The Importance of Social Cues When Browsing Appearance-Focused Social Media Content: A Think Aloud Protocol Analysis Using Fitspiration Images and Instagram Feed Browsing

ABSTRACT Previous research has demonstrated an impact of exposure to appearance-focused social media images on measures of female body image. However, the processes through which social media users consider and respond to appearance-focused content whilst browsing, or the degree to which this is reflective of their reactions to images typically encountered during every day browsing, are not fully understood. Think Aloud Protocol (TAP) was used to collect real-time, concurrent responses to Instagram content during two browsing tasks. Female participants browsed a set of 20 predefined fitspiration images, embedded within an online profile, and their own Instagram feed. Content analyses revealed that patterns of attentional focus in both tasks indicated the natural evaluation of images to extend far beyond simple comparison with the physical appearance of women. Participants engaged in more appearance appraisals when browsing fitspiration images compared to when browsing their own feed. A framework for understanding natural engagement with Instagram content is described which identifies the importance of extended social information when observing and judging appearance-focused images, via social media, that may be harmful to body image.


Body Image & Social Media
Social media use has been linked to negative body image outcomes (Saiphoo & Vahedi, 2019).Studies have reported that exposure to attractive peer and celebrity images (Brown & Tiggemann, 2016) and images promoting thinideals (Mingoia et al., 2017) or fit-ideals (Tiggemann & Zaccardo, 2015) are associated with subsequent increases in body dissatisfaction for women.Investigations into different types of social media use have identified that appearance-focused use, rather than general use, is associated with poorer body image outcomes and that images depicting idealized appearance are the most damaging (de Valle et al., 2021).However, existing academic literature is yet to explore the thoughts that women experience while engaging in Instagram use in real-time.Given the negative effect of social media use on body image, it is important to better understand how women respond to appearance-focused content online and the thought processes they experience when viewing such content in order to comprehend how, and the processes through which, their body image is negatively impacted.
Much of what is known about the impact of social media on body dissatisfaction has largely been determined using experimental paradigms where outcome measures are taken before and after exposure to specific types of images (e.g.Tiggemann & Zaccardo, 2015) or particular accounts (e.g.Hogue & Mills, 2019).Such methods have been effective at demonstrating a direct effect of exposure on body image outcome measures, but may not capture the everyday experience of engaging with and forming impressions from social media content.Whilst passive social media use (without engagement in direct social interactions) is thought to have more negative consequences than active use (engagement in social interactions), this dichotomy may oversimplify the nature of social media exchanges (Meier & Krause, 2022).Focus groups have been used to provide more holistic insights into how women construct personal narratives around social media (e.g.Baker et al., 2019), although such retrospective accounts are typically abstracted from the experience of browsing.Methodologies involving naturalistic interactions aimed at understanding how participants process appearance-focused content in real-time are largely absent from the existing literature.
Previous analyses of social media content have been instrumental in highlighting key features of appearance-focused social media posts through purposively sampled content portraying appearance-ideals (e.g.Tiggemann & Zaccardo, 2018).Instagram is believed to place greater emphasis on physical appearance than other platforms, and cause more appearance-related thoughts (Engeln et al., 2020).However, the extent to which reactions to fixed sets of appearance-ideal images might also be applicable to women's experiences and reactions while browsing their own feeds, which will contain other types of images, is not clear.Whilst forced exposure to appearance-ideal images is important for understanding their effects on body image, research should also consider the extent to which this is reflective of browsing that is more typical of daily use.

Theoretical Perspectives
Studies of online impression formation suggest that social media users tend to place inordinate importance on limited social cues, which are used to make rapid inferences about others online.These can include assumptions about a person's confidence, likability, intelligence, sociability, status, and outlook on life, which are derived from observable features of their life, interests, interaction with others, physical appearance, image or style portrayed online (Bacev-Giles & Haji, 2017).According to the Hyper-personal model (HPM; Walther, 1996), online impressions can lead to greater levels of liking and attraction than typically occur face-to-face, where inferences about others are deduced from partial social information provided online.Self-presentation techniques are thought to exacerbate this effect because social cues are often strategically manipulated online to encourage positive impression formation (Walther, 2007).In practice, it has been found that cues including facial expressions and image backgrounds shown within selfies are used consistently by observers to form judgments of the selfie-taker (Qiu et al., 2015).However, little is known about the process by which these immediate cues, used to judge others online, might be internalized by observers or impact satisfaction with their own appearance.
One of the most commonly used frameworks to explain female body dissatisfaction has been Festinger's (1954) Social Comparison Theory (SCT).Its application in this context derives from the reported tendency of women to compare their appearance to that of other women (e.g.Myers & Crowther, 2009).Women high in appearance comparison tendency have been reported to show a greater desire to change physical aspects of their appearance after viewing social media content compared to other appearance-neutral websites (Fardouly et al., 2015).A broadly applied adaption of SCT, used to understand body image, has been to differentiate between upward and downward comparisons.Upward comparisons, to those we perceive as more attractive, are thought to be more common and are associated with negative outcomes for women (Fardouly et al., 2017).The tendency to form upward comparisons is known to statistically mediate the relationship between body image concern and the use of social media platforms (e.g.Jung et al., 2022).However, further studies are needed to explore how comparisons are made in real-time and the circumstances which underpin them, that may lead to negative outcomes.As Tiggemann (2022) notes for example, research needs to consider the automatic or intentional nature of comparisons to help inform the rationale for, and assess the suitability of, future interventions which attempt to reduce instances of appearance comparison to protect body image.
A stipulation of the original SCT was that comparison should target similar others to facilitate accurate self-evaluations (Festinger, 1954).However, much body image research has explored the impact of appearance comparisons made with women who possess ideal bodies which represent a minority of women (Dignard & Jarry, 2021).Social media sites present a unique environment in which users have routine opportunities to compare themselves to members of their social group (e.g.friends and acquaintances) as well as individuals who are less personally known and more often associated with appearance-ideals (e.g.social media models and celebrities).Fardouly et al. (2021) found that upward comparisons made to close friends, acquaintances, strangers and celebrities or models all lead to lower appearance satisfaction, and that no one target type led to worse outcomes compared to the others.However, the perceived unattainability of the appearance of celebrities and models was linked to less appearance satisfaction.Conversely, Stieger et al. (2022) have shown that comparisons with peers and acquaintances were associated with poorer appearance satisfaction, compared to those made with celebrities and models.The authors speculated that this could be because while celebrities and models represent appearance-ideals which may be seen as unrealistic, the appearance of attractive peers may be seen as more normative or attainable.It is therefore unclear how the degree of familiarity with an attractive comparison target might influence the nature of the underlying comparisons when targets are known or unknown.
An important, but somewhat overlooked, consideration is the extent to which the application of SCT that assumes all upward comparisons lead to negative outcomes omits other potentially important considerations, such as the role of situational factors and self-target similarity during the comparison process.The Selective Accessibility Model (SAM; Mussweiler, 2003) suggests that comparisons with a target individual can lead to improved self-evaluation when the comparer perceives themselves to be similar to the target.When first seeing a person online, the observer makes a rapid judgment of similarity or dissimilarity to themselves.This initial hypothesis is then used to selectively search for cues to confirm this initial self-evaluation.In support of SAM, perceived moderate to high similarity with a social media target has been shown to mediate the effect of upward comparisons, with those who see themselves as being similar to the target rating their own physical appearance and financial status more positively (Kang & Liu, 2019).It has also been suggested that social comparisons via Instagram which elicit "benign envy" can generate more benevolent, self-improvement motives which inspire observers to achieve greater similarity to the comparison target (Meier & Schäfer, 2018).The tendency for comparisons to inspire rather than create negative emotional outcomes is also reportedly greater when an observer perceives the similarity between themselves and their own social network to be greater (Noon & Meier, 2019), and their own personal identity online to be clearer (Latif et al., 2021).It is therefore important not just to examine the general direction of social comparisons, but also to explore the thought processes through which women assess the wider context of social media images, and how this might relate to their evaluation of comparison targets (e.g. as similar or dissimilar to themselves).This will help to inform current understanding of the situational factors which influence the development of appearance comparisons and improve knowledge of the conditions which may lead to negative body image outcomes.
Another important limitation of the work on social media-based appearance comparisons to date has been the tendency to assess social comparison at the trait level.For example, dispositional ratings are gathered to different hypothetical situations, or the general tendency to make comparisons is measured after viewing a group of images, which may overlook important situational factors.Evidence from a thought elicitation study where participants were asked to write short sentences that came to mind when viewing six thin-ideal images showed that, on average, upward social comparisons were made by less than a quarter of women, and to fewer than one of the six images viewed (Anixiadis et al., 2019).It follows that measures of general comparison tendency may not capture the impact of particular images, and that not all appearance-focused images will have the same individual effect.To this extent, a more nuanced exploration of women's unique thoughts and reactions to Instagram content, which occur in response to specific attractive targets, may help to further our understanding of the factors which influence the formation of comparisons and their personal meaning to women.

Aims of the Current Study
The current study explored the spontaneous thoughts and feelings expressed by women in relation to viewing Instagram content, with the aim of improving our understanding of the impact of viewing appearance-ideal content, and personal nature of appearance comparison.Using a Think Aloud Protocol (TAP; Ericsson & Simon, 1984), reactions during two Instagram browsing tasks are explored: (i) how participants respond to a set of appearance-ideal fitspiration images and (ii) how they respond to the full range of content encountered on their own Instagram feeds.The study seeks to explore the nature of social comparisons and circumstances which underpin them, during real-time interaction with appearance-ideal content.The study also seeks to explore and draw parallels between the subjective experiences of participants when browsing appearance-ideal content and the range of images typical to their own feeds, which may contain varying levels of appearance-focus.

RQ1
What do female participants' real-time reactions while browsing fitspiration Instagram posts tell us about the thought processes that occur when they are presented with attractive comparison targets, and to what extent do these reactions reveal evidence of social comparison?RQ2 Are the thoughts and feelings identifiable when browsing fitspiration posts similar to those that are experienced when browsing the content on female participants' own Instagram feeds?

Participants
Participants were 20 females aged 18 to 25 years (M = 19.30,SD = 1.87) who regularly used Instagram (average of 1.75 hours per day) and were current residents of the UK.Females in this age range were purposively sampled as belonging to the target demographic thought to be the most influenced by social media (Royal Society for Public Health, 2017).Key engagement data for the sample showed the participants, on average, followed 784 accounts, had 869 followers and had posted 69 images to their Instagram accounts.The average BMI of the sample (M = 22.76, SD = 3.47) was within the "normal" weight range (18.5-25;World Health Organization, 2016).State body dissatisfaction was recorded, following Heinberg and Thompson (1995).Visual Analogue Scales (VAS) measured four aspects of body dissatisfaction (body fat, body size, body shape and physical attractiveness) on 100-point sliding scales.The four items were scaled into one measure of body dissatisfaction, out of 100 (where high scores indicated higher dissatisfaction).At baseline participants had an average score of 60.19 (SD = 22.78), suggesting slight-moderate dissatisfaction.Half of the participants (10 out of 20) reported engaging in regular exercise.

Think-Aloud Interviews
TAP is a method developed to provide insights into the thought patterns and behaviors of participants while completing different tasks.As a wellestablished method for assessing user experiences of computer interfaces, TAP allows real-time data collection of reactions to online environments while avoiding the loss of information which might occur as a result of memory constraints where data are collected retrospectively (Jaspers et al., 2004).TAP has been successfully used with digital media (e.g.Davies, 2018) and appearance-focused content (e.g.Ogden & Russell, 2013), and provides a means to capture potential processes underlying state body image responses during real-time social media use (Tiggemann & Velissaris, 2020).
The current study used TAP during two Instagram browsing tasks, preceded by a warm up task.The first (fitspiration) task asked participants to browse a set of 20 predefined fitspiration posts.To imitate feed browsing, the fitspiration posts were embedded within a single, real Instagram profile, through which participants were able to browse each image in turn.Participants were told that the images were real posts taken from various public profiles.The second (feed) task asked participants to browse their own Instagram feed for five minutes, interacting with content as they would normally when using the platform in their own time.TAP instructions were standardized across both tasks and asked participants to engage in typical browsing behaviors (including active engagement with content), while also speaking aloud their thoughts and feelings as they occurred, in relation to the content they were viewing (Ericsson & Simon, 1984).In the fitspiration browsing task, participants were asked to view all 20 images contained within the profile but were told they could spend as much or as little time considering each individual image.This number of images was chosen through testing trials to be equivalent to approximately five minutes of feed browsing.
The warm-up task asked participants to search for #Travel (a popular search) on Instagram and used the same speak-aloud instructions, requiring participants to express their thoughts and feelings in relation to the first 5-posts located from the search.The warm-up continued until participants understood what was required and were comfortable with the technique, with additional searches (e.g.#nature, #university) being used if further practice was needed.During all think aloud tasks, the researcher sat out of view and spoke only to prompt participants to "keep talking" if they fell silent for more than 20 seconds.After the warm-up, all participants completed the two Instagram browsing tasks in the same order.The feed browsing task was completed before the fitspiration task to avoid the possibility of appearance themes in the fitspiration images influencing participants' reactions to their feeds.Immediately following each task, participants were asked to reflect back over the task they had just completed through a short semi-structured interview (Taylor & Dionne, 2000) lasting around 15 minutes.All participants were asked two specific questions as part of this interview: "were there any posts that stood out to you and why do you think that is?" and "were there any posts that made you feel anything in particular and why do you think that is?".These initial follow-up questions were broadly worded to allow participants to identify aspects of their browsing experience which were most salient to them.These were followed by further bespoke questions which were dependent on the specific features or points raised in participants' answers.For example, "I noticed you said . . .could you tell me a bit more about this please?".Prompts and further questions were used where necessary to encourage participants to provide detailed accounts of their reactions during the tasks.In this manner, in-depth insights could be gained that were grounded in participants' personal experiences of each task.This study was approved by the Science and Health Faculty Ethics Committee at our institution (SFEC 2019-109).Informed consent was provided by participants prior to taking part.

Materials
Participants completed the browsing tasks on their own personal mobile device.During the tasks the screen content was simultaneously recorded, so that the types of post viewed could be coded and the dwell time spent viewing each individual post measured.Concurrent verbal data were captured using a Digital Voice Recorder and transcribed verbatim, and where relevant, emotional responses (e.g.laughter) were coded within the transcripts.
For the fitspiration browsing task, 20 images were selected from an initial pool of 50 images identified using the Instagram hashtag, #fitspiration.Pilot testing with 10 females of similar age and social media experience to the study sample was used to select images which were considered to best fit the definition of fitspiration (following Tiggemann & Zaccardo, 2018).Of the 20 images selected, 4 were full-body selfies, 15 were posed images of a single individual taken by someone else and 1 was an image of two women taken by someone else.Four of the 20 images were taken when the individual was midexercise, consistent with findings that roughly a quarter of fitspiration images are of individuals actively engaged in exercise (Tiggemann & Zaccardo, 2018).The remaining 16 images, showed women posing in gym clothes and exercise settings.

Body Dissatisfaction Manipulation Check
To explore whether participants experienced changes to their state body dissatisfaction as a result of the browsing tasks, the measure of state body dissatisfaction was completed twice, once at the beginning and end of the study.A paired samples t-test revealed that the pre-and post-exposure measures of body dissatisfaction were significantly different (t(19) = 3.94, p < .001,d = 1.81).On average, post exposure scores were 4.70 points higher than prescores (95%CI, 2.20 to 7.20) on the 100-point scale.This suggests that participants felt more dissatisfied with their bodies after completing the two browsing tasks.

Video data analysis
The screen recordings were used to measure how long participants paused to view each post.A participant was considered to have attended to the image if they paused and did not scroll past an image for 800 ms or longer.This dwell time was determined from previous research regarding the speed at which conceptual understanding can be achieved when presented with a novel image (Potter et al., 2014).
The feed task screen recordings were also used to collect data on the type of posts viewed by participants during this task, to contextualize the verbal data and gauge the extent to which this content was appearance-focused.Video data were coded using both inductive and deductive means.Nine initial image codes were taken from Slater et al. (2017) and expanded where additional codes were identified by the researcher during analysis.The final image codes can be found in Table 2.A Cohen's Kappa analysis of the image coding, showed strong agreement between the primary researcher and an independent rater in their categorization of each image type based on the 17 codes used (K (n = 135) = .947,p < .001).
All posts viewed on participants' own feeds were additionally coded based on how appearance-focused each image was, using the following definition: "posts in which the predominant aim of the image was to portray the physical attractiveness of the individual or individuals in the image, whereby the viewers" attention is drawn to the physical features of those in the image'.An image was considered appearance-focused if it contained any of the following features: thin and/or toned ideals (Boepple et al., 2016;Cohen et al., 2019), strategic body pose (Boepple et al., 2016), body fragmentation (close ups of body parts rather than whole bodies) (Tiggemann & Zaccardo, 2018;Webb et al., 2017), fashion (Boepple et al., 2016;Cohen et al., 2019) and facial beauty (Boepple et al., 2016;Cohen et al., 2019).For this purpose, images of men or images that did not contain people were not considered to be appearance-focused.Any images that were categorized as containing appearance-focused content were then rated on a scale between 1 (not very) and 10 (extremely) regarding how strongly appearance-focused the image was.Based on a Cohen's Kappa analysis of the images in participants' feeds, strong agreement was found between the primary researcher and an independent rater in their categorization of images within the 5-primary appearance-focused codes (K(n = 135) = .791,p < .001).Similarly, good agreement was found regarding the strength of the appearance-focused ratings (out of 10) between the two raters, using Pearson's correlation (r(133df) = .982,p < .001).

Verbal data analysis
The verbalizations produced by participants during each browsing task were analyzed using two separate content analyses, to allow the themes independently produced from each condition to be contrasted in the analysis.Following Elo and Kyngäs (2008), data were first transcribed and read through a number of times until the researcher felt fully immersed in the dataset.Data from the concurrent and retrospective aspects of the browsing task were coded together.The initial analysis of the data took an inductive approach describing its manifest content (Graneheim & Lundman, 2004).During the analysis phase, data were coded, using NVivo, by adding labels to recurrent data points, creating lists of the codes and creating open themes to re-contextualize the codes into homogenous groups (Bengtsson, 2016).Themes were then collapsed where appropriate to reduce the overall number and arranged into groups.At each point the themes and groupings were compared to the original dataset to ensure that an accurate depiction of the data was constructed.The themes were then used to formulate a description of the dataset through abstraction (Graneheim & Lundman, 2004).In the final phase, theme descriptions were formed to create a structured understanding of the utterances produced by participants in response to the different types of Instagram content viewed.Inter-rater reliability using Cohen's Kappa was calculated for the themes in the verbal data produced during the feed browsing task (K (n = 89) = .746,p < .001)and the fitspiration browsing task (K(n = 79) = .780,p < .001),which suggested good agreement between the codes allocated by the primary researcher and those of an independent rater.

Real-Time Reactions While Browsing Fitspiration Posts
Participants spent between 1.4 min and 10.3 min (M = 3.9 min; 95%CI, 2.9 to 5.0 min) browsing the 20 fitspiration images, with an average of 10.5 s (95%CI, 9.6 to 11.2 s) being spent per image.

Reaction themes
A total of 528 reactions (data instances) were coded from participants' verbal responses.Table 1 presents the key themes and sub-themes identified from this analysis including the frequency of reactions within each.Percentage values in the text below reflect the number of verbal responses as a proportion of the total data instances coded from this task.

Making judgment themes.
When presented with Instagram images of attractive comparison targets, participants tended to express quick judgments over the salient features shown in each image in 61% (n = 324) of data instances coded.Participants typically engaged with the images by expressing their preferences toward the specific content of the images, either by direct reference to the person in the image or other social cues that were implied within the image.
Most commonly, participants expressed favorable opinions toward the physical appearance of the women in the fitspiration images, making appearance comments (18%) about their bodies or facial attractiveness: "Her body is really nice and her hair, and she's really tanned."(May, p.8).Only in a minority of instances were participants critical of the women's bodies: "Her body looks weird."(Lucy, p.7). Participants also commonly picked up on fashion (15%) cues in the images, by expressing opinions over the clothing worn by the women, which reflected a mixture of approval: "I was just looking at this one longer because I noticed that I like her outfit, the pink."(Maya, p.2), and disapproval: "Don't really like what they are wearing to be honest, wouldn't wear that myself" (Katie, p.8).
Participants also questioned the authenticity of some of the images, considering them to be fake (10%).Participants picked up on the social strategies being used in some images that they felt were posed to emphasize attractiveness: Table 1.Themes relating to participants' real-time reactions while browsing fitspiration posts and the full range of posts typical of participants' own feeds, including how many participants reacted in each way and total number of data instances.Other images were thought to be staged to resemble exercise rather than reflecting real exercise: All these girls' hair looks really nice as well, like considering they are meant to be being active and sporting and stuff.When I go to the gym my hair (laughter) just doesn't look like that.(Hattie, p.12) Other images were interpreted as exhibitionism; with women showing off (2%) their bodies: "Your picture is just showing off your body, it's not promoting anything healthy."(Taylor,p.23).Interestingly, some participants used cues in the images to make positive assumptions about how the women were feeling (how they feel; 2%): What I do like about the fitness ones is that some of them do look really happy with their body, and like motivated, and just really like content, and like they've got that positive outlook.(Zoe, p.17) Participants also identified other features which were less central to the focus of the images, including forming preferences toward the scenery (10%) shown in the images: "That's a really pretty view" (Lucy, p.1), sometimes preferring it " . . .more than the actual person."(Hannah, p.9).Other cues were interpreted as indicators of consumerism (3%), such as attempts to promote or advertise products: "She's obviously branding something because the bottle logo is right toward the camera."(Charlotte, p.4).

Personal association themes.
A common response in the women sampled was seeking to identify with the females shown in the images or specific cues in the images, which occurred in 26% (n = 137) of instances.Participants specifically assessed the images based on their similarity to their own lives, appearances and exercise behaviors.
Appearance comparisons (10%) were made directly by the participant toward the women, or with how the attributes of the women in the images compared to the participant's notion of the "ideal" body.This typically involved the identification of particular body parts (e.g."abs", "bum" etc.) that participants found desirable: I would like to look like this, slim with the perky bum, tanned, toned arms . . .Yeah this is someone I would like to look like.I really like her toned arms, toned body.(Katie p.8) Participants also made exercise comparisons (8%) between their own habits and those assumed of the women in the images, including how the women in the images made them feel more motivated to exercise.
I think the ones in the gym that kind of made me want to go to the gym . . .because you see that figure and then because they are in the gym you associate it with going to the gym.(Hannah,p.22)Additionally, participants made associations to the women in the images, by expressing their own relatability (8%) to specific details or assumed characteristics depicted in the images shown: For some reason I can relate to her a lot more than the other ones . . . it looks like she is having fun and it's just like a silly picture.(Rose, p.4) Image impact themes.Participants expressed the emotional impact of the images on themselves in 13% (n = 67) of instances.This revealed variable consequences and a trade-off between positive and negative reactions.
Images were sometimes interpreted as conveying meaningful messages (5%): "her caption is quite motivating and positive."(Ellen, p.10), while other participants noted that messages lacked depth or sincerity: "the caption is very cheesy" (Katie, p.9). Participants also described how they were feeling (7%) when looking at the images, for example "motivated and inspired" (Hattie, p.12), "jealous" (Lauren, p.19), "Lazy" (Mel, p.24), or explained why they were feeling negative as a result of seeing the images: Quite negative, I just wanted to look like them and that's not where my body is.[I felt] like 'why do I not just go to the gym it's so easy?I could just work out.' Yeah, kind of put me down.(Katie, p. 24)

Type of images on participants' feeds
Before considering participants' qualitative reactions while browsing their own feeds, content analysis was first used to categorize the range of images encountered by participants in the sample, and the extent to which these images were appearance-focused.This helped to contextualize participants' reactions to the types of posts typical of their feeds.Participants attended to between 15 and 83 images during the five minutes of feed browsing (M = 51.0,95% CI 44.0 to 58.0).Table 2 shows the frequency with which different image types were seen by participants.Memes were the most commonly occurring single category of image that was attended to within participants' feeds, with 27% of all images falling into this category.By contrast, body positive themes were largely absent from the feeds of participants, with only 2 instances of body positive messages viewed.Collectively, selfies, images of people alone, or in groups, accounted for 42% of all images with a high proportion (59 to 85%) of the images within these categories being classified as appearance-focused.Overall, 39% of all images attended to during feed browsing were categorized as appearancefocused.
With respect to the thematic content of appearance-focused images, chisquare analysis showed a significant difference in the frequency with which different appearance-focused image types occurred (X 2 (4) = 134.33,p < .001).Images emphasizing facial beauty were found to be the most common type of appearance-focused image seen by participants (37%), with images emphasizing thin and/or fit ideals (28%), and strategic body poses (18%), fashion focused images (13%) or images depicting body fragmentation (3%) occurring less frequently.

Reaction themes
A total of 643 reactions (data instances) were coded from participants' verbal responses during the five minutes of feed browsing.Table 1 presents the themes identified from this analysis including the frequency of reactions within each.Percentage values given below reflect the number of verbal responses as a proportion of the total data instances coded from this task.

Making judgments themes.
As with the fitspiration browsing task, when viewing images on their own feed participants expressed quick judgments over the key features of the images in 34% (n = 220) of instances.These judgments covered a diverse range of cues shown in the images, including a focus on physical appearance.Opinions (14%) were formed from specific social cues and additional information shown in the images, including items, activities and people: I love her pictures, love the posts.I like aesthetically pleasing posts and pictures.She takes lovely pictures -like her next to a plant with the sunlight on her (Katie,p.46)As with fitspiration browsing, participants often spontaneously appraised the physical appearance of people in the images by making appearance comments (7%).These comments were almost exclusively directed toward the women depicted on participants' feeds, and assessed facial and body appearance: "She's really pretty" (Zoe, p.18) and "she has a really nice body" (Samira, p.28).Fashion (9%) cues were also appraised where participants expressed their opinions over make-up and clothing: Picture of a group of girls and their all wearing bodysuits, dressed up and going out.But I don't think I would ever wear anything like that.(Hannah,p. 38) Finally, some images were interpreted as fake (4%), which was felt to reduce the authenticity of the people or lifestyles portrayed: "She doesn't really look like that in real-life though.She looks kind of weird with a lot of make-up."(Mia,p.33).

Personal association themes.
As with fitspiration browsing, participants also explored their own personal association with the content of images, in 20% (n = 132) of instances.Participants assessed the cues shown in the images in relation to views of themselves, their own experiences and appearance.
Relatability (11%) was explored through cues depicting similarity between the participant and the people, activities or items shown in the images: " . . .because they were all wearing them [bodysuits], I feel like I should be confident enough to wear it too."(Hannah,p.39).Some participants also made direct upward appearance comparisons (4%) with women in the images: If they look more attractive than I do in a post, or in the picture that I post, then I feel worse about my appearance.(Megan,p.13)Exercise comparisons (2%) were also made toward images of women which depicted exercise themes: When I see things that I can achieve, I feel good about them.I could do that, it's given me inspiration, I want to go to the gym now.(Katie,p.46)Additionally, participants also made upward lifestyle comparisons (3%) regarding the financial situation, social life or travel activities of the individuals portrayed in the images: . . .other people are having more fun elsewhere.I feel like, is it me?Do you know what I mean?Am I not the life and soul [of the party]?. (Hattie, p.6) Image impact themes.As with the fitspiration task, participants' reactions to their feed images also described the emotional impact of images on them in 21% (n = 136) of instances.This revealed both positive and negative consequences during feed browsing.
Expressions of humor (10%) were verbalized or often expressed through laughter, commonly in response to memes: "I pay more attention to the funny memes."(Jamila, p.62).Other expressions of how participants were feeling (8%) during browsing encapsulated a range of emotions including annoyance, motivation, confusion, pride, boredom, sadness, jealousy, intrigue and frustration: "She's all about positivity . . .when I see her posts it kinda makes me feel better."(Hattie, p.5).Finally, themes of negative body image (3%) were discussed, where participants reflected on the negative influence of some accounts and image types that appeared on their feed: The only celebrities that I do follow are people that I know won't be posting half naked pictures of themselves.I make conscious decisions not to follow certain people.(Rose,p.3)Browsing behavior themes.Unique to the feed browsing task, participants also verbally described their normative browsing behaviors, in 24% (n = 155) of data instances.
Narratives explained typical patterns of use (20%) such as the circumstances under which participants would like, save or share content: "I'll 'like' the picture because she's pretty."(Afia, p.25).Some participants described staying informed (3%) about current events through news channel and celebrity accounts as a key driver of their browsing behavior: "This account is useful for that; so I can be updated with politics."(Katie, p.44).Moreover, Instagram was described as a key method for being "in the loop" (1%) with friends they no longer see in person: It's quite nice catching up with her via Instagram because I don't talk to her anymore, but she was quite a close friend when I was in school.(Zoe, p.16)

Browsing Task Comparison
The mean number of appearance comments and reports of appearance comparisons, corrected for the total number of images viewed, were compared between the two browsing tasks.This was done to gauge whether viewing appearance-ideal images only, during fitspiration browsing, would lead to significantly different rates of appearance appraisals compared to viewing the range of different images typical to the participants' feeds.Participants were found to make significantly more appearance related comments in response to the fitspiration images (M = 0.24, SD = 0.21) compared to when viewing their own feed (M = 0.04, SD = 0.04), (t(20) = 4.696, p < .001,d = 1.32).Additionally, a significant difference in appearance comparisons was found (t(20) = 2.88, p = .010,d = 0.82), with more comparisons made in the fitspiration browsing task (M = 0.13, SD = 0.17) compared to feed browsing (M = 0.03, SD = 0.03).

Discussion
The current study explored participants' real-time reactions to appearanceideal Instagram content, and whether the thoughts and feelings reported would reveal more about the processes associated with how participants evaluate attractive comparison targets.The study also explored whether similar interpretive patterns applied when participants browse content on their own Instagram feeds.
With respect to RQ1, analysis of the thoughts and feelings verbalized by participants during the fitspiration task revealed three common response patterns.In broad terms, these suggested that participants engaged with and interpreted appearance-ideal images by making automatic judgments, trying to form personal associations with image content, and expressing the personal impact of engagement with images.When browsing appearanceideal fitspiration images, participants reported active, real-time social comparisons with the appearance and exercise behaviors of the women that they encountered.Importantly, this evidences that comparisons may, at least in part, be a conscious process for some women during social media use.This lends some support to the idea that body image interventions designed to target the formation of appearance comparisons online might provide a possible means of reducing negative body image effects of social media use (Tiggemann, 2022), although these findings are limited to the small sample used in the current study and would need to be confirmed in further research.We speculate that participants' reactions to the images through the formation of judgments and personal associations reveals the interpretative schema through which participants assess and engage with images and the people shown within them, and that this reveals evidence of the nature of social comparisons and the circumstances that underpin them.Firstly, the sub-processes used to form judgments may reflect the mechanism for classifying potential human comparison targets and the assessment of preference toward the other social cues that they convey (i.e.whether they or their lifestyle are genuine and likeable, or seen as inauthentic).This may be guided by participants' assessment of whether or not the target possessed desirable traits (e.g.physical attractiveness).As physical attractiveness and personal happiness are likely enhanced online (Pounders et al., 2016), this might partly explain why the majority of appraisals made surrounding appearance and personal attributes were positive.
Secondly, the personal association themes, where participants appeared to actively assess their similarity to the content of images during browsing, supports the application of SAM in the context of this study.Participants' reactions actively verbalized and considered their similarity to, and degree to which they could identify with, the women in the fitspiration images, reporting that it was important for them to relate to an image in order to be able to engage with it.Participants made judgments of similarity based on the available cues presented in the images which extended beyond the consideration of physical attractiveness (e.g.activity taking place or the age of the target).Consistent with the assumptions of SAM, this highlights the importance of participants' decisionmaking at an individual level, since each participant drew different personal associations with the images.It follows that predicting the outcomes of comparisons made by the participants is more complex than physical comparison direction (e.g.upward) alone, since the relevance of other social cues to the individual also needs to be considered.Thus, participants' assessment of the wider context of the images, such as location or fashion, suggests that social media users are influenced by more than just the presence of an attractive comparison target.While caution should be taken with the generalizability of these results, due to the limited number of Instagram users sampled, this study suggests that more research is needed to consider how social cues moderate the impact of an attractive target's appearance on body dissatisfaction.Since previous research has indicated that selfies on Instagram receive a greater number of "likes" when social cues revealing information about the selfie taker are included in the image (Hong et al., 2020), understanding of the wider context of appearance-ideal images may be important for our understanding of social comparison, and the self-evaluative consequences for the person making comparisons.In addition, the image impact theme reflected participants' experiences of the variable consequences of exposure to appearance-ideal images over a short period of time.The sub-themes of impact depict a trade-off between both positive and negative outcomes, where participants expressed negative reactions such as envy or despondency and positive emotions such as inspiration and motivation.If, as has been argued elsewhere, the tendency for comparisons to inspire, rather than create negative emotional outcomes, is associated to the perception of self-target similarity (Noon & Meier, 2019), it could be speculated that the different expressions of positive and negative consequences observed in the present study was driven by the relatability and ability to identify with cues in the comparison target's image and their assumed lifestyle.
RQ2 aimed to explore whether patterns in participants' reactions to fitspiration content would be similar to their responses to other content typically seen on their Instagram feeds.This was because, while forced exposure to fitspiration images was useful for understanding how participants responded to appearance-ideal images of attractive comparison targets, some consideration was needed for the fact that this may not be reflective of their everyday browsing experiences.Content analysis of the posts our sample encountered indicated that participants were exposed to a broad range of images on their own Instagram feed, with approximately two thirds of these not containing appearance-focused themes.Participants in the study experienced an average of 20 appearance-focused images over the five minutes of feed browsing, which emphasized different physical attributes.Whilst on the surface this may seem a small number of images, this may still represent a significant cumulative effect over time in the context of everyday browsing, where social media users are typically exposed to a large number of images on a daily basis (around 1.3 billion images are shared on Instagram a day; Broz, 2023).Given the small sample size used to explore qualitative reactions to images within the current TAP methodology, caution should be taken when generalizing the absolute levels of exposure to different image types seen in the present analysis, and it cannot be assumed that the exposure levels to appearance-focused content observed will be typical of all social media user groups.To determine this, future research could usefully explore engagement with appearancefocused content with larger sample sizes, using other experience sampling methods, to provide a broader understanding of the possible cumulative effects of exposure.Of the appearance-focused content participants viewed in the current sample, facial attractiveness was found to be the most prevalent theme.The impact of appearance-focused Instagram images on facial dissatisfaction has received comparatively less research attention than overall body dissatisfaction (Sampson et al., 2020), although, some research has examined the impact of facial attractiveness (e.g."ideal smile" images; Sampson et al., 2020) or considered facial appearance as an integral component of body dissatisfaction (e.g. Brown & Tiggemann, 2016).The current study suggests that on some women's feeds, facial beauty may be a key appearance-focus theme, above that of body-related appearance-ideals.
Examination of participants' qualitative reactions while browsing their feeds revealed that, while participants' feeds contained a mix of images of which around one-third were appearance-focused, the same broad themes of judgments, personal associations and image impact noted during the fitspiration task were also found during the feed browsing task.Since less than 2% of images viewed during feed browsing were fitspiration images, it follows that these themes may reflect a common framework for how participants engage with Instagram content.
Based on the theme clusters identified across both browsing tasks, we propose an Instagram Engagement Framework (Figure 1), which suggests that participants engage with and interpret a broad range of Instagram content-types by first forming judgments and personal associations with image content, before expressing the personal impact of engagement.Whilst it should be noted that exposing participants to one type of appearancefocused image in the fitspiration browsing condition, places exaggerated emphasis on appearance themes than was likely on their typical feeds, the range of judgment sub-themes identified across both browsing tasks suggests that participants were actively aware of many other social cues during browsing, besides cues to physical attractiveness.For example, in the judgments theme, participants were motivated to form impressions of other women by inferring positive attributes from cues such as fashion, personal lifestyle and even the assumed happiness of the people present in the images.In line with the HPM of impression formation, attitudes toward this social information and subsequent inferences about the women encountered during browsing were largely positive.
In many cases, across both browsing tasks, participants appeared to actively search for ways in which they could personally associate to cues within image content including activities, locations, items shown, beauty practices, meme meaning, as well as the people shown in the images.This was interesting since it suggests the tendency to seek similarity and identification with content, in accordance with the SAM, and not just to focus on an individual's appearance.It should be noted that participants may be more likely to achieve a perception of similarity during feed browsing, compared to fitspiration browsing, since their feeds are likely curated to their specific social groups and interests.This effect may be reflected in the data since reactions pertaining to the relatability of the content were approximately twice as common in the feed task compared to the fitspiration task.Whilst the fitspiration task exposed participants to images of women not personally known to them, the feed browsing task is likely to have included exposure to a mix of close friends, celebrities, strangers and acquaintances (whether known beforehand outside of social media or formed through social media).Previous research findings have been inconclusive regarding whether comparisons with known or unknown targets are more harmful for body image (e.g.Fardouly et al., 2021;Stieger et al., 2022).Although no formal classification was obtained from the TAP tasks regarding the degree of familiarity, or nature of the relationship between the participant and the people seen in images during the study, broadly similar themes did emerge in response to the fitspiration task and feed browsing task.It could therefore be speculated that a similar underlying engagement process arises when exposed to both known and unknown social media targets.Future research could usefully explore the shared thematic components identified in the current study that were common to forming judgments during fitspiration and feed browsing (e.g.fashion, realism, relatability, emotional reaction, etc.) as a basis for understanding the circumstances under which comparisons to known or unknown targets may potentially be more damaging to body image.
The impact of browsing led to a complex mix of gratifying emotions (e.g.humor, pride), but also negative reactions (e.g.jealousy, diverted attention), which at times were attributed directly toward the initial impressions formed of individuals shown in the images.Taken together, the proposed framework for engaging with Instagram content provides an insight into how the participants of this study interacted and evaluated appearance-ideal content as well as the full range of content typical of their feeds.While participants responses in the image impact themes reflected a range of positive and negative consequences during both tasks, overall it is noted that a significant increase in body dissatisfaction was recorded subsequent to the browsing tasks.This suggests that the themes included within the framework may represent the types of thoughts participants had simultaneous to experiencing negative body image consequences from browsing Instagram.Among this sample, it implies that overall negative body image effects outweighed the potential benefits or positive impact of individual images generating humor, inspiration or personal meaning seen in the current study.Previous research has suggested that negative emotions have greater valence than positive emotions (Vaish et al., 2008).Furthermore, individuals tend to be more affected by negative entities than positive ones, and invest more in avoiding a "feared-self" than striving toward a "hoped-for self" (Baumeister et al., 2001).Such tendencies might partially account for the decrease in body dissatisfaction seen in the current study.In the context of body image research, it has been found that feared body image outcomes have a stronger mediating influence on the relationship between BMI and self-esteem than hoped-for body image outcomes (Dalley et al., 2013).Likewise, the fear-of-fat is a more powerful motivator for dieting than a hoped-for thin-ideal (Dalley & Buunk, 2011).Therefore, whilst the proposed engagement framework evidenced both positive and negative reactions to Instagram use, these elements did not necessarily carry equal weight, with respect to their potential influence on psychological outcomes.
The findings of the current study should be taken with the following limitations in mind.Firstly, this study consisted of a qualitative exploration of the reactions of a small group of women during Instagram use, therefore caution should be taken when generalizing these results.The current sample were chosen for their regularity of Instagram use, as well as belonging to a specific demographic group who are thought to be most at risk of negative body image outcomes due to social media.We do not seek to claim, therefore, that the range of content the present sample encountered when browsing Instagram, or even their reactions to content will be typical to all users.Rather, the findings of this study identify a number of important avenues for future research exploration, and broad range of possible content to which people can be exposed.Whilst the sufficiency of sample sizes within qualitative health research is a source of ongoing academic debate (e.g.Vasileiou et al., 2018), we confirm that the recruitment of participants, and identification of the themes within the present study continued until data saturation was reached, that being when no new themes emerged.The themes contained within the engagement framework presented can therefore be considered to provide a robust description of the participants' accounts sampled.
Secondly, it cannot be guaranteed that participants' verbal data gathered using TAP provides a true reflection of their thoughts and feelings.Whilst this has often been a common criticism of the TAP methodology, previous research does though lend support to the idea that TAP provides useful insights into participants' conscious thought processes while interacting with digital media (Perski et al., 2017), and it was observed in the present study that participants did not appear to encounter any difficulties with the use of the technique.Thirdly, since the browsing tasks were completed one after the other, it is possible that the order might have impacted upon participants' responses during the second browsing task.This deliberate ordering of tasks was intended to limit the impact of the fitspiration task, which may more likely have primed participants to appearance-related themes if completed prior to the feed browsing task.The feed browsing task, as being indicative of participants' regular Instagram use, was deemed less likely to bias participants.Fourthly, where participants' feeds are likely to have included images posted by users who are well known to them, social cues within these images may have less impact on the impressions formed of these individuals.However, since social media platforms foster environments through which users are encouraged to encounter new people, who are not well known to them (e.g.social media models or celebrities), impression formation and the use of social cues are still pertinent to everyday social media use.To this extent, the fitspiration browsing task helped to ensure that our data captured reactions to individuals not personally known to participants.Finally, based on the data collected in this study, whilst we cannot measure the specific self-evaluative consequences of social cues with respect to their impact on body dissatisfaction, these findings do provide a theoretical basis for future exploration.
In conclusion, the current study provides an insight into participants' realtime reactions to appearance-ideal content and how their assessment of the wider context of the images might influence the impact of appearance comparisons.The study also evidenced that while participants feeds contained varying levels of appearance-focused images, similar themes of judgments, associations and impacts were recorded in response to both fitspiration and feed browsing.This contributes to our understanding of appearance comparisons, by highlighting the importance of examining the impact of social cues, included in appearance-focused images, on body image outcomes.

Disclosure Statement
No potential conflict of interest was reported by the authors.

Funding
The research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Figure 1 .
Figure 1.The Instagram Engagement Framework.Three common themes identified in verbal responses to images viewed during fitspiration and feed browsing.Key: Sub-theme groupings indicate responses that were shared or unique to each browsing task.-Verbal reactions to fitspiration images; ---Verbal reactions to feed images.Note The circles represent Venn diagrams where each theme of verbal reaction is depicted in relation to which task it was made in response to.The area of overlap depicts reactions which were present during both tasks.
Then there's another girl who's really skinny with a thick bum.But that's because of the pose, everyone knows that pose, it makes you look like you've got a bum when you don't.(Charlotte,p.3)

Table 2 .
Frequency at which each type of image appeared during feed browsing, including level of attendance and appearance-focus.
1 How many of the 20 participants encountered at least 1 of this type of image. 2 % based on total occurrences.