived sense of unity between artworks (M = 5.17). Participants entered the exhibition in a rather positive (M = 5.71, see Figure 5) and low negative mood (M = 2.05). Paired t-tests comparing pre-/post-viewing reports suggested that participants showed significantly decreased positive mood (M = 4.85, t(40) = 3.622, p = .001, d = .566) and increased negative mood after the visit (M = 2.90, t(40) = -3.404, p = .002, d = .532). No significant change was found for emotional arousal (t(40) = 0.675, p = .504, d = .105).

Pre-Post Attitude Change

To assess the attitude measures, we calculated composite scores for all scales or sub-scales following their original papers’ methodologies. To ensure comparability across measures, we used average scores instead of sum scores. For Social Value Orientation (SVO) scoring, three participants appeared to misunderstand the task (i.e., selecting no or more than one possible money divisions), and were excluded from analyses. No other scales had missing data.

The results are shown as box plots/averages in Figure 5 (see also Table A1, Appendix and Pelowski et al., 2024 for statistical information). In pre-viewing baselines, most participants reported positive scores across all scales (e.g., relatively higher self-assessed agreement of Empathic Concern, Prosocialness; lowish Xenophobia), notably with no suggested ceiling or floor effects. Post-exhibition, most factors tended to show small general decreases—suggesting lower agreement that one was as empathically concerned or prosocial as they had previously professed before the exhibition, but also lower xenophobia. Repeated measures t-tests conducted across the sample, however, revealed only one significant decrease for Xenophobia with a smallish effect (d = 0.425; see Cohen, 1988; see Table A1 for Bonferroni correction information).

Emotion

Figure 4 displays means and boxplots of reported felt emotions for all 38 items. The figure also denotes those emotions specifically identified by the curator as intended (denoted by a red circle). Participants tended, in general, to report a rather moving, reflective, epistemic emotional experience, with the highest average magnitude emotions—across the participant sample—being “moved,” “self aware,” “insight,” “shock,” etcetera. Notably, as can be seen from the leftward clustering of red circles in Figure 4, the emotions identified by the curator tended to be among those with the highest magnitudes. This trend was particularly evident for the emotions added by the curator (“vulnerable,” “compassion”). At the same time, and similar to the exhibition ratings, when looking across the entire sample of viewers, the emotional reports also showed a rather large spread, with individuals reporting answers across most scale points.

Felt/Identified Intended Emotions and Attitude Change

Finally, to test whether the degree to which viewers felt or identified the emotions intended by the curator coincided with greater changes in attitudes, we again looked to correlations between the felt intended- versus non-intended-emotion scores, the d’, and the change scores for the attitude measures. As above, we also considered total felt emotion, and, as a potential additional indicator of impact, answers to the post-test question regarding whether the visit had made individuals reflect or think about the world and themselves.

Results (see Table 1) suggested, among the target emotion factors, only a significant relation between feeling more the intended curator emotions and agreement that the exhibition had caused one to rethink or reflect on oneself or one's expectations (note again, that this was only assessed post-intervention). A positive, albeit non-significant, trend was also found between feeling more intended emotions and increased (pre- vs. post-) Social Value Orientation scores. Significant relations were also found between feeling more emotions in general (regardless of whether these were intended) and increased Prosocialness as well as for agreement with the post-intervention self reflection question. Only the first finding for the intended feelings above survived a Bonferroni correction (see Table 1 footnote). No relations were found regarding ability to identify curator emotion intentions (d’).

OPEN IN VIEWER

Table 1.

Study 1: Correlations Between Viewers’ Felt Curator-Intended Emotion, Ability to Pick out Curator Emotion Intentions, Magnitude of Emotion Felt, and Post-Exhibition Attitude Change.

	Felt Intended vs. Not-Intended Emotion	Identified Intended vs. Not-Intended Emotions (d’)	Total Felt Emotions
	r (p)	r (p)	r (p)
Change Xenophobia	.093 (.564)	−.084 (.603)	.290 (.066)
Change Empathic Concern	−.069 (.670)	.087 (.598)	−.153 (.340)
Change Prosocialness	.031 (.845)	.109 (.499)	.435 (.004)*
Change Social Value Orientation	.279 (.090)	.184 (.268)	−.101 (.545)
Exhibition caused me to reflect about self, expectations, or world (post-intervention only)	.513 (.001)*	.251 (.113)	.385 (.013)*

Note. Results based on aggregate scores of 41 participants. * Correlation significant at p < .05. All reported p values shown uncorrected for multiple comparisons. Correlations that would survive a Bonferroni correction shown in bold (adjusted p value = .0033 (.05/15)).

Stimuli/Setting—“The Mine Gives, the Mine Takes,” Galerie Wedding, Berlin

The setting of the second study was the exhibition The Mine Gives, the Mine Takes (“Die Mine gibt, die Mine nimmt” in German), curated by S. H. Ovesen and by the artist Ana Alenso, held at Galerie Wedding in Berlin (4 December 2020–27 February 2021; Figure 6). The exhibition had the overall aim of addressing the ecological and socio-economic crisis affecting the Amazonian area of Venezuela, stemming from widespread, mostly-uncontrolled, gold or other metal mining. As communicated in the accompanying brochure prepared by the artist and curator (Alenso, 2020), “the mines in South America are not just a well of resources—they are permanently robbing the surrounding environment and the local population of their livelihood,” imperiling workers through hazardous use of mercury, breaking down the surrounding society, and leaving the land stripped and polluted. In collaboration with the environmental organization SOS Orinoco (https://sosorinoco.org), the artist aimed to share first-hand accounts of this reality and to raise the public's awareness, even among individuals (e.g., Berlin residents) who might never visit Venezuela, on their co-responsibility derived from inaction or purchase of precious metals.

OPEN IN VIEWER

Figure 6.

Overview of Exhibition and Testing Locations, “The Mine Gives, The Mine Takes,” 2020, Galerie Wedding, Berlin. (A, Exhibition Poster (Alenso, 2020). B, Members of Researcher Team. C, Gallery Floorplan, Testing Location, And Location of Artworks. D, Still from Ana Alenso, We are Satellites, 2020, Video Installation Showing Intertwined Images and Videos Shot from Satellites with Documentary Films about Mining in the Southern Orinoco of the Venezuelan Amazon. E, Viewers Engaging the Exhibition from the Public Sidewalk. F, Researcher Team Collecting Surveys of Visitor Experience. G, Installation view, Lo que la mina te da, la mina te quita, 2020, Showing Gold Mining Apparatus. H, Exhibition as Seen from Outside Street. (All Photos by Authors. Images used with Artist and Curator Permission)).

The exhibition was situated in the main two rooms of the gallery (Figure 6, C) and included three components: Lo que la mina te da, la mina te quita (“The Mine Gives, the Mine Takes,” 2020, Figure 6, G) an installation of a makeshift gold mining apparatus used in the region and showing, especially through the flowing of muddy wastewater, the unstable and dangerous impact on the environment; We are Satellites, 2020 (Figure 6, D), a video installation showing intertwined images and videos shot from satellites with documentary films about mines in the Orinoco region; and Gold Exploration, 2020, another video connected explicitly to the commercial gold business. Due to the Covid-19 pandemic, which forced the closures of most galleries and museums, the exhibition was designed so that it could be viewed entirely from the outside at street level and with viewers able to engage from the pedestrian square and sidewalk looking through the large windows of the gallery. This also allowed us to conduct testing, operating in the same public outside space, while following all regulations.

Part 1: Curator/Artist Intentions for Attitude Change and Emotional Experience

The procedure for assessing target attitudes as well as targeted emotions was the same as in Study 1. However, we had the opportunity to interview both curator and artist, who had worked together on the exhibition, thus allowing us to use both of their reports as a basis for the forthcoming assessment. The interviews were conducted on Zoom (Zoom Video Communications, Inc., San Jose, California). Both the artist and curator gave informed consent and participated in a brief introduction and then moved to two, separate breakout rooms so that we could collect answers simulteneously.

Both artist and curator showed high agreement regarding targets for specific attitude changes. Thus, both were combined for Part 2. These involved increasing awareness of environmental issues in Venezuela but also globally, as well as increasing awareness of each individuals’ connection to environmental and planetary health through their daily actions (such as gold purchasing or other consumption). Both also noted that they would like to see a shifting of individuals’ priorities or values, away from material possessions to appreciation of the environment.

To assess target emotions, we employed an expanded 46-item list (see Figure 7). This included the items from Study 1, with the addition of eight new items (see grey text in figure) covering especially some more generally positive states often connected to art engagements and which had been omitted in the prior study. The artist identified 11 emotions, including largely positive or reflective responses (“absorbed,” “compassion,” “gratitude,” “moved,” “profundity,” “self-aware,” “stimulated,” “surprise,” “transformed,” “awe,” “overwhelmed”). These showed overlap with the curator, with all but the last two items also included in her own list. At the same time, the curator also indicated a more expansive 30-item list, including also more negative states (“angry,” “anxiety,” “confused,” “disgusted,” “vulnerable,” “powerless,” “sad”), social emotions (“guilt,” “shame”), epistemic reactions (“changed mind about art meaning,” “changed something about myself,” “epiphany,” “insight,” “novelty”), as well as additional positive outcomes (“happy,” “hopeful,” “playfulness,” “sublime,” “thrilled,” “wonder”). One additional emotion (“disappointed”) was suggested as not aimed at, per se, but expected, and thus included. No additional items were added, in the present study, by curator or artist.

OPEN IN VIEWER

Figure 7.

Emotions Personally Felt When Viewing Artworks and also Emotions Perceived to be Intended by the Curator/Artist (“The Mine Gives, the Mine Takes,” Galerie Wedding, Berlin). (Results Based Reports from N = 49 Participants. Dotted Line Represents Group Mean of Reported Subjectively Felt Emotions; Ovals Represent % of Viewers who Selected each Emotion as Probably Intended by the Curator/Artist to be Evoked in Viewers, Regardless of Whether they Actually Felt this Way; Red Circles Represent Emotions Intended by Curator; Blue Solid Circles Represent Emotions Intended by Artist. * Difference Between Mean Reported Felt Emotions (Average of all Intended and Non-Intended as Specified by Curator or Artist) Significant at p < .001 (Paired T-Test)).

Part 2: Visitor Assessment

Pre-Post Attitude Change

The pre-visit target attitudes (Figure 8, bottom) indicated that participants rated their sense of Nature Connectedness in the middle of the scale. They showed comparatively higher agreement with Nature Awareness, as well as Empathic Concern (reflecting similar levels to Study 1). Participants also reported relatively high Biospheric values as defining their selves. Hedonic values, similarly, showed high agreement. Whereas, Egoistic values were comparatively lower.

Paired t-tests, across the entire sample (Table A5), suggested significant increases, following the exhibition, in Nature Awareness (t(48) = -3.851, p < .001, d = .550) and a significant decrease in Hedonic values (t(48) = 2.356, p = .023, d = .337). Biospheric values showed an increase, although not reaching significance (p = .096). Only increased Nature Awareness survived a Bonferroni correction. Correlations between attitude changes, baseline attitude scores, and demographic factors (Table S6–8, Supp. Mat.) were largely in line with Study 1, with low-to-moderate correlations between the change scores of the various attitudes, and low relations between baselines and changes for specific scales. As in Study 1, art interest, as well as the exhibition's ratings, showed no significant correlation to attitude change. Comparison between individuals who had or had not read the accompanying informational brochure showed, unlike Study 1, a significantly greater decrease in Egoistic values and increase in Nature Awareness, post visit, among those who had read.

Emotion

Emotions reported by participants (group-wide boxplots and means) are shown in Figure 7, which also denotes those that had been singled out by the curator (red circles) and by the artist (blue solid circles). In general, the exhibition tended to lead to a mix of reported epistemic and also reflective emotions, with––much as in Study 1––some of the highest magnitudes reported for being “moved,” “insight,” “self-aware,” “stimulated”; “absorbed”. At the same time, the exhibition also led to other emotions that had not shown such high magnitudes in Study 1, including “sad,” “fear,” “angry,” and “powerless”. As in Study 1, most emotions also showed a wide spread across the sample of participants.

Felt/Identified Curator-Intended Emotions and Attitude Change

Correlations between the relative amount of felt or identified intended emotions, as selected by both curator or artist, and changes in attitudes are shown in Table 2. These detected a significant correlation between feeling relatively more intended emotions, as selected by the curator, and decreased Nature Connectedness, post visit, as well as reduction in Hedonic values. As in Study 1, similar correlations were also found for the same above attitude changes and feeling more emotions in general, which also correlated to increased Nature Awareness. Ability to detect curator intentions (d’), as in Study 1, showed few notable results, with only higher sensitivity coinciding with a reduction in Empathic Concern, but also with less professed agreement to act pro-environmentally in the future (assessed post-visit only). Using the artist's profile revealed no significant relations with any attitude change, both for feeling or identifying emotion intentions.

OPEN IN VIEWER

Table 2.

Study 2: Correlations Between Viewers’ Felt Curator-/Artist-Intended Emotion, Ability to Pick out Emotion Intentions, and Magnitude of Total Emotion Felt, and Attitude Change (“The Mine Gives, the Mine Takes,” Galerie Wedding, Berlin).

		Curator Intentions		Artist Intentions
	Total Felt Emotions	Felt Intended vs. Not-Intended	Identified Intended vs. Not-Intended (d’)	Felt Intended vs. Not-Intended	Identified Intended vs. Not-Intended (d’)
	r (p)	r (p)	r (p)	r (p)	r (p)
Attitude Change
Nature Connectedness	−.437 (.002)*	−.342 (.017)*	.020 (.895)	−.137 (.354)	−.203 (.181)
Nature Awareness	.323 (.024)*	.126 (.390)	−.138 (.361)	.067 (.649)	.100 (.509)
Empathic Concern	.202 (.165)	.075 (.611)	−.329 (.026)*	.132 (.366)	.119 (.433)
Biospheric Values	−.217 (.134)	−.106 (.468)	−.021 (.890)	−.148 (.310)	.013 (.933)
Hedonic Values	−.293 (.041)*	−.297 (.038)*	.050 (.741)	−.044 (.765)	−.048 (.754)
Altruistic Values	−.019 (.895)	−.058 (.692)	.027 (.857)	−.014 (.926)	.110 (.465)
Egoistic Values	.055 (.710)	.116 (.429)	.091 (.548)	.191 (.189)	−.050 (.743)

Note. Results based on aggregate scores of N = 49 participants using their alignment with intended emotion profiles provided by both curator and artist, separately. * Correlation significant at p < .05. All reported p values shown uncorrected for multiple comparisons. Results that would survive family-wise Bonferroni correction shown in bold (note, none; corrected alpha = .0014 (.05/35).

A final similar consideration of the same attitude changes and the individual emotions considered in Study 1 showed (see Table A8) significant negative relations between Nature Connectedness and feeling “shame,” “anger,” “fear,” “sadness,” as well as “compassion.” “Compassion” and “wonder” showed correlation to increased Nature Awareness, and “compassion” correlated also to Empathic Concern, with effect sizes for all findings comparable to those for Study 1, but again generally lower than those for total feeling as averaged across all possible emotion items.

Artist/Curator Reports: Evidence for Clear Attitude Targets and Intended Emotional Response

First, beginning with the basic reports from the exhibitions’ designers, it is notable that, in both studies, both artist and curators did have clear targets for specific changes that they wished to achieve via their art or exhibitions. These impacts––increased refugee acceptance, compassion, prosocialness, and lower xenophobia in Study 1; increased awareness of our connection to and our impacts on the environment, as well as a rethinking of values, especially in regards to gold mining, in Study 2––were very much in line with areas of emerging discourse on arts-based interventions. These reports provided supporting examples that such specific attitudes or value changes can be salient features in designers’ intentions. These could be clearly articulated, and, at least here, do appear to play a key aspect of the design process. Even more, we also found clear evidence that the designers did look to emotion as a means of mediating effects. In both studies, both curators and the artist reported specific sets of emotions or other phenomenal states that they wished, or at least envisioned, a visitor would experience, suggesting that designers did see these as a key element for achieving their chosen didactic, value-related, and/or attitude/behavioral-related outcomes in the art experience.

The identified emotion terms, especially regarding the number and variety of items, also raise implications. In Study 1, the curator noted 17 specific items––roughly half of the 36 possible provided. These included a mix of social (e.g., “vulnerable,” “compassion”), generally negative (“shocked,” “sad,” “overwhelmed,” “disgusted,” “angry”), self-reflective/epistemic (“self-aware,” “epiphany,” “changed something about myself,” “transformed,” “novelty,” “insight”) as well as more positive or even classic ‘aesthetic’ (“absorbed,” “moved”) states. In Study 2, from an expanded list of 46, we found a similarly nuanced list from both the artist, who identified 11 feelings, many of which overlapped with Study 1 (“absorbed,” “compassion,” “gratitude,” “moved,” “profundity,” “self-aware,” “stimulated,” “surprise,” “transformed,” “awe,” “overwhelmed”), and from the curator, who––in addition to showing very high overlap with the list of the artist (9 of the 11 items)––provided a 30 positive, negative (additionally, e.g.,: “angry,” “anxiety,” “confused,” “disgusted,” “vulnerable,” “powerless,” “sad,” “disappointed”), social (“guilt,” “shame”), and epistemic reactions (“epiphany,” “insight,” “novelty,” “changed mind about myself”).

The emotion profiles go beyond the single or small handful of items currently considered in most empirical research (though see Rodriguez et al., 2024 and Cotter et al., 2024 for other analytic methods considering emotional complexity in arts experiences). While sets may include a mix feelings that could align with previous studies––in the present case, notably negative feelings such as, for example, ‘anger’ (Rees et al., 2015; Sommer et al., 2019), ‘sadness’ (Rees et al., 2015; Sommer et al., 2019), or social feelings such as ‘guilt’ (Curtis, 2010; Rees et al., 2015; Sommer et al., 2019)––the reports also tended to include a rich combination of positive, self-reflective, as well as even more profound responses that might best work in concert or even be particularly salient with art. Although the present study was not designed to dig into the specific combinations themselves (however, we do note that these combinations do tend to coincide generally with previous arguments for reflective, transformative experiences; see Pelowski & Akiba, 2011; Miller et al., 2025), this finding provides new empirical verification of what to date had been a well-acknowledged but not actually empirically considered topic of the curatorial/artistic or even more general intervention design (Curtis et al., 2014; Falk, 2021; Klöckner & Sommer, 2021; Munroe, 2014; Norris & Tisdale, 2017; Varutti, 2023; Weibel, 2015). It is also of note that the present finding is, to our knowledge, the first to actually elicit the actual intentions, at the emotion level, from curators or artists in the designs of their interventions.

Exhibitions’ Impact: Evidence for General, but Limited Attitude Change in Socially-Focused Art

Turning to the resulting visitors’ experience, we uncovered a number of findings that support many of this paper's arguments. First, before considering emotions, in the basic question of whether and to what degree individuals tended to be affected, in general, by their engagement with the art, we did find some evidence for change. In Study 1 (see again Pelowski et al., 2024 for further discussion), when considered across the entire participant cohorts, a significant decrease was found in Xenophobia. This was concomitant also with a decrease in mood and increased reports of self-reflection as a result of the visit. In Study 2, we found similar significant cohort-wide impacts on mood and increases in Nature Awareness as well as a shift in values, with a significant decrease in Hedonic values, or professed importance and pursuit of material possessions or basic pleasure and, albeit not significant, a notable increase in Biospheric values or importance of nature, with both Study 1 and 2 findings also in line with the main targets as identified by the curators/artists. From two different interventions, with different main targets and design features, we find that the art could create a general impact on visitors, supporting such applications or the general efficacy of interventions showcasing the arts.

At the same time, the detected effects also suggested similar issues as had been noted in previous research (Pelowski et al., 2024; Trupp et al., in press). Although cases of notable effects could be found, in our general assessments, the actual effect sizes in both studies tended to be small (see Tables A1 and A4), with inconconsistent impacts on some but not all expected scales, and with large inter-individual variance. Notably in the present data, only two results––decreased Xenophobia in Study 1 (Cohen's d = 0.425) and increased Nature Awareness in Study 2 (d = 0.550)––would have survived correction for multiple comparisons and only when using rather liberal family-wise approaches (see also Pelowski et al., 2024 for discussion on Study 1 involving this topic). These findings, in conjunction with a generally low relation between the attitude changes and other factors such as overall exhibition appraisals (Study 1–2), levels of art interest (Study 1–2), art knowledge (Study 1), political orientation (Study 1), or even reading of the accompanying information (Study 1 but, in this case, not 2; see Tables S2–4, S6–8 Supp. Mat.), suggested again the difficulty with creating and detecting truly robust, consistent attitude change and also highlighted the need to consider other potentially mediating factors.

Felt/Intuited Emotions––Was There Evidence for Relations to Art Visits and Attitude Change?

Turning, then, to our target research questions, we did find some support for the role of emotion, and especially the intended combinations in visitors’ experience. There was consistent evidence that participants generally reported feeling more the emotions that had been intended by curators and artist. When averaging all intended emotions (e.g., the above emotion ‘profiles’) and comparing to the non-selected emotion terms, the former was higher for 85.37% of Study 1 participants (curator profile) and for 87.76% when using the curator's and for 71.43% when using the artist's, intended profile in Study 2. Participants were also adept at identifying those emotions that they thought might be intended. In Study 1, all participants selected at least one correct emotion and 87.8% showed positive d’ scores, indicating detection ability in their yes/no answers (see Dubal et al., 2014; Pelowski et al., 2020, 2022). Similar results were found in Study 2, with positive d’ scores in both artist and curator cases found for 83.7% of participants.

These findings are in line with, and even at the higher end of, previous reports by Pelowski et al. (2020, 2022), adding further evidence for a connection between the intentions of a designer and the experience of a viewer. Also of note, in both present studies very few participants (0% in Study 1; 6.1% in Study 2) reported that they thought that no emotions had been intended, suggesting this to be a quite intuitive feature in many viewers’ expectations or reactions to their visits. Also in keeping with past studies (Pelowski et al., 2020, 2022), individuals who felt more the intended emotions rated the exhibitions more positively, with significant correlations to interestingness, meaningfulness, potency, and willingness to pay in Study 1. Correlation were found between liking, interest, beauty, willingness to pay and feeling both the curator's and the artist's profiles in Study 2. Whereas, in both studies, and also in line with the previous papers, ability to guess intentions (d’) did not show significant correlations with ratings (see Pelowski et al., 2020, 2022 for similar museum results).

At the same time, when then considering whether the degree to which individuals felt relatively more, or could better identify, the intended emotion profiles related to actual higher attitude changes, we found some, but rather mixed, evidence. In Study 1, a signficant correlation was found between feeling more the curator-intended emotions and higher agreement that the exhibition had caused one to rethink or reflect on themselves (assessed post-visit only), as well as a non-significant trend with increased (pre- vs- post-visit) Social Value Orientation, or willingness to allocate more money from oneself to an unknown other (Murphy et al., 2011). At the same time, relations with the other target aspects (Xenophobia, Prosocialness, Empathic Concern) were not found. In Study 2, significant correlations were similarly found between feeling more intended emotions, using the curator's (but not the artist's) profile, and decreased post-visit Nature Connectedness (a greater reported sense of actual separation between nature and the viewer) but also in Hedonic values. ⁹

These findings do provide some support for this paper's main target, regarding the relation between such guided profiles and target change. These relations could be tracked across two studies and did suggest that, when such feelings as intended and designed for by artist or curator are more present for viewers, this may coincide with some higher likelihood of intended effects at the level of attitudes or future behaviors. This does give some credence to the general suggestions, in both belletristic discussions and in current artistic or curatorial practice for this role of emotional guidance (e.g., Belfiore & Bennett, 2007; Keltner & Oatley, 2022; Stamkou, 2022; see also Curtis et al., 2014; Klöckner & Sommer, 2021; Marks et al., 2016; Sommer & Klöckner, 2021).

On the other hand, as with the general exhibition ratings above, less evidence was found for ability to guess intentions (d’), with no significant relations in Study 1 and higher sensitivity in Study 2 only coinciding with a reduction in self-assessed Empathic Concern and, interestingly, with less professed agreement to act pro-environmentally in the future (albeit assessed post-visit only). This lack of, or even negative, relations with identifying emotions also fits past findings, suggesting that, while such a basic ability may be present for most participants (see Pelowski et al., 2020), this does not coincide with positive changes or ratings of art, institution, or experience. Rather, it appears that actually feeling such emotion and viscerally responding to art or perhaps to other types of interventions is actually most important. This is again reflected in literature and lab-based studies on attitude change, contrasting against intellectual ‘understanding,’ with emotion perhaps a better driver of impact (O’Keefe, 2002; Moser & Dilling, 2007; for review O’Neill & Nicholson-Cole, 2009). Note as well that, for example, the aspect of reading extra information also showed non significant relations to attitude changes, suggesting against an alternative informational or didactic approach (see also, e.g., Curtis, 2010).

The findings also support arguments that mixed sets of emotions or quite diverse and nuanced responses may be quite common in individuals’ reactions (Myers et al., 2023; Rees et al., 2015) and should be accounted for in research. More broadly, by considering the full designer-viewer pipeline, this aspect of the guided emotional experience may provide an intriguing example of a specific mechanism (Belfiore & Bennett, 2007; Fancourt et al., 2021; Trupp et al., 2023) mediating desired outputs via interventions, and suggesting the potential at least for this line of research.

‘Profiles,’ Specific Emotions, or General Feeling––What is a Best Mediator for Changes from Art?

The above findings also leave open questions. Although we did find some evidence for significant relations between feeling intended emotion and attitude change, this was still low in effect size and the consistency with which they led to desired impacts. As a point of comparison, we considered several individual emotions that had been oft-targeted in intervention research with similar (climate-related, refugee acceptance) topics: “guilt,” “shame,” “anger,” “fear,” “sadness,” “hope” (see Curtis, 2010; Hastings et al., 2004; O’Neill & Nicholson-Cole, 2009; Rees et al., 2015; Reser & Bradley, 2017; Sommer et al., 2019; Swim & Bloodhart, 2015), as well as “wonder” and “compassion,”’ with the latter noted by both studies’ curators and the Study 2 artist. These did also reveal several significant correlations, between “guilt,” “shame,” “fear,” “wonder” and Prosocialness in Study 1, as well as between “compassion” and Social Value Orientation. In Study 2, significant correlations were found between “shame,” “fear,” and “compassion,” as well as “anger” and “sadness,” and (reduced) Nature Connectedness. “Compassion” and “wonder” also showed correlation to increased Nature Awareness and “compassion” to Empathic Concern. For these significant outcomes, the effect sizes did also tend to be slightly higher than for feeling more the full intended profiles considered above (e.g., with rs in Study 1 of .32 to .40 for individual emotions as compared to a max of r = .28 for relations to feeling the intended profile; in Study 2, r = .3–.43 compared to .29–.34; see again Tables A3/A6).

These results provide somewhat of a counter-example to the main target above, suggesting that, instead of large combinations of emotions as a predictor of attitude change, it may be that certain specific responses could be important to consider in interventions. It is of note also that our designers did include several of these terms––including “sad” (Study 1 but not 2), “guilt,” “shame,” “anger,” “hope” (Study 2 curator, but not Study 1). The artist in Study 2 did not identify any of these, and also, notably, did not show any significant effects with attitude change.

The individual emotion findings themselves can connect to past research and suggestions for intervention design. Especially for example, “shame” and “fear,” which were found to be significantly related in both studies, might provide a key candidate and validating past findings with these terms (Allpress et al., 2010; Hastings et al., 2004; O’Neill & Nicholson-Cole, 2009; Reser & Bradley, 2017). Climate change researchers have suggested that fear-based interventions may not be effective as tools for inspiring action (Reser & Bradley, 2017), whereas, hope (Swim & Bloodhart, 2015) has been argued to be better, as the former may block constructive engagement (O'Neill & Nicholson-Cole, 2009; but see also Ojala, 2023 for potential differences with this response regarding cognitive/emotional framing or relative amount of constructiveness in relation to target attitude or behavioral changes). See also Rees and Bamberg (2014) for arguments against anger.

Similarly, shame, and contrary to the present findings, has been suggested as potentially shutting off potential for personal change due to a relation to feeling oneself to simply be a ‘bad person’ rather than opening the desire to repair any damage done (Allpress et al., 2010). Guilt, as well, which has been both questioned but also argued to be a “mechansim of persuation,” (O’Keefe, 2002; see also Rees et al., 2015) finds support in our Study 2 (see also Harth et al., 2013; or Rees et al., 2015; Sommer et al., 2019 for previous findings with anger). The present results do suggest many of these to be viable targets.

“Compassion” also would appear to be a key candidate. Although to our knowledge this has not been previously considered, this feeling was both identified by all designers and also showed a range of significant relations in both exhibitions. At the same time, these findings suggest against arguments that social/negative, or generally ‘unwanted’ emotions (Moser & Dilling, 2007; Slovic et al., 2007; Swim & Bloodhart, 2015) are generally not suited to attitude changes (note, our findings are contrary to, e.g., Sommer et al.'s, 2019 reports from their own art-based study).

Although slightly higher than the profiles, the individual emotions still showed rather smallish effects, with no finding surviving Bonferroni correction. The different specific emotions from the above list also still tended to show rather differential results across the studies, leading to the suggestion, once again, that these factors are not so systematic or, perhaps more likely, that different emotions may work better in different contexts (see similar arguments from Myers et al., 2023 reporting differential effects from many of the same emotions across non-arts-based climate change interventions). When looking to the visitor reports, especially in our present studies, it should also be remembered that these emotions clearly did not occur in isolation, but were rather blended into a multivariate experience. This could partially explain why these emotions tended to show similar consistency of effects or even some of the negative responses still showed significant relations to changes where previously more focused lab studies may have not.

On the other hand, the different emotions and the intended profiles may, in concert, have partially revealed what did appear to be the strongest predictor of consistent attitude changes––total magnitude of emotion felt. In Study 1, the total felt emotion, averaged across all individual items, and irrespective of their intendedness, showed significant correlations to the widest range of factors––increased Prosocialness (pre-versus-post visit) and to reporting the exhibition had caused viewers to rethink (post-visit), and also showed a non-significant trend with Xenophobia. In Study 2, similarly, feeling more emotion in general correlated to significantly decreased Hedonic values, decreased sense that individuals were connected to nature (Nature Connectedness), as well as general increased sense of Nature Awareness. In both studies, these findings also tended to show the highest relative effect sizes (e.g., r = .43).

This result, in conjunction especially also with the similarly significant correlations with the positive ratings of the exhhibitions (see e.g., Pelowski et al., 2020, 2022), is in fact in keeping with our past findings from general museum studies. These papers had also noted that general magnitude of reported emotions, even above specific intentions, showed the highest relation to evaluations, sense of meaning, or willingness to revisit. In Pelowski et al. (2022), we had suggested that this may specifically speak to the basic importance of evoking emotion with art, as well as for considering this as an individual-centered, unique experience. While feelings might be anticipated and even shared by many visitors, perhaps as a factor of specific artworks or design decisions, it may be most important that individuals have some feelings, in general, in their engagement. This suggestion would provide some contextualization for the above emotion findings, which did overlap as part of this global emotion experience, suggesting why many emotions or profiles did show some general relations to change.

Thus, in sum, these studies may provide a final suggestion for intervention research, suggesting again for the basic importance of emotion and that these should be evoked or actually felt in participants. Such emotions may well occur in concert across many different, interrelated specific feeling states. They may also be guided or anticipated as part of the design process. However, it may be most important that the aim of an artist, curator, or other designer is to create the grounds for the basic feelings of emotions in the first place. One could further interpret the results to connect to ongoing discussions for providing a necessary “open-endedness” (Bourriaud, 2020; Hegenbart, 2021; Marini et al., 2012; Nakajima, 2012), with individuals left open to find their own experiences, and which could, once again, be particularly resonant in cases such as with art. But, see again also Rodriguez et al. (2024) suggesting that merely a complex mix of emotions by itself may also not lead to positive impacts, but rather these may still need to be in some way cohesive and targeted.

Caveats and Future Targets

This paper is of course not without caveats and targets for future study. This was an exploratory, first analysis, with a wide, and rather speculative, range of considered factors. Future studies should, following from these findings, employ more controlled, directed means of follow-up testings. The two empirical studies further employed a sample of participants from a homogeneous cultural background, and the two exhibitions chosen were highly thematized and addressed very specific behavioral traits. In addition, due to the methods adopted (Pearson's and Kendall's correlations), no claim on causality could be made. Future studies are needed to examine the effects of emotional sharing and understanding on a bigger and more culturally varied sample, employing a broad range of art media and contents, and adopting a hypothesis-testing approach to allow for assessing of consistency or discrepancy with the present results.

Research could also benefit from a much more detailed and systematic assessment of specific design features and how these relate to individual emotions, as well as whether certain profiles or combinations might be ideal candidates for change or reveal or map to certain cognitive or behavioral states. This paper did not explore whether specific patterns are ‘best,’ beyond simply comparing between that of the curator and artist in Study 2. Such a nuanced mapping of feelings or other factors and consideration of their interaction with formal or contextual factors, while beyond the aim of the present paper, is only now emerging (e.g., see Miller et al., 2025), but would be an important target.

Questions can also be built around potential differences between artists, curators, or other stakeholders in their designs and intentions. One of the notable aspects of especially Study 2 was that the profile of the artist, in regards to intended feelings, did not show any specific relationships when considering the degree to which this was felt by participants. Whereas, that of the curator (and similarly for Study 1) did show effects. This was also despite the fact that both curator and artist showed high agreement in their attitude change targets. Most probably, this difference was due to some combination of a more limited set of chosen responses, which did not include any of the specific emotions considered independently above. One speculation may also be that curators, whose job precisely consists in mediating artworks’ message to the public, may generally be accomplished at eliciting or anticipating those reactions in an exhibition framework. The present paper was again one of the first, to our knowledge, to even look into such a double perspective of curators and artists along with the emotion-related results. Future study could explore even more fully the specific design decisions and making process as a component of the overall intervention impact.

Finally, it is also important to note that the present study, in considering its target prosocial aspects, focused on often standard batteries for largely hypothetical attitudes or constructs. While it is informative that we did fid pre-post changes here, the actual relation between hypothetical changes in self-reports and in either intentions to act or, even better, actual changes in concrete actions, is not often so clear (see e.g., Pelowski et al., 2024). Especially impacting the latter is an outstanding challenge in intervention research and might also be targeted via the approaches of the present paper.